Skip to contentSkip to conversation

OOM detection rework v4

PATCH In review Michal Hocko 13 participants 152 messages Dec 15, 2015 – Apr 4, 2016 Page 1 of 2

Subsystems (8)
lore
Share & tools
lore

Michal Hocko

author reviewer
#0 lore raw

cover OOM detection rework v4

Hi,

This is v4 of the series. The previous version was posted [1]. I have dropped the RFC because this has been sitting and waiting for the fundamental objections for quite some time and there were none. I still do not think we should rush this and merge it no sooner than 4.6. Having this in the mmotm and thus linux-next would open it to a much larger testing coverage. I will iron out issues as they come but hopefully there will no serious ones.

* Changes since v3
- factor out the new heuristic into its own function as suggested by Johannes (no functional changes)
* Changes since v2
- rebased on top of mmotm-2015-11-25-17-08 which includes wait_iff_congested related changes which needed refresh in patch#1 and patch#2
- use zone_page_state_snapshot for NR_FREE_PAGES per David
- shrink_zones doesn't need to return anything per David
- retested because the major kernel version has changed since the last time (4.2 -> 4.3 based kernel + mmotm patches)

* Changes since v1
- backoff calculation was de-obfuscated by using DIV_ROUND_UP
- __GFP_NOFAIL high order migh fail fixed - theoretical bug

as pointed by Linus [2][3] relying on zone_reclaimable as a way to communicate the reclaim progress is rater dubious. I tend to agree, not only it is really obscure, it is not hard to imagine cases where a single page freed in the loop keeps all the reclaimers looping without getting any progress because their gfp_mask wouldn't allow to get that page anyway (e.g. single GFP_ATOMIC alloc and free loop). This is rather rare so it doesn't happen in the practice but the current logic which we have is rather obscure and hard to follow a also non-deterministic.

This is an attempt to make the OOM detection more deterministic and easier to follow because each reclaimer basically tracks its own progress which is implemented at the page allocator layer rather spread out between the allocator and the reclaim. The more on the implementation is described in the first patch.

I have tested several different scenarios but it should be clear that testing OOM killer is quite hard to be representative. There is usually a tiny gap between almost OOM and full blown OOM which is often time sensitive. Anyway, I have tested the following 3 scenarios and I would appreciate if there are more to test.

Testing environment: a virtual machine with 2G of RAM and 2CPUs without any swap to make the OOM more deterministic.

1) 2 writers (each doing dd with 4M blocks to an xfs partition with 1G size, removes the files and starts over again) running in parallel for 10s to build up a lot of dirty pages when 100 parallel mem_eaters (anon private populated mmap which waits until it gets signal) with 80M each.

This causes an OOM flood of course and I have compared both patched and unpatched kernels. The test is considered finished after there are no OOM conditions detected. This should tell us whether there are any excessive kills or some of them premature:

I have performed two runs this time each after a fresh boot.

* base kernel

$ grep "Killed process" base-oom-run1.log | tail -n1
[  211.824379] Killed process 3086 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:332kB, shmem-rss:0kB
$ grep "Killed process" base-oom-run2.log | tail -n1
[  157.188326] Killed process 3094 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:368kB, shmem-rss:0kB
$ grep "invoked oom-killer" base-oom-run1.log | wc -l
78
$ grep "invoked oom-killer" base-oom-run2.log | wc -l
76

The number of OOM invocations is consistent with my last measurements but the runtime is way too different (it took 800+s). One thing that could have skewed results was that I was tail -f the serial log on the host system to see the progress. I have stopped doing that. The results are more consistent now but still too different from the last time. This is really weird so I've retested with the last 4.2 mmotm again and I am getting consistent ~220s which is really close to the above. If I apply the WQ vmstat patch on top I am getting close to 160s so the stale vmstat counters made a difference which is to be expected. I have a new SSD in my laptop which migh have made a difference but I wouldn't expect it to be that large.

$ grep "DMA32.*all_unreclaimable? no" base-oom-run1.log | wc -l
4
$ grep "DMA32.*all_unreclaimable? no" base-oom-run2.log | wc -l
1

* patched kernel

$ grep "Killed process" patched-oom-run1.log | tail -n1
[  341.164930] Killed process 3099 (mem_eater) total-vm:85852kB, anon-rss:82000kB, file-rss:336kB, shmem-rss:0kB
$ grep "Killed process" patched-oom-run2.log | tail -n1
[  349.111539] Killed process 3082 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:4kB, shmem-rss:0kB
$ grep "invoked oom-killer" patched-oom-run1.log | wc -l
78
$ grep "invoked oom-killer" patched-oom-run2.log | wc -l
77
$ grep "DMA32.*all_unreclaimable? no" patched-oom-run1.log | wc -l
1
$ grep "DMA32.*all_unreclaimable? no" patched-oom-run2.log | wc -l
0

So the number of OOM killer invocation is the same but the overall runtime of the test was much longer with the patched kernel. This can be attributed to more retries in general. The results from the base kernel are quite inconsitent and I think that consistency is better here.

2) 2 writers again with 10s of run and then 10 mem_eaters to consume as much memory as possible without triggering the OOM killer. This required a lot of tuning but I've considered 3 consecutive runs without OOM as a success.

* base kernel
size=$(awk '/MemFree/{printf "%dK", ($2/10)-(15*1024)}' /proc/meminfo)

* patched kernel
size=$(awk '/MemFree/{printf "%dK", ($2/10)-(9*1024)}' /proc/meminfo)

It was -14M for the base 4.2 kernel and -7500M for the patched 4.2 kernel in my last measurements.
The patched kernel handled the low mem conditions better and fired OOM killer later.

3) Costly high-order allocations with a limited amount of memory. Start 10 memeaters in parallel each with
size=$(awk '/MemTotal/{printf "%d\n", $2/10}' /proc/meminfo) This will cause an OOM killer which will kill one of them which will free up 200M and then try to use all the remaining space for hugetlb pages. See how many of them will pass kill everything, wait 2s and try again. This tests whether we do not fail __GFP_REPEAT costly allocations too early now.
* base kernel

$ sort base-hugepages.log | uniq -c
      1 64
     13 65
      6 66
     20 Trying to allocate 73

* patched kernel

$ sort patched-hugepages.log | uniq -c
     17 65
      3 66
     20 Trying to allocate 73

This also doesn't look very bad but this particular test is quite timing sensitive.

The above results do seem optimistic but more loads should be tested obviously. I would really appreciate a feedback on the approach I have chosen before I go into more tuning. Is this viable way to go?

[1] http://lkml.kernel.org/r/[email protected]
[2] http://lkml.kernel.org/r/CA+55aFwapaED7JV6zm-NVkP-jKie+eQ1vDXWrKD=SkbshZSgmw@mail.gmail.com
[3] http://lkml.kernel.org/r/CA+55aFxwg=vS2nrXsQhAUzPQDGb8aQpZi0M7UUh21ftBo-z46Q@mail.gmail.com

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#1 lore raw

PATCH 1/3 mm, oom: rework oom detection

From: Michal Hocko <[email protected]>

__alloc_pages_slowpath has traditionally relied on the direct reclaim and did_some_progress as an indicator that it makes sense to retry allocation rather than declaring OOM. shrink_zones had to rely on zone_reclaimable if shrink_zone didn't make any progress to prevent from a premature OOM killer invocation - the LRU might be full of dirty or writeback pages and direct reclaim cannot clean those up.

zone_reclaimable allows to rescan the reclaimable lists several times and restart if a page is freed. This is really subtle behavior and it might lead to a livelock when a single freed page keeps allocator looping but the current task will not be able to allocate that single page. OOM killer would be more appropriate than looping without any progress for unbounded amount of time.

This patch changes OOM detection logic and pulls it out from shrink_zone which is too low to be appropriate for any high level decisions such as OOM which is per zonelist property. It is __alloc_pages_slowpath which knows how many attempts have been done and what was the progress so far therefore it is more appropriate to implement this logic.

The new heuristic is implemented in should_reclaim_retry helper called from __alloc_pages_slowpath. It tries to be more deterministic and easier to follow. It builds on an assumption that retrying makes sense only if the currently reclaimable memory + free pages would allow the current allocation request to succeed (as per __zone_watermark_ok) at least for one zone in the usable zonelist.

This alone wouldn't be sufficient, though, because the writeback might get stuck and reclaimable pages might be pinned for a really long time or even depend on the current allocation context. Therefore there is a feedback mechanism implemented which reduces the reclaim target after each reclaim round without any progress. This means that we should eventually converge to only NR_FREE_PAGES as the target and fail on the wmark check and proceed to OOM. The backoff is simple and linear with 1/16 of the reclaimable pages for each round without any progress. We are optimistic and reset counter for successful reclaim rounds.

Costly high order pages mostly preserve their semantic and those without __GFP_REPEAT fail right away while those which have the flag set will back off after the amount of reclaimable pages reaches equivalent of the requested order. The only difference is that if there was no progress during the reclaim we rely on zone watermark check. This is more logical thing to do than previous 1<<order attempts which were a result of zone_reclaimable faking the progress.

[[email protected]: separate the heuristic into should_reclaim_retry]
[[email protected]: use zone_page_state_snapshot for NR_FREE_PAGES]
[[email protected]: shrink_zones doesn't need to return anything]

factor out the retry logic into separate function - per Johannes

---
 include/linux/swap.h |  1 +
 mm/page_alloc.c      | 91 +++++++++++++++++++++++++++++++++++++++++++++++-----
 mm/vmscan.c          | 25 +++------------
 3 files changed, 88 insertions(+), 29 deletions(-)
include/linux/swap.h +1 -0
index 457181844b6e..738ae2206635 100644
@@ -316,6 +316,7 @@ extern void lru_cache_add_active_or_unevictable(struct page *page,
 						struct vm_area_struct *vma);
 
 /* linux/mm/vmscan.c */
+extern unsigned long zone_reclaimable_pages(struct zone *zone);
 extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 					gfp_t gfp_mask, nodemask_t *mask);
 extern int __isolate_lru_page(struct page *page, isolate_mode_t mode);
mm/page_alloc.c +83 -8

118 lines not on this page · raw message

mm/vmscan.c +4 -21
index 4589cfdbe405..489212252cd6 100644
@@ -192,7 +192,7 @@ static bool sane_reclaim(struct scan_control *sc)
 }
 #endif
 
-static unsigned long zone_reclaimable_pages(struct zone *zone)
+unsigned long zone_reclaimable_pages(struct zone *zone)
 {
 	unsigned long nr;
 
@@ -2516,10 +2516,8 @@ static inline bool compaction_ready(struct zone *zone, int order)
  *
  * If a zone is deemed to be full of pinned pages then just give it a light
  * scan then give up on it.
- *
- * Returns true if a zone was reclaimable.
  */
-static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
+static void shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 {
 	struct zoneref *z;
 	struct zone *zone;
@@ -2527,7 +2525,6 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 	unsigned long nr_soft_scanned;
 	gfp_t orig_mask;
 	enum zone_type requested_highidx = gfp_zone(sc->gfp_mask);
-	bool reclaimable = false;
 
 	/*
 	 * If the number of buffer_heads in the machine exceeds the maximum
@@ -2592,17 +2589,10 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 						&nr_soft_scanned);
 			sc->nr_reclaimed += nr_soft_reclaimed;
 			sc->nr_scanned += nr_soft_scanned;
-			if (nr_soft_reclaimed)
-				reclaimable = true;
 			/* need some check for avoid more shrink_zone() */
 		}
 
-		if (shrink_zone(zone, sc, zone_idx(zone) == classzone_idx))
-			reclaimable = true;
-
-		if (global_reclaim(sc) &&
-		    !reclaimable && zone_reclaimable(zone))
-			reclaimable = true;
+		shrink_zone(zone, sc, zone_idx(zone));
 	}
 
 	/*
@@ -2610,8 +2600,6 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 	 * promoted it to __GFP_HIGHMEM.
 	 */
 	sc->gfp_mask = orig_mask;
-
-	return reclaimable;
 }
 
 /*
@@ -2636,7 +2624,6 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 	int initial_priority = sc->priority;
 	unsigned long total_scanned = 0;
 	unsigned long writeback_threshold;
-	bool zones_reclaimable;
 retry:
 	delayacct_freepages_start();
 
@@ -2647,7 +2634,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		vmpressure_prio(sc->gfp_mask, sc->target_mem_cgroup,
 				sc->priority);
 		sc->nr_scanned = 0;
-		zones_reclaimable = shrink_zones(zonelist, sc);
+		shrink_zones(zonelist, sc);
 
 		total_scanned += sc->nr_scanned;
 		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
@@ -2694,10 +2681,6 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
 		goto retry;
 	}
 
-	/* Any of the zones still reclaimable?  Don't OOM. */
-	if (zones_reclaimable)
-		return 1;
-
 	return 0;
 }
 
Signature · 7 lines
2.6.2

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#2 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

Quoted 75 lines · Michal Hocko
include/linux/swap.h +1 -0
index 457181844b6e..738ae2206635 100644
@@ -316,6 +316,7 @@ extern void lru_cache_add_active_or_unevictable(struct page *page,
 						struct vm_area_struct *vma);
 
 /* linux/mm/vmscan.c */
+extern unsigned long zone_reclaimable_pages(struct zone *zone);
 extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 					gfp_t gfp_mask, nodemask_t *mask);
 extern int __isolate_lru_page(struct page *page, isolate_mode_t mode);
mm/page_alloc.c +55 -0
index e267faad4649..f77e283fb8c6 100644
@@ -2984,6 +2984,75 @@ static inline bool is_thp_gfp_mask(gfp_t gfp_mask)
 	return (gfp_mask & (GFP_TRANSHUGE | __GFP_KSWAPD_RECLAIM)) == GFP_TRANSHUGE;
 }
 
+/*
+ * Maximum number of reclaim retries without any progress before OOM killer
+ * is consider as the only way to move forward.
+ */
+#define MAX_RECLAIM_RETRIES 16
+
+/*
+ * Checks whether it makes sense to retry the reclaim to make a forward progress
+ * for the given allocation request.
+ * The reclaim feedback represented by did_some_progress (any progress during
+ * the last reclaim round), pages_reclaimed (cumulative number of reclaimed
+ * pages) and no_progress_loops (number of reclaim rounds without any progress
+ * in a row) is considered as well as the reclaimable pages on the applicable
+ * zone list (with a backoff mechanism which is a function of no_progress_loops).
+ *
+ * Returns true if a retry is viable or false to enter the oom path.
+ */
+static inline bool
+should_reclaim_retry(gfp_t gfp_mask, unsigned order,
+		     struct alloc_context *ac, int alloc_flags,
+		     bool did_some_progress, unsigned long pages_reclaimed,
+		     int no_progress_loops)
+{
+	struct zone *zone;
+	struct zoneref *z;
+
+	/*
+	 * Make sure we converge to OOM if we cannot make any progress
+	 * several times in the row.
+	 */
+	if (no_progress_loops > MAX_RECLAIM_RETRIES)
+		return false;
+
+	/* Do not retry high order allocations unless they are __GFP_REPEAT */
+	if (order > PAGE_ALLOC_COSTLY_ORDER) {
+		if (!(gfp_mask & __GFP_REPEAT) || pages_reclaimed >= (1<<order))
+			return false;
+
+		if (did_some_progress)
+			return true;
+	}
+
+	/*
+	 * Keep reclaiming pages while there is a chance this will lead somewhere.
+	 * If none of the target zones can satisfy our allocation request even
+	 * if all reclaimable pages are considered then we are screwed and have
+	 * to go OOM.
+	 */
+	for_each_zone_zonelist_nodemask(zone, z, ac->zonelist, ac->high_zoneidx, ac->nodemask) {
+		unsigned long available;
+
+		available = zone_reclaimable_pages(zone);
+		available -= DIV_ROUND_UP(no_progress_loops * available, MAX_RECLAIM_RETRIES);
+		available += zone_page_state_snapshot(zone, NR_FREE_PAGES);
+

mm/page_alloc.c static inline bool is_thp_gfp_mask(gfp_t gfp_mask) · line 3051

Quoted 10 lines
mm/page_alloc.c +10 -0
+		/*
+		 * Would the allocation succeed if we reclaimed the whole available?
+		 */
+		if (__zone_watermark_ok(zone, order, min_wmark_pages(zone),
+				ac->high_zoneidx, alloc_flags, available)) {
+			/* Wait for some write requests to complete then retry */
+			wait_iff_congested(zone, BLK_RW_ASYNC, HZ/50);
+			return true;
+		}
+	}

Tetsuo's log of an early oom in this thread shows that this check is wrong. The allocation in question is an order-2 GFP_KERNEL on a system with only ZONE_DMA and ZONE_DMA32:

	zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
	zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50

and the watermarks:

	Node 0 DMA free:6908kB min:44kB low:52kB high:64kB ...
	lowmem_reserve[]: 0 1714 1714 1714
	Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB  ...
	lowmem_reserve[]: 0 0 0 0

and the scary thing is that this triggers when no_progress_loops == 0, so this is the first time trying the allocation after progress has been made.

Watermarks clearly indicate that memory is available, the problem is fragmentation for the order-2 allocation. This is not a situation where we want to immediately call the oom killer to solve since we have no guarantee it is going to free contiguous memory (in fact it wouldn't be used at all for PAGE_ALLOC_COSTLY_ORDER).

There is order-2 memory available however:

	Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB

The failure for ZONE_DMA makes sense for the lowmem_reserve ratio, it's oom for this allocation. ZONE_DMA32 is not, however.

I'm wondering if this has to do with the z->nr_reserved_highatomic estimate. ZONE_DMA32 present pages is 2080640kB, so this would be limited to 1%, or 20806kB. That failure would make sense if free is 17996kB.

Tetsuo, would it be possible to try your workload with just this match and also show z->nr_reserved_highatomic?

This patch would need to at least have knowledge of the heuristics used by __zone_watermark_ok() since it's making an inference on reclaimability based on numbers that include pageblocks that are reserved from usage.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#3 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

Quoted 30 lines · David Rientjes

Tetsuo's log of an early oom in this thread shows that this check is wrong. The allocation in question is an order-2 GFP_KERNEL on a system with only ZONE_DMA and ZONE_DMA32:

	zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
	zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50

and the watermarks:

	Node 0 DMA free:6908kB min:44kB low:52kB high:64kB ...
	lowmem_reserve[]: 0 1714 1714 1714
	Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB  ...
	lowmem_reserve[]: 0 0 0 0

and the scary thing is that this triggers when no_progress_loops == 0, so this is the first time trying the allocation after progress has been made.

Watermarks clearly indicate that memory is available, the problem is fragmentation for the order-2 allocation. This is not a situation where we want to immediately call the oom killer to solve since we have no guarantee it is going to free contiguous memory (in fact it wouldn't be used at all for PAGE_ALLOC_COSTLY_ORDER).

There is order-2 memory available however:

	Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB

The failure for ZONE_DMA makes sense for the lowmem_reserve ratio, it's oom for this allocation. ZONE_DMA32 is not, however.

Quoted 6 lines

I'm wondering if this has to do with the z->nr_reserved_highatomic estimate. ZONE_DMA32 present pages is 2080640kB, so this would be limited to 1%, or 20806kB. That failure would make sense if free is 17996kB.

Tetsuo, would it be possible to try your workload with just this match and also show z->nr_reserved_highatomic?

I don't know what "try your workload with just this match" expects, but zone->nr_reserved_highatomic is always 0.

----------

[  178.058803] zone=DMA32 reclaimable=367474 available=369923 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0
[  178.061350] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0
[  178.132174] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:3256kB inactive_anon:172kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:56kB shmem:180kB slab_reclaimable:2056kB slab_unreclaimable:1096kB kernel_stack:192kB pagetables:180kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  178.145589] Node 0 DMA32 free:11532kB min:5564kB low:6952kB high:8344kB active_anon:133896kB inactive_anon:8204kB active_file:1001828kB inactive_file:462944kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021064kB mlocked:0kB dirty:8kB writeback:0kB mapped:8572kB shmem:8468kB slab_reclaimable:57136kB slab_unreclaimable:86380kB kernel_stack:50080kB pagetables:83600kB unstable:0kB bounce:0kB free_pcp:1268kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:356 all_unreclaimable? no
[  198.457718] zone=DMA32 reclaimable=381991 available=386237 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0
[  198.460111] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0
[  198.507204] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:3088kB inactive_anon:172kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:92kB shmem:180kB slab_reclaimable:976kB slab_unreclaimable:1468kB kernel_stack:672kB pagetables:336kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  198.507209] Node 0 DMA32 free:19992kB min:5564kB low:6952kB high:8344kB active_anon:104176kB inactive_anon:8204kB active_file:905320kB inactive_file:617264kB unevictable:0kB isolated(anon):0kB isolated(file):116kB present:2080640kB managed:2021064kB mlocked:0kB dirty:176kB writeback:0kB mapped:12772kB shmem:8468kB slab_reclaimable:60372kB slab_unreclaimable:77856kB kernel_stack:44144kB pagetables:69180kB unstable:0kB bounce:0kB free_pcp:1104kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  198.647075] zone=DMA32 reclaimable=374429 available=378945 no_progress_loops=0 did_some_progress=61 nr_reserved_highatomic=0
[  198.647076] zone=DMA reclaimable=1 available=1983 no_progress_loops=0 did_some_progress=61 nr_reserved_highatomic=0
[  198.652177] Node 0 DMA free:7928kB min:40kB low:48kB high:60kB active_anon:588kB inactive_anon:172kB active_file:0kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:88kB shmem:180kB slab_reclaimable:1008kB slab_unreclaimable:2576kB kernel_stack:1840kB pagetables:408kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  198.652182] Node 0 DMA32 free:17608kB min:5564kB low:6952kB high:8344kB active_anon:89528kB inactive_anon:8204kB active_file:1025084kB inactive_file:472512kB unevictable:0kB isolated(anon):0kB isolated(file):120kB present:2080640kB managed:2021064kB mlocked:0kB dirty:176kB writeback:0kB mapped:12848kB shmem:8468kB slab_reclaimable:60372kB slab_unreclaimable:86628kB kernel_stack:50880kB pagetables:82336kB unstable:0kB bounce:0kB free_pcp:236kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  207.045450] zone=DMA32 reclaimable=386923 available=392299 no_progress_loops=0 did_some_progress=38 nr_reserved_highatomic=0
[  207.045451] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=38 nr_reserved_highatomic=0
[  207.050241] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:732kB inactive_anon:336kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:140kB shmem:436kB slab_reclaimable:456kB slab_unreclaimable:3536kB kernel_stack:1584kB pagetables:188kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  207.050246] Node 0 DMA32 free:20092kB min:5564kB low:6952kB high:8344kB active_anon:91600kB inactive_anon:18620kB active_file:921896kB inactive_file:626544kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021064kB mlocked:0kB dirty:964kB writeback:0kB mapped:17016kB shmem:24584kB slab_reclaimable:51908kB slab_unreclaimable:72792kB kernel_stack:40832kB pagetables:67396kB unstable:0kB bounce:0kB free_pcp:472kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  221.034713] zone=DMA32 reclaimable=389283 available=393245 no_progress_loops=0 did_some_progress=40 nr_reserved_highatomic=0
[  221.037103] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=40 nr_reserved_highatomic=0
[  221.105952] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:416kB inactive_anon:304kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:132kB shmem:436kB slab_reclaimable:424kB slab_unreclaimable:3156kB kernel_stack:2352kB pagetables:212kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  221.119016] Node 0 DMA32 free:7220kB min:5564kB low:6952kB high:8344kB active_anon:74480kB inactive_anon:23544kB active_file:946560kB inactive_file:618900kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021064kB mlocked:0kB dirty:1056kB writeback:0kB mapped:14760kB shmem:32768kB slab_reclaimable:51328kB slab_unreclaimable:75692kB kernel_stack:42960kB pagetables:66732kB unstable:0kB bounce:0kB free_pcp:196kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:248 all_unreclaimable? no
[  224.072875] zone=DMA32 reclaimable=397667 available=401058 no_progress_loops=0 did_some_progress=56 nr_reserved_highatomic=0
[  224.075212] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=56 nr_reserved_highatomic=0
[  224.133813] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:664kB inactive_anon:296kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:436kB slab_reclaimable:424kB slab_unreclaimable:3760kB kernel_stack:1136kB pagetables:376kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  224.145691] Node 0 DMA32 free:12160kB min:5564kB low:6952kB high:8344kB active_anon:69352kB inactive_anon:23140kB active_file:1191992kB inactive_file:399408kB unevictable:0kB isolated(anon):0kB isolated(file):104kB present:2080640kB managed:2021064kB mlocked:0kB dirty:844kB writeback:0kB mapped:4916kB shmem:32768kB slab_reclaimable:51288kB slab_unreclaimable:68392kB kernel_stack:38560kB pagetables:61820kB unstable:0kB bounce:0kB free_pcp:184kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  234.291285] zone=DMA32 reclaimable=403563 available=407626 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0
[  234.293557] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0
[  234.357091] Node 0 DMA free:7920kB min:40kB low:48kB high:60kB active_anon:312kB inactive_anon:296kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:144kB shmem:436kB slab_reclaimable:424kB slab_unreclaimable:2596kB kernel_stack:2992kB pagetables:204kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  234.370106] Node 0 DMA32 free:6804kB min:5564kB low:6952kB high:8344kB active_anon:77364kB inactive_anon:23140kB active_file:1168356kB inactive_file:454384kB unevictable:0kB isolated(anon):0kB isolated(file):128kB present:2080640kB managed:2021064kB mlocked:0kB dirty:0kB writeback:0kB mapped:11884kB shmem:32768kB slab_reclaimable:51292kB slab_unreclaimable:61492kB kernel_stack:32016kB pagetables:49248kB unstable:0kB bounce:0kB free_pcp:760kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:696 all_unreclaimable? no
[  246.183836] zone=DMA32 reclaimable=405496 available=410200 no_progress_loops=0 did_some_progress=59 nr_reserved_highatomic=0
[  246.186069] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=59 nr_reserved_highatomic=0
[  246.246157] Node 0 DMA free:7920kB min:40kB low:48kB high:60kB active_anon:1144kB inactive_anon:284kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:124kB shmem:436kB slab_reclaimable:424kB slab_unreclaimable:2404kB kernel_stack:1392kB pagetables:660kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  246.260159] Node 0 DMA32 free:11564kB min:5564kB low:6952kB high:8344kB active_anon:74360kB inactive_anon:23036kB active_file:1173248kB inactive_file:456000kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021064kB mlocked:0kB dirty:732kB writeback:0kB mapped:14812kB shmem:32768kB slab_reclaimable:51292kB slab_unreclaimable:59884kB kernel_stack:31824kB pagetables:47960kB unstable:0kB bounce:0kB free_pcp:136kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  258.994846] zone=DMA32 reclaimable=403441 available=407544 no_progress_loops=0 did_some_progress=61 nr_reserved_highatomic=0
[  258.997488] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=61 nr_reserved_highatomic=0
[  259.055818] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:848kB inactive_anon:284kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:136kB shmem:436kB slab_reclaimable:428kB slab_unreclaimable:2692kB kernel_stack:1872kB pagetables:476kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  259.067950] Node 0 DMA32 free:29136kB min:5564kB low:6952kB high:8344kB active_anon:71476kB inactive_anon:23032kB active_file:1129276kB inactive_file:485324kB unevictable:0kB isolated(anon):0kB isolated(file):112kB present:2080640kB managed:2021064kB mlocked:0kB dirty:0kB writeback:0kB mapped:14340kB shmem:32768kB slab_reclaimable:51312kB slab_unreclaimable:61680kB kernel_stack:34704kB pagetables:44856kB unstable:0kB bounce:0kB free_pcp:1996kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  271.392099] zone=DMA32 reclaimable=399774 available=406049 no_progress_loops=0 did_some_progress=59 nr_reserved_highatomic=0
[  271.394646] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=59 nr_reserved_highatomic=0
[  271.459049] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:832kB inactive_anon:284kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:124kB shmem:436kB slab_reclaimable:428kB slab_unreclaimable:2824kB kernel_stack:2320kB pagetables:180kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  271.472413] Node 0 DMA32 free:21848kB min:5564kB low:6952kB high:8344kB active_anon:77144kB inactive_anon:23032kB active_file:1148420kB inactive_file:462308kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021064kB mlocked:0kB dirty:664kB writeback:0kB mapped:14700kB shmem:32768kB slab_reclaimable:51312kB slab_unreclaimable:61672kB kernel_stack:32064kB pagetables:50888kB unstable:0kB bounce:0kB free_pcp:848kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  274.428858] zone=DMA32 reclaimable=404186 available=408756 no_progress_loops=0 did_some_progress=52 nr_reserved_highatomic=0
[  274.431146] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=52 nr_reserved_highatomic=0
[  274.487864] Node 0 DMA free:7924kB min:40kB low:48kB high:60kB active_anon:600kB inactive_anon:284kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:436kB slab_reclaimable:428kB slab_unreclaimable:3504kB kernel_stack:1120kB pagetables:532kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  274.499779] Node 0 DMA32 free:17040kB min:5564kB low:6952kB high:8344kB active_anon:60480kB inactive_anon:23032kB active_file:1277956kB inactive_file:339528kB unevictable:0kB isolated(anon):0kB isolated(file):68kB present:2080640kB managed:2021064kB mlocked:0kB dirty:664kB writeback:0kB mapped:5912kB shmem:32768kB slab_reclaimable:51312kB slab_unreclaimable:64216kB kernel_stack:37520kB pagetables:52096kB unstable:0kB bounce:0kB free_pcp:308kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no

----------
Complete log is at http://I-love.SAKURA.ne.jp/tmp/serial-20160116.txt.xz .

Quoted 5 lines

This patch would need to at least have knowledge of the heuristics used by __zone_watermark_ok() since it's making an inference on reclaimability based on numbers that include pageblocks that are reserved from usage.

#4 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

Quoted 34 lines · Tetsuo Handa
Quoted 34 lines

Tetsuo's log of an early oom in this thread shows that this check is wrong. The allocation in question is an order-2 GFP_KERNEL on a system with only ZONE_DMA and ZONE_DMA32:

	zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
	zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50

and the watermarks:

	Node 0 DMA free:6908kB min:44kB low:52kB high:64kB ...
	lowmem_reserve[]: 0 1714 1714 1714
	Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB  ...
	lowmem_reserve[]: 0 0 0 0

and the scary thing is that this triggers when no_progress_loops == 0, so this is the first time trying the allocation after progress has been made.

Watermarks clearly indicate that memory is available, the problem is fragmentation for the order-2 allocation. This is not a situation where we want to immediately call the oom killer to solve since we have no guarantee it is going to free contiguous memory (in fact it wouldn't be used at all for PAGE_ALLOC_COSTLY_ORDER).

There is order-2 memory available however:

	Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB

The failure for ZONE_DMA makes sense for the lowmem_reserve ratio, it's oom for this allocation. ZONE_DMA32 is not, however.

I'm wondering if this has to do with the z->nr_reserved_highatomic estimate. ZONE_DMA32 present pages is 2080640kB, so this would be limited to 1%, or 20806kB. That failure would make sense if free is 17996kB.

Quoted 6 lines
Quoted 2 lines

Tetsuo, would it be possible to try your workload with just this match and also show z->nr_reserved_highatomic?

I don't know what "try your workload with just this match" expects, but zone->nr_reserved_highatomic is always 0.

My point about z->nr_reserved_highatomic still stands, specifically that pageblocks may be reserved from allocation and __zone_watermark_ok() may fail, which would cause a premature oom condition, for this patch's calculation of "available". It may not have caused a problem on your specific workload, however.

Are you able to precisely identify why __zone_watermark_ok() is failing and triggering the oom in the log you posted January 3?

[  154.829582] zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
[  154.831562] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50
// here //
[  154.838499] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  154.841167] fork cpuset=/ mems_allowed=0
[  154.842348] CPU: 1 PID: 9599 Comm: fork Tainted: G        W       4.4.0-rc7-next-20151231+ #273
...
[  154.852386] Call Trace:
[  154.853350]  [<ffffffff81398b83>] dump_stack+0x4b/0x68
[  154.854731]  [<ffffffff811bc81c>] dump_header+0x5b/0x3b0
[  154.856309]  [<ffffffff810bdd79>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[  154.858046]  [<ffffffff810bde4d>] ? trace_hardirqs_on+0xd/0x10
[  154.859593]  [<ffffffff81143d36>] oom_kill_process+0x366/0x540
[  154.861142]  [<ffffffff8114414f>] out_of_memory+0x1ef/0x5a0
[  154.862655]  [<ffffffff8114420d>] ? out_of_memory+0x2ad/0x5a0
[  154.864194]  [<ffffffff81149c72>] __alloc_pages_nodemask+0xda2/0xde0
[  154.865852]  [<ffffffff810bdd00>] ? trace_hardirqs_on_caller+0x80/0x1c0
[  154.867844]  [<ffffffff81149e6c>] alloc_kmem_pages_node+0x4c/0xc0
[  154.868726] zone=DMA32 reclaimable=309003 available=312677 no_progress_loops=0 did_some_progress=48
[  154.868727] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=48

// and also here, if we didn't serialize the oom killer //

I think that would help in fixing the issue you reported.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#5 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

Quoted 23 lines · David Rientjes

Are you able to precisely identify why __zone_watermark_ok() is failing and triggering the oom in the log you posted January 3?

[  154.829582] zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
[  154.831562] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50
// here //
[  154.838499] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  154.841167] fork cpuset=/ mems_allowed=0
[  154.842348] CPU: 1 PID: 9599 Comm: fork Tainted: G        W       4.4.0-rc7-next-20151231+ #273
...
[  154.852386] Call Trace:
[  154.853350]  [<ffffffff81398b83>] dump_stack+0x4b/0x68
[  154.854731]  [<ffffffff811bc81c>] dump_header+0x5b/0x3b0
[  154.856309]  [<ffffffff810bdd79>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[  154.858046]  [<ffffffff810bde4d>] ? trace_hardirqs_on+0xd/0x10
[  154.859593]  [<ffffffff81143d36>] oom_kill_process+0x366/0x540
[  154.861142]  [<ffffffff8114414f>] out_of_memory+0x1ef/0x5a0
[  154.862655]  [<ffffffff8114420d>] ? out_of_memory+0x2ad/0x5a0
[  154.864194]  [<ffffffff81149c72>] __alloc_pages_nodemask+0xda2/0xde0
[  154.865852]  [<ffffffff810bdd00>] ? trace_hardirqs_on_caller+0x80/0x1c0
[  154.867844]  [<ffffffff81149e6c>] alloc_kmem_pages_node+0x4c/0xc0
[  154.868726] zone=DMA32 reclaimable=309003 available=312677 no_progress_loops=0 did_some_progress=48
[  154.868727] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=48
Quoted 4 lines

// and also here, if we didn't serialize the oom killer //

I think that would help in fixing the issue you reported.

Does "why __zone_watermark_ok() is failing" mean "which 'return false;' statement in __zone_watermark_ok() I'm hitting on my specific workload"? Then, answer is the former for DMA zone and the latter for DMA32 zone.

----------

mm/page_alloc.c +17 -7
index 9d70a80..dd36f01 100644
@@ -2390,7 +2390,7 @@ static inline bool should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
  */
 static bool __zone_watermark_ok(struct zone *z, unsigned int order,
 			unsigned long mark, int classzone_idx, int alloc_flags,
-			long free_pages)
+				long free_pages, bool *no_free)
 {
 	long min = mark;
 	int o;
@@ -2423,6 +2423,7 @@ static bool __zone_watermark_ok(struct zone *z, unsigned int order,
 	 * are not met, then a high-order request also cannot go ahead
 	 * even if a suitable page happened to be free.
 	 */
+	*no_free = false;
 	if (free_pages <= min + z->lowmem_reserve[classzone_idx])
 		return false;
 
@@ -2453,26 +2454,30 @@ static bool __zone_watermark_ok(struct zone *z, unsigned int order,
 		}
 #endif
 	}
+	*no_free = true;
 	return false;
 }
 
 bool zone_watermark_ok(struct zone *z, unsigned int order, unsigned long mark,
 		      int classzone_idx, int alloc_flags)
 {
+	bool unused;
+
 	return __zone_watermark_ok(z, order, mark, classzone_idx, alloc_flags,
-					zone_page_state(z, NR_FREE_PAGES));
+				   zone_page_state(z, NR_FREE_PAGES), &unused);
 }
 
 bool zone_watermark_ok_safe(struct zone *z, unsigned int order,
 			unsigned long mark, int classzone_idx)
 {
+	bool unused;
 	long free_pages = zone_page_state(z, NR_FREE_PAGES);
 
 	if (z->percpu_drift_mark && free_pages < z->percpu_drift_mark)
 		free_pages = zone_page_state_snapshot(z, NR_FREE_PAGES);
 
 	return __zone_watermark_ok(z, order, mark, classzone_idx, 0,
-								free_pages);
+				   free_pages, &unused);
 }
 
 #ifdef CONFIG_NUMA
@@ -3014,7 +3019,7 @@ static inline bool is_thp_gfp_mask(gfp_t gfp_mask)
 static inline bool
 should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		     struct alloc_context *ac, int alloc_flags,
-		     bool did_some_progress,
+		     unsigned long did_some_progress,
 		     int no_progress_loops)
 {
 	struct zone *zone;
@@ -3024,8 +3029,10 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 	 * Make sure we converge to OOM if we cannot make any progress
 	 * several times in the row.
 	 */
-	if (no_progress_loops > MAX_RECLAIM_RETRIES)
+	if (no_progress_loops > MAX_RECLAIM_RETRIES) {
+		printk(KERN_INFO "Reached MAX_RECLAIM_RETRIES.\n");
 		return false;
+	}
 
 	/* Do not retry high order allocations unless they are __GFP_REPEAT */
 	if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
@@ -3039,6 +3046,7 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 	 */
 	for_each_zone_zonelist_nodemask(zone, z, ac->zonelist,
 			ac->high_zoneidx, ac->nodemask) {
+		bool no_free;
 		unsigned long available;
 		unsigned long reclaimable;
 
@@ -3052,7 +3060,7 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		 * available?
 		 */
 		if (__zone_watermark_ok(zone, order, min_wmark_pages(zone),
-				ac->high_zoneidx, alloc_flags, available)) {
+					ac->high_zoneidx, alloc_flags, available, &no_free)) {
 			unsigned long writeback;
 			unsigned long dirty;
 
@@ -3086,6 +3094,8 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 
 			return true;
 		}
+		printk(KERN_INFO "zone=%s reclaimable=%lu available=%lu no_progress_loops=%u did_some_progress=%lu nr_reserved_highatomic=%lu no_free=%u\n",
+		       zone->name, reclaimable, available, no_progress_loops, did_some_progress, zone->nr_reserved_highatomic, no_free);
 	}
 
 	return false;
@@ -3273,7 +3283,7 @@ retry:
 		no_progress_loops++;
 
 	if (should_reclaim_retry(gfp_mask, order, ac, alloc_flags,
-				 did_some_progress > 0, no_progress_loops))
+				 did_some_progress, no_progress_loops))
 		goto retry;
 
 	/* Reclaim has failed us, start killing things */

----------

Complete log is at http://I-love.SAKURA.ne.jp/tmp/serial-20160120.txt.xz .
----------

[  141.987548] zone=DMA32 reclaimable=367085 available=371232 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=1
[  141.990091] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=0
[  141.997360] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  142.055908] Node 0 DMA free:7920kB min:40kB low:48kB high:60kB active_anon:3208kB inactive_anon:188kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB\
 dirty:0kB writeback:0kB mapped:60kB shmem:188kB slab_reclaimable:2792kB slab_unreclaimable:360kB kernel_stack:224kB pagetables:260kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_\
unreclaimable? no
[  142.066690] lowmem_reserve[]: 0 1970 1970 1970
[  142.914557] zone=DMA32 reclaimable=345975 available=348821 no_progress_loops=0 did_some_progress=58 nr_reserved_highatomic=0 no_free=1
[  142.914558] zone=DMA reclaimable=2 available=1980 no_progress_loops=0 did_some_progress=58 nr_reserved_highatomic=0 no_free=0
[  142.921113] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  153.615466] zone=DMA32 reclaimable=385567 available=389678 no_progress_loops=0 did_some_progress=36 nr_reserved_highatomic=0 no_free=1
[  153.615467] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=36 nr_reserved_highatomic=0 no_free=0
[  153.620507] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  153.658621] zone=DMA32 reclaimable=384064 available=388833 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0 no_free=1
[  153.658623] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0 no_free=0
[  153.663401] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  159.614894] zone=DMA32 reclaimable=356635 available=361925 no_progress_loops=0 did_some_progress=32 nr_reserved_highatomic=0 no_free=1
[  159.614895] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=32 nr_reserved_highatomic=0 no_free=0
[  159.622374] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  164.781516] zone=DMA32 reclaimable=393457 available=397561 no_progress_loops=0 did_some_progress=40 nr_reserved_highatomic=0 no_free=1
[  164.781518] zone=DMA reclaimable=1 available=1983 no_progress_loops=0 did_some_progress=40 nr_reserved_highatomic=0 no_free=0
[  164.786560] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  171.006952] zone=DMA32 reclaimable=405821 available=410137 no_progress_loops=0 did_some_progress=34 nr_reserved_highatomic=0 no_free=1
[  171.006954] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=34 nr_reserved_highatomic=0 no_free=0
[  171.010690] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  171.030121] zone=DMA32 reclaimable=405016 available=409801 no_progress_loops=0 did_some_progress=34 nr_reserved_highatomic=0 no_free=1
[  171.030123] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=34 nr_reserved_highatomic=0 no_free=0
[  171.033530] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  184.631660] zone=DMA32 reclaimable=356652 available=359338 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=1
[  184.634207] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=0
[  184.642800] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  190.499877] zone=DMA32 reclaimable=382152 available=384996 no_progress_loops=0 did_some_progress=32 nr_reserved_highatomic=0 no_free=1
[  190.499878] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=32 nr_reserved_highatomic=0 no_free=0
[  190.504901] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  196.146728] zone=DMA32 reclaimable=371941 available=374605 no_progress_loops=0 did_some_progress=61 nr_reserved_highatomic=0 no_free=1
[  196.146730] zone=DMA reclaimable=1 available=1982 no_progress_loops=0 did_some_progress=61 nr_reserved_highatomic=0 no_free=0
[  196.152546] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  201.837825] zone=DMA32 reclaimable=364569 available=370359 no_progress_loops=0 did_some_progress=59 nr_reserved_highatomic=0 no_free=1
[  201.837826] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=59 nr_reserved_highatomic=0 no_free=0
[  201.844879] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  212.862325] zone=DMA32 reclaimable=381542 available=387785 no_progress_loops=0 did_some_progress=39 nr_reserved_highatomic=0 no_free=1
[  212.862327] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=39 nr_reserved_highatomic=0 no_free=0
[  212.866857] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  212.866914] Node 0 DMA free:7920kB min:40kB low:48kB high:60kB active_anon:440kB inactive_anon:196kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB \
dirty:8kB writeback:0kB mapped:0kB shmem:280kB slab_reclaimable:480kB slab_unreclaimable:3856kB kernel_stack:1776kB pagetables:240kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_u\
nreclaimable? no
[  212.866915] lowmem_reserve[]: 0 1970 1970 1970

----------

author reviewer
#6 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

On Wed 20-01-16 20:13:32, Tetsuo Handa wrote:
[...]

Quoted 3 lines

Complete log is at http://I-love.SAKURA.ne.jp/tmp/serial-20160120.txt.xz .

[  141.987548] zone=DMA32 reclaimable=367085 available=371232 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=1

Ok, so we really do not have _any_ pages on the order 2+ free lists and that is why __zone_watermark_ok failed.

Quoted 1 line
[  141.990091] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=0

DMA zone is not even interesting because it is fully protected by the lowmem reserves.

Quoted 1 line
[  141.997360] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  142.086897] Node 0 DMA32: 1796*4kB (M) 763*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 13288kB

And indeed we still do not have any order-2+ available. OOM seems reasonable.

Quoted 3 lines
[  142.914557] zone=DMA32 reclaimable=345975 available=348821 no_progress_loops=0 did_some_progress=58 nr_reserved_highatomic=0 no_free=1
[  142.914558] zone=DMA reclaimable=2 available=1980 no_progress_loops=0 did_some_progress=58 nr_reserved_highatomic=0 no_free=0
[  142.921113] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  142.921192] Node 0 DMA32: 1794*4kB (UME) 464*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 10888kB

Ditto

Quoted 3 lines
[  153.615466] zone=DMA32 reclaimable=385567 available=389678 no_progress_loops=0 did_some_progress=36 nr_reserved_highatomic=0 no_free=1
[  153.615467] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=36 nr_reserved_highatomic=0 no_free=0
[  153.620507] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  153.620582] Node 0 DMA32: 1241*4kB (UME) 1280*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15204kB

Ditto

Quoted 3 lines
[  153.658621] zone=DMA32 reclaimable=384064 available=388833 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0 no_free=1
[  153.658623] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=37 nr_reserved_highatomic=0 no_free=0
[  153.663401] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  153.663480] Node 0 DMA32: 554*4kB (UME) 2148*8kB (UM) 3*16kB (M) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 19448kB

Now we have __zone_watermark_ok claiming no order 2+ blocks available but oom report little bit later sees 3 blocks. This would suggest that this is just a matter of timing when the children exit and free their stacks which are order-2.

Quoted 3 lines
[  159.614894] zone=DMA32 reclaimable=356635 available=361925 no_progress_loops=0 did_some_progress=32 nr_reserved_highatomic=0 no_free=1
[  159.614895] zone=DMA reclaimable=2 available=1983 no_progress_loops=0 did_some_progress=32 nr_reserved_highatomic=0 no_free=0
[  159.622374] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  159.622451] Node 0 DMA32: 2141*4kB (UM) 1435*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20044kB

Again no high order pages.

Quoted 3 lines
[  164.781516] zone=DMA32 reclaimable=393457 available=397561 no_progress_loops=0 did_some_progress=40 nr_reserved_highatomic=0 no_free=1
[  164.781518] zone=DMA reclaimable=1 available=1983 no_progress_loops=0 did_some_progress=40 nr_reserved_highatomic=0 no_free=0
[  164.786560] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  164.786643] Node 0 DMA32: 2961*4kB (UME) 432*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15300kB

Ditto

Quoted 3 lines
[  184.631660] zone=DMA32 reclaimable=356652 available=359338 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=1
[  184.634207] zone=DMA reclaimable=2 available=1982 no_progress_loops=0 did_some_progress=60 nr_reserved_highatomic=0 no_free=0
[  184.642800] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)

[...]

Quoted 1 line
[  184.728695] Node 0 DMA32: 3144*4kB (UME) 971*8kB (UME) 43*16kB (UM) 3*32kB (M) 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 21128kB

Again we have order >=2 pages available here after the allocator has seen none earlier. And the pattern repeats later on. So I would say that in this particular load it is a timing which plays the role. I am not sure we can tune for such a load beause any difference in the timing would result in a different behavior and basically breaking such a tuning.

The current heuristic is based on an assumption that retrying for high order allocations only makes sense if they are hidden behind the min watermark and the currently reclaimable pages would get us above the watermark. We cannot assume that the order-0 reclaimable pages will form the required high order blocks because there is no such guarantee. I think such a heuristic makes sense because we have passed the direct reclaim and also compaction at the time when we check for the retry so chances to get the required block from the reclaim are not that high.

So I am not really sure what to do here now. On one hand the previous heuristic would happen to work here probably better because we would be looping in the allocator, exiting processes would rest the counter and keep the retries and sooner or later the fork would be lucky and see its order-2 block and continue. We could starve in this state for basically unbounded amount of time though which is excatly what I would like to get rid of. I guess we might want to give few attempts to retry for all order>0. Let me think about it some more.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#7 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

In reply to Michal Hocko

On Tue, Dec 15, 2015 at 07:19:44PM +0100, Michal Hocko wrote:
...

Quoted 6 lines
mm/vmscan.c +0 -2
@@ -2592,17 +2589,10 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 						&nr_soft_scanned);
 			sc->nr_reclaimed += nr_soft_reclaimed;
 			sc->nr_scanned += nr_soft_scanned;
-			if (nr_soft_reclaimed)
-				reclaimable = true;

mm/vmscan.c static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc) · line 2595

Quoted 10 lines
mm/vmscan.c +1 -6
 			/* need some check for avoid more shrink_zone() */
 		}
 
-		if (shrink_zone(zone, sc, zone_idx(zone) == classzone_idx))
-			reclaimable = true;
-
-		if (global_reclaim(sc) &&
-		    !reclaimable && zone_reclaimable(zone))
-			reclaimable = true;
+		shrink_zone(zone, sc, zone_idx(zone));

Shouldn't it be

		shrink_zone(zone, sc, zone_idx(zone) == classzone_idx);

?

Review static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc) · line 2598

Quoted 3 lines
(quoted diff) +0 -0
 	}
 
 	/*

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#8 lore raw

Re: [PATCH 1/3] mm, oom: rework oom detection

Quoted 21 lines · Vladimir Davydov

On Tue, Dec 15, 2015 at 07:19:44PM +0100, Michal Hocko wrote:
...

Quoted 6 lines
mm/vmscan.c +0 -2
@@ -2592,17 +2589,10 @@ static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc)
 						&nr_soft_scanned);
 			sc->nr_reclaimed += nr_soft_reclaimed;
 			sc->nr_scanned += nr_soft_scanned;
-			if (nr_soft_reclaimed)
-				reclaimable = true;

mm/vmscan.c static bool shrink_zones(struct zonelist *zonelist, struct scan_control *sc) · line 2595

Quoted 10 lines
mm/vmscan.c +1 -6
 			/* need some check for avoid more shrink_zone() */
 		}
 
-		if (shrink_zone(zone, sc, zone_idx(zone) == classzone_idx))
-			reclaimable = true;
-
-		if (global_reclaim(sc) &&
-		    !reclaimable && zone_reclaimable(zone))
-			reclaimable = true;
+		shrink_zone(zone, sc, zone_idx(zone));

Shouldn't it be

Quoted 3 lines
		shrink_zone(zone, sc, zone_idx(zone) == classzone_idx);

?

I cannot remember the reason why I have removed it so it is more likely this was unintentional. Thanks for catching this. I will fold it into the original patch before I repost the full series (this week hopefully).

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#9 lore raw

PATCH 2/3 mm: throttle on IO only when there are too many dirty and writeback pages

In reply to Michal Hocko

From: Michal Hocko <[email protected]>

wait_iff_congested has been used to throttle allocator before it retried another round of direct reclaim to allow the writeback to make some progress and prevent reclaim from looping over dirty/writeback pages without making any progress. We used to do congestion_wait before 0e093d99763e ("writeback: do not sleep on the congestion queue if there are no congested BDIs or if significant congestion is not being encountered in the current zone") but that led to undesirable stalls and sleeping for the full timeout even when the BDI wasn't congested. Hence wait_iff_congested was used instead. But it seems that even wait_iff_congested doesn't work as expected. We might have a small file LRU list with all pages dirty/writeback and yet the bdi is not congested so this is just a cond_resched in the end and can end up triggering pre mature OOM.

This patch replaces the unconditional wait_iff_congested by congestion_wait which is executed only if we _know_ that the last round of direct reclaim didn't make any progress and dirty+writeback pages are more than a half of the reclaimable pages on the zone which might be usable for our target allocation. This shouldn't reintroduce stalls fixed by 0e093d99763e because congestion_wait is called only when we are getting hopeless when sleeping is a better choice than OOM with many pages under IO.

We have to preserve logic introduced by "mm, vmstat: allow WQ concurrency to discover memory reclaim doesn't make any progress" into the __alloc_pages_slowpath now that wait_iff_congested is not used anymore. As the only remaining user of wait_iff_congested is shrink_inactive_list we can remove the WQ specific short sleep from wait_iff_congested because the sleep is needed to be done only once in the allocation retry cycle.

---
 mm/backing-dev.c | 19 +++----------------
 mm/page_alloc.c  | 36 +++++++++++++++++++++++++++++++++---
 2 files changed, 36 insertions(+), 19 deletions(-)
mm/backing-dev.c +3 -16
index 7340353f8aea..d2473ce9cc57 100644
@@ -957,9 +957,8 @@ EXPORT_SYMBOL(congestion_wait);
  * jiffies for either a BDI to exit congestion of the given @sync queue
  * or a write to complete.
  *
- * In the absence of zone congestion, a short sleep or a cond_resched is
- * performed to yield the processor and to allow other subsystems to make
- * a forward progress.
+ * In the absence of zone congestion, cond_resched() is called to yield
+ * the processor if necessary but otherwise does not sleep.
  *
  * The return value is 0 if the sleep is for the full timeout. Otherwise,
  * it is the number of jiffies that were still remaining when the function
@@ -980,19 +979,7 @@ long wait_iff_congested(struct zone *zone, int sync, long timeout)
 	if (atomic_read(&nr_wb_congested[sync]) == 0 ||
 	    !test_bit(ZONE_CONGESTED, &zone->flags)) {
 
-		/*
-		 * Memory allocation/reclaim might be called from a WQ
-		 * context and the current implementation of the WQ
-		 * concurrency control doesn't recognize that a particular
-		 * WQ is congested if the worker thread is looping without
-		 * ever sleeping. Therefore we have to do a short sleep
-		 * here rather than calling cond_resched().
-		 */
-		if (current->flags & PF_WQ_WORKER)
-			schedule_timeout(1);
-		else
-			cond_resched();
-
+		cond_resched();
 		/* In case we scheduled, work out time remaining */
 		ret = timeout - (jiffies - start);
 		if (ret < 0)
mm/page_alloc.c +33 -3
index f77e283fb8c6..b2de8c8761ad 100644
@@ -3034,8 +3034,9 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 	 */
 	for_each_zone_zonelist_nodemask(zone, z, ac->zonelist, ac->high_zoneidx, ac->nodemask) {
 		unsigned long available;
+		unsigned long reclaimable;
 
-		available = zone_reclaimable_pages(zone);
+		available = reclaimable = zone_reclaimable_pages(zone);
 		available -= DIV_ROUND_UP(no_progress_loops * available, MAX_RECLAIM_RETRIES);
 		available += zone_page_state_snapshot(zone, NR_FREE_PAGES);
 
@@ -3044,8 +3045,37 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		 */
 		if (__zone_watermark_ok(zone, order, min_wmark_pages(zone),
 				ac->high_zoneidx, alloc_flags, available)) {
-			/* Wait for some write requests to complete then retry */
-			wait_iff_congested(zone, BLK_RW_ASYNC, HZ/50);
+			unsigned long writeback;
+			unsigned long dirty;
+
+			writeback = zone_page_state_snapshot(zone, NR_WRITEBACK);
+			dirty = zone_page_state_snapshot(zone, NR_FILE_DIRTY);
+
+			/*
+			 * If we didn't make any progress and have a lot of
+			 * dirty + writeback pages then we should wait for
+			 * an IO to complete to slow down the reclaim and
+			 * prevent from pre mature OOM
+			 */
+			if (!did_some_progress && 2*(writeback + dirty) > reclaimable) {
+				congestion_wait(BLK_RW_ASYNC, HZ/10);
+				return true;
+			}
+
+			/*
+			 * Memory allocation/reclaim might be called from a WQ
+			 * context and the current implementation of the WQ
+			 * concurrency control doesn't recognize that
+			 * a particular WQ is congested if the worker thread is
+			 * looping without ever sleeping. Therefore we have to
+			 * do a short sleep here rather than calling
+			 * cond_resched().
+			 */
+			if (current->flags & PF_WQ_WORKER)
+				schedule_timeout(1);
+			else
+				cond_resched();
+
 			return true;
 		}
 	}
Signature · 7 lines
2.6.2

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#10 lore raw

Re: [PATCH 2/3] mm: throttle on IO only when there are too many dirty and writeback pages

Today I was testing

----------

fs/fs-writeback.c +1 -1
index 6915c950e6e8..aa52e23ac280 100644
@@ -887,7 +887,7 @@ void wb_start_writeback(struct bdi_writeback *wb, long nr_pages,
 {
 	struct wb_writeback_work *work;
 
-	if (!wb_has_dirty_io(wb))
+	if (!wb_has_dirty_io(wb) || writeback_in_progress(wb))
 		return;
 
 	/*

----------

using next-20160317, and I got below results.

Complete log is at http://I-love.SAKURA.ne.jp/tmp/serial-20160317.txt.xz . ---------- console log ----------

[ 1354.048836] Out of memory: Kill process 3641 (file_io.02) score 1000 or sacrifice child
[ 1354.054773] Killed process 3641 (file_io.02) total-vm:4308kB, anon-rss:104kB, file-rss:1264kB, shmem-rss:0kB
[ 1593.471245] sysrq: SysRq : Show State
(...snipped...)
[ 1595.944649] kswapd0         D ffff88003681f760     0    53      2 0x00000000
[ 1595.949872]  ffff88003681f760 ffff88003fbfa140 ffff88003681a040 ffff880036820000
[ 1595.955342]  ffff88002b5e0750 ffff88002b5e0768 ffff88003681f958 0000000000000001
[ 1595.960826]  ffff88003681f778 ffffffff81660570 ffff88003681a040 ffff88003681f7d8
[ 1595.966319] Call Trace:
[ 1595.968662]  [<ffffffff81660570>] schedule+0x30/0x80
[ 1595.972552]  [<ffffffff81663fd6>] rwsem_down_read_failed+0xd6/0x140
[ 1595.977199]  [<ffffffff81322d98>] call_rwsem_down_read_failed+0x18/0x30
[ 1595.982087]  [<ffffffff810b8b4b>] down_read_nested+0x3b/0x50
[ 1595.986370]  [<ffffffffa024bcbb>] ? xfs_ilock+0x4b/0xe0 [xfs]
[ 1595.990681]  [<ffffffffa024bcbb>] xfs_ilock+0x4b/0xe0 [xfs]
[ 1595.994898]  [<ffffffffa0236330>] xfs_map_blocks+0x80/0x150 [xfs]
[ 1595.999441]  [<ffffffffa02372db>] xfs_do_writepage+0x15b/0x500 [xfs]
[ 1596.004138]  [<ffffffffa02376b6>] xfs_vm_writepage+0x36/0x70 [xfs]
[ 1596.008692]  [<ffffffff811538ef>] pageout.isra.43+0x18f/0x240
[ 1596.012938]  [<ffffffff81155253>] shrink_page_list+0x803/0xae0
[ 1596.017247]  [<ffffffff81155c8b>] shrink_inactive_list+0x1fb/0x460
[ 1596.021771]  [<ffffffff81156896>] shrink_zone_memcg+0x5b6/0x780
[ 1596.026103]  [<ffffffff81156b34>] shrink_zone+0xd4/0x2f0
[ 1596.030111]  [<ffffffff811579e1>] kswapd+0x441/0x830
[ 1596.033847]  [<ffffffff811575a0>] ? mem_cgroup_shrink_node_zone+0xb0/0xb0
[ 1596.038786]  [<ffffffff8109196e>] kthread+0xee/0x110
[ 1596.042546]  [<ffffffff81665672>] ret_from_fork+0x22/0x50
[ 1596.046591]  [<ffffffff81091880>] ? kthread_create_on_node+0x230/0x230
(...snipped...)
[ 1596.216946] kworker/u128:1  D ffff8800368eaf78     0    70      2 0x00000000
[ 1596.222105] Workqueue: writeback wb_workfn (flush-8:0)
[ 1596.226009]  ffff8800368eaf78 ffff88003aa4c040 ffff88003686c0c0 ffff8800368ec000
[ 1596.231502]  ffff8800368eafb0 ffff88003d610300 000000010013c47d ffff88003ffdf100
[ 1596.237003]  ffff8800368eaf90 ffffffff81660570 ffff88003d610300 ffff8800368eb038
[ 1596.242505] Call Trace:
[ 1596.244750]  [<ffffffff81660570>] schedule+0x30/0x80
[ 1596.248519]  [<ffffffff816645f7>] schedule_timeout+0x117/0x1c0
[ 1596.252841]  [<ffffffff810bc5c6>] ? mark_held_locks+0x66/0x90
[ 1596.257153]  [<ffffffff810df270>] ? init_timer_key+0x40/0x40
[ 1596.261424]  [<ffffffff810e60f7>] ? ktime_get+0xa7/0x130
[ 1596.265390]  [<ffffffff8165fab1>] io_schedule_timeout+0xa1/0x110
[ 1596.269836]  [<ffffffff8116104d>] congestion_wait+0x7d/0xd0
[ 1596.273978]  [<ffffffff810b6620>] ? wait_woken+0x80/0x80
[ 1596.278153]  [<ffffffff8114a982>] __alloc_pages_nodemask+0xb42/0xd50
[ 1596.283301]  [<ffffffff81193876>] alloc_pages_current+0x96/0x1b0
[ 1596.287737]  [<ffffffffa0270d70>] xfs_buf_allocate_memory+0x170/0x2ab [xfs]
[ 1596.292829]  [<ffffffffa023c9aa>] xfs_buf_get_map+0xfa/0x160 [xfs]
[ 1596.297457]  [<ffffffffa023cea9>] xfs_buf_read_map+0x29/0xe0 [xfs]
[ 1596.302034]  [<ffffffffa02670e7>] xfs_trans_read_buf_map+0x97/0x1a0 [xfs]
[ 1596.307004]  [<ffffffffa02171b3>] xfs_btree_read_buf_block.constprop.29+0x73/0xc0 [xfs]
[ 1596.312736]  [<ffffffffa021727b>] xfs_btree_lookup_get_block+0x7b/0xf0 [xfs]
[ 1596.317859]  [<ffffffffa021b981>] xfs_btree_lookup+0xc1/0x580 [xfs]
[ 1596.322448]  [<ffffffffa0205dcc>] ? xfs_allocbt_init_cursor+0x3c/0xc0 [xfs]
[ 1596.327478]  [<ffffffffa0204290>] xfs_alloc_ag_vextent_near+0xb0/0x880 [xfs]
[ 1596.332841]  [<ffffffffa0204b57>] xfs_alloc_ag_vextent+0xf7/0x130 [xfs]
[ 1596.338547]  [<ffffffffa02056a2>] xfs_alloc_vextent+0x3b2/0x480 [xfs]
[ 1596.343706]  [<ffffffffa021316f>] xfs_bmap_btalloc+0x3bf/0x710 [xfs]
[ 1596.348841]  [<ffffffffa02134c9>] xfs_bmap_alloc+0x9/0x10 [xfs]
[ 1596.353988]  [<ffffffffa0213eba>] xfs_bmapi_write+0x47a/0xa10 [xfs]
[ 1596.359255]  [<ffffffffa02493cd>] xfs_iomap_write_allocate+0x16d/0x380 [xfs]
[ 1596.365138]  [<ffffffffa02363ed>] xfs_map_blocks+0x13d/0x150 [xfs]
[ 1596.370046]  [<ffffffffa02372db>] xfs_do_writepage+0x15b/0x500 [xfs]
[ 1596.375322]  [<ffffffff8114d756>] write_cache_pages+0x1f6/0x490
[ 1596.380014]  [<ffffffffa0237180>] ? xfs_aops_discard_page+0x140/0x140 [xfs]
[ 1596.385220]  [<ffffffffa0236fa6>] xfs_vm_writepages+0x66/0xa0 [xfs]
[ 1596.389823]  [<ffffffff8114e8bc>] do_writepages+0x1c/0x30
[ 1596.393865]  [<ffffffff811ed543>] __writeback_single_inode+0x33/0x170
[ 1596.398583]  [<ffffffff811ede3e>] writeback_sb_inodes+0x2ce/0x570
[ 1596.403200]  [<ffffffff811ee167>] __writeback_inodes_wb+0x87/0xc0
[ 1596.407955]  [<ffffffff811ee38b>] wb_writeback+0x1eb/0x220
[ 1596.412037]  [<ffffffff811eea2f>] wb_workfn+0x1df/0x2b0
[ 1596.416133]  [<ffffffff8108b2c5>] process_one_work+0x1a5/0x400
[ 1596.420437]  [<ffffffff8108b261>] ? process_one_work+0x141/0x400
[ 1596.424836]  [<ffffffff8108b646>] worker_thread+0x126/0x490
[ 1596.428948]  [<ffffffff8108b520>] ? process_one_work+0x400/0x400
[ 1596.433635]  [<ffffffff8109196e>] kthread+0xee/0x110
[ 1596.437346]  [<ffffffff81665672>] ret_from_fork+0x22/0x50
[ 1596.441325]  [<ffffffff81091880>] ? kthread_create_on_node+0x230/0x230
(...snipped...)
[ 1599.581883] kworker/0:2     D ffff880036743878     0  3476      2 0x00000080
[ 1599.587099] Workqueue: events_freezable_power_ disk_events_workfn
[ 1599.591615]  ffff880036743878 ffffffff81c0d540 ffff880039c02040 ffff880036744000
[ 1599.597112]  ffff8800367438b0 ffff88003d610300 000000010013d1a9 ffff88003ffdf100
[ 1599.602613]  ffff880036743890 ffffffff81660570 ffff88003d610300 ffff880036743938
[ 1599.608068] Call Trace:
[ 1599.610155]  [<ffffffff81660570>] schedule+0x30/0x80
[ 1599.613996]  [<ffffffff816645f7>] schedule_timeout+0x117/0x1c0
[ 1599.618285]  [<ffffffff810bc5c6>] ? mark_held_locks+0x66/0x90
[ 1599.622537]  [<ffffffff810df270>] ? init_timer_key+0x40/0x40
[ 1599.626721]  [<ffffffff810e60f7>] ? ktime_get+0xa7/0x130
[ 1599.630666]  [<ffffffff8165fab1>] io_schedule_timeout+0xa1/0x110
[ 1599.635108]  [<ffffffff8116104d>] congestion_wait+0x7d/0xd0
[ 1599.639234]  [<ffffffff810b6620>] ? wait_woken+0x80/0x80
[ 1599.643156]  [<ffffffff8114a982>] __alloc_pages_nodemask+0xb42/0xd50
[ 1599.647774]  [<ffffffff810bc500>] ? mark_lock+0x620/0x680
[ 1599.651785]  [<ffffffff81193876>] alloc_pages_current+0x96/0x1b0
[ 1599.656235]  [<ffffffff812e108d>] ? bio_alloc_bioset+0x20d/0x2d0
[ 1599.660662]  [<ffffffff812e2454>] bio_copy_kern+0xc4/0x180
[ 1599.664702]  [<ffffffff812ed070>] blk_rq_map_kern+0x70/0x130
[ 1599.668864]  [<ffffffff8144c4bd>] scsi_execute+0x12d/0x160
[ 1599.672950]  [<ffffffff8144c5e4>] scsi_execute_req_flags+0x84/0xf0
[ 1599.677784]  [<ffffffffa01e8762>] sr_check_events+0xb2/0x2a0 [sr_mod]
[ 1599.682744]  [<ffffffffa01ce163>] cdrom_check_events+0x13/0x30 [cdrom]
[ 1599.687747]  [<ffffffffa01e8ba5>] sr_block_check_events+0x25/0x30 [sr_mod]
[ 1599.692752]  [<ffffffff812f874b>] disk_check_events+0x5b/0x150
[ 1599.697130]  [<ffffffff812f8857>] disk_events_workfn+0x17/0x20
[ 1599.701783]  [<ffffffff8108b2c5>] process_one_work+0x1a5/0x400
[ 1599.706347]  [<ffffffff8108b261>] ? process_one_work+0x141/0x400
[ 1599.710809]  [<ffffffff8108b646>] worker_thread+0x126/0x490
[ 1599.715005]  [<ffffffff8108b520>] ? process_one_work+0x400/0x400
[ 1599.719427]  [<ffffffff8109196e>] kthread+0xee/0x110
[ 1599.723220]  [<ffffffff81665672>] ret_from_fork+0x22/0x50
[ 1599.727240]  [<ffffffff81091880>] ? kthread_create_on_node+0x230/0x230
(...snipped...)
[ 1698.163933] 1 lock held by kswapd0/53:
[ 1698.166948]  #0:  (&xfs_nondir_ilock_class){++++--}, at: [<ffffffffa024bcbb>] xfs_ilock+0x4b/0xe0 [xfs]
[ 1698.174361] 5 locks held by kworker/u128:1/70:
[ 1698.177849]  #0:  ("writeback"){.+.+.+}, at: [<ffffffff8108b261>] process_one_work+0x141/0x400
[ 1698.184626]  #1:  ((&(&wb->dwork)->work)){+.+.+.}, at: [<ffffffff8108b261>] process_one_work+0x141/0x400
[ 1698.191670]  #2:  (&type->s_umount_key#30){++++++}, at: [<ffffffff811c35d6>] trylock_super+0x16/0x50
[ 1698.198449]  #3:  (sb_internal){.+.+.?}, at: [<ffffffff811c35ac>] __sb_start_write+0xcc/0xe0
[ 1698.204743]  #4:  (&xfs_nondir_ilock_class){++++--}, at: [<ffffffffa024bcef>] xfs_ilock+0x7f/0xe0 [xfs]
(...snipped...)
[ 1698.222061] 2 locks held by kworker/0:2/3476:
[ 1698.225546]  #0:  ("events_freezable_power_efficient"){.+.+.+}, at: [<ffffffff8108b261>] process_one_work+0x141/0x400
[ 1698.233350]  #1:  ((&(&ev->dwork)->work)){+.+.+.}, at: [<ffffffff8108b261>] process_one_work+0x141/0x400
(...snipped...)
[ 1718.427909] Showing busy workqueues and worker pools:
[ 1718.432224] workqueue events: flags=0x0
[ 1718.435754]   pwq 4: cpus=2 node=0 flags=0x0 nice=0 active=3/256
[ 1718.440769]     in-flight: 52:mptspi_dv_renegotiate_work [mptspi]
[ 1718.445766]     pending: vmpressure_work_fn, cache_reap
[ 1718.450227] workqueue events_power_efficient: flags=0x80
[ 1718.454645]   pwq 4: cpus=2 node=0 flags=0x0 nice=0 active=1/256
[ 1718.459663]     pending: fb_flashcursor
[ 1718.463133] workqueue events_freezable_power_: flags=0x84
[ 1718.467620]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[ 1718.472552]     in-flight: 3476:disk_events_workfn
[ 1718.476643] workqueue writeback: flags=0x4e
[ 1718.480197]   pwq 128: cpus=0-63 flags=0x4 nice=0 active=1/256
[ 1718.484977]     in-flight: 70:wb_workfn
[ 1718.488671] workqueue vmstat: flags=0xc
[ 1718.492312]   pwq 4: cpus=2 node=0 flags=0x0 nice=0 active=1/256 MAYDAY
[ 1718.497665]     pending: vmstat_update
[ 1718.501304] pool 0: cpus=0 node=0 flags=0x0 nice=0 hung=0s workers=3 idle: 3451 3501
[ 1718.507471] pool 4: cpus=2 node=0 flags=0x0 nice=0 hung=15s workers=2 manager: 3490
[ 1718.513528] pool 128: cpus=0-63 flags=0x4 nice=0 hung=0s workers=3 manager: 3471 idle: 6
[ 1745.495540] sysrq: SysRq : Show Memory
[ 1745.508581] Mem-Info:
[ 1745.516772] active_anon:182211 inactive_anon:12238 isolated_anon:0
[ 1745.516772]  active_file:6978 inactive_file:19887 isolated_file:32
[ 1745.516772]  unevictable:0 dirty:19697 writeback:214 unstable:0
[ 1745.516772]  slab_reclaimable:2382 slab_unreclaimable:8786
[ 1745.516772]  mapped:6820 shmem:12582 pagetables:1311 bounce:0
[ 1745.516772]  free:1877 free_pcp:132 free_cma:0
[ 1745.563639] Node 0 DMA free:3868kB min:60kB low:72kB high:84kB active_anon:6184kB inactive_anon:1120kB active_file:644kB inactive_file:1784kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:1784kB writeback:0kB mapped:644kB shmem:1172kB slab_reclaimable:220kB slab_unreclaimable:660kB kernel_stack:496kB pagetables:252kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:15392 all_unreclaimable? yes
[ 1745.595872] lowmem_reserve[]: 0 953 953 953
[ 1745.599508] Node 0 DMA32 free:3640kB min:3780kB low:4752kB high:5724kB active_anon:722660kB inactive_anon:47832kB active_file:27268kB inactive_file:77764kB unevictable:0kB isolated(anon):0kB isolated(file):128kB present:1032064kB managed:980852kB mlocked:0kB dirty:77004kB writeback:856kB mapped:26636kB shmem:49156kB slab_reclaimable:9308kB slab_unreclaimable:34484kB kernel_stack:19760kB pagetables:4992kB unstable:0kB bounce:0kB free_pcp:528kB local_pcp:60kB free_cma:0kB writeback_tmp:0kB pages_scanned:1387692 all_unreclaimable? yes
[ 1745.633558] lowmem_reserve[]: 0 0 0 0
[ 1745.636871] Node 0 DMA: 25*4kB (UME) 9*8kB (UME) 7*16kB (UME) 2*32kB (ME) 3*64kB (ME) 4*128kB (UE) 3*256kB (UME) 4*512kB (UE) 0*1024kB 0*2048kB 0*4096kB = 3868kB
[ 1745.648828] Node 0 DMA32: 886*4kB (UE) 8*8kB (UM) 2*16kB (U) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 3640kB
[ 1745.658179] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB
[ 1745.664712] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 1745.671127] 39477 total pagecache pages
[ 1745.674392] 0 pages in swap cache
[ 1745.677315] Swap cache stats: add 0, delete 0, find 0/0
[ 1745.681493] Free swap  = 0kB
[ 1745.684113] Total swap = 0kB
[ 1745.686786] 262013 pages RAM
[ 1745.689386] 0 pages HighMem/MovableOnly
[ 1745.692883] 12824 pages reserved
[ 1745.695779] 0 pages cma reserved
[ 1745.698763] 0 pages hwpoisoned
[ 1746.841678] BUG: workqueue lockup - pool cpus=2 node=0 flags=0x0 nice=0 stuck for 44s!
[ 1746.866634] Showing busy workqueues and worker pools:
[ 1746.881055] workqueue events: flags=0x0
[ 1746.887480]   pwq 4: cpus=2 node=0 flags=0x0 nice=0 active=3/256
[ 1746.894205]     in-flight: 52:mptspi_dv_renegotiate_work [mptspi]
[ 1746.900892]     pending: vmpressure_work_fn, cache_reap
[ 1746.906938] workqueue events_power_efficient: flags=0x80
[ 1746.912780]   pwq 4: cpus=2 node=0 flags=0x0 nice=0 active=1/256
[ 1746.917657]     pending: fb_flashcursor
[ 1746.920983] workqueue events_freezable_power_: flags=0x84
[ 1746.925304]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[ 1746.930114]     in-flight: 3476:disk_events_workfn
[ 1746.934076] workqueue writeback: flags=0x4e
[ 1746.937546]   pwq 128: cpus=0-63 flags=0x4 nice=0 active=1/256
[ 1746.942258]     in-flight: 70:wb_workfn
[ 1746.945978] pool 0: cpus=0 node=0 flags=0x0 nice=0 hung=0s workers=3 idle: 3451 3501
[ 1746.952268] pool 4: cpus=2 node=0 flags=0x0 nice=0 hung=44s workers=2 manager: 3490
[ 1746.958276] pool 128: cpus=0-63 flags=0x4 nice=0 hung=0s workers=3 manager: 3471 idle: 6

---------- console log ----------

This is an OOM-livelocked situation where kswapd got stuck and allocating tasks are sleeping at

	/*
	 * If we didn't make any progress and have a lot of
	 * dirty + writeback pages then we should wait for
	 * an IO to complete to slow down the reclaim and
	 * prevent from pre mature OOM
	 */
	if (!did_some_progress && 2*(writeback + dirty) > reclaimable) {
		congestion_wait(BLK_RW_ASYNC, HZ/10);
		return true;
	}

in should_reclaim_retry(). Presumably out_of_memory() is called (I didn't confirm it using kmallocwd), and this is a situation where "we need to select next OOM-victim" or "fail !__GFP_FS && !__GFP_NOFAIL allocation requests".

But what I felt strange is what should_reclaim_retry() is doing.

Quoted 16 lines · Michal Hocko
mm/page_alloc.c +6 -2
index f77e283fb8c6..b2de8c8761ad 100644
@@ -3044,8 +3045,37 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		 */
 		if (__zone_watermark_ok(zone, order, min_wmark_pages(zone),
 				ac->high_zoneidx, alloc_flags, available)) {
-			/* Wait for some write requests to complete then retry */
-			wait_iff_congested(zone, BLK_RW_ASYNC, HZ/50);
+			unsigned long writeback;
+			unsigned long dirty;
+
+			writeback = zone_page_state_snapshot(zone, NR_WRITEBACK);
+			dirty = zone_page_state_snapshot(zone, NR_FILE_DIRTY);
+

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3063

Quoted 10 lines
mm/page_alloc.c +10 -0
+			/*
+			 * If we didn't make any progress and have a lot of
+			 * dirty + writeback pages then we should wait for
+			 * an IO to complete to slow down the reclaim and
+			 * prevent from pre mature OOM
+			 */
+			if (!did_some_progress && 2*(writeback + dirty) > reclaimable) {
+				congestion_wait(BLK_RW_ASYNC, HZ/10);
+				return true;
+			}

writeback and dirty are used only when did_some_progress == 0. Thus, we don't need to calculate writeback and dirty using zone_page_state_snapshot() unless did_some_progress == 0.

But, does it make sense to take writeback and dirty into account when disk_events_workfn (trace shown above) is doing GFP_NOIO allocation and wb_workfn (trace shown above) is doing (presumably) GFP_NOFS allocation? Shouldn't we use different threshold for GFP_NOIO / GFP_NOFS / GFP_KERNEL?

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3075

Quoted 12 lines
mm/page_alloc.c +12 -0
+
+			/*
+			 * Memory allocation/reclaim might be called from a WQ
+			 * context and the current implementation of the WQ
+			 * concurrency control doesn't recognize that
+			 * a particular WQ is congested if the worker thread is
+			 * looping without ever sleeping. Therefore we have to
+			 * do a short sleep here rather than calling
+			 * cond_resched().
+			 */
+			if (current->flags & PF_WQ_WORKER)
+				schedule_timeout(1);

This schedule_timeout(1) does not sleep. You lost the fix as of next-20160317. Please update.

Quoted 7 lines
mm/page_alloc.c +3 -0
+			else
+				cond_resched();
+
 			return true;
 		}
 	}
author reviewer
#11 lore raw

Re: [PATCH 2/3] mm: throttle on IO only when there are too many dirty and writeback pages

On Thu 17-03-16 20:35:23, Tetsuo Handa wrote:
[...]

Quoted 25 lines

But what I felt strange is what should_reclaim_retry() is doing.

Quoted 22 lines · Michal Hocko
mm/page_alloc.c +12 -2
index f77e283fb8c6..b2de8c8761ad 100644
@@ -3044,8 +3045,37 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		 */
 		if (__zone_watermark_ok(zone, order, min_wmark_pages(zone),
 				ac->high_zoneidx, alloc_flags, available)) {
-			/* Wait for some write requests to complete then retry */
-			wait_iff_congested(zone, BLK_RW_ASYNC, HZ/50);
+			unsigned long writeback;
+			unsigned long dirty;
+
+			writeback = zone_page_state_snapshot(zone, NR_WRITEBACK);
+			dirty = zone_page_state_snapshot(zone, NR_FILE_DIRTY);
+
+			/*
+			 * If we didn't make any progress and have a lot of
+			 * dirty + writeback pages then we should wait for
+			 * an IO to complete to slow down the reclaim and
+			 * prevent from pre mature OOM
+			 */
Quoted 8 lines

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3063

Quoted 4 lines
mm/page_alloc.c +4 -0
+			if (!did_some_progress && 2*(writeback + dirty) > reclaimable) {
+				congestion_wait(BLK_RW_ASYNC, HZ/10);
+				return true;
+			}

writeback and dirty are used only when did_some_progress == 0. Thus, we don't need to calculate writeback and dirty using zone_page_state_snapshot() unless did_some_progress == 0.

OK, I will move this into if !did_some_progress.

Quoted 4 lines

But, does it make sense to take writeback and dirty into account when disk_events_workfn (trace shown above) is doing GFP_NOIO allocation and wb_workfn (trace shown above) is doing (presumably) GFP_NOFS allocation? Shouldn't we use different threshold for GFP_NOIO / GFP_NOFS / GFP_KERNEL?

I have considered skiping the throttling part for GFP_NOFS/GFP_NOIO previously but I couldn't have convinced myself it would make any difference. We know there was no progress in the reclaim and even if the current context is doing FS/IO allocation potentially then it obviously cannot get its memory so it cannot proceed. So now we are in the state where we either busy loop or sleep for a while. So I ended up not complicating the code even more. If you have a use case where busy waiting makes a difference then I would vote for a separate patch with a clear description.

Quoted 7 lines
Quoted 7 lines
mm/page_alloc.c +7 -0
+
+			/*
+			 * Memory allocation/reclaim might be called from a WQ
+			 * context and the current implementation of the WQ
+			 * concurrency control doesn't recognize that
+			 * a particular WQ is congested if the worker thread is
+			 * looping without ever sleeping. Therefore we have to
Quoted 8 lines

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3075

Quoted 5 lines
mm/page_alloc.c +5 -0
+			 * do a short sleep here rather than calling
+			 * cond_resched().
+			 */
+			if (current->flags & PF_WQ_WORKER)
+				schedule_timeout(1);

This schedule_timeout(1) does not sleep. You lost the fix as of next-20160317. Please update.

Yeah, I have that updated in my local patch already.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#12 lore raw

PATCH 3/3 mm: use watermak checks for __GFP_REPEAT high order allocations

In reply to Michal Hocko

From: Michal Hocko <[email protected]>

__alloc_pages_slowpath retries costly allocations until at least order worth of pages were reclaimed or the watermark check for at least one zone would succeed after all reclaiming all pages if the reclaim hasn't made any progress.

The first condition was added by a41f24ea9fd6 ("page allocator: smarter retry of costly-order allocations) and it assumed that lumpy reclaim could have created a page of the sufficient order. Lumpy reclaim, has been removed quite some time ago so the assumption doesn't hold anymore. It would be more appropriate to check the compaction progress instead but this patch simply removes the check and relies solely on the watermark check.

To prevent from too many retries the no_progress_loops is not reseted after a reclaim which made progress because we cannot assume it helped high order situation. Only costly allocation requests depended on pages_reclaimed so we can drop it.

---
 mm/page_alloc.c | 34 +++++++++++++++-------------------
 1 file changed, 15 insertions(+), 19 deletions(-)
mm/page_alloc.c +15 -19
index b2de8c8761ad..268de1654128 100644
@@ -2994,17 +2994,17 @@ static inline bool is_thp_gfp_mask(gfp_t gfp_mask)
  * Checks whether it makes sense to retry the reclaim to make a forward progress
  * for the given allocation request.
  * The reclaim feedback represented by did_some_progress (any progress during
- * the last reclaim round), pages_reclaimed (cumulative number of reclaimed
- * pages) and no_progress_loops (number of reclaim rounds without any progress
- * in a row) is considered as well as the reclaimable pages on the applicable
- * zone list (with a backoff mechanism which is a function of no_progress_loops).
+ * the last reclaim round) and no_progress_loops (number of reclaim rounds without
+ * any progress in a row) is considered as well as the reclaimable pages on the
+ * applicable zone list (with a backoff mechanism which is a function of
+ * no_progress_loops).
  *
  * Returns true if a retry is viable or false to enter the oom path.
  */
 static inline bool
 should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		     struct alloc_context *ac, int alloc_flags,
-		     bool did_some_progress, unsigned long pages_reclaimed,
+		     bool did_some_progress,
 		     int no_progress_loops)
 {
 	struct zone *zone;
@@ -3018,13 +3018,8 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		return false;
 
 	/* Do not retry high order allocations unless they are __GFP_REPEAT */
-	if (order > PAGE_ALLOC_COSTLY_ORDER) {
-		if (!(gfp_mask & __GFP_REPEAT) || pages_reclaimed >= (1<<order))
-			return false;
-
-		if (did_some_progress)
-			return true;
-	}
+	if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
+		return false;
 
 	/*
 	 * Keep reclaiming pages while there is a chance this will lead somewhere.
@@ -3090,7 +3085,6 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	bool can_direct_reclaim = gfp_mask & __GFP_DIRECT_RECLAIM;
 	struct page *page = NULL;
 	int alloc_flags;
-	unsigned long pages_reclaimed = 0;
 	unsigned long did_some_progress;
 	enum migrate_mode migration_mode = MIGRATE_ASYNC;
 	bool deferred_compaction = false;
@@ -3255,16 +3249,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	if (gfp_mask & __GFP_NORETRY)
 		goto noretry;
 
-	if (did_some_progress) {
+	/*
+	 * Costly allocations might have made a progress but this doesn't mean
+	 * their order will become available due to high fragmentation so do
+	 * not reset the no progress counter for them
+	 */
+	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
 		no_progress_loops = 0;
-		pages_reclaimed += did_some_progress;
-	} else {
+	else
 		no_progress_loops++;
-	}
 
 	if (should_reclaim_retry(gfp_mask, order, ac, alloc_flags,
-				 did_some_progress > 0, pages_reclaimed,
-				 no_progress_loops))
+				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
 	/* Reclaim has failed us, start killing things */
Signature · 7 lines
2.6.2

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
maintainer
#13 lore raw
In reply to Michal Hocko
Quoted 5 lines · On Tue, 15 Dec 2015 19:19:43 +0100 Michal Hocko <[email protected]> wrote:

This is an attempt to make the OOM detection more deterministic and easier to follow because each reclaimer basically tracks its own progress which is implemented at the page allocator layer rather spread out between the allocator and the reclaim. The more on the implementation is described in the first patch.

We've been futzing with this stuff for many years and it still isn't working well. This makes me expect that the new implementation will take a long time to settle in.

To aid and accelerate this process I suggest we lard this code up with lots of debug info, so when someone reports an issue we have the best possible chance of understanding what went wrong.

This is easy in the case of oom-too-early - it's all slowpath code and we can just do printk(everything). It's not so easy in the case of oom-too-late-or-never. The reporter's machine just hangs or it twiddles thumbs for five minutes then goes oom. But there are things we can do here as well, such as:

- add an automatic "nearly oom" detection which detects when things start going wrong and turns on diagnostics (this would need an enable knob, possibly in debugfs).

- forget about an autodetector and simply add a debugfs knob to turn on the diagnostics.

- sprinkle tracepoints everywhere and provide a set of instructions/scripts so that people who know nothing about kernel internals or tracing can easily gather the info we need to understand issues.

- add a sysrq key to turn on diagnostics. Pretty essential when the machine is comatose and doesn't respond to keystrokes.

- something else

So... please have a think about it? What can we add in here to make it as easy as possible for us (ie: you ;)) to get this code working well? At this time, too much developer support code will be better than too little. We can take it out later on.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#14 lore raw

On Wed 16-12-15 15:35:13, Andrew Morton wrote:
[...]

Quoted 4 lines

So... please have a think about it? What can we add in here to make it as easy as possible for us (ie: you ;)) to get this code working well? At this time, too much developer support code will be better than too little. We can take it out later on.

Sure. I will think about this and get back to it early next year. I will be mostly offline starting next week.

Thanks for looking into this!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
maintainer
#15 lore raw
In reply to Michal Hocko
Quoted 9 lines · On Tue, 15 Dec 2015 19:19:43 +0100 Michal Hocko <[email protected]> wrote:

...

* base kernel

$ grep "Killed process" base-oom-run1.log | tail -n1
[  211.824379] Killed process 3086 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:332kB, shmem-rss:0kB
$ grep "Killed process" base-oom-run2.log | tail -n1
[  157.188326] Killed process 3094 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:368kB, shmem-rss:0kB
Quoted 7 lines
$ grep "invoked oom-killer" base-oom-run1.log | wc -l
78
$ grep "invoked oom-killer" base-oom-run2.log | wc -l
76

The number of OOM invocations is consistent with my last measurements but the runtime is way too different (it took 800+s).

I'm seeing 211 seconds vs 157 seconds? If so, that's not toooo bad. I assume the 800+s is sum-across-multiple-CPUs? Given that all the CPUs are pounding away at the same data and the same disk, that doesn't sound like very interesting info - the overall elapsed time is the thing to look at in this case.

Quoted 16 lines

One thing that
could have skewed results was that I was tail -f the serial log on the host system to see the progress. I have stopped doing that. The results are more consistent now but still too different from the last time. This is really weird so I've retested with the last 4.2 mmotm again and I am getting consistent ~220s which is really close to the above. If I apply the WQ vmstat patch on top I am getting close to 160s so the stale vmstat counters made a difference which is to be expected. I have a new SSD in my laptop which migh have made a difference but I wouldn't expect it to be that large.

$ grep "DMA32.*all_unreclaimable? no" base-oom-run1.log | wc -l
4
$ grep "DMA32.*all_unreclaimable? no" base-oom-run2.log | wc -l
1
Quoted 5 lines

* patched kernel

$ grep "Killed process" patched-oom-run1.log | tail -n1
[  341.164930] Killed process 3099 (mem_eater) total-vm:85852kB, anon-rss:82000kB, file-rss:336kB, shmem-rss:0kB
$ grep "Killed process" patched-oom-run2.log | tail -n1
[  349.111539] Killed process 3082 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:4kB, shmem-rss:0kB

Even better.

Quoted 10 lines
$ grep "invoked oom-killer" patched-oom-run1.log | wc -l
78
$ grep "invoked oom-killer" patched-oom-run2.log | wc -l
77
$ grep "DMA32.*all_unreclaimable? no" patched-oom-run1.log | wc -l
1
$ grep "DMA32.*all_unreclaimable? no" patched-oom-run2.log | wc -l
0
Quoted 4 lines

So the number of OOM killer invocation is the same but the overall runtime of the test was much longer with the patched kernel. This can be attributed to more retries in general. The results from the base kernel are quite inconsitent and I think that consistency is better here.

It's hard to say how long declaration of oom should take. Correctness comes first. But what is "correct"? oom isn't a binary condition - there's a chance that if we keep churning away for another 5 minutes we'll be able to satisfy this allocation (but probably not the next one). There are tradeoffs between promptness-of-declaring-oom and exhaustiveness-in-avoiding-it.

Quoted 4 lines

2) 2 writers again with 10s of run and then 10 mem_eaters to consume as much memory as possible without triggering the OOM killer. This required a lot of tuning but I've considered 3 consecutive runs without OOM as a success.

"a lot of tuning" sounds bad. It means that the tuning settings you have now for a particular workload on a particular machine will be wrong for other workloads and machines. uh-oh.

Quoted 1 line

...

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#16 lore raw
Quoted 16 lines · Andrew Morton
Quoted 14 lines · On Tue, 15 Dec 2015 19:19:43 +0100 Michal Hocko <[email protected]> wrote:

...

* base kernel

$ grep "Killed process" base-oom-run1.log | tail -n1
[  211.824379] Killed process 3086 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:332kB, shmem-rss:0kB
$ grep "Killed process" base-oom-run2.log | tail -n1
[  157.188326] Killed process 3094 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:368kB, shmem-rss:0kB
$ grep "invoked oom-killer" base-oom-run1.log | wc -l
78
$ grep "invoked oom-killer" base-oom-run2.log | wc -l
76
Quoted 5 lines
Quoted 2 lines

The number of OOM invocations is consistent with my last measurements but the runtime is way too different (it took 800+s).

I'm seeing 211 seconds vs 157 seconds? If so, that's not toooo bad. I assume the 800+s is sum-across-multiple-CPUs?

This is the time until the oom situation settled down. And I really suspect that the new SSD made a difference here.

Quoted 4 lines

Given that all the CPUs
are pounding away at the same data and the same disk, that doesn't sound like very interesting info - the overall elapsed time is the thing to look at in this case.

Which is what I was looking at when checking the timestamp in the log.

[...]

Quoted 23 lines
Quoted 5 lines

* patched kernel

$ grep "Killed process" patched-oom-run1.log | tail -n1
[  341.164930] Killed process 3099 (mem_eater) total-vm:85852kB, anon-rss:82000kB, file-rss:336kB, shmem-rss:0kB
$ grep "Killed process" patched-oom-run2.log | tail -n1
[  349.111539] Killed process 3082 (mem_eater) total-vm:85852kB, anon-rss:81996kB, file-rss:4kB, shmem-rss:0kB

Even better.

Quoted 10 lines
$ grep "invoked oom-killer" patched-oom-run1.log | wc -l
78
$ grep "invoked oom-killer" patched-oom-run2.log | wc -l
77
$ grep "DMA32.*all_unreclaimable? no" patched-oom-run1.log | wc -l
1
$ grep "DMA32.*all_unreclaimable? no" patched-oom-run2.log | wc -l
0
Quoted 4 lines

So the number of OOM killer invocation is the same but the overall runtime of the test was much longer with the patched kernel. This can be attributed to more retries in general. The results from the base kernel are quite inconsitent and I think that consistency is better here.

Quoted 6 lines

It's hard to say how long declaration of oom should take. Correctness comes first. But what is "correct"? oom isn't a binary condition - there's a chance that if we keep churning away for another 5 minutes we'll be able to satisfy this allocation (but probably not the next one). There are tradeoffs between promptness-of-declaring-oom and exhaustiveness-in-avoiding-it.

Yes, this is really hard to tell. What I wanted to achieve here is a determinism - the same load should give comparable results. It seems that there is an improvement in this regards. The time to settle is much more consistent than with the original implementation.

Quoted 7 lines
Quoted 3 lines

2) 2 writers again with 10s of run and then 10 mem_eaters to consume as much memory as possible without triggering the OOM killer. This required a lot of tuning but I've considered 3 consecutive runs without OOM as a success.

"a lot of tuning" sounds bad. It means that the tuning settings you have now for a particular workload on a particular machine will be wrong for other workloads and machines. uh-oh.

Well, I had to tune the test to see how close to the edge I can get. I haven't done any decisions based on this test.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
maintainer
#17 lore raw
Quoted 8 lines · Michal Hocko
Quoted 6 lines · Andrew Morton

It's hard to say how long declaration of oom should take. Correctness comes first. But what is "correct"? oom isn't a binary condition - there's a chance that if we keep churning away for another 5 minutes we'll be able to satisfy this allocation (but probably not the next one). There are tradeoffs between promptness-of-declaring-oom and exhaustiveness-in-avoiding-it.

Quoted 4 lines

Yes, this is really hard to tell. What I wanted to achieve here is a determinism - the same load should give comparable results. It seems that there is an improvement in this regards. The time to settle is much more consistent than with the original implementation.

+1

Before that we couldn't even really make a meaningful statement about how long we are going to try - "as long as reclaim thinks it can maybe do some more, depending on heuristics". I think the best thing we can strive for with OOM is to make the rules simple and predictable.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#18 lore raw
In reply to Michal Hocko

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

[ 3902.430630] kthreadd invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[ 3902.432780] kthreadd cpuset=/ mems_allowed=0
[ 3902.433904] CPU: 3 PID: 2 Comm: kthreadd Not tainted 4.4.0-rc6-next-20151222 #255
[ 3902.435463] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[ 3902.437541]  0000000000000000 000000009cc7eb67 ffff88007cc1faa0 ffffffff81395bc3
[ 3902.439129]  0000000000000000 ffff88007cc1fb40 ffffffff811babac 0000000000000206
[ 3902.440779]  ffffffff81810470 ffff88007cc1fae0 ffffffff810bce29 0000000000000206
[ 3902.442436] Call Trace:
[ 3902.443094]  [<ffffffff81395bc3>] dump_stack+0x4b/0x68
[ 3902.444188]  [<ffffffff811babac>] dump_header+0x5b/0x3b0
[ 3902.445301]  [<ffffffff810bce29>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[ 3902.446656]  [<ffffffff810bcefd>] ? trace_hardirqs_on+0xd/0x10
[ 3902.447881]  [<ffffffff81142646>] oom_kill_process+0x366/0x540
[ 3902.449093]  [<ffffffff81142a5f>] out_of_memory+0x1ef/0x5a0
[ 3902.450266]  [<ffffffff81142b1d>] ? out_of_memory+0x2ad/0x5a0
[ 3902.451430]  [<ffffffff8114836d>] __alloc_pages_nodemask+0xb9d/0xd90
[ 3902.452757]  [<ffffffff810bce00>] ? trace_hardirqs_on_caller+0xd0/0x1c0
[ 3902.454468]  [<ffffffff8114871c>] alloc_kmem_pages_node+0x4c/0xc0
[ 3902.455756]  [<ffffffff8106c451>] copy_process.part.31+0x131/0x1b40
[ 3902.457076]  [<ffffffff8108f590>] ? kthread_create_on_node+0x230/0x230
[ 3902.458396]  [<ffffffff8106e02b>] _do_fork+0xdb/0x5d0
[ 3902.459480]  [<ffffffff81094a8a>] ? finish_task_switch+0x6a/0x2b0
[ 3902.460775]  [<ffffffff8106e544>] kernel_thread+0x24/0x30
[ 3902.461894]  [<ffffffff8109007c>] kthreadd+0x1bc/0x220
[ 3902.463035]  [<ffffffff816fc89f>] ? ret_from_fork+0x3f/0x70
[ 3902.464230]  [<ffffffff8108fec0>] ? kthread_create_on_cpu+0x60/0x60
[ 3902.465502]  [<ffffffff816fc89f>] ret_from_fork+0x3f/0x70
[ 3902.466648]  [<ffffffff8108fec0>] ? kthread_create_on_cpu+0x60/0x60
[ 3902.467953] Mem-Info:
[ 3902.468537] active_anon:20817 inactive_anon:2098 isolated_anon:0
[ 3902.468537]  active_file:145434 inactive_file:145453 isolated_file:0
[ 3902.468537]  unevictable:0 dirty:20613 writeback:7248 unstable:0
[ 3902.468537]  slab_reclaimable:86363 slab_unreclaimable:14905
[ 3902.468537]  mapped:6670 shmem:2167 pagetables:1497 bounce:0
[ 3902.468537]  free:5422 free_pcp:75 free_cma:0
[ 3902.476541] Node 0 DMA free:6904kB min:44kB low:52kB high:64kB active_anon:3268kB inactive_anon:200kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:36kB shmem:216kB slab_reclaimable:3708kB slab_unreclaimable:456kB kernel_stack:48kB pagetables:160kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 3902.486494] lowmem_reserve[]: 0 1714 1714 1714
[ 3902.487659] Node 0 DMA32 free:13760kB min:5172kB low:6464kB high:7756kB active_anon:80000kB inactive_anon:8192kB active_file:581780kB inactive_file:581848kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:1758960kB mlocked:0kB dirty:82312kB writeback:29588kB mapped:26648kB shmem:8452kB slab_reclaimable:341744kB slab_unreclaimable:59496kB kernel_stack:3456kB pagetables:5828kB unstable:0kB bounce:0kB free_pcp:732kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:560 all_unreclaimable? no
[ 3902.500438] lowmem_reserve[]: 0 0 0 0
[ 3902.502373] Node 0 DMA: 42*4kB (UME) 84*8kB (UM) 57*16kB (UM) 15*32kB (UM) 11*64kB (M) 9*128kB (UME) 1*256kB (M) 1*512kB (M) 2*1024kB (UM) 0*2048kB 0*4096kB = 6904kB
[ 3902.507561] Node 0 DMA32: 3788*4kB (UME) 184*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 16624kB
[ 3902.511236] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 3902.513938] 292144 total pagecache pages
[ 3902.515609] 0 pages in swap cache
[ 3902.517139] Swap cache stats: add 0, delete 0, find 0/0
[ 3902.519153] Free swap  = 0kB
[ 3902.520587] Total swap = 0kB
[ 3902.522095] 524157 pages RAM
[ 3902.523511] 0 pages HighMem/MovableOnly
[ 3902.525091] 80441 pages reserved
[ 3902.526580] 0 pages hwpoisoned
[ 3902.528169] Out of memory: Kill process 687 (firewalld) score 11 or sacrifice child
[ 3902.531017] Killed process 687 (firewalld) total-vm:323600kB, anon-rss:17032kB, file-rss:4896kB, shmem-rss:0kB
[ 5262.901161] smbd invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[ 5262.903629] smbd cpuset=/ mems_allowed=0
[ 5262.904725] CPU: 2 PID: 3935 Comm: smbd Not tainted 4.4.0-rc6-next-20151222 #255
[ 5262.906401] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[ 5262.908679]  0000000000000000 00000000eaa24b41 ffff88007c37faf8 ffffffff81395bc3
[ 5262.910459]  0000000000000000 ffff88007c37fb98 ffffffff811babac 0000000000000206
[ 5262.912224]  ffffffff81810470 ffff88007c37fb38 ffffffff810bce29 0000000000000206
[ 5262.914019] Call Trace:
[ 5262.914839]  [<ffffffff81395bc3>] dump_stack+0x4b/0x68
[ 5262.916118]  [<ffffffff811babac>] dump_header+0x5b/0x3b0
[ 5262.917493]  [<ffffffff810bce29>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[ 5262.919131]  [<ffffffff810bcefd>] ? trace_hardirqs_on+0xd/0x10
[ 5262.920690]  [<ffffffff81142646>] oom_kill_process+0x366/0x540
[ 5262.922204]  [<ffffffff81142a5f>] out_of_memory+0x1ef/0x5a0
[ 5262.923863]  [<ffffffff81142b1d>] ? out_of_memory+0x2ad/0x5a0
[ 5262.925386]  [<ffffffff8114836d>] __alloc_pages_nodemask+0xb9d/0xd90
[ 5262.927121]  [<ffffffff8114871c>] alloc_kmem_pages_node+0x4c/0xc0
[ 5262.928738]  [<ffffffff8106c451>] copy_process.part.31+0x131/0x1b40
[ 5262.930438]  [<ffffffff8111c4da>] ? __audit_syscall_entry+0xaa/0xf0
[ 5262.932110]  [<ffffffff8106e02b>] _do_fork+0xdb/0x5d0
[ 5262.933410]  [<ffffffff8111c4da>] ? __audit_syscall_entry+0xaa/0xf0
[ 5262.935016]  [<ffffffff810030c1>] ? do_audit_syscall_entry+0x61/0x70
[ 5262.936632]  [<ffffffff81003254>] ? syscall_trace_enter_phase1+0x134/0x150
[ 5262.938383]  [<ffffffff81003017>] ? trace_hardirqs_on_thunk+0x17/0x19
[ 5262.940024]  [<ffffffff8106e5a4>] SyS_clone+0x14/0x20
[ 5262.941465]  [<ffffffff816fc532>] entry_SYSCALL_64_fastpath+0x12/0x76
[ 5262.943137] Mem-Info:
[ 5262.944068] active_anon:37901 inactive_anon:2095 isolated_anon:0
[ 5262.944068]  active_file:134812 inactive_file:135474 isolated_file:0
[ 5262.944068]  unevictable:0 dirty:257 writeback:0 unstable:0
[ 5262.944068]  slab_reclaimable:90770 slab_unreclaimable:12759
[ 5262.944068]  mapped:4223 shmem:2166 pagetables:1428 bounce:0
[ 5262.944068]  free:3738 free_pcp:49 free_cma:0
[ 5262.953176] Node 0 DMA free:6904kB min:44kB low:52kB high:64kB active_anon:900kB inactive_anon:200kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:32kB shmem:216kB slab_reclaimable:5556kB slab_unreclaimable:712kB kernel_stack:48kB pagetables:152kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 5262.963749] lowmem_reserve[]: 0 1714 1714 1714
[ 5262.965434] Node 0 DMA32 free:8048kB min:5172kB low:6464kB high:7756kB active_anon:150704kB inactive_anon:8180kB active_file:539244kB inactive_file:541892kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:1758960kB mlocked:0kB dirty:1028kB writeback:0kB mapped:16860kB shmem:8448kB slab_reclaimable:357524kB slab_unreclaimable:50324kB kernel_stack:3232kB pagetables:5560kB unstable:0kB bounce:0kB free_pcp:184kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:132 all_unreclaimable? no
[ 5262.976879] lowmem_reserve[]: 0 0 0 0
[ 5262.978586] Node 0 DMA: 58*4kB (UME) 60*8kB (UME) 73*16kB (UME) 23*32kB (UME) 13*64kB (UME) 5*128kB (UM) 5*256kB (UME) 3*512kB (UE) 0*1024kB 0*2048kB 0*4096kB = 6904kB
[ 5262.983496] Node 0 DMA32: 1987*4kB (UME) 14*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 8060kB
[ 5262.987124] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 5262.989532] 272459 total pagecache pages
[ 5262.991203] 0 pages in swap cache
[ 5262.992583] Swap cache stats: add 0, delete 0, find 0/0
[ 5262.994334] Free swap  = 0kB
[ 5262.995787] Total swap = 0kB
[ 5262.997038] 524157 pages RAM
[ 5262.998270] 0 pages HighMem/MovableOnly
[ 5262.999683] 80441 pages reserved
[ 5263.001153] 0 pages hwpoisoned
[ 5263.002612] Out of memory: Kill process 26226 (genxref) score 54 or sacrifice child
[ 5263.004648] Killed process 26226 (genxref) total-vm:130348kB, anon-rss:94680kB, file-rss:4756kB, shmem-rss:0kB
[ 5269.764580] kthreadd invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[ 5269.767289] kthreadd cpuset=/ mems_allowed=0
[ 5269.768904] CPU: 2 PID: 2 Comm: kthreadd Not tainted 4.4.0-rc6-next-20151222 #255
[ 5269.770956] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[ 5269.773754]  0000000000000000 000000009cc7eb67 ffff88007cc1faa0 ffffffff81395bc3
[ 5269.776088]  0000000000000000 ffff88007cc1fb40 ffffffff811babac 0000000000000206
[ 5269.778213]  ffffffff81810470 ffff88007cc1fae0 ffffffff810bce29 0000000000000206
[ 5269.780497] Call Trace:
[ 5269.781796]  [<ffffffff81395bc3>] dump_stack+0x4b/0x68
[ 5269.783634]  [<ffffffff811babac>] dump_header+0x5b/0x3b0
[ 5269.786116]  [<ffffffff810bce29>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[ 5269.788495]  [<ffffffff810bcefd>] ? trace_hardirqs_on+0xd/0x10
[ 5269.790538]  [<ffffffff81142646>] oom_kill_process+0x366/0x540
[ 5269.792755]  [<ffffffff81142a5f>] out_of_memory+0x1ef/0x5a0
[ 5269.794784]  [<ffffffff81142b1d>] ? out_of_memory+0x2ad/0x5a0
[ 5269.796848]  [<ffffffff8114836d>] __alloc_pages_nodemask+0xb9d/0xd90
[ 5269.799038]  [<ffffffff810bce00>] ? trace_hardirqs_on_caller+0xd0/0x1c0
[ 5269.801073]  [<ffffffff8114871c>] alloc_kmem_pages_node+0x4c/0xc0
[ 5269.803186]  [<ffffffff8106c451>] copy_process.part.31+0x131/0x1b40
[ 5269.805249]  [<ffffffff8108f590>] ? kthread_create_on_node+0x230/0x230
[ 5269.807374]  [<ffffffff8106e02b>] _do_fork+0xdb/0x5d0
[ 5269.809089]  [<ffffffff81094a8a>] ? finish_task_switch+0x6a/0x2b0
[ 5269.811146]  [<ffffffff8106e544>] kernel_thread+0x24/0x30
[ 5269.812944]  [<ffffffff8109007c>] kthreadd+0x1bc/0x220
[ 5269.814698]  [<ffffffff816fc89f>] ? ret_from_fork+0x3f/0x70
[ 5269.816330]  [<ffffffff8108fec0>] ? kthread_create_on_cpu+0x60/0x60
[ 5269.818088]  [<ffffffff816fc89f>] ret_from_fork+0x3f/0x70
[ 5269.819685]  [<ffffffff8108fec0>] ? kthread_create_on_cpu+0x60/0x60
[ 5269.821399] Mem-Info:
[ 5269.822430] active_anon:14280 inactive_anon:2095 isolated_anon:0
[ 5269.822430]  active_file:134344 inactive_file:134515 isolated_file:0
[ 5269.822430]  unevictable:0 dirty:2 writeback:0 unstable:0
[ 5269.822430]  slab_reclaimable:96214 slab_unreclaimable:22185
[ 5269.822430]  mapped:3512 shmem:2166 pagetables:1368 bounce:0
[ 5269.822430]  free:12388 free_pcp:51 free_cma:0
[ 5269.831310] Node 0 DMA free:6892kB min:44kB low:52kB high:64kB active_anon:856kB inactive_anon:200kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:32kB shmem:216kB slab_reclaimable:5556kB slab_unreclaimable:768kB kernel_stack:48kB pagetables:152kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 5269.840580] lowmem_reserve[]: 0 1714 1714 1714
[ 5269.842107] Node 0 DMA32 free:42660kB min:5172kB low:6464kB high:7756kB active_anon:56264kB inactive_anon:8180kB active_file:537372kB inactive_file:538056kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:1758960kB mlocked:0kB dirty:8kB writeback:0kB mapped:14020kB shmem:8448kB slab_reclaimable:379300kB slab_unreclaimable:87972kB kernel_stack:3232kB pagetables:5320kB unstable:0kB bounce:0kB free_pcp:204kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 5269.852375] lowmem_reserve[]: 0 0 0 0
[ 5269.853784] Node 0 DMA: 67*4kB (ME) 60*8kB (UME) 72*16kB (ME) 22*32kB (ME) 13*64kB (UME) 5*128kB (UM) 5*256kB (UME) 3*512kB (UE) 0*1024kB 0*2048kB 0*4096kB = 6892kB
[ 5269.858330] Node 0 DMA32: 10648*4kB (UME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 42592kB
[ 5269.861551] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 5269.863676] 271012 total pagecache pages
[ 5269.865100] 0 pages in swap cache
[ 5269.866366] Swap cache stats: add 0, delete 0, find 0/0
[ 5269.867996] Free swap  = 0kB
[ 5269.869363] Total swap = 0kB
[ 5269.870593] 524157 pages RAM
[ 5269.871857] 0 pages HighMem/MovableOnly
[ 5269.873604] 80441 pages reserved
[ 5269.874937] 0 pages hwpoisoned
[ 5269.876207] Out of memory: Kill process 2710 (tuned) score 7 or sacrifice child
[ 5269.878265] Killed process 2710 (tuned) total-vm:553052kB, anon-rss:10596kB, file-rss:2776kB, shmem-rss:0kB
#19 lore raw
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Well, current patch invokes OOM killers when more than 75% of memory is used for file cache (active_file: + inactive_file:). I think this is a surprising thing for administrators and we want to retry more harder (but not forever, please).

Complete log is at http://I-love.SAKURA.ne.jp/tmp/serial-20151228.txt.xz .
----------

[  277.863985] Node 0 DMA32 free:20128kB min:5564kB low:6952kB high:8344kB active_anon:108332kB inactive_anon:8252kB active_file:985160kB inactive_file:615436kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:5904kB shmem:8524kB slab_reclaimable:52088kB slab_unreclaimable:59748kB kernel_stack:31280kB pagetables:55708kB unstable:0kB bounce:0kB free_pcp:1056kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  277.884512] Node 0 DMA32: 3438*4kB (UME) 791*8kB (UME) 3*16kB (UM) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20128kB
[  291.331040] Node 0 DMA32 free:29500kB min:5564kB low:6952kB high:8344kB active_anon:126756kB inactive_anon:8252kB active_file:821500kB inactive_file:604016kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:0kB writeback:0kB mapped:12684kB shmem:8524kB slab_reclaimable:56808kB slab_unreclaimable:99804kB kernel_stack:58448kB pagetables:92552kB unstable:0kB bounce:0kB free_pcp:2004kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  291.349097] Node 0 DMA32: 4221*4kB (UME) 1971*8kB (UME) 436*16kB (UME) 141*32kB (UME) 8*64kB (UM) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 44652kB
[  302.897985] Node 0 DMA32 free:28240kB min:5564kB low:6952kB high:8344kB active_anon:79344kB inactive_anon:8248kB active_file:1016568kB inactive_file:604696kB unevictable:0kB isolated(anon):0kB isolated(file):120kB present:2080640kB managed:2021100kB mlocked:0kB dirty:80kB writeback:0kB mapped:13004kB shmem:8520kB slab_reclaimable:52076kB slab_unreclaimable:64064kB kernel_stack:35168kB pagetables:48552kB unstable:0kB bounce:0kB free_pcp:1384kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  302.916334] Node 0 DMA32: 4304*4kB (UM) 1181*8kB (UME) 59*16kB (UME) 7*32kB (ME) 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 27832kB
[  311.014501] Node 0 DMA32 free:22820kB min:5564kB low:6952kB high:8344kB active_anon:56852kB inactive_anon:11976kB active_file:1142936kB inactive_file:582040kB unevictable:0kB isolated(anon):0kB isolated(file):116kB present:2080640kB managed:2021100kB mlocked:0kB dirty:160kB writeback:0kB mapped:10796kB shmem:16640kB slab_reclaimable:48608kB slab_unreclaimable:41912kB kernel_stack:16560kB pagetables:30876kB unstable:0kB bounce:0kB free_pcp:948kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:128 all_unreclaimable? no
[  311.034251] Node 0 DMA32: 6*4kB (U) 2401*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 19232kB
[  314.293371] Node 0 DMA32 free:15244kB min:5564kB low:6952kB high:8344kB active_anon:82496kB inactive_anon:11976kB active_file:1110984kB inactive_file:467400kB unevictable:0kB isolated(anon):0kB isolated(file):88kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:9440kB shmem:16640kB slab_reclaimable:53684kB slab_unreclaimable:72536kB kernel_stack:40048kB pagetables:67672kB unstable:0kB bounce:0kB free_pcp:1076kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:12 all_unreclaimable? no
[  314.314336] Node 0 DMA32: 1180*4kB (UM) 1449*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 16312kB
[  322.774181] Node 0 DMA32 free:19780kB min:5564kB low:6952kB high:8344kB active_anon:68264kB inactive_anon:17816kB active_file:1155724kB inactive_file:470216kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:8kB writeback:0kB mapped:9744kB shmem:24708kB slab_reclaimable:52540kB slab_unreclaimable:63216kB kernel_stack:32464kB pagetables:51856kB unstable:0kB bounce:0kB free_pcp:1076kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  322.796256] Node 0 DMA32: 86*4kB (UME) 2474*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20136kB
[  330.804341] Node 0 DMA32 free:22076kB min:5564kB low:6952kB high:8344kB active_anon:47616kB inactive_anon:17816kB active_file:1063272kB inactive_file:685848kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:216kB writeback:0kB mapped:9708kB shmem:24708kB slab_reclaimable:48536kB slab_unreclaimable:36844kB kernel_stack:12048kB pagetables:25992kB unstable:0kB bounce:0kB free_pcp:776kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  330.826190] Node 0 DMA32: 1637*4kB (UM) 1354*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 17380kB
[  332.828224] Node 0 DMA32 free:15544kB min:5564kB low:6952kB high:8344kB active_anon:63184kB inactive_anon:17784kB active_file:1215752kB inactive_file:468872kB unevictable:0kB isolated(anon):0kB isolated(file):68kB present:2080640kB managed:2021100kB mlocked:0kB dirty:312kB writeback:0kB mapped:9116kB shmem:24708kB slab_reclaimable:49912kB slab_unreclaimable:50068kB kernel_stack:21600kB pagetables:42384kB unstable:0kB bounce:0kB free_pcp:1364kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  332.846805] Node 0 DMA32: 4108*4kB (UME) 897*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 23608kB
[  341.054731] Node 0 DMA32 free:20512kB min:5564kB low:6952kB high:8344kB active_anon:76796kB inactive_anon:23792kB active_file:1053836kB inactive_file:618588kB unevictable:0kB isolated(anon):0kB isolated(file):96kB present:2080640kB managed:2021100kB mlocked:0kB dirty:1656kB writeback:0kB mapped:19768kB shmem:32784kB slab_reclaimable:49000kB slab_unreclaimable:47636kB kernel_stack:21664kB pagetables:37188kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  341.073722] Node 0 DMA32: 3309*4kB (UM) 1124*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 22228kB
[  360.075472] Node 0 DMA32 free:17856kB min:5564kB low:6952kB high:8344kB active_anon:117872kB inactive_anon:25588kB active_file:1022532kB inactive_file:466856kB unevictable:0kB isolated(anon):0kB isolated(file):116kB present:2080640kB managed:2021100kB mlocked:0kB dirty:420kB writeback:0kB mapped:25300kB shmem:40976kB slab_reclaimable:57804kB slab_unreclaimable:79416kB kernel_stack:46784kB pagetables:78044kB unstable:0kB bounce:0kB free_pcp:1100kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  360.093794] Node 0 DMA32: 2719*4kB (UM) 97*8kB (UM) 14*16kB (UM) 37*32kB (UME) 27*64kB (UME) 3*128kB (UM) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15172kB
[  368.853099] Node 0 DMA32 free:22524kB min:5564kB low:6952kB high:8344kB active_anon:79156kB inactive_anon:24876kB active_file:872972kB inactive_file:738900kB unevictable:0kB isolated(anon):0kB isolated(file):96kB present:2080640kB managed:2021100kB mlocked:0kB dirty:0kB writeback:0kB mapped:25708kB shmem:40976kB slab_reclaimable:50820kB slab_unreclaimable:62880kB kernel_stack:32048kB pagetables:49656kB unstable:0kB bounce:0kB free_pcp:524kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  368.871173] Node 0 DMA32: 5042*4kB (UM) 248*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 22152kB
[  379.261759] Node 0 DMA32 free:15888kB min:5564kB low:6952kB high:8344kB active_anon:89928kB inactive_anon:23780kB active_file:1295512kB inactive_file:358284kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:1608kB writeback:0kB mapped:25376kB shmem:40976kB slab_reclaimable:47972kB slab_unreclaimable:50848kB kernel_stack:22320kB pagetables:42360kB unstable:0kB bounce:0kB free_pcp:248kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  379.279344] Node 0 DMA32: 2994*4kB (ME) 503*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 16000kB
[  387.367409] Node 0 DMA32 free:15320kB min:5564kB low:6952kB high:8344kB active_anon:76364kB inactive_anon:28712kB active_file:1061180kB inactive_file:596956kB unevictable:0kB isolated(anon):0kB isolated(file):120kB present:2080640kB managed:2021100kB mlocked:0kB dirty:20kB writeback:0kB mapped:27700kB shmem:49168kB slab_reclaimable:51236kB slab_unreclaimable:51096kB kernel_stack:22912kB pagetables:40920kB unstable:0kB bounce:0kB free_pcp:700kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  387.385740] Node 0 DMA32: 3638*4kB (UM) 115*8kB (UM) 1*16kB (U) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15488kB
[  391.207543] Node 0 DMA32 free:15224kB min:5564kB low:6952kB high:8344kB active_anon:115956kB inactive_anon:28392kB active_file:1117532kB inactive_file:359656kB unevictable:0kB isolated(anon):0kB isolated(file):116kB present:2080640kB managed:2021100kB mlocked:0kB dirty:0kB writeback:0kB mapped:29348kB shmem:49168kB slab_reclaimable:56028kB slab_unreclaimable:85168kB kernel_stack:48592kB pagetables:81620kB unstable:0kB bounce:0kB free_pcp:1124kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:356 all_unreclaimable? no
[  391.228084] Node 0 DMA32: 3374*4kB (UME) 221*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15264kB
[  395.663881] Node 0 DMA32 free:12820kB min:5564kB low:6952kB high:8344kB active_anon:98924kB inactive_anon:27520kB active_file:1105780kB inactive_file:494760kB unevictable:0kB isolated(anon):4kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:1412kB writeback:12kB mapped:29588kB shmem:49168kB slab_reclaimable:49836kB slab_unreclaimable:60524kB kernel_stack:32176kB pagetables:50356kB unstable:0kB bounce:0kB free_pcp:1500kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:388 all_unreclaimable? no
[  395.683137] Node 0 DMA32: 3794*4kB (ME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15176kB
[  399.871655] Node 0 DMA32 free:18432kB min:5564kB low:6952kB high:8344kB active_anon:99156kB inactive_anon:26780kB active_file:1150532kB inactive_file:408872kB unevictable:0kB isolated(anon):68kB isolated(file):80kB present:2080640kB managed:2021100kB mlocked:0kB dirty:3492kB writeback:0kB mapped:30924kB shmem:49168kB slab_reclaimable:54236kB slab_unreclaimable:68184kB kernel_stack:37392kB pagetables:63708kB unstable:0kB bounce:0kB free_pcp:784kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  399.890082] Node 0 DMA32: 4155*4kB (UME) 200*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18220kB
[  408.447006] Node 0 DMA32 free:12684kB min:5564kB low:6952kB high:8344kB active_anon:74296kB inactive_anon:25960kB active_file:1086404kB inactive_file:605660kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:264kB writeback:0kB mapped:30604kB shmem:49168kB slab_reclaimable:50200kB slab_unreclaimable:45212kB kernel_stack:19184kB pagetables:34500kB unstable:0kB bounce:0kB free_pcp:740kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  408.465169] Node 0 DMA32: 2804*4kB (ME) 203*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 12840kB
[  416.426931] Node 0 DMA32 free:15396kB min:5564kB low:6952kB high:8344kB active_anon:98836kB inactive_anon:32120kB active_file:964808kB inactive_file:666224kB unevictable:0kB isolated(anon):0kB isolated(file):116kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:33628kB shmem:57332kB slab_reclaimable:51048kB slab_unreclaimable:51824kB kernel_stack:23328kB pagetables:41896kB unstable:0kB bounce:0kB free_pcp:988kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  416.447247] Node 0 DMA32: 5158*4kB (UME) 68*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 21176kB
[  418.780159] Node 0 DMA32 free:8876kB min:5564kB low:6952kB high:8344kB active_anon:86544kB inactive_anon:31516kB active_file:965016kB inactive_file:654444kB unevictable:0kB isolated(anon):0kB isolated(file):116kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:8408kB shmem:57332kB slab_reclaimable:48856kB slab_unreclaimable:61116kB kernel_stack:30224kB pagetables:48636kB unstable:0kB bounce:0kB free_pcp:980kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:260 all_unreclaimable? no
[  418.799643] Node 0 DMA32: 3093*4kB (UME) 1043*8kB (UME) 2*16kB (M) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20748kB
[  428.087913] Node 0 DMA32 free:22760kB min:5564kB low:6952kB high:8344kB active_anon:94544kB inactive_anon:38936kB active_file:1013576kB inactive_file:564976kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2021100kB mlocked:0kB dirty:0kB writeback:0kB mapped:36096kB shmem:65376kB slab_reclaimable:52196kB slab_unreclaimable:60576kB kernel_stack:29888kB pagetables:56364kB unstable:0kB bounce:0kB free_pcp:852kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  428.109005] Node 0 DMA32: 2943*4kB (UME) 458*8kB (UME) 20*16kB (UME) 11*32kB (UME) 11*64kB (ME) 4*128kB (UME) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 17324kB
[  439.014180] Node 0 DMA32 free:11232kB min:5564kB low:6952kB high:8344kB active_anon:82868kB inactive_anon:38872kB active_file:1189912kB inactive_file:439592kB unevictable:0kB isolated(anon):12kB isolated(file):40kB present:2080640kB managed:2021100kB mlocked:0kB dirty:0kB writeback:1152kB mapped:35948kB shmem:65376kB slab_reclaimable:51224kB slab_unreclaimable:56664kB kernel_stack:27696kB pagetables:43180kB unstable:0kB bounce:0kB free_pcp:380kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  439.032446] Node 0 DMA32: 2761*4kB (UM) 28*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 11268kB
[  441.731001] Node 0 DMA32 free:15056kB min:5564kB low:6952kB high:8344kB active_anon:90532kB inactive_anon:42716kB active_file:1204248kB inactive_file:377196kB unevictable:0kB isolated(anon):12kB isolated(file):116kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:5552kB shmem:73568kB slab_reclaimable:52956kB slab_unreclaimable:68304kB kernel_stack:39936kB pagetables:47472kB unstable:0kB bounce:0kB free_pcp:624kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  441.731018] Node 0 DMA32: 3130*4kB (UM) 338*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15224kB
[  442.070851] Node 0 DMA32 free:8852kB min:5564kB low:6952kB high:8344kB active_anon:90412kB inactive_anon:42664kB active_file:1179304kB inactive_file:371316kB unevictable:0kB isolated(anon):108kB isolated(file):268kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:5544kB shmem:73568kB slab_reclaimable:55136kB slab_unreclaimable:80080kB kernel_stack:55456kB pagetables:52692kB unstable:0kB bounce:0kB free_pcp:312kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:348 all_unreclaimable? no
[  442.070867] Node 0 DMA32: 590*4kB (ME) 827*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 8976kB
[  442.245192] Node 0 DMA32 free:10832kB min:5564kB low:6952kB high:8344kB active_anon:97756kB inactive_anon:42664kB active_file:1082048kB inactive_file:417012kB unevictable:0kB isolated(anon):108kB isolated(file):268kB present:2080640kB managed:2021100kB mlocked:0kB dirty:4kB writeback:0kB mapped:5248kB shmem:73568kB slab_reclaimable:62816kB slab_unreclaimable:88964kB kernel_stack:61408kB pagetables:62908kB unstable:0kB bounce:0kB free_pcp:696kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  442.245208] Node 0 DMA32: 1902*4kB (UME) 410*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 10888kB

----------

Since I cannot establish workload that caused December 24's natural OOM killers, I used the following stressor for generating similar situation.

The fileio.c fills up all memory with file cache and tries to keep them on memory. The fork.c is flood of order-2 allocation generator because December 24's OOM killers were triggered by copy_process() which involves order-2 allocation request.

---------- fileio.c start ----------
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>

int main(int argc, char *argv[])

{
	int i;
	static char buffer[4096];
	signal(SIGCHLD, SIG_IGN);
	for (i = 0; i < 2; i++) {
		int fd;
		int j;
		snprintf(buffer, sizeof(buffer), "/tmp/file.%u", i);
		fd = open(buffer, O_RDWR | O_CREAT, 0600);
		memset(buffer, 0, sizeof(buffer));
		for (j = 0; j < 1048576 * 1000 / 4096; j++) /* 1000 is MemTotal / 2 */
			write(fd, buffer, sizeof(buffer));
		close(fd);
	}
	for (i = 0; i < 2; i++) {
		if (fork() == 0) {
			int fd;
			snprintf(buffer, sizeof(buffer), "/tmp/file.%u", i);
			fd = open(buffer, O_RDWR);
			memset(buffer, 0, sizeof(buffer));
			while (fd != EOF) {
				lseek(fd, 0, SEEK_SET);
				while (read(fd, buffer, sizeof(buffer)) == sizeof(buffer));
			}
			_exit(0);
		}
	}
	if (fork() == 0) {
		execl("./fork", "./fork", NULL);
		_exit(1);
	}
	if (fork() == 0) {
		sleep(1);
		execl("./fork", "./fork", NULL);
		_exit(1);
	}
	while (1)
		system("pidof fork | wc");
	return 0;
}

---------- fileio.c end ----------

---------- fork.c start ----------
#include <unistd.h>
#include <signal.h>

int main(int argc, char *argv[])

{
	int i;
	signal(SIGCHLD, SIG_IGN);
	while (1) {
		sleep(5);
		for (i = 0; i < 2000; i++) {
			if (fork() == 0) {
				sleep(3);
				_exit(0);
			}
		}
	}
}

---------- fork.c end ----------

This reproducer also showed that once the OOM killer is invoked, subsequent OOM killers tend to occur shortly because file cache do not decrease.

#20 lore raw
Quoted 8 lines · Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Since I cannot establish workload that caused December 24's natural OOM killers, I used the following stressor for generating similar situation.

I came to feel that I am observing a different problem which is currently hidden behind the "too small to fail" memory-allocation rule. That is, tasks requesting order > 0 pages are continuously losing the competition when tasks requesting order = 0 pages dominate, for reclaimed pages are stolen by tasks requesting order = 0 pages before reclaimed pages are combined to order > 0 pages (or maybe order > 0 pages are immediately split into order = 0 pages due to tasks requesting order = 0 pages).

Currently, order <= PAGE_ALLOC_COSTLY_ORDER allocations implicitly retry unless chosen by the OOM killer. Therefore, even if tasks requesting order = 2 pages lost the competition when there are tasks requesting order = 0 pages, the order = 2 allocation request is implicitly retried and therefore the OOM killer is not invoked (though there is a problem that tasks requesting order > 0 allocation will stall as long as tasks requesting order = 0 pages dominate).

But this patchset introduced a limit of 16 retries. Thus, if tasks requesting order = 2 pages lost the competition for 16 times due to tasks requesting order = 0 pages, tasks requesting order = 2 pages invoke the OOM killer. To avoid the OOM killer, we need to make sure that pages reclaimed for order > 0 allocations will not be stolen by tasks requesting order = 0 allocations.

Is my feeling plausible?

#21 lore raw
Quoted 10 lines · Tetsuo Handa
Quoted 8 lines · Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Since I cannot establish workload that caused December 24's natural OOM killers, I used the following stressor for generating similar situation.

Quoted 7 lines

I came to feel that I am observing a different problem which is currently hidden behind the "too small to fail" memory-allocation rule. That is, tasks requesting order > 0 pages are continuously losing the competition when tasks requesting order = 0 pages dominate, for reclaimed pages are stolen by tasks requesting order = 0 pages before reclaimed pages are combined to order > 0 pages (or maybe order > 0 pages are immediately split into order = 0 pages due to tasks requesting order = 0 pages).

Hm I would expect that as long as there are some reserves left that your reproducer cannot grab, there are some free pages left and the allocator should thus preserve the order-2 pages that combine, since order-0 allocations will get existing order-0 pages before splitting higher orders. Compaction should also be able to successfully combine order-2 without racing allocators thanks to per-cpu caching (but I'd have to check).

So I think the problem is not higher-order page itself, but that order-2 needs 4 pages and thus needs to pass a bit higher watermark, thus being at disadvantage to order-0 allocations. Thus I would expect the order-2 pages to be there, but not available for allocation due to watermarks.

Quoted 15 lines

Currently, order <= PAGE_ALLOC_COSTLY_ORDER allocations implicitly retry unless chosen by the OOM killer. Therefore, even if tasks requesting order = 2 pages lost the competition when there are tasks requesting order = 0 pages, the order = 2 allocation request is implicitly retried and therefore the OOM killer is not invoked (though there is a problem that tasks requesting order > 0 allocation will stall as long as tasks requesting order = 0 pages dominate).

But this patchset introduced a limit of 16 retries. Thus, if tasks requesting order = 2 pages lost the competition for 16 times due to tasks requesting order = 0 pages, tasks requesting order = 2 pages invoke the OOM killer. To avoid the OOM killer, we need to make sure that pages reclaimed for order > 0 allocations will not be stolen by tasks requesting order = 0 allocations.

Quoted 8 lines

Is my feeling plausible?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#22 lore raw
In reply to Tetsuo Handa
Quoted 18 lines · Tetsuo Handa
Quoted 8 lines · Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Since I cannot establish workload that caused December 24's natural OOM killers, I used the following stressor for generating similar situation.

I came to feel that I am observing a different problem which is currently hidden behind the "too small to fail" memory-allocation rule. That is, tasks requesting order > 0 pages are continuously losing the competition when tasks requesting order = 0 pages dominate, for reclaimed pages are stolen by tasks requesting order = 0 pages before reclaimed pages are combined to order > 0 pages (or maybe order > 0 pages are immediately split into order = 0 pages due to tasks requesting order = 0 pages).

Quoted 7 lines

Currently, order <= PAGE_ALLOC_COSTLY_ORDER allocations implicitly retry unless chosen by the OOM killer. Therefore, even if tasks requesting order = 2 pages lost the competition when there are tasks requesting order = 0 pages, the order = 2 allocation request is implicitly retried and therefore the OOM killer is not invoked (though there is a problem that tasks requesting order > 0 allocation will stall as long as tasks requesting order = 0 pages dominate).

Yes this is possible and nothing new. High order allocations (even small orders) are never for free and more expensive than order-0. I have seen an OOM killer striking while there were megs of free memory on a larger machine just because of the high fragmentation.

Quoted 1 line

But this patchset introduced a limit of 16 retries.

We retry 16 times _only_ if the reclaim hasn't made _any_ progress which means it hasn't reclaimed a single page. We can still fail due to watermarks check for the required order but I think this is a correct and desirable behavior because there is no guarantee that lower order pages will get coalesced after more retries. The primary point of this rework is to make the whole thing more deterministic.

So we can see some OOM reports for high orders (<COSTLY) which would survive before just because we have retried so many times that we end up allocating that single high order page but this was a pure luck and indeterministic behavior. That being said I agree we might end up doing some more tuning for non-costly high order allocation but it should be bounded as well and based on failures on some reasonable workloads. I haven't got to OOM reports you have posted yet but I definitely plan to check them soon.

[...]

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#23 lore raw
In reply to Tetsuo Handa
Quoted 9 lines · Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Well, current patch invokes OOM killers when more than 75% of memory is used for file cache (active_file: + inactive_file:). I think this is a surprising thing for administrators and we want to retry more harder (but not forever, please).

Here again, it would be good to see what is the comparision between the original and the new behavior. 75% of a page cache is certainly unexpected but those pages might be pinned for other reasons and so unreclaimable and basically IO bound. This is hard to optimize for without causing any undesirable side effects for other loads. I will have a look at the oom reports later but having a comparision would be a great start.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#24 lore raw
Quoted 11 lines · Michal Hocko
Quoted 9 lines · Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Well, current patch invokes OOM killers when more than 75% of memory is used for file cache (active_file: + inactive_file:). I think this is a surprising thing for administrators and we want to retry more harder (but not forever, please).

Quoted 7 lines

Here again, it would be good to see what is the comparision between the original and the new behavior. 75% of a page cache is certainly unexpected but those pages might be pinned for other reasons and so unreclaimable and basically IO bound. This is hard to optimize for without causing any undesirable side effects for other loads. I will have a look at the oom reports later but having a comparision would be a great start.

Prior to "mm, oom: rework oom detection" patch (the original), this stressor never invoked the OOM killer. After this patch (the new), this stressor easily invokes the OOM killer. Both the original and the new case, active_file: +
inactive_file: occupies nearly 75%. I think we lost invisible retry logic for order > 0 allocation requests.

#25 lore raw
Quoted 20 lines · Tetsuo Handa
Quoted 11 lines · Michal Hocko
Quoted 9 lines · Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

Well, current patch invokes OOM killers when more than 75% of memory is used for file cache (active_file: + inactive_file:). I think this is a surprising thing for administrators and we want to retry more harder (but not forever, please).

Quoted 7 lines

Here again, it would be good to see what is the comparision between the original and the new behavior. 75% of a page cache is certainly unexpected but those pages might be pinned for other reasons and so unreclaimable and basically IO bound. This is hard to optimize for without causing any undesirable side effects for other loads. I will have a look at the oom reports later but having a comparision would be a great start.

Quoted 6 lines

Prior to "mm, oom: rework oom detection" patch (the original), this stressor never invoked the OOM killer. After this patch (the new), this stressor easily invokes the OOM killer. Both the original and the new case, active_file: +
inactive_file: occupies nearly 75%. I think we lost invisible retry logic for order > 0 allocation requests.

I retested with below debug printk() patch.

----------

mm/page_alloc.c +7 -3
index 9d70a80..e433504 100644
@@ -3014,7 +3014,7 @@ static inline bool is_thp_gfp_mask(gfp_t gfp_mask)
 static inline bool
 should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		     struct alloc_context *ac, int alloc_flags,
-		     bool did_some_progress,
+		     unsigned long did_some_progress,
 		     int no_progress_loops)
 {
 	struct zone *zone;
@@ -3024,8 +3024,10 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 	 * Make sure we converge to OOM if we cannot make any progress
 	 * several times in the row.
 	 */
-	if (no_progress_loops > MAX_RECLAIM_RETRIES)
+	if (no_progress_loops > MAX_RECLAIM_RETRIES) {
+		printk(KERN_INFO "Reached MAX_RECLAIM_RETRIES.\n");
 		return false;
+	}
 
 	/* Do not retry high order allocations unless they are __GFP_REPEAT */
 	if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
@@ -3086,6 +3088,8 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 
 			return true;
 		}
+		printk(KERN_INFO "zone=%s reclaimable=%lu available=%lu no_progress_loops=%u did_some_progress=%lu\n",
+		       zone->name, reclaimable, available, no_progress_loops, did_some_progress);
 	}
 
 	return false;
@@ -3273,7 +3277,7 @@ retry:
 		no_progress_loops++;
 
 	if (should_reclaim_retry(gfp_mask, order, ac, alloc_flags,
-				 did_some_progress > 0, no_progress_loops))
+				 did_some_progress, no_progress_loops))
 		goto retry;
 
 	/* Reclaim has failed us, start killing things */

----------

The output showed that __zone_watermark_ok() returning false on both DMA32 and DMA zones is the trigger of the OOM killer invocation. Direct reclaim is constantly reclaiming some pages, but I guess freelist for 2 <= order < MAX_ORDER are empty. That trigger was introduced by commit 97a16fc82a7c5b0c ("mm, page_alloc: only enforce watermarks for order-0 allocations"), and "mm, oom: rework oom detection" patch hits the trigger.

----------

[  154.547143] zone=DMA32 reclaimable=323478 available=325894 no_progress_loops=0 did_some_progress=58
[  154.551119] zone=DMA32 reclaimable=323153 available=325770 no_progress_loops=0 did_some_progress=58
[  154.571983] zone=DMA32 reclaimable=319582 available=322161 no_progress_loops=0 did_some_progress=56
[  154.576121] zone=DMA32 reclaimable=319647 available=322016 no_progress_loops=0 did_some_progress=56
[  154.583523] zone=DMA32 reclaimable=319467 available=321801 no_progress_loops=0 did_some_progress=55
[  154.593948] zone=DMA32 reclaimable=317400 available=320988 no_progress_loops=0 did_some_progress=56
[  154.730880] zone=DMA32 reclaimable=312385 available=313952 no_progress_loops=0 did_some_progress=48
[  154.733226] zone=DMA32 reclaimable=312337 available=313919 no_progress_loops=0 did_some_progress=48
[  154.737270] zone=DMA32 reclaimable=312417 available=313871 no_progress_loops=0 did_some_progress=48
[  154.739569] zone=DMA32 reclaimable=312369 available=313844 no_progress_loops=0 did_some_progress=48
[  154.743195] zone=DMA32 reclaimable=312385 available=313790 no_progress_loops=0 did_some_progress=48
[  154.745534] zone=DMA32 reclaimable=312365 available=313813 no_progress_loops=0 did_some_progress=48
[  154.748431] zone=DMA32 reclaimable=312272 available=313728 no_progress_loops=0 did_some_progress=48
[  154.750973] zone=DMA32 reclaimable=312273 available=313760 no_progress_loops=0 did_some_progress=48
[  154.753503] zone=DMA32 reclaimable=312289 available=313958 no_progress_loops=0 did_some_progress=48
[  154.753584] zone=DMA32 reclaimable=312241 available=313958 no_progress_loops=0 did_some_progress=48
[  154.753660] zone=DMA32 reclaimable=312193 available=313958 no_progress_loops=0 did_some_progress=48
[  154.781574] zone=DMA32 reclaimable=312147 available=314095 no_progress_loops=0 did_some_progress=48
[  154.784281] zone=DMA32 reclaimable=311539 available=314015 no_progress_loops=0 did_some_progress=49
[  154.786639] zone=DMA32 reclaimable=311498 available=314040 no_progress_loops=0 did_some_progress=49
[  154.788761] zone=DMA32 reclaimable=311432 available=314040 no_progress_loops=0 did_some_progress=49
[  154.791047] zone=DMA32 reclaimable=311366 available=314040 no_progress_loops=0 did_some_progress=49
[  154.793388] zone=DMA32 reclaimable=311300 available=314040 no_progress_loops=0 did_some_progress=49
[  154.795802] zone=DMA32 reclaimable=311153 available=314006 no_progress_loops=0 did_some_progress=49
[  154.804685] zone=DMA32 reclaimable=309950 available=313140 no_progress_loops=0 did_some_progress=49
[  154.807039] zone=DMA32 reclaimable=309867 available=313138 no_progress_loops=0 did_some_progress=49
[  154.809440] zone=DMA32 reclaimable=309761 available=313080 no_progress_loops=0 did_some_progress=49
[  154.811583] zone=DMA32 reclaimable=309735 available=313120 no_progress_loops=0 did_some_progress=49
[  154.814090] zone=DMA32 reclaimable=309561 available=313068 no_progress_loops=0 did_some_progress=49
[  154.817381] zone=DMA32 reclaimable=309463 available=313030 no_progress_loops=0 did_some_progress=49
[  154.824387] zone=DMA32 reclaimable=309414 available=313030 no_progress_loops=0 did_some_progress=49
[  154.829582] zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
[  154.831562] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50
[  154.838499] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  154.841167] fork cpuset=/ mems_allowed=0
[  154.842348] CPU: 1 PID: 9599 Comm: fork Tainted: G        W       4.4.0-rc7-next-20151231+ #273
[  154.844308] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[  154.846654]  0000000000000000 0000000045061c6b ffff88007a5dbb00 ffffffff81398b83
[  154.848559]  0000000000000000 ffff88007a5dbba0 ffffffff811bc81c 0000000000000206
[  154.850488]  ffffffff818104b0 ffff88007a5dbb40 ffffffff810bdd79 0000000000000206
[  154.852386] Call Trace:
[  154.853350]  [<ffffffff81398b83>] dump_stack+0x4b/0x68
[  154.854731]  [<ffffffff811bc81c>] dump_header+0x5b/0x3b0
[  154.856309]  [<ffffffff810bdd79>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[  154.858046]  [<ffffffff810bde4d>] ? trace_hardirqs_on+0xd/0x10
[  154.859593]  [<ffffffff81143d36>] oom_kill_process+0x366/0x540
[  154.861142]  [<ffffffff8114414f>] out_of_memory+0x1ef/0x5a0
[  154.862655]  [<ffffffff8114420d>] ? out_of_memory+0x2ad/0x5a0
[  154.864194]  [<ffffffff81149c72>] __alloc_pages_nodemask+0xda2/0xde0
[  154.865852]  [<ffffffff810bdd00>] ? trace_hardirqs_on_caller+0x80/0x1c0
[  154.867844]  [<ffffffff81149e6c>] alloc_kmem_pages_node+0x4c/0xc0
[  154.868726] zone=DMA32 reclaimable=309003 available=312677 no_progress_loops=0 did_some_progress=48
[  154.868727] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=48
[  154.875357]  [<ffffffff8106d441>] copy_process.part.31+0x131/0x1b40
[  154.877845]  [<ffffffff8111d8da>] ? __audit_syscall_entry+0xaa/0xf0
[  154.880397]  [<ffffffff8106f01b>] _do_fork+0xdb/0x5d0
[  154.882259]  [<ffffffff8111d8da>] ? __audit_syscall_entry+0xaa/0xf0
[  154.884722]  [<ffffffff810030c1>] ? do_audit_syscall_entry+0x61/0x70
[  154.887201]  [<ffffffff81003254>] ? syscall_trace_enter_phase1+0x134/0x150
[  154.889666]  [<ffffffff81003017>] ? trace_hardirqs_on_thunk+0x17/0x19
[  154.891519]  [<ffffffff8106f594>] SyS_clone+0x14/0x20
[  154.893059]  [<ffffffff816feeb2>] entry_SYSCALL_64_fastpath+0x12/0x76
[  154.894859] Mem-Info:
[  154.895851] active_anon:31807 inactive_anon:2093 isolated_anon:0
[  154.895851]  active_file:242656 inactive_file:67266 isolated_file:0
[  154.895851]  unevictable:0 dirty:8 writeback:0 unstable:0
[  154.895851]  slab_reclaimable:15100 slab_unreclaimable:20839
[  154.895851]  mapped:1681 shmem:2162 pagetables:18491 bounce:0
[  154.895851]  free:4243 free_pcp:343 free_cma:0
[  154.905459] Node 0 DMA free:6908kB min:44kB low:52kB high:64kB active_anon:3408kB inactive_anon:120kB active_file:4kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:64kB shmem:124kB slab_reclaimable:872kB slab_unreclaimable:3032kB kernel_stack:176kB pagetables:328kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  154.916097] lowmem_reserve[]: 0 1714 1714 1714
[  154.917857] Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB active_anon:121688kB inactive_anon:8252kB active_file:970620kB inactive_file:269060kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:1758944kB mlocked:0kB dirty:32kB writeback:0kB mapped:6660kB shmem:8524kB slab_reclaimable:59528kB slab_unreclaimable:80460kB kernel_stack:47312kB pagetables:70972kB unstable:0kB bounce:0kB free_pcp:1356kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  154.929908] lowmem_reserve[]: 0 0 0 0
[  154.931918] Node 0 DMA: 107*4kB (UME) 72*8kB (ME) 47*16kB (UME) 19*32kB (UME) 9*64kB (ME) 1*128kB (M) 3*256kB (M) 2*512kB (E) 2*1024kB (UM) 0*2048kB 0*4096kB = 6908kB
[  154.937453] Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB
[  154.941617] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[  154.944167] 312171 total pagecache pages
[  154.945926] 0 pages in swap cache
[  154.947521] Swap cache stats: add 0, delete 0, find 0/0
[  154.949436] Free swap  = 0kB
[  154.950920] Total swap = 0kB
[  154.952531] 524157 pages RAM
[  154.954063] 0 pages HighMem/MovableOnly
[  154.955785] 80445 pages reserved
[  154.957362] 0 pages hwpoisoned

----------

author reviewer
#26 lore raw

On Sun 03-01-16 00:47:30, Tetsuo Handa wrote:
[...]

Quoted 3 lines

The output showed that __zone_watermark_ok() returning false on both DMA32 and DMA zones is the trigger of the OOM killer invocation. Direct reclaim is constantly reclaiming some pages, but I guess freelist for 2 <= order < MAX_ORDER are empty.

Yes and this is to be expected. Direct reclaim doesn't guarantee any progress for high order allocations. We might be reclaiming pages which cannot be coalesced.

Quoted 3 lines

That trigger was introduced by commit 97a16fc82a7c5b0c ("mm, page_alloc: only enforce watermarks for order-0 allocations"), and "mm, oom: rework oom detection" patch hits the trigger.

[....]

Quoted 4 lines
[  154.829582] zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
[  154.831562] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50
[  154.838499] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  154.841167] fork cpuset=/ mems_allowed=0

[...]

Quoted 1 line
[  154.917857] Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB ....

[...]

Quoted 2 lines
[  154.931918] Node 0 DMA: 107*4kB (UME) 72*8kB (ME) 47*16kB (UME) 19*32kB (UME) 9*64kB (ME) 1*128kB (M) 3*256kB (M) 2*512kB (E) 2*1024kB (UM) 0*2048kB 0*4096kB = 6908kB
[  154.937453] Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB

It is really strange that __zone_watermark_ok claimed DMA32 unusable here. With the target of 312734 which should easilly pass the wmark check for the particular order and there are 116*16kB 15*32kB 1*64kB blocks "usable" for our request because GFP_KERNEL can use both Unmovable and Movable blocks. So it makes sense to wait for more order-0 allocations to pass the basic (NR_FREE_MEMORY) watermark and continue with this particular allocation request.

The nr_reserved_highatomic might be too high to matter but then you see
[1] the reserce being 0. So this doesn't make much sense to me. I will dig into it some more.

[1] http://lkml.kernel.org/r/[email protected]

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#27 lore raw
Quoted 22 lines · Michal Hocko
Quoted 3 lines

That trigger was introduced by commit 97a16fc82a7c5b0c ("mm, page_alloc: only enforce watermarks for order-0 allocations"), and "mm, oom: rework oom detection" patch hits the trigger.

[....]

Quoted 4 lines
[  154.829582] zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
[  154.831562] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50
[  154.838499] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  154.841167] fork cpuset=/ mems_allowed=0

[...]

Quoted 1 line
[  154.917857] Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB ....

[...]

Quoted 2 lines
[  154.931918] Node 0 DMA: 107*4kB (UME) 72*8kB (ME) 47*16kB (UME) 19*32kB (UME) 9*64kB (ME) 1*128kB (M) 3*256kB (M) 2*512kB (E) 2*1024kB (UM) 0*2048kB 0*4096kB = 6908kB
[  154.937453] Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB

It is really strange that __zone_watermark_ok claimed DMA32 unusable here. With the target of 312734 which should easilly pass the wmark check for the particular order and there are 116*16kB 15*32kB 1*64kB blocks "usable" for our request because GFP_KERNEL can use both Unmovable and Movable blocks. So it makes sense to wait for more order-0 allocations to pass the basic (NR_FREE_MEMORY) watermark and continue with this particular allocation request.

Quoted 5 lines

The nr_reserved_highatomic might be too high to matter but then you see
[1] the reserce being 0. So this doesn't make much sense to me. I will dig into it some more.

[1] http://lkml.kernel.org/r/[email protected]

There's another issue in the use of zone_reclaimable_pages(). I think should_reclaim_retry() using zone_page_state_snapshot() is approrpriate, as I indicated before, but notice that zone_reclaimable_pages() only uses zone_page_state(). It means that the heuristic is based on some up-to-date members and some stale members. If we are relying on NR_ISOLATED_* to be accurate, for example, in zone_reclaimable_pages(), then it may take up to 1s for that to actually occur and may quickly exhaust the retry counter in should_reclaim_retry() before that happens.

This is the same issue that Joonsoo reported with the use of zone_page_state(NR_ISOLATED_*) in the too_many_isolated() loops of reclaim and compaction.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#28 lore raw
Quoted 30 lines · David Rientjes
Quoted 22 lines · Michal Hocko
Quoted 3 lines

That trigger was introduced by commit 97a16fc82a7c5b0c ("mm, page_alloc: only enforce watermarks for order-0 allocations"), and "mm, oom: rework oom detection" patch hits the trigger.

[....]

Quoted 4 lines
[  154.829582] zone=DMA32 reclaimable=308907 available=312734 no_progress_loops=0 did_some_progress=50
[  154.831562] zone=DMA reclaimable=2 available=1728 no_progress_loops=0 did_some_progress=50
[  154.838499] fork invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[  154.841167] fork cpuset=/ mems_allowed=0

[...]

Quoted 1 line
[  154.917857] Node 0 DMA32 free:17996kB min:5172kB low:6464kB high:7756kB ....

[...]

Quoted 2 lines
[  154.931918] Node 0 DMA: 107*4kB (UME) 72*8kB (ME) 47*16kB (UME) 19*32kB (UME) 9*64kB (ME) 1*128kB (M) 3*256kB (M) 2*512kB (E) 2*1024kB (UM) 0*2048kB 0*4096kB = 6908kB
[  154.937453] Node 0 DMA32: 1113*4kB (UME) 1400*8kB (UME) 116*16kB (UM) 15*32kB (UM) 1*64kB (M) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18052kB

It is really strange that __zone_watermark_ok claimed DMA32 unusable here. With the target of 312734 which should easilly pass the wmark check for the particular order and there are 116*16kB 15*32kB 1*64kB blocks "usable" for our request because GFP_KERNEL can use both Unmovable and Movable blocks. So it makes sense to wait for more order-0 allocations to pass the basic (NR_FREE_MEMORY) watermark and continue with this particular allocation request.

Quoted 5 lines

The nr_reserved_highatomic might be too high to matter but then you see
[1] the reserce being 0. So this doesn't make much sense to me. I will dig into it some more.

[1] http://lkml.kernel.org/r/[email protected]

Quoted 8 lines

There's another issue in the use of zone_reclaimable_pages(). I think should_reclaim_retry() using zone_page_state_snapshot() is approrpriate, as I indicated before, but notice that zone_reclaimable_pages() only uses zone_page_state(). It means that the heuristic is based on some up-to-date members and some stale members. If we are relying on NR_ISOLATED_* to be accurate, for example, in zone_reclaimable_pages(), then it may take up to 1s for that to actually occur and may quickly exhaust the retry counter in should_reclaim_retry() before that happens.

You are right. I will post a patch to fix that.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#29 lore raw
In reply to Tetsuo Handa
Quoted 3 lines · Tetsuo Handa

I got OOM killers while running heavy disk I/O (extracting kernel source, running lxr's genxref command). (Environ: 4 CPUs / 2048MB RAM / no swap / XFS) Do you think these OOM killers reasonable? Too weak against fragmentation?

I will have a look at the oom report more closely early next week (I am still in holiday mode) but it would be good to compare how the same load behaves with the original implementation. It would be also interesting to see how stable are the results (is there any variability in multiple runs?).

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#30 lore raw

[PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

In reply to Michal Hocko

From: Michal Hocko <[email protected]>

__alloc_pages_may_oom has been doing get_page_from_freelist with ALLOC_WMARK_HIGH target before going out_of_memory and invoking the oom killer. This has two reasons as explained by Andrea:
"
: the reason for the high wmark is to reduce the likelihood of livelocks : and be sure to invoke the OOM killer, if we're still under pressure : and reclaim just failed. The high wmark is used to be sure the failure : of reclaim isn't going to be ignored. If using the min wmark like : you propose there's risk of livelock or anyway of delayed OOM killer : invocation.
:
: The reason for doing one last wmark check (regardless of the wmark : used) before invoking the oom killer, was just to be sure another OOM : killer invocation hasn't already freed a ton of memory while we were : stuck in reclaim. A lot of free memory generated by the OOM killer, : won't make a parallel reclaim more likely to succeed, it just creates : free memory, but reclaim only succeeds when it finds "freeable" memory : and it makes progress in converting it to free memory. So for the : purpose of this last check, the high wmark would work fine as lots of : free memory would have been generated in such case.
"

This is no longer a concern after "mm, oom: rework oom detection" because should_reclaim_retry performs the water mark check right before __alloc_pages_may_oom is invoked. Remove the last moment allocation request as it just makes the code more confusing and doesn't really serve any purpose because a success is basically impossible otherwise should_reclaim_retry would force the reclaim to retry. So this is merely a code cleanup rather than a functional change.

---
 mm/page_alloc.c | 10 ----------
 1 file changed, 10 deletions(-)
mm/page_alloc.c +0 -10
index 268de1654128..f82941c0ac4e 100644
@@ -2743,16 +2743,6 @@ __alloc_pages_may_oom(gfp_t gfp_mask, unsigned int order,
 		return NULL;
 	}
 
-	/*
-	 * Go through the zonelist yet one more time, keep very high watermark
-	 * here, this is only to catch a parallel oom killing, we must fail if
-	 * we're still under heavy pressure.
-	 */
-	page = get_page_from_freelist(gfp_mask | __GFP_HARDWALL, order,
-					ALLOC_WMARK_HIGH|ALLOC_CPUSET, ac);
-	if (page)
-		goto out;
-
 	if (!(gfp_mask & __GFP_NOFAIL)) {
 		/* Coredumps can quickly deplete all memory reserves */
 		if (current->flags & PF_DUMPCORE)
Signature · 7 lines
2.7.0.rc3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
maintainer
#31 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

Quoted 32 lines · Michal Hocko

From: Michal Hocko <[email protected]>

__alloc_pages_may_oom has been doing get_page_from_freelist with ALLOC_WMARK_HIGH target before going out_of_memory and invoking the oom killer. This has two reasons as explained by Andrea:
"
: the reason for the high wmark is to reduce the likelihood of livelocks : and be sure to invoke the OOM killer, if we're still under pressure : and reclaim just failed. The high wmark is used to be sure the failure : of reclaim isn't going to be ignored. If using the min wmark like : you propose there's risk of livelock or anyway of delayed OOM killer : invocation.
:
: The reason for doing one last wmark check (regardless of the wmark : used) before invoking the oom killer, was just to be sure another OOM : killer invocation hasn't already freed a ton of memory while we were : stuck in reclaim. A lot of free memory generated by the OOM killer, : won't make a parallel reclaim more likely to succeed, it just creates : free memory, but reclaim only succeeds when it finds "freeable" memory : and it makes progress in converting it to free memory. So for the : purpose of this last check, the high wmark would work fine as lots of : free memory would have been generated in such case.
"

This is no longer a concern after "mm, oom: rework oom detection" because should_reclaim_retry performs the water mark check right before __alloc_pages_may_oom is invoked. Remove the last moment allocation request as it just makes the code more confusing and doesn't really serve any purpose because a success is basically impossible otherwise should_reclaim_retry would force the reclaim to retry. So this is merely a code cleanup rather than a functional change.

Quoted 1 line

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

Please drop this patch.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#32 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

Quoted 4 lines · Johannes Weiner

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

Right, we need to try with ALLOC_WMARK_HIGH after oom_lock has been acquired.

The situation is still somewhat fragile, however, but I think it's tangential to this patch series. If the ALLOC_WMARK_HIGH allocation fails because an oom victim hasn't freed its memory yet, and then the TIF_MEMDIE thread isn't visible during the oom killer's tasklist scan because it has exited, we still end up killing more than we should. The likelihood of this happening grows with the length of the tasklist.

Perhaps we should try testing watermarks after a victim has been selected and immediately before killing? (Aside: we actually carry an internal patch to test mem_cgroup_margin() in the memcg oom path after selecting a victim because we have been hit with this before in the memcg path.)

I would think that retrying with ALLOC_WMARK_HIGH would be enough memory to deem that we aren't going to immediately reenter an oom condition so the deferred killing is a waste of time.

The downside is how sloppy this would be because it's blurring the line between oom killer and page allocator. We'd need the oom killer to return the selected victim to the page allocator, try the allocation, and then call oom_kill_process() if necessary.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

maintainer
#33 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

Quoted 22 lines · David Rientjes
Quoted 4 lines · Johannes Weiner

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

Right, we need to try with ALLOC_WMARK_HIGH after oom_lock has been acquired.

The situation is still somewhat fragile, however, but I think it's tangential to this patch series. If the ALLOC_WMARK_HIGH allocation fails because an oom victim hasn't freed its memory yet, and then the TIF_MEMDIE thread isn't visible during the oom killer's tasklist scan because it has exited, we still end up killing more than we should. The likelihood of this happening grows with the length of the tasklist.

Perhaps we should try testing watermarks after a victim has been selected and immediately before killing? (Aside: we actually carry an internal patch to test mem_cgroup_margin() in the memcg oom path after selecting a victim because we have been hit with this before in the memcg path.)

Quoted 8 lines

I would think that retrying with ALLOC_WMARK_HIGH would be enough memory to deem that we aren't going to immediately reenter an oom condition so the deferred killing is a waste of time.

The downside is how sloppy this would be because it's blurring the line between oom killer and page allocator. We'd need the oom killer to return the selected victim to the page allocator, try the allocation, and then call oom_kill_process() if necessary.

https://lkml.org/lkml/2015/3/25/40

We could have out_of_memory() wait until the number of outstanding OOM victims drops to 0. Then __alloc_pages_may_oom() doesn't relinquish the lock until its kill has been finalized:

mm/oom_kill.c +3 -1
index 914451a..4dc5b9d 100644
@@ -892,7 +892,9 @@ bool out_of_memory(struct oom_control *oc)
 		 * Give the killed process a good chance to exit before trying
 		 * to allocate memory again.
 		 */
-		schedule_timeout_killable(1);
+		if (!test_thread_flag(TIF_MEMDIE))
+			wait_event_timeout(oom_victims_wait,
+					   !atomic_read(&oom_victims), HZ);
 	}
 	return true;
 }

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#34 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

Quoted 18 lines · Johannes Weiner
Quoted 17 lines · David Rientjes
Quoted 4 lines · Johannes Weiner

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

Right, we need to try with ALLOC_WMARK_HIGH after oom_lock has been acquired.

The situation is still somewhat fragile, however, but I think it's tangential to this patch series. If the ALLOC_WMARK_HIGH allocation fails because an oom victim hasn't freed its memory yet, and then the TIF_MEMDIE thread isn't visible during the oom killer's tasklist scan because it has exited, we still end up killing more than we should. The likelihood of this happening grows with the length of the tasklist.

Quoted 4 lines
Quoted 4 lines

Perhaps we should try testing watermarks after a victim has been selected and immediately before killing? (Aside: we actually carry an internal patch to test mem_cgroup_margin() in the memcg oom path after selecting a victim because we have been hit with this before in the memcg path.)

Yes. Moving final testing to after selecting an OOM victim can reduce the possibility of killing more OOM victims than we need. But unfortunately, it is likely that memory becomes available (i.e. get_page_from_freelist() succeeds) during dump_header() is printing OOM messages using printk(), for printk() is a slow operation compared to selecting a victim. This happens very much later counted from the moment the victim cleared TIF_MEMDIE.

We can avoid killing more OOM victims than we need if we move final testing to after printing OOM messages, but we can't avoid printing OOM messages when we don't kill a victim. Maybe this is not a problem if we do

pr_err("But did not kill any process ...")

instead of

  do_send_sig_info(SIGKILL);
  mark_oom_victim();
  pr_err("Killed process %d (%s) ...")

when final testing succeeded.

Quoted 9 lines
Quoted 9 lines

I would think that retrying with ALLOC_WMARK_HIGH would be enough memory to deem that we aren't going to immediately reenter an oom condition so the deferred killing is a waste of time.

The downside is how sloppy this would be because it's blurring the line between oom killer and page allocator. We'd need the oom killer to return the selected victim to the page allocator, try the allocation, and then call oom_kill_process() if necessary.

I assumed that Michal wants to preserve the boundary between the OOM killer and the page allocator. Therefore, I proposed a patch
( http://lkml.kernel.org/r/[email protected] ) which tries to manage it without returning a victim and without depending on TIF_MEMDIE or oom_victims.

Quoted 13 lines

https://lkml.org/lkml/2015/3/25/40

We could have out_of_memory() wait until the number of outstanding OOM victims drops to 0. Then __alloc_pages_may_oom() doesn't relinquish the lock until its kill has been finalized:

mm/oom_kill.c +0 -0
index 914451a..4dc5b9d 100644

mm/oom_kill.c bool out_of_memory(struct oom_control *oc) · line 900

Quoted 10 lines
mm/oom_kill.c +3 -1
@@ -892,7 +892,9 @@ bool out_of_memory(struct oom_control *oc)
 		 * Give the killed process a good chance to exit before trying
 		 * to allocate memory again.
 		 */
-		schedule_timeout_killable(1);
+		if (!test_thread_flag(TIF_MEMDIE))
+			wait_event_timeout(oom_victims_wait,
+					   !atomic_read(&oom_victims), HZ);
 	}
 	return true;
 }

oom_victims became 0 does not mean that memory became available (i.e. get_page_from_freelist() will succeed). I think this patch wants some effort for trying to reduce possibility of killing more OOM victims than we need.

author reviewer
#35 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

In reply to Johannes Weiner
Quoted 43 lines · Johannes Weiner
Quoted 22 lines · David Rientjes
Quoted 4 lines · Johannes Weiner

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

Right, we need to try with ALLOC_WMARK_HIGH after oom_lock has been acquired.

The situation is still somewhat fragile, however, but I think it's tangential to this patch series. If the ALLOC_WMARK_HIGH allocation fails because an oom victim hasn't freed its memory yet, and then the TIF_MEMDIE thread isn't visible during the oom killer's tasklist scan because it has exited, we still end up killing more than we should. The likelihood of this happening grows with the length of the tasklist.

Perhaps we should try testing watermarks after a victim has been selected and immediately before killing? (Aside: we actually carry an internal patch to test mem_cgroup_margin() in the memcg oom path after selecting a victim because we have been hit with this before in the memcg path.)

Quoted 8 lines

I would think that retrying with ALLOC_WMARK_HIGH would be enough memory to deem that we aren't going to immediately reenter an oom condition so the deferred killing is a waste of time.

The downside is how sloppy this would be because it's blurring the line between oom killer and page allocator. We'd need the oom killer to return the selected victim to the page allocator, try the allocation, and then call oom_kill_process() if necessary.

https://lkml.org/lkml/2015/3/25/40

We could have out_of_memory() wait until the number of outstanding OOM victims drops to 0. Then __alloc_pages_may_oom() doesn't relinquish the lock until its kill has been finalized:

mm/oom_kill.c +0 -0
index 914451a..4dc5b9d 100644

mm/oom_kill.c bool out_of_memory(struct oom_control *oc) · line 900

Quoted 10 lines
mm/oom_kill.c +3 -1
@@ -892,7 +892,9 @@ bool out_of_memory(struct oom_control *oc)
 		 * Give the killed process a good chance to exit before trying
 		 * to allocate memory again.
 		 */
-		schedule_timeout_killable(1);
+		if (!test_thread_flag(TIF_MEMDIE))
+			wait_event_timeout(oom_victims_wait,
+					   !atomic_read(&oom_victims), HZ);
 	}
 	return true;
 }

Yes this makes sense to me

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#36 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

Quoted 15 lines · Michal Hocko
Quoted 15 lines

https://lkml.org/lkml/2015/3/25/40

We could have out_of_memory() wait until the number of outstanding OOM victims drops to 0. Then __alloc_pages_may_oom() doesn't relinquish the lock until its kill has been finalized:

mm/oom_kill.c +0 -1
index 914451a..4dc5b9d 100644
@@ -892,7 +892,9 @@ bool out_of_memory(struct oom_control *oc)
 		 * Give the killed process a good chance to exit before trying
 		 * to allocate memory again.
 		 */
-		schedule_timeout_killable(1);
Quoted 8 lines

mm/oom_kill.c bool out_of_memory(struct oom_control *oc) · line 900

Quoted 6 lines
mm/oom_kill.c +3 -0
+		if (!test_thread_flag(TIF_MEMDIE))
+			wait_event_timeout(oom_victims_wait,
+					   !atomic_read(&oom_victims), HZ);
 	}
 	return true;
 }

Yes this makes sense to me

I think schedule_timeout_killable(1) was used for handling cases where current thread did not get TIF_MEMDIE but got SIGKILL due to sharing the victim's memory. If current thread is blocking TIF_MEMDIE thread, this can become a needless delay.

Also, I don't know whether using wait_event_*() helps handling a problem that schedule_timeout_killable(1) can sleep for many minutes with oom_lock held when there are a lot of tasks. Detail is explained in my proposed patch.

author reviewer
#37 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

In reply to David Rientjes
Quoted 10 lines · David Rientjes
Quoted 4 lines · Johannes Weiner

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

Right, we need to try with ALLOC_WMARK_HIGH after oom_lock has been acquired.

Quoted 6 lines

The situation is still somewhat fragile, however, but I think it's tangential to this patch series. If the ALLOC_WMARK_HIGH allocation fails because an oom victim hasn't freed its memory yet, and then the TIF_MEMDIE thread isn't visible during the oom killer's tasklist scan because it has exited, we still end up killing more than we should. The likelihood of this happening grows with the length of the tasklist.

Yes exactly the point I made in the original thread which brought the question about ALLOC_WMARK_HIGH originally. The race window after the last attempt is much larger than between the last wmark check and the attempt.

Quoted 5 lines

Perhaps we should try testing watermarks after a victim has been selected and immediately before killing? (Aside: we actually carry an internal patch to test mem_cgroup_margin() in the memcg oom path after selecting a victim because we have been hit with this before in the memcg path.)

Quoted 8 lines

I would think that retrying with ALLOC_WMARK_HIGH would be enough memory to deem that we aren't going to immediately reenter an oom condition so the deferred killing is a waste of time.

The downside is how sloppy this would be because it's blurring the line between oom killer and page allocator. We'd need the oom killer to return the selected victim to the page allocator, try the allocation, and then call oom_kill_process() if necessary.

Yes the layer violation is definitely not nice.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#38 lore raw

Re: [PATCH 4/3] mm, oom: drop the last allocation attempt before out_of_memory

In reply to Johannes Weiner
Quoted 33 lines · Johannes Weiner
Quoted 32 lines · Michal Hocko

From: Michal Hocko <[email protected]>

__alloc_pages_may_oom has been doing get_page_from_freelist with ALLOC_WMARK_HIGH target before going out_of_memory and invoking the oom killer. This has two reasons as explained by Andrea:
"
: the reason for the high wmark is to reduce the likelihood of livelocks : and be sure to invoke the OOM killer, if we're still under pressure : and reclaim just failed. The high wmark is used to be sure the failure : of reclaim isn't going to be ignored. If using the min wmark like : you propose there's risk of livelock or anyway of delayed OOM killer : invocation.
:
: The reason for doing one last wmark check (regardless of the wmark : used) before invoking the oom killer, was just to be sure another OOM : killer invocation hasn't already freed a ton of memory while we were : stuck in reclaim. A lot of free memory generated by the OOM killer, : won't make a parallel reclaim more likely to succeed, it just creates : free memory, but reclaim only succeeds when it finds "freeable" memory : and it makes progress in converting it to free memory. So for the : purpose of this last check, the high wmark would work fine as lots of : free memory would have been generated in such case.
"

This is no longer a concern after "mm, oom: rework oom detection" because should_reclaim_retry performs the water mark check right before __alloc_pages_may_oom is invoked. Remove the last moment allocation request as it just makes the code more confusing and doesn't really serve any purpose because a success is basically impossible otherwise should_reclaim_retry would force the reclaim to retry. So this is merely a code cleanup rather than a functional change.

Quoted 5 lines
Quoted 1 line

The check has to happen while holding the OOM lock, otherwise we'll end up killing much more than necessary when there are many racing allocations.

My testing shows that this doesn't trigger even during oom flood testing. So I am not really convinced it does anything useful.

Quoted 1 line

Please drop this patch.

Sure I do not insist...

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#39 lore raw

[PATCH 5/3] mm, vmscan: make zone_reclaimable_pages more precise

In reply to Michal Hocko

From: Michal Hocko <[email protected]>

zone_reclaimable_pages is used in should_reclaim_retry which uses it to calculate the target for the watermark check. This means that precise numbers are important for the correct decision. zone_reclaimable_pages uses zone_page_state which can contain stale data with per-cpu diffs not synced yet (the last vmstat_update might have run 1s in the past).

Use zone_page_state_snapshot in zone_reclaimable_pages instead. None of the current callers is in a hot path where getting the precise value (which involves per-cpu iteration) would cause an unreasonable overhead.

---
 mm/vmscan.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
mm/vmscan.c +7 -7
index 489212252cd6..9145e3f89eab 100644
@@ -196,21 +196,21 @@ unsigned long zone_reclaimable_pages(struct zone *zone)
 {
 	unsigned long nr;
 
-	nr = zone_page_state(zone, NR_ACTIVE_FILE) +
-	     zone_page_state(zone, NR_INACTIVE_FILE) +
-	     zone_page_state(zone, NR_ISOLATED_FILE);
+	nr = zone_page_state_snapshot(zone, NR_ACTIVE_FILE) +
+	     zone_page_state_snapshot(zone, NR_INACTIVE_FILE) +
+	     zone_page_state_snapshot(zone, NR_ISOLATED_FILE);
 
 	if (get_nr_swap_pages() > 0)
-		nr += zone_page_state(zone, NR_ACTIVE_ANON) +
-		      zone_page_state(zone, NR_INACTIVE_ANON) +
-		      zone_page_state(zone, NR_ISOLATED_ANON);
+		nr += zone_page_state_snapshot(zone, NR_ACTIVE_ANON) +
+		      zone_page_state_snapshot(zone, NR_INACTIVE_ANON) +
+		      zone_page_state_snapshot(zone, NR_ISOLATED_ANON);
 
 	return nr;
 }
 
 bool zone_reclaimable(struct zone *zone)
 {
-	return zone_page_state(zone, NR_PAGES_SCANNED) <
+	return zone_page_state_snapshot(zone, NR_PAGES_SCANNED) <
 		zone_reclaimable_pages(zone) * 6;
 }
 
Signature · 7 lines
2.7.0.rc3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
reviewer Acked-by
#40 lore raw

Re: [PATCH 5/3] mm, vmscan: make zone_reclaimable_pages more precise

Quoted 8 lines · Michal Hocko

From: Michal Hocko <[email protected]>

zone_reclaimable_pages is used in should_reclaim_retry which uses it to calculate the target for the watermark check. This means that precise numbers are important for the correct decision. zone_reclaimable_pages uses zone_page_state which can contain stale data with per-cpu diffs not synced yet (the last vmstat_update might have run 1s in the past).

Quoted 6 lines

Use zone_page_state_snapshot in zone_reclaimable_pages instead. None of the current callers is in a hot path where getting the precise value (which involves per-cpu iteration) would cause an unreasonable overhead.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Acked-by
#41 lore raw

Re: [PATCH 5/3] mm, vmscan: make zone_reclaimable_pages more precise

In reply to Michal Hocko
Quoted 9 lines

From: Michal Hocko <[email protected]>

zone_reclaimable_pages is used in should_reclaim_retry which uses it to calculate the target for the watermark check. This means that precise numbers are important for the correct decision. zone_reclaimable_pages uses zone_page_state which can contain stale data with per-cpu diffs not synced yet (the last vmstat_update might have run 1s in the past).

Quoted 7 lines

Use zone_page_state_snapshot in zone_reclaimable_pages instead. None of the current callers is in a hot path where getting the precise value (which involves per-cpu iteration) would cause an unreasonable overhead.

---

Quoted 37 lines
 mm/vmscan.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
mm/vmscan.c +7 -8
index 489212252cd6..9145e3f89eab 100644
@@ -196,21 +196,21 @@ unsigned long zone_reclaimable_pages(struct zone *zone)
 {
 	unsigned long nr;
 
-	nr = zone_page_state(zone, NR_ACTIVE_FILE) +
-	     zone_page_state(zone, NR_INACTIVE_FILE) +
-	     zone_page_state(zone, NR_ISOLATED_FILE);
+	nr = zone_page_state_snapshot(zone, NR_ACTIVE_FILE) +
+	     zone_page_state_snapshot(zone, NR_INACTIVE_FILE) +
+	     zone_page_state_snapshot(zone, NR_ISOLATED_FILE);
 
 	if (get_nr_swap_pages() > 0)
-		nr += zone_page_state(zone, NR_ACTIVE_ANON) +
-		      zone_page_state(zone, NR_INACTIVE_ANON) +
-		      zone_page_state(zone, NR_ISOLATED_ANON);
+		nr += zone_page_state_snapshot(zone, NR_ACTIVE_ANON) +
+		      zone_page_state_snapshot(zone, NR_INACTIVE_ANON) +
+		      zone_page_state_snapshot(zone, NR_ISOLATED_ANON);
 
 	return nr;
 }
 
 bool zone_reclaimable(struct zone *zone)
 {
-	return zone_page_state(zone, NR_PAGES_SCANNED) <
+	return zone_page_state_snapshot(zone, NR_PAGES_SCANNED) <
 		zone_reclaimable_pages(zone) * 6;
 }
 
--
Quoted 1 line

2.7.0.rc3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Acked-by
#42 lore raw

Re: [PATCH 5/3] mm, vmscan: make zone_reclaimable_pages more precise

In reply to Michal Hocko
Quoted 12 lines · Michal Hocko

From: Michal Hocko <[email protected]>

zone_reclaimable_pages is used in should_reclaim_retry which uses it to calculate the target for the watermark check. This means that precise numbers are important for the correct decision. zone_reclaimable_pages uses zone_page_state which can contain stale data with per-cpu diffs not synced yet (the last vmstat_update might have run 1s in the past).

Use zone_page_state_snapshot in zone_reclaimable_pages instead. None of the current callers is in a hot path where getting the precise value (which involves per-cpu iteration) would cause an unreasonable overhead.

Quoted 6 lines
---
 mm/vmscan.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

I didn't know http://lkml.kernel.org/r/[email protected]
was forgotten. Anyway,

author reviewer
#43 lore raw

Re: [PATCH 5/3] mm, vmscan: make zone_reclaimable_pages more precise

Quoted 20 lines · Tetsuo Handa
Quoted 12 lines · Michal Hocko

From: Michal Hocko <[email protected]>

zone_reclaimable_pages is used in should_reclaim_retry which uses it to calculate the target for the watermark check. This means that precise numbers are important for the correct decision. zone_reclaimable_pages uses zone_page_state which can contain stale data with per-cpu diffs not synced yet (the last vmstat_update might have run 1s in the past).

Use zone_page_state_snapshot in zone_reclaimable_pages instead. None of the current callers is in a hot path where getting the precise value (which involves per-cpu iteration) would cause an unreasonable overhead.

Quoted 6 lines
---
 mm/vmscan.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
Quoted 2 lines

I didn't know http://lkml.kernel.org/r/[email protected]
was forgotten. Anyway,

OK, that explains why this sounded so familiar. Sorry I comepletely forgot about it.

Quoted 1 line

Can I change it to your Signed-off-by?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#44 lore raw

Re: [PATCH 5/3] mm, vmscan: make zone_reclaimable_pages more precise

Quoted 24 lines · Michal Hocko
Quoted 20 lines · Tetsuo Handa
Quoted 12 lines · Michal Hocko

From: Michal Hocko <[email protected]>

zone_reclaimable_pages is used in should_reclaim_retry which uses it to calculate the target for the watermark check. This means that precise numbers are important for the correct decision. zone_reclaimable_pages uses zone_page_state which can contain stale data with per-cpu diffs not synced yet (the last vmstat_update might have run 1s in the past).

Use zone_page_state_snapshot in zone_reclaimable_pages instead. None of the current callers is in a hot path where getting the precise value (which involves per-cpu iteration) would cause an unreasonable overhead.

Quoted 6 lines
---
 mm/vmscan.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
Quoted 2 lines

I didn't know http://lkml.kernel.org/r/[email protected]
was forgotten. Anyway,

Quoted 6 lines

OK, that explains why this sounded so familiar. Sorry I comepletely forgot about it.

Quoted 1 line

Can I change it to your Signed-off-by?

No problem.

Quoted 5 lines
Signature · 3 lines
-- 
Michal Hocko
SUSE Labs
author reviewer
#45 lore raw
In reply to Michal Hocko

Hi,
this thread went mostly quite. Are all the main concerns clarified? Are there any new concerns? Are there any objections to targeting this for the next merge window?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#46 lore raw
Quoted 4 lines · Michal Hocko

Hi,
this thread went mostly quite. Are all the main concerns clarified? Are there any new concerns? Are there any objections to targeting this for the next merge window?

Did we ever figure out what was causing the oom killer to be called much earlier in Tetsuo's http://marc.info/?l=linux-kernel&m=145096089726481 and
http://marc.info/?l=linux-kernel&m=145130454913757 ? I'd like to take a look at the patch(es) that fixed it.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#47 lore raw
Quoted 11 lines · David Rientjes
Quoted 4 lines · Michal Hocko

Hi,
this thread went mostly quite. Are all the main concerns clarified? Are there any new concerns? Are there any objections to targeting this for the next merge window?

Did we ever figure out what was causing the oom killer to be called much earlier in Tetsuo's http://marc.info/?l=linux-kernel&m=145096089726481 and

From the OOM report:

[ 3902.430630] kthreadd invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[ 3902.507561] Node 0 DMA32: 3788*4kB (UME) 184*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 16624kB
[ 5262.901161] smbd invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[ 5262.983496] Node 0 DMA32: 1987*4kB (UME) 14*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 8060kB
[ 5269.764580] kthreadd invoked oom-killer: order=2, oom_score_adj=0, gfp_mask=0x27000c0(GFP_KERNEL|GFP_NOTRACK|0x100000)
[ 5269.858330] Node 0 DMA32: 10648*4kB (UME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 42592kB
Quoted 1 line

http://marc.info/?l=linux-kernel&m=145130454913757 ?

[  277.884512] Node 0 DMA32: 3438*4kB (UME) 791*8kB (UME) 3*16kB (UM) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20128kB
[  291.349097] Node 0 DMA32: 4221*4kB (UME) 1971*8kB (UME) 436*16kB (UME) 141*32kB (UME) 8*64kB (UM) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 44652kB
[  302.916334] Node 0 DMA32: 4304*4kB (UM) 1181*8kB (UME) 59*16kB (UME) 7*32kB (ME) 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 27832kB
[  311.034251] Node 0 DMA32: 6*4kB (U) 2401*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 19232kB
[  314.314336] Node 0 DMA32: 1180*4kB (UM) 1449*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 16312kB
[  322.796256] Node 0 DMA32: 86*4kB (UME) 2474*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20136kB
[  330.826190] Node 0 DMA32: 1637*4kB (UM) 1354*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 17380kB
[  332.846805] Node 0 DMA32: 4108*4kB (UME) 897*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 23608kB
[  341.073722] Node 0 DMA32: 3309*4kB (UM) 1124*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 22228kB
[  360.093794] Node 0 DMA32: 2719*4kB (UM) 97*8kB (UM) 14*16kB (UM) 37*32kB (UME) 27*64kB (UME) 3*128kB (UM) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15172kB
[  368.871173] Node 0 DMA32: 5042*4kB (UM) 248*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 22152kB
[  379.279344] Node 0 DMA32: 2994*4kB (ME) 503*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 16000kB
[  387.385740] Node 0 DMA32: 3638*4kB (UM) 115*8kB (UM) 1*16kB (U) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15488kB
[  391.228084] Node 0 DMA32: 3374*4kB (UME) 221*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15264kB
[  395.683137] Node 0 DMA32: 3794*4kB (ME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15176kB
[  399.890082] Node 0 DMA32: 4155*4kB (UME) 200*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18220kB
[  408.465169] Node 0 DMA32: 2804*4kB (ME) 203*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 12840kB
[  416.447247] Node 0 DMA32: 5158*4kB (UME) 68*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 21176kB
[  418.799643] Node 0 DMA32: 3093*4kB (UME) 1043*8kB (UME) 2*16kB (M) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 20748kB
[  428.109005] Node 0 DMA32: 2943*4kB (UME) 458*8kB (UME) 20*16kB (UME) 11*32kB (UME) 11*64kB (ME) 4*128kB (UME) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 17324kB
[  439.032446] Node 0 DMA32: 2761*4kB (UM) 28*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 11268kB
[  441.731018] Node 0 DMA32: 3130*4kB (UM) 338*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 15224kB
[  442.070867] Node 0 DMA32: 590*4kB (ME) 827*8kB (ME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 8976kB
[  442.245208] Node 0 DMA32: 1902*4kB (UME) 410*8kB (UME) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 10888kB

There are cases where order-2 has some pages but I have commented on that here [1]

Quoted 1 line

I'd like to take a look at the patch(es) that fixed it.

I am not sure we can fix these pathological loads where we hit the higher order depletion and there is a chance that one of the thousands tasks terminates in an unpredictable way which happens to race with the OOM killer. As I've pointed out in [1] once the watermark check for the higher order allocation fails for the given order then we cannot rely on the reclaimable pages ever construct the required order. The current zone_reclaimable approach just happens to work for this particular load because the NR_PAGES_SCANNED gets reseted too often with a side effect of an undeterministic behavior.

[1] http://lkml.kernel.org/r/[email protected]

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#48 lore raw
Quoted 4 lines · Michal Hocko

I am not sure we can fix these pathological loads where we hit the higher order depletion and there is a chance that one of the thousands tasks terminates in an unpredictable way which happens to race with the OOM killer.

When I hit this problem on Dec 24th, I didn't run thousands of tasks. I think there were less than one hundred tasks in the system and only a few tasks were running. Not a pathological load at all.

I'm running thousands of tasks only for increasing the possibility in the reproducer.

author reviewer
#49 lore raw
Quoted 9 lines · Tetsuo Handa
Quoted 4 lines · Michal Hocko

I am not sure we can fix these pathological loads where we hit the higher order depletion and there is a chance that one of the thousands tasks terminates in an unpredictable way which happens to race with the OOM killer.

When I hit this problem on Dec 24th, I didn't run thousands of tasks. I think there were less than one hundred tasks in the system and only a few tasks were running. Not a pathological load at all.

But as the OOM report clearly stated there were no > order-1 pages available in that particular case. And that happened after the direct reclaim and compaction were already invoked.

As I've mentioned in the referenced email, we can try to do multiple retries e.g. do not give up on the higher order requests until we hit the maximum number of retries but I consider it quite ugly to be honest. I think that a proper communication with compaction is a more appropriate way to go long term. E.g. I find it interesting that try_to_compact_pages doesn't even care about PAGE_ALLOC_COSTLY_ORDER and treat is as any other high order request.

Something like the following:
---

mm/page_alloc.c +16 -4
index 269a04f20927..1ae5b7da821b 100644
@@ -3106,6 +3106,18 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		}
 	}
 
+	/*
+	 * OK, so the watermak check has failed. Make sure we do all the
+	 * retries for !costly high order requests and hope that multiple
+	 * runs of compaction will generate some high order ones for us.
+	 *
+	 * XXX: ideally we should teach the compaction to try _really_ hard
+	 * if we are in the retry path - something like priority 0 for the
+	 * reclaim
+	 */
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
+
 	return false;
 }
 
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		goto noretry;
 
 	/*
-	 * Costly allocations might have made a progress but this doesn't mean
-	 * their order will become available due to high fragmentation so do
-	 * not reset the no progress counter for them
+	 * High order allocations might have made a progress but this doesn't
+	 * mean their order will become available due to high fragmentation so
+	 * do not reset the no progress counter for them
 	 */
-	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+	if (did_some_progress && !order)
 		no_progress_loops = 0;
 	else
 		no_progress_loops++;
Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#50 lore raw
Quoted 23 lines · Michal Hocko
Quoted 9 lines · Tetsuo Handa
Quoted 4 lines · Michal Hocko

I am not sure we can fix these pathological loads where we hit the higher order depletion and there is a chance that one of the thousands tasks terminates in an unpredictable way which happens to race with the OOM killer.

When I hit this problem on Dec 24th, I didn't run thousands of tasks. I think there were less than one hundred tasks in the system and only a few tasks were running. Not a pathological load at all.

But as the OOM report clearly stated there were no > order-1 pages available in that particular case. And that happened after the direct reclaim and compaction were already invoked.

As I've mentioned in the referenced email, we can try to do multiple retries e.g. do not give up on the higher order requests until we hit the maximum number of retries but I consider it quite ugly to be honest. I think that a proper communication with compaction is a more appropriate way to go long term. E.g. I find it interesting that try_to_compact_pages doesn't even care about PAGE_ALLOC_COSTLY_ORDER and treat is as any other high order request.

Quoted 1 line

Something like the following:

With the patch description. Please note I haven't tested this yet so this is more a RFC than something I am really convinced about. I can live with it because the number of retries is nicely bounded but it sounds too hackish because it makes the decision rather blindly. I will talk to Vlastimil and Mel whether they see some way how to communicate the compaction state in a reasonable way. But I guess this is something that can come up later. What do you think?
---

#51 lore raw
In reply to Michal Hocko
Quoted 15 lines · Michal Hocko
Quoted 9 lines · Tetsuo Handa
Quoted 4 lines · Michal Hocko

I am not sure we can fix these pathological loads where we hit the higher order depletion and there is a chance that one of the thousands tasks terminates in an unpredictable way which happens to race with the OOM killer.

When I hit this problem on Dec 24th, I didn't run thousands of tasks. I think there were less than one hundred tasks in the system and only a few tasks were running. Not a pathological load at all.

But as the OOM report clearly stated there were no > order-1 pages available in that particular case. And that happened after the direct reclaim and compaction were already invoked.

Quoted 8 lines

As I've mentioned in the referenced email, we can try to do multiple retries e.g. do not give up on the higher order requests until we hit the maximum number of retries but I consider it quite ugly to be honest. I think that a proper communication with compaction is a more appropriate way to go long term. E.g. I find it interesting that try_to_compact_pages doesn't even care about PAGE_ALLOC_COSTLY_ORDER and treat is as any other high order request.

FYI, I again hit unexpected OOM-killer during genxref on linux-4.5-rc2 source. I think current patchset is too fragile to merge.
----------------------------------------

[ 3101.626995] smbd invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[ 3101.629148] smbd cpuset=/ mems_allowed=0
[ 3101.630332] CPU: 1 PID: 3941 Comm: smbd Not tainted 4.5.0-rc2-next-20160205 #293
[ 3101.632335] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[ 3101.634567]  0000000000000286 000000005784a8f9 ffff88007c47bad0 ffffffff8139abbd
[ 3101.636533]  0000000000000000 ffff88007c47bd00 ffff88007c47bb70 ffffffff811bdc6c
[ 3101.638381]  0000000000000206 ffffffff81810b30 ffff88007c47bb10 ffffffff810be079
[ 3101.640215] Call Trace:
[ 3101.641169]  [<ffffffff8139abbd>] dump_stack+0x85/0xc8
[ 3101.642560]  [<ffffffff811bdc6c>] dump_header+0x5b/0x3b0
[ 3101.643983]  [<ffffffff810be079>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[ 3101.645616]  [<ffffffff810be14d>] ? trace_hardirqs_on+0xd/0x10
[ 3101.647081]  [<ffffffff81143fb6>] oom_kill_process+0x366/0x550
[ 3101.648631]  [<ffffffff811443df>] out_of_memory+0x1ef/0x5a0
[ 3101.650081]  [<ffffffff8114449d>] ? out_of_memory+0x2ad/0x5a0
[ 3101.651624]  [<ffffffff81149d0d>] __alloc_pages_nodemask+0xbad/0xd90
[ 3101.653207]  [<ffffffff8114a0ac>] alloc_kmem_pages_node+0x4c/0xc0
[ 3101.654767]  [<ffffffff8106d5c1>] copy_process.part.31+0x131/0x1b40
[ 3101.656381]  [<ffffffff8111d9ea>] ? __audit_syscall_entry+0xaa/0xf0
[ 3101.657952]  [<ffffffff810e8119>] ? current_kernel_time64+0xa9/0xc0
[ 3101.659492]  [<ffffffff8106f19b>] _do_fork+0xdb/0x5d0
[ 3101.660814]  [<ffffffff810030c1>] ? do_audit_syscall_entry+0x61/0x70
[ 3101.662305]  [<ffffffff81003254>] ? syscall_trace_enter_phase1+0x134/0x150
[ 3101.663988]  [<ffffffff81703d2c>] ? return_from_SYSCALL_64+0x2d/0x7a
[ 3101.665572]  [<ffffffff810035ec>] ? do_syscall_64+0x1c/0x180
[ 3101.667067]  [<ffffffff8106f714>] SyS_clone+0x14/0x20
[ 3101.668510]  [<ffffffff8100362d>] do_syscall_64+0x5d/0x180
[ 3101.669931]  [<ffffffff81703cff>] entry_SYSCALL64_slow_path+0x25/0x25
[ 3101.671642] Mem-Info:
[ 3101.672612] active_anon:46842 inactive_anon:2094 isolated_anon:0
 active_file:108974 inactive_file:131350 isolated_file:0
 unevictable:0 dirty:1174 writeback:0 unstable:0
 slab_reclaimable:107536 slab_unreclaimable:14287
 mapped:4199 shmem:2166 pagetables:1524 bounce:0
 free:6260 free_pcp:31 free_cma:0
[ 3101.681294] Node 0 DMA free:6884kB min:44kB low:52kB high:64kB active_anon:3488kB inactive_anon:100kB active_file:0kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:4kB shmem:100kB slab_reclaimable:3852kB slab_unreclaimable:444kB kernel_stack:80kB pagetables:112kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 3101.691319] lowmem_reserve[]: 0 1714 1714 1714
[ 3101.692847] Node 0 DMA32 free:18156kB min:5172kB low:6464kB high:7756kB active_anon:183880kB inactive_anon:8276kB active_file:435896kB inactive_file:525396kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:1759480kB mlocked:0kB dirty:4696kB writeback:0kB mapped:16792kB shmem:8564kB slab_reclaimable:426292kB slab_unreclaimable:56704kB kernel_stack:3328kB pagetables:5984kB unstable:0kB bounce:0kB free_pcp:120kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 3101.704239] lowmem_reserve[]: 0 0 0 0
[ 3101.705887] Node 0 DMA: 75*4kB (UME) 69*8kB (UME) 43*16kB (UM) 23*32kB (UME) 8*64kB (UM) 4*128kB (UME) 2*256kB (UM) 0*512kB 1*1024kB (U) 1*2048kB (M) 0*4096kB = 6884kB
[ 3101.710581] Node 0 DMA32: 4513*4kB (UME) 15*8kB (U) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18172kB
[ 3101.713857] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 3101.716332] 242517 total pagecache pages
[ 3101.717878] 0 pages in swap cache
[ 3101.719332] Swap cache stats: add 0, delete 0, find 0/0
[ 3101.721577] Free swap  = 0kB
[ 3101.722980] Total swap = 0kB
[ 3101.724364] 524157 pages RAM
[ 3101.725697] 0 pages HighMem/MovableOnly
[ 3101.727165] 80311 pages reserved
[ 3101.728482] 0 pages hwpoisoned
[ 3101.729754] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[ 3101.732071] [  492]     0   492     9206      975      20       4        0             0 systemd-journal
[ 3101.734357] [  520]     0   520    10479      631      22       3        0         -1000 systemd-udevd
[ 3101.737036] [  527]     0   527    12805      682      24       3        0         -1000 auditd
[ 3101.739505] [ 1174]     0  1174     4830      556      14       3        0             0 irqbalance
[ 3101.741876] [ 1180]    81  1180     6672      604      20       3        0          -900 dbus-daemon
[ 3101.744728] [ 1817]     0  1817    56009      880      40       4        0             0 rsyslogd
[ 3101.747164] [ 1818]     0  1818     1096      349       8       3        0             0 rngd
[ 3101.749788] [ 1820]     0  1820    52575     1074      56       3        0             0 abrtd
[ 3101.752135] [ 1821]     0  1821    80901     5160      80       4        0             0 firewalld
[ 3101.754532] [ 1823]     0  1823     6602      681      20       3        0             0 systemd-logind
[ 3101.757342] [ 1825]    70  1825     6999      458      20       3        0             0 avahi-daemon
[ 3101.759784] [ 1827]     0  1827    51995      986      55       3        0             0 abrt-watch-log
[ 3101.762465] [ 1838]     0  1838    31586      647      21       3        0             0 crond
[ 3101.764797] [ 1946]    70  1946     6999       58      19       3        0             0 avahi-daemon
[ 3101.767262] [ 2043]     0  2043    65187      858      43       3        0             0 vmtoolsd
[ 3101.769665] [ 2618]     0  2618    27631     3112      53       3        0             0 dhclient
[ 3101.772203] [ 2622]   999  2622   130827     2570      56       3        0             0 polkitd
[ 3101.774645] [ 2704]     0  2704   138263     3351      91       4        0             0 tuned
[ 3101.777114] [ 2709]     0  2709    20640      773      45       3        0         -1000 sshd
[ 3101.779428] [ 2711]     0  2711     7328      551      19       3        0             0 xinetd
[ 3101.782016] [ 3883]     0  3883    22785      827      45       3        0             0 master
[ 3101.784576] [ 3884]    89  3884    22811      924      46       4        0             0 pickup
[ 3101.786898] [ 3885]    89  3885    22828      886      44       3        0             0 qmgr
[ 3101.789287] [ 3916]     0  3916    23203      736      50       3        0             0 login
[ 3101.791666] [ 3927]     0  3927    27511      381      13       3        0             0 agetty
[ 3101.794116] [ 3930]     0  3930    79392     1063     105       3        0             0 nmbd
[ 3101.796387] [ 3941]     0  3941    96485     1544     138       4        0             0 smbd
[ 3101.798602] [ 3944]     0  3944    96485     1290     131       4        0             0 smbd
[ 3101.800783] [ 7471]     0  7471    28886      732      15       3        0             0 bash
[ 3101.803013] [ 7580]     0  7580     2380      613      10       3        0             0 makelxr.sh
[ 3101.805147] [ 7786]     0  7786    27511      395      10       3        0             0 agetty
[ 3101.807198] [ 8139]     0  8139    35888      974      72       3        0             0 sshd
[ 3101.809255] [ 8144]     0  8144    28896      761      15       4        0             0 bash
[ 3101.811335] [15286]     0 15286    38294    30474      81       3        0             0 genxref
[ 3101.813512] Out of memory: Kill process 15286 (genxref) score 66 or sacrifice child
[ 3101.815659] Killed process 15286 (genxref) total-vm:153176kB, anon-rss:117092kB, file-rss:4804kB, shmem-rss:0kB

----------------------------------------

Quoted 1 line

Something like the following:

Yes, I do think we need something like it.

author reviewer
#52 lore raw

On Sun 07-02-16 13:09:33, Tetsuo Handa wrote:
[...]

Quoted 5 lines

FYI, I again hit unexpected OOM-killer during genxref on linux-4.5-rc2 source. I think current patchset is too fragile to merge.
----------------------------------------

[ 3101.626995] smbd invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[ 3101.629148] smbd cpuset=/ mems_allowed=0

[...]

Quoted 2 lines
[ 3101.705887] Node 0 DMA: 75*4kB (UME) 69*8kB (UME) 43*16kB (UM) 23*32kB (UME) 8*64kB (UM) 4*128kB (UME) 2*256kB (UM) 0*512kB 1*1024kB (U) 1*2048kB (M) 0*4096kB = 6884kB
[ 3101.710581] Node 0 DMA32: 4513*4kB (UME) 15*8kB (U) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18172kB

How come this is an unexpected OOM? There is clearly no order-2+ page available for the allocation request.

Quoted 2 lines
Quoted 1 line

Something like the following:

Yes, I do think we need something like it.

Was the patch applied?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#53 lore raw
Quoted 7 lines · Michal Hocko

On Sun 07-02-16 13:09:33, Tetsuo Handa wrote:
[...]

Quoted 5 lines

FYI, I again hit unexpected OOM-killer during genxref on linux-4.5-rc2 source. I think current patchset is too fragile to merge.
----------------------------------------

[ 3101.626995] smbd invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[ 3101.629148] smbd cpuset=/ mems_allowed=0
Quoted 6 lines

[...]

Quoted 2 lines
[ 3101.705887] Node 0 DMA: 75*4kB (UME) 69*8kB (UME) 43*16kB (UM) 23*32kB (UME) 8*64kB (UM) 4*128kB (UME) 2*256kB (UM) 0*512kB 1*1024kB (U) 1*2048kB (M) 0*4096kB = 6884kB
[ 3101.710581] Node 0 DMA32: 4513*4kB (UME) 15*8kB (U) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18172kB

How come this is an unexpected OOM? There is clearly no order-2+ page available for the allocation request.

I used "unexpected" because there were only 35 userspace processes and genxref was the only process which did a lot of memory allocation (modulo kernel threads woken by file I/O) and most memory is reclaimable.

Quoted 5 lines
Quoted 3 lines
Quoted 2 lines

Something like the following:

Yes, I do think we need something like it.

Was the patch applied?

No for above result.

A result with the patch ([email protected]) applied on today's linux-next is shown below. It seems that protection is not enough.

----------

[  118.584571] fork invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[  118.586684] fork cpuset=/ mems_allowed=0
[  118.588254] CPU: 2 PID: 9565 Comm: fork Not tainted 4.5.0-rc4-next-20160216+ #306
[  118.589795] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[  118.591941]  0000000000000286 0000000085a9ed62 ffff88007b3d3ad0 ffffffff8139e82d
[  118.593616]  0000000000000000 ffff88007b3d3d00 ffff88007b3d3b70 ffffffff811bedec
[  118.595273]  0000000000000206 ffffffff81810b70 ffff88007b3d3b10 ffffffff810be8f9
[  118.596970] Call Trace:
[  118.597634]  [<ffffffff8139e82d>] dump_stack+0x85/0xc8
[  118.598787]  [<ffffffff811bedec>] dump_header+0x5b/0x3b0
[  118.599979]  [<ffffffff810be8f9>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[  118.601421]  [<ffffffff810be9cd>] ? trace_hardirqs_on+0xd/0x10
[  118.602713]  [<ffffffff811447f6>] oom_kill_process+0x366/0x550
[  118.604882]  [<ffffffff81144c1f>] out_of_memory+0x1ef/0x5a0
[  118.606940]  [<ffffffff81144cdd>] ? out_of_memory+0x2ad/0x5a0
[  118.608275]  [<ffffffff8114a63b>] __alloc_pages_nodemask+0xb3b/0xd80
[  118.609698]  [<ffffffff810be800>] ? mark_held_locks+0x90/0x90
[  118.611166]  [<ffffffff8114aa3c>] alloc_kmem_pages_node+0x4c/0xc0
[  118.612589]  [<ffffffff8106d661>] copy_process.part.33+0x131/0x1be0
[  118.614203]  [<ffffffff8111e20a>] ? __audit_syscall_entry+0xaa/0xf0
[  118.615689]  [<ffffffff810e8939>] ? current_kernel_time64+0xa9/0xc0
[  118.617151]  [<ffffffff8106f2db>] _do_fork+0xdb/0x5d0
[  118.618391]  [<ffffffff810030c1>] ? do_audit_syscall_entry+0x61/0x70
[  118.619875]  [<ffffffff81003254>] ? syscall_trace_enter_phase1+0x134/0x150
[  118.621642]  [<ffffffff810bae1a>] ? up_read+0x1a/0x40
[  118.622920]  [<ffffffff817093ce>] ? retint_user+0x18/0x23
[  118.624262]  [<ffffffff810035ec>] ? do_syscall_64+0x1c/0x180
[  118.625661]  [<ffffffff8106f854>] SyS_clone+0x14/0x20
[  118.626959]  [<ffffffff8100362d>] do_syscall_64+0x5d/0x180
[  118.628340]  [<ffffffff81708abf>] entry_SYSCALL64_slow_path+0x25/0x25
[  118.630002] Mem-Info:
[  118.630853] active_anon:27270 inactive_anon:2094 isolated_anon:0
[  118.630853]  active_file:253575 inactive_file:89021 isolated_file:22
[  118.630853]  unevictable:0 dirty:0 writeback:0 unstable:0
[  118.630853]  slab_reclaimable:14202 slab_unreclaimable:13906
[  118.630853]  mapped:1622 shmem:2162 pagetables:10587 bounce:0
[  118.630853]  free:5328 free_pcp:356 free_cma:0
[  118.639774] Node 0 DMA free:6904kB min:44kB low:52kB high:64kB active_anon:3280kB inactive_anon:156kB active_file:684kB inactive_file:2292kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:420kB shmem:164kB slab_reclaimable:564kB slab_unreclaimable:800kB kernel_stack:256kB pagetables:200kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  118.650132] lowmem_reserve[]: 0 1714 1714 1714
[  118.651763] Node 0 DMA32 free:14256kB min:5172kB low:6464kB high:7756kB active_anon:105924kB inactive_anon:8220kB active_file:1026268kB inactive_file:340844kB unevictable:0kB isolated(anon):0kB isolated(file):88kB present:2080640kB managed:1759460kB mlocked:0kB dirty:0kB writeback:0kB mapped:6436kB shmem:8484kB slab_reclaimable:56740kB slab_unreclaimable:54824kB kernel_stack:28112kB pagetables:42148kB unstable:0kB bounce:0kB free_pcp:1440kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  118.663101] lowmem_reserve[]: 0 0 0 0
[  118.664704] Node 0 DMA: 83*4kB (ME) 51*8kB (UME) 9*16kB (UME) 2*32kB (UM) 1*64kB (M) 4*128kB (UME) 5*256kB (UME) 2*512kB (UM) 1*1024kB (E) 1*2048kB (M) 0*4096kB = 6900kB
[  118.670166] Node 0 DMA32: 2327*4kB (ME) 621*8kB (M) 1*16kB (M) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 14292kB
[  118.673742] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB
[  118.676297] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[  118.678610] 344508 total pagecache pages
[  118.680163] 0 pages in swap cache
[  118.681567] Swap cache stats: add 0, delete 0, find 0/0
[  118.681567] Free swap  = 0kB
[  118.681568] Total swap = 0kB
[  118.681625] 524157 pages RAM
[  118.681625] 0 pages HighMem/MovableOnly
[  118.681625] 80316 pages reserved
[  118.681626] 0 pages hwpoisoned
[  120.117093] fork invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[  120.117097] fork cpuset=/ mems_allowed=0
[  120.117099] CPU: 0 PID: 9566 Comm: fork Not tainted 4.5.0-rc4-next-20160216+ #306
[  120.117100] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[  120.117102]  0000000000000286 00000000be6c9129 ffff880035dabad0 ffffffff8139e82d
[  120.117103]  0000000000000000 ffff880035dabd00 ffff880035dabb70 ffffffff811bedec
[  120.117104]  0000000000000206 ffffffff81810b70 ffff880035dabb10 ffffffff810be8f9
[  120.117104] Call Trace:
[  120.117111]  [<ffffffff8139e82d>] dump_stack+0x85/0xc8
[  120.117113]  [<ffffffff811bedec>] dump_header+0x5b/0x3b0
[  120.117116]  [<ffffffff810be8f9>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[  120.117117]  [<ffffffff810be9cd>] ? trace_hardirqs_on+0xd/0x10
[  120.117119]  [<ffffffff811447f6>] oom_kill_process+0x366/0x550
[  120.117121]  [<ffffffff81144c1f>] out_of_memory+0x1ef/0x5a0
[  120.117122]  [<ffffffff81144cdd>] ? out_of_memory+0x2ad/0x5a0
[  120.117123]  [<ffffffff8114a63b>] __alloc_pages_nodemask+0xb3b/0xd80
[  120.117124]  [<ffffffff810be800>] ? mark_held_locks+0x90/0x90
[  120.117125]  [<ffffffff8114aa3c>] alloc_kmem_pages_node+0x4c/0xc0
[  120.117128]  [<ffffffff8106d661>] copy_process.part.33+0x131/0x1be0
[  120.117130]  [<ffffffff8111e20a>] ? __audit_syscall_entry+0xaa/0xf0
[  120.117132]  [<ffffffff810e8939>] ? current_kernel_time64+0xa9/0xc0
[  120.117133]  [<ffffffff8106f2db>] _do_fork+0xdb/0x5d0
[  120.117136]  [<ffffffff810030c1>] ? do_audit_syscall_entry+0x61/0x70
[  120.117137]  [<ffffffff81003254>] ? syscall_trace_enter_phase1+0x134/0x150
[  120.117139]  [<ffffffff810bae1a>] ? up_read+0x1a/0x40
[  120.117142]  [<ffffffff817093ce>] ? retint_user+0x18/0x23
[  120.117143]  [<ffffffff810035ec>] ? do_syscall_64+0x1c/0x180
[  120.117144]  [<ffffffff8106f854>] SyS_clone+0x14/0x20
[  120.117145]  [<ffffffff8100362d>] do_syscall_64+0x5d/0x180
[  120.117147]  [<ffffffff81708abf>] entry_SYSCALL64_slow_path+0x25/0x25
[  120.117147] Mem-Info:
[  120.117150] active_anon:30895 inactive_anon:2094 isolated_anon:0
[  120.117150]  active_file:183306 inactive_file:118692 isolated_file:18
[  120.117150]  unevictable:0 dirty:47 writeback:0 unstable:0
[  120.117150]  slab_reclaimable:14405 slab_unreclaimable:22372
[  120.117150]  mapped:3101 shmem:2162 pagetables:20154 bounce:0
[  120.117150]  free:7231 free_pcp:108 free_cma:0
[  120.117154] Node 0 DMA free:6904kB min:44kB low:52kB high:64kB active_anon:1172kB inactive_anon:156kB active_file:684kB inactive_file:1356kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:420kB shmem:164kB slab_reclaimable:564kB slab_unreclaimable:2244kB kernel_stack:1376kB pagetables:436kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  120.117156] lowmem_reserve[]: 0 1714 1714 1714
[  120.117172] Node 0 DMA32 free:22020kB min:5172kB low:6464kB high:7756kB active_anon:122408kB inactive_anon:8220kB active_file:732540kB inactive_file:473412kB unevictable:0kB isolated(anon):0kB isolated(file):72kB present:2080640kB managed:1759460kB mlocked:0kB dirty:188kB writeback:0kB mapped:11984kB shmem:8484kB slab_reclaimable:57056kB slab_unreclaimable:87244kB kernel_stack:52048kB pagetables:80180kB unstable:0kB bounce:0kB free_pcp:432kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  120.117230] lowmem_reserve[]: 0 0 0 0
[  120.117238] Node 0 DMA: 46*4kB (UME) 82*8kB (ME) 37*16kB (UME) 13*32kB (M) 3*64kB (UM) 2*128kB (ME) 2*256kB (ME) 2*512kB (UM) 1*1024kB (E) 1*2048kB (M) 0*4096kB = 6904kB
[  120.117242] Node 0 DMA32: 709*4kB (UME) 2374*8kB (UME) 0*16kB 10*32kB (E) 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 22148kB
[  120.117244] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB
[  120.117244] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[  120.117245] 304244 total pagecache pages
[  120.117246] 0 pages in swap cache
[  120.117246] Swap cache stats: add 0, delete 0, find 0/0
[  120.117247] Free swap  = 0kB
[  120.117247] Total swap = 0kB
[  120.117248] 524157 pages RAM
[  120.117248] 0 pages HighMem/MovableOnly
[  120.117248] 80316 pages reserved
[  120.117249] 0 pages hwpoisoned
[  126.034913] fork invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[  126.034918] fork cpuset=/ mems_allowed=0
[  126.034920] CPU: 2 PID: 9566 Comm: fork Not tainted 4.5.0-rc4-next-20160216+ #306
[  126.034921] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/31/2013
[  126.034923]  0000000000000286 00000000be6c9129 ffff880035dabad0 ffffffff8139e82d
[  126.034925]  0000000000000000 ffff880035dabd00 ffff880035dabb70 ffffffff811bedec
[  126.034926]  0000000000000206 ffffffff81810b70 ffff880035dabb10 ffffffff810be8f9
[  126.034926] Call Trace:
[  126.034932]  [<ffffffff8139e82d>] dump_stack+0x85/0xc8
[  126.034935]  [<ffffffff811bedec>] dump_header+0x5b/0x3b0
[  126.034938]  [<ffffffff810be8f9>] ? trace_hardirqs_on_caller+0xf9/0x1c0
[  126.034939]  [<ffffffff810be9cd>] ? trace_hardirqs_on+0xd/0x10
[  126.034941]  [<ffffffff811447f6>] oom_kill_process+0x366/0x550
[  126.034943]  [<ffffffff81144c1f>] out_of_memory+0x1ef/0x5a0
[  126.034944]  [<ffffffff81144cdd>] ? out_of_memory+0x2ad/0x5a0
[  126.034945]  [<ffffffff8114a63b>] __alloc_pages_nodemask+0xb3b/0xd80
[  126.034947]  [<ffffffff810be800>] ? mark_held_locks+0x90/0x90
[  126.034948]  [<ffffffff8114aa3c>] alloc_kmem_pages_node+0x4c/0xc0
[  126.034950]  [<ffffffff8106d661>] copy_process.part.33+0x131/0x1be0
[  126.034952]  [<ffffffff8111e20a>] ? __audit_syscall_entry+0xaa/0xf0
[  126.034954]  [<ffffffff810e8939>] ? current_kernel_time64+0xa9/0xc0
[  126.034956]  [<ffffffff8106f2db>] _do_fork+0xdb/0x5d0
[  126.034958]  [<ffffffff810030c1>] ? do_audit_syscall_entry+0x61/0x70
[  126.034959]  [<ffffffff81003254>] ? syscall_trace_enter_phase1+0x134/0x150
[  126.034961]  [<ffffffff810bae1a>] ? up_read+0x1a/0x40
[  126.034965]  [<ffffffff817093ce>] ? retint_user+0x18/0x23
[  126.034965]  [<ffffffff810035ec>] ? do_syscall_64+0x1c/0x180
[  126.034967]  [<ffffffff8106f854>] SyS_clone+0x14/0x20
[  126.034968]  [<ffffffff8100362d>] do_syscall_64+0x5d/0x180
[  126.034969]  [<ffffffff81708abf>] entry_SYSCALL64_slow_path+0x25/0x25
[  126.034970] Mem-Info:
[  126.034973] active_anon:27060 inactive_anon:2093 isolated_anon:0
[  126.034973]  active_file:206123 inactive_file:85224 isolated_file:32
[  126.034973]  unevictable:0 dirty:47 writeback:0 unstable:0
[  126.034973]  slab_reclaimable:13214 slab_unreclaimable:26604
[  126.034973]  mapped:2421 shmem:2161 pagetables:24889 bounce:0
[  126.034973]  free:4649 free_pcp:30 free_cma:0
[  126.034986] Node 0 DMA free:6924kB min:44kB low:52kB high:64kB active_anon:1156kB inactive_anon:156kB active_file:728kB inactive_file:1060kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:368kB shmem:164kB slab_reclaimable:468kB slab_unreclaimable:2496kB kernel_stack:832kB pagetables:704kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:4 all_unreclaimable? no
[  126.034988] lowmem_reserve[]: 0 1714 1714 1714
[  126.034992] Node 0 DMA32 free:11672kB min:5172kB low:6464kB high:7756kB active_anon:107084kB inactive_anon:8216kB active_file:823764kB inactive_file:339836kB unevictable:0kB isolated(anon):0kB isolated(file):128kB present:2080640kB managed:1759460kB mlocked:0kB dirty:188kB writeback:0kB mapped:9316kB shmem:8480kB slab_reclaimable:52388kB slab_unreclaimable:103920kB kernel_stack:66016kB pagetables:98852kB unstable:0kB bounce:0kB free_pcp:120kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  126.034993] lowmem_reserve[]: 0 0 0 0
[  126.035000] Node 0 DMA: 70*4kB (UME) 16*8kB (UME) 59*16kB (UME) 34*32kB (ME) 14*64kB (UME) 2*128kB (UE) 1*256kB (E) 2*512kB (M) 2*1024kB (ME) 0*2048kB 0*4096kB = 6920kB
[  126.035005] Node 0 DMA32: 2372*4kB (UME) 290*8kB (UM) 3*16kB (U) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 11856kB
[  126.035006] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB
[  126.035006] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[  126.035007] 293674 total pagecache pages
[  126.035008] 0 pages in swap cache
[  126.035008] Swap cache stats: add 0, delete 0, find 0/0
[  126.035009] Free swap  = 0kB
[  126.035009] Total swap = 0kB
[  126.035010] 524157 pages RAM
[  126.035010] 0 pages HighMem/MovableOnly
[  126.035010] 80316 pages reserved
[  126.035011] 0 pages hwpoisoned

----------

author reviewer
#54 lore raw
Quoted 12 lines · Tetsuo Handa
Quoted 11 lines · Michal Hocko

On Sun 07-02-16 13:09:33, Tetsuo Handa wrote:
[...]

Quoted 5 lines

FYI, I again hit unexpected OOM-killer during genxref on linux-4.5-rc2 source. I think current patchset is too fragile to merge.
----------------------------------------

[ 3101.626995] smbd invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[ 3101.629148] smbd cpuset=/ mems_allowed=0

[...]

Quoted 2 lines
[ 3101.705887] Node 0 DMA: 75*4kB (UME) 69*8kB (UME) 43*16kB (UM) 23*32kB (UME) 8*64kB (UM) 4*128kB (UME) 2*256kB (UM) 0*512kB 1*1024kB (U) 1*2048kB (M) 0*4096kB = 6884kB
[ 3101.710581] Node 0 DMA32: 4513*4kB (UME) 15*8kB (U) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 18172kB
Quoted 6 lines
Quoted 2 lines

How come this is an unexpected OOM? There is clearly no order-2+ page available for the allocation request.

I used "unexpected" because there were only 35 userspace processes and genxref was the only process which did a lot of memory allocation (modulo kernel threads woken by file I/O) and most memory is reclaimable.

The memory is reclaimable but that doesn't mean that order-2 page block will get formed even if all of it gets reclaimed. The memory is simply too fragmented. That is why I think the OOM makes sense.

Quoted 12 lines
Quoted 4 lines
Quoted 2 lines
Quoted 1 line

Something like the following:

Yes, I do think we need something like it.

Was the patch applied?

No for above result.

A result with the patch ([email protected]) applied on today's linux-next is shown below. It seems that protection is not enough.

----------

[  118.584571] fork invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0

[...]

Quoted 2 lines
[  118.664704] Node 0 DMA: 83*4kB (ME) 51*8kB (UME) 9*16kB (UME) 2*32kB (UM) 1*64kB (M) 4*128kB (UME) 5*256kB (UME) 2*512kB (UM) 1*1024kB (E) 1*2048kB (M) 0*4096kB = 6900kB
[  118.670166] Node 0 DMA32: 2327*4kB (ME) 621*8kB (M) 1*16kB (M) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 14292kB

[...]

Quoted 1 line
[  120.117093] fork invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0

[...]

Quoted 2 lines
[  120.117238] Node 0 DMA: 46*4kB (UME) 82*8kB (ME) 37*16kB (UME) 13*32kB (M) 3*64kB (UM) 2*128kB (ME) 2*256kB (ME) 2*512kB (UM) 1*1024kB (E) 1*2048kB (M) 0*4096kB = 6904kB
[  120.117242] Node 0 DMA32: 709*4kB (UME) 2374*8kB (UME) 0*16kB 10*32kB (E) 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 22148kB

[...]

Quoted 1 line
[  126.034913] fork invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0

[...]

Quoted 2 lines
[  126.035000] Node 0 DMA: 70*4kB (UME) 16*8kB (UME) 59*16kB (UME) 34*32kB (ME) 14*64kB (UME) 2*128kB (UE) 1*256kB (E) 2*512kB (M) 2*1024kB (ME) 0*2048kB 0*4096kB = 6920kB
[  126.035005] Node 0 DMA32: 2372*4kB (UME) 290*8kB (UM) 3*16kB (U) 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 11856kB

As you can see, in all cases we had order-2 requests and no order-2+ free blocks even after all the retries. I think the OOM is appropriate at that time. We could have tried N+1 times but we have to draw a line at some point of time. The reason why we do not have any high order block available is a completely different question IMO. Maybe the compaction just gets deferred and doesn't do anything. This would be interesting to investigate further of course. Anyway my point is that going OOM with the current fragmentation is simply the only choice.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#55 lore raw
In reply to Michal Hocko
Quoted 4 lines · Michal Hocko

Hi,
this thread went mostly quite. Are all the main concerns clarified? Are there any new concerns? Are there any objections to targeting this for the next merge window?

Sorry to say at this late date, but I do have one concern: hopefully you can tweak something somewhere, or point me to some tunable that I can adjust (I've not studied the patches, sorry).

This rework makes it impossible to run my tmpfs swapping loads:
they're soon OOM-killed when they ran forever before, so swapping does not get the exercise on mmotm that it used to. (But I'm not so arrogant as to expect you to optimize for my load!)

Maybe it's just that I'm using tmpfs, and there's code that's conscious of file and anon, but doesn't cope properly with the awkward shmem case.

(Of course, tmpfs is and always has been a problem for OOM-killing, given that it takes up memory, but none is freed by killing processes:
but although that is a tiresome problem, it's not what either of us is attacking here.)

Taking many of the irrelevancies out of my load, here's something you could try, first on v4.5-rc5 and then on mmotm.

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

This OOM detection rework significantly lowers the number of jobs which can be run in parallel without being OOM-killed. Which would be welcome if it were choosing to abort in place of thrashing, but the system was far from thrashing: j20 took a few seconds more than j6, and even j30 didn't take 50% longer.

(I have /proc/sys/vm/swappiness 100, if that matters.)

I hope there's an easy answer to this: thanks!
Hugh

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#56 lore raw

Hello,

Quoted 18 lines · Hugh Dickins
Quoted 4 lines · Michal Hocko

Hi,
this thread went mostly quite. Are all the main concerns clarified? Are there any new concerns? Are there any objections to targeting this for the next merge window?

Sorry to say at this late date, but I do have one concern: hopefully you can tweak something somewhere, or point me to some tunable that I can adjust (I've not studied the patches, sorry).

This rework makes it impossible to run my tmpfs swapping loads:
they're soon OOM-killed when they ran forever before, so swapping does not get the exercise on mmotm that it used to. (But I'm not so arrogant as to expect you to optimize for my load!)

Maybe it's just that I'm using tmpfs, and there's code that's conscious of file and anon, but doesn't cope properly with the awkward shmem case.

Quoted 8 lines

(Of course, tmpfs is and always has been a problem for OOM-killing, given that it takes up memory, but none is freed by killing processes:
but although that is a tiresome problem, it's not what either of us is attacking here.)

Taking many of the irrelevancies out of my load, here's something you could try, first on v4.5-rc5 and then on mmotm.

FWIW,

I have recently noticed the same change while testing zram-zsmalloc. next/mmots are much more likely to OOM-kill apps now. and, unlike before, I don't see a lot of shrinker->zsmalloc->zs_shrinker_scan() calls or swapouts, the kernel just oom-kills Xorg, etc.

the test script just creates a zram device (ext4 fs, lzo compression) and fills it with some data, nothing special.

OOM example:

[ 2392.663170] zram-test.sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[ 2392.663175] CPU: 1 PID: 9517 Comm: zram-test.sh Not tainted 4.5.0-rc5-next-20160225-dbg-00009-g334f687-dirty #190
[ 2392.663178]  0000000000000000 ffff88000b4efb88 ffffffff81237bac 0000000000000000
[ 2392.663181]  ffff88000b4efd28 ffff88000b4efbf8 ffffffff8113a077 ffff88000b4efba8
[ 2392.663184]  ffffffff81080e24 ffff88000b4efbc8 ffffffff8151584e ffffffff81a48460
[ 2392.663187] Call Trace:
[ 2392.663191]  [<ffffffff81237bac>] dump_stack+0x67/0x90
[ 2392.663195]  [<ffffffff8113a077>] dump_header.isra.5+0x54/0x351
[ 2392.663197]  [<ffffffff81080e24>] ? trace_hardirqs_on+0xd/0xf
[ 2392.663201]  [<ffffffff8151584e>] ? _raw_spin_unlock_irqrestore+0x4b/0x60
[ 2392.663204]  [<ffffffff810f7ae7>] oom_kill_process+0x89/0x4ff
[ 2392.663206]  [<ffffffff810f8319>] out_of_memory+0x36c/0x387
[ 2392.663208]  [<ffffffff810fc9c2>] __alloc_pages_nodemask+0x9ba/0xaa8
[ 2392.663211]  [<ffffffff810fcca8>] alloc_kmem_pages_node+0x1b/0x1d
[ 2392.663213]  [<ffffffff81040216>] copy_process.part.9+0xfe/0x183f
[ 2392.663216]  [<ffffffff81041aea>] _do_fork+0xbd/0x5f1
[ 2392.663218]  [<ffffffff81117402>] ? __might_fault+0x40/0x8d
[ 2392.663220]  [<ffffffff81515f52>] ? entry_SYSCALL_64_fastpath+0x5/0xa8
[ 2392.663223]  [<ffffffff81001844>] ? do_syscall_64+0x18/0xe6
[ 2392.663224]  [<ffffffff810420a4>] SyS_clone+0x19/0x1b
[ 2392.663226]  [<ffffffff81001886>] do_syscall_64+0x5a/0xe6
[ 2392.663228]  [<ffffffff8151601a>] entry_SYSCALL64_slow_path+0x25/0x25
[ 2392.663230] Mem-Info:
[ 2392.663233] active_anon:87788 inactive_anon:69289 isolated_anon:0
                active_file:161111 inactive_file:320022 isolated_file:0
                unevictable:0 dirty:51 writeback:0 unstable:0
                slab_reclaimable:80335 slab_unreclaimable:5920
                mapped:30115 shmem:29235 pagetables:2589 bounce:0
                free:10949 free_pcp:189 free_cma:0
[ 2392.663239] DMA free:15096kB min:28kB low:40kB high:52kB active_anon:0kB inactive_anon:0kB active_file:32kB inactive_file:120kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15984kB managed:15900kB mlocked:0kB dirty:0kB writeback:0kB mapped:136kB shmem:0kB slab_reclaimable:48kB slab_unreclaimable:92kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 2392.663240] lowmem_reserve[]: 0 3031 3855 3855
[ 2392.663247] DMA32 free:22876kB min:6232kB low:9332kB high:12432kB active_anon:316384kB inactive_anon:172076kB active_file:512592kB inactive_file:1011992kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3194880kB managed:3107516kB mlocked:0kB dirty:148kB writeback:0kB mapped:93284kB shmem:90904kB slab_reclaimable:248836kB slab_unreclaimable:14620kB kernel_stack:2208kB pagetables:7796kB unstable:0kB bounce:0kB free_pcp:628kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:256 all_unreclaimable? no
[ 2392.663249] lowmem_reserve[]: 0 0 824 824
[ 2392.663256] Normal free:5824kB min:1696kB low:2540kB high:3384kB active_anon:34768kB inactive_anon:105080kB active_file:131820kB inactive_file:267720kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:917504kB managed:844512kB mlocked:0kB dirty:56kB writeback:0kB mapped:27040kB shmem:26036kB slab_reclaimable:72456kB slab_unreclaimable:8968kB kernel_stack:1296kB pagetables:2560kB unstable:0kB bounce:0kB free_pcp:128kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:128 all_unreclaimable? no
[ 2392.663257] lowmem_reserve[]: 0 0 0 0
[ 2392.663260] DMA: 4*4kB (M) 1*8kB (M) 4*16kB (ME) 1*32kB (M) 2*64kB (UE) 2*128kB (UE) 3*256kB (UME) 3*512kB (UME) 2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 15096kB
[ 2392.663284] DMA32: 5809*4kB (UME) 3*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 23260kB
[ 2392.663293] Normal: 1515*4kB (UME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 6060kB
[ 2392.663302] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 2392.663303] 510384 total pagecache pages
[ 2392.663305] 31 pages in swap cache
[ 2392.663306] Swap cache stats: add 113, delete 82, find 47/62
[ 2392.663307] Free swap  = 8388268kB
[ 2392.663308] Total swap = 8388604kB
[ 2392.663308] 1032092 pages RAM
[ 2392.663309] 0 pages HighMem/MovableOnly
[ 2392.663310] 40110 pages reserved
[ 2392.663311] 0 pages hwpoisoned
[ 2392.663312] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[ 2392.663316] [  149]     0   149     9683     1612      20       3        4             0 systemd-journal
[ 2392.663319] [  183]     0   183     8598     1103      19       3       18         -1000 systemd-udevd
[ 2392.663321] [  285]    81   285     8183      911      20       3        0          -900 dbus-daemon
[ 2392.663323] [  288]     0   288     3569      653      13       3        0             0 crond
[ 2392.663326] [  289]     0   289     3855      649      12       3        0             0 systemd-logind
[ 2392.663328] [  291]     0   291    22469      967      48       3        0             0 login
[ 2392.663330] [  299]  1000   299     8493     1140      21       3        0             0 systemd
[ 2392.663332] [  301]  1000   301    24226      416      47       3       20             0 (sd-pam)
[ 2392.663334] [  306]  1000   306     4471     1126      14       3        0             0 bash
[ 2392.663336] [  313]  1000   313     3717      739      13       3        0             0 startx
[ 2392.663339] [  335]  1000   335     3981      236      14       3        0             0 xinit
[ 2392.663341] [  336]  1000   336    47841    19104      94       3        0             0 Xorg
[ 2392.663343] [  338]  1000   338    39714     4302      80       3        0             0 openbox
[ 2392.663345] [  349]  1000   349    43472     3280      88       3        0             0 tint2
[ 2392.663347] [  355]  1000   355    34168     5710      57       3        0             0 urxvt
[ 2392.663349] [  356]  1000   356     4533     1248      15       3        0             0 bash
[ 2392.663351] [  435]     0   435     3691     2168      10       3        0             0 dhclient
[ 2392.663353] [  451]  1000   451     4445     1111      14       4        0             0 bash
[ 2392.663355] [  459]  1000   459    45577     6121      59       3        0             0 urxvt
[ 2392.663357] [  460]  1000   460     4445     1070      15       3        0             0 bash
[ 2392.663359] [  463]  1000   463     5207      728      16       3        0             0 tmux
[ 2392.663362] [  465]  1000   465     6276     1299      18       3        0             0 tmux
[ 2392.663364] [  466]  1000   466     4445     1113      14       3        0             0 bash
[ 2392.663366] [  473]  1000   473     4445     1087      15       3        0             0 bash
[ 2392.663368] [  476]  1000   476     5207      760      15       3        0             0 tmux
[ 2392.663370] [  477]  1000   477     4445     1080      14       3        0             0 bash
[ 2392.663372] [  484]  1000   484     4445     1076      14       3        0             0 bash
[ 2392.663374] [  487]  1000   487     4445     1129      14       3        0             0 bash
[ 2392.663376] [  490]  1000   490     4445     1115      14       3        0             0 bash
[ 2392.663378] [  493]  1000   493    10206     1135      24       3        0             0 top
[ 2392.663380] [  495]  1000   495     4445     1146      15       3        0             0 bash
[ 2392.663382] [  502]  1000   502     3745      814      13       3        0             0 coretemp-sensor
[ 2392.663385] [  536]  1000   536    27937     4429      53       3        0             0 urxvt
[ 2392.663387] [  537]  1000   537     4445     1092      14       3        0             0 bash
[ 2392.663389] [  543]  1000   543    29981     4138      53       3        0             0 urxvt
[ 2392.663391] [  544]  1000   544     4445     1095      14       3        0             0 bash
[ 2392.663393] [  549]  1000   549    29981     4132      53       3        0             0 urxvt
[ 2392.663395] [  550]  1000   550     4445     1121      13       3        0             0 bash
[ 2392.663397] [  555]  1000   555    45194     5728      62       3        0             0 urxvt
[ 2392.663399] [  556]  1000   556     4445     1116      14       3        0             0 bash
[ 2392.663401] [  561]  1000   561    30173     4317      51       3        0             0 urxvt
[ 2392.663403] [  562]  1000   562     4445     1075      14       3        0             0 bash
[ 2392.663405] [  586]  1000   586    57178     7499      65       4        0             0 urxvt
[ 2392.663408] [  587]  1000   587     4478     1156      14       3        0             0 bash
[ 2392.663410] [  593]     0   593    17836     1213      39       3        0             0 sudo
[ 2392.663412] [  594]     0   594   136671     1794     188       4        0             0 journalctl
[ 2392.663414] [  616]  1000   616    29981     4140      54       3        0             0 urxvt
[ 2392.663416] [  617]  1000   617     4445     1122      14       3        0             0 bash
[ 2392.663418] [  622]  1000   622    34169     8473      60       3        0             0 urxvt
[ 2392.663420] [  623]  1000   623     4445     1116      14       3        0             0 bash
[ 2392.663422] [  646]  1000   646     4445     1124      15       3        0             0 bash
[ 2392.663424] [  668]  1000   668     4445     1090      15       3        0             0 bash
[ 2392.663426] [  671]  1000   671     4445     1090      13       3        0             0 bash
[ 2392.663429] [  674]  1000   674     4445     1083      13       3        0             0 bash
[ 2392.663431] [  677]  1000   677     4445     1124      15       3        0             0 bash
[ 2392.663433] [  720]  1000   720     3717      707      12       3        0             0 build99
[ 2392.663435] [  721]  1000   721     9107     1244      21       3        0             0 ssh
[ 2392.663437] [  768]     0   768    17827     1292      40       3        0             0 sudo
[ 2392.663439] [  771]     0   771     4640      622      14       3        0             0 screen
[ 2392.663441] [  772]     0   772     4673      505      11       3        0             0 screen
[ 2392.663443] [  775]  1000   775     4445     1120      14       3        0             0 bash
[ 2392.663445] [  778]  1000   778     4445     1097      14       3        0             0 bash
[ 2392.663447] [  781]  1000   781     4445     1088      13       3        0             0 bash
[ 2392.663449] [  784]  1000   784     4445     1109      13       3        0             0 bash
[ 2392.663451] [  808]  1000   808   341606    79367     532       5        0             0 firefox
[ 2392.663454] [  845]  1000   845     8144      799      20       3        0             0 dbus-daemon
[ 2392.663456] [  852]  1000   852    83828     1216      31       4        0             0 at-spi-bus-laun
[ 2392.663458] [ 9064]  1000  9064     4478     1154      13       3        0             0 bash
[ 2392.663460] [ 9068]  1000  9068     4478     1135      15       3        0             0 bash
[ 2392.663462] [ 9460]  1000  9460    11128      767      26       3        0             0 su
[ 2392.663464] [ 9463]     0  9463     4474     1188      14       4        0             0 bash
[ 2392.663482] [ 9517]     0  9517     3750      830      13       3        0             0 zram-test.sh
[ 2392.663485] [ 9917]  1000  9917     4444     1124      14       3        0             0 bash
[ 2392.663487] [13623]  1000 13623     1764      186       9       3        0             0 sleep
[ 2392.663489] Out of memory: Kill process 808 (firefox) score 25 or sacrifice child
[ 2392.663769] Killed process 808 (firefox) total-vm:1366424kB, anon-rss:235572kB, file-rss:82320kB, shmem-rss:8kB
[ 2400.152464] zram-test.sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[ 2400.152470] CPU: 1 PID: 9517 Comm: zram-test.sh Not tainted 4.5.0-rc5-next-20160225-dbg-00009-g334f687-dirty #190
[ 2400.152473]  0000000000000000 ffff88000b4efb88 ffffffff81237bac 0000000000000000
[ 2400.152476]  ffff88000b4efd28 ffff88000b4efbf8 ffffffff8113a077 ffff88000b4efba8
[ 2400.152479]  ffffffff81080e24 ffff88000b4efbc8 ffffffff8151584e ffffffff81a48460
[ 2400.152481] Call Trace:
[ 2400.152487]  [<ffffffff81237bac>] dump_stack+0x67/0x90
[ 2400.152490]  [<ffffffff8113a077>] dump_header.isra.5+0x54/0x351
[ 2400.152493]  [<ffffffff81080e24>] ? trace_hardirqs_on+0xd/0xf
[ 2400.152496]  [<ffffffff8151584e>] ? _raw_spin_unlock_irqrestore+0x4b/0x60
[ 2400.152500]  [<ffffffff810f7ae7>] oom_kill_process+0x89/0x4ff
[ 2400.152502]  [<ffffffff810f8319>] out_of_memory+0x36c/0x387
[ 2400.152504]  [<ffffffff810fc9c2>] __alloc_pages_nodemask+0x9ba/0xaa8
[ 2400.152506]  [<ffffffff810fcca8>] alloc_kmem_pages_node+0x1b/0x1d
[ 2400.152509]  [<ffffffff81040216>] copy_process.part.9+0xfe/0x183f
[ 2400.152511]  [<ffffffff81083178>] ? lock_acquire+0x11f/0x1c7
[ 2400.152513]  [<ffffffff81041aea>] _do_fork+0xbd/0x5f1
[ 2400.152515]  [<ffffffff81117402>] ? __might_fault+0x40/0x8d
[ 2400.152517]  [<ffffffff81515f52>] ? entry_SYSCALL_64_fastpath+0x5/0xa8
[ 2400.152520]  [<ffffffff81001844>] ? do_syscall_64+0x18/0xe6
[ 2400.152522]  [<ffffffff810420a4>] SyS_clone+0x19/0x1b
[ 2400.152524]  [<ffffffff81001886>] do_syscall_64+0x5a/0xe6
[ 2400.152526]  [<ffffffff8151601a>] entry_SYSCALL64_slow_path+0x25/0x25
[ 2400.152527] Mem-Info:
[ 2400.152531] active_anon:37648 inactive_anon:59709 isolated_anon:0
                active_file:160072 inactive_file:275086 isolated_file:0
                unevictable:0 dirty:49 writeback:0 unstable:0
                slab_reclaimable:54096 slab_unreclaimable:5978
                mapped:13650 shmem:29234 pagetables:2058 bounce:0
                free:13017 free_pcp:134 free_cma:0
[ 2400.152536] DMA free:15096kB min:28kB low:40kB high:52kB active_anon:0kB inactive_anon:0kB active_file:32kB inactive_file:120kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15984kB managed:15900kB mlocked:0kB dirty:0kB writeback:0kB mapped:136kB shmem:0kB slab_reclaimable:48kB slab_unreclaimable:92kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 2400.152537] lowmem_reserve[]: 0 3031 3855 3855
[ 2400.152545] DMA32 free:31504kB min:6232kB low:9332kB high:12432kB active_anon:129548kB inactive_anon:172076kB active_file:508480kB inactive_file:872492kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3194880kB managed:3107516kB mlocked:0kB dirty:132kB writeback:0kB mapped:42296kB shmem:90900kB slab_reclaimable:165548kB slab_unreclaimable:14964kB kernel_stack:1712kB pagetables:6176kB unstable:0kB bounce:0kB free_pcp:428kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:424 all_unreclaimable? no
[ 2400.152546] lowmem_reserve[]: 0 0 824 824
[ 2400.152553] Normal free:5468kB min:1696kB low:2540kB high:3384kB active_anon:21044kB inactive_anon:66760kB active_file:131776kB inactive_file:227732kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:917504kB managed:844512kB mlocked:0kB dirty:64kB writeback:0kB mapped:12168kB shmem:26036kB slab_reclaimable:50788kB slab_unreclaimable:8856kB kernel_stack:912kB pagetables:2056kB unstable:0kB bounce:0kB free_pcp:108kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:160 all_unreclaimable? no
[ 2400.152555] lowmem_reserve[]: 0 0 0 0
[ 2400.152558] DMA: 4*4kB (M) 1*8kB (M) 4*16kB (ME) 1*32kB (M) 2*64kB (UE) 2*128kB (UE) 3*256kB (UME) 3*512kB (UME) 2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 15096kB
[ 2400.152573] DMA32: 7835*4kB (UME) 55*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 31780kB
[ 2400.152582] Normal: 1383*4kB (UM) 22*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 5708kB
[ 2400.152592] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 2400.152593] 464295 total pagecache pages
[ 2400.152594] 31 pages in swap cache
[ 2400.152595] Swap cache stats: add 113, delete 82, find 47/62
[ 2400.152596] Free swap  = 8388268kB
[ 2400.152597] Total swap = 8388604kB
[ 2400.152598] 1032092 pages RAM
[ 2400.152599] 0 pages HighMem/MovableOnly
[ 2400.152600] 40110 pages reserved
[ 2400.152600] 0 pages hwpoisoned
[ 2400.152601] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[ 2400.152605] [  149]     0   149     9683     1990      20       3        4             0 systemd-journal
[ 2400.152608] [  183]     0   183     8598     1103      19       3       18         -1000 systemd-udevd
[ 2400.152610] [  285]    81   285     8183      911      20       3        0          -900 dbus-daemon
[ 2400.152613] [  288]     0   288     3569      653      13       3        0             0 crond
[ 2400.152615] [  289]     0   289     3855      649      12       3        0             0 systemd-logind
[ 2400.152617] [  291]     0   291    22469      967      48       3        0             0 login
[ 2400.152619] [  299]  1000   299     8493     1140      21       3        0             0 systemd
[ 2400.152621] [  301]  1000   301    24226      416      47       3       20             0 (sd-pam)
[ 2400.152623] [  306]  1000   306     4471     1126      14       3        0             0 bash
[ 2400.152626] [  313]  1000   313     3717      739      13       3        0             0 startx
[ 2400.152628] [  335]  1000   335     3981      236      14       3        0             0 xinit
[ 2400.152630] [  336]  1000   336    47713    19103      93       3        0             0 Xorg
[ 2400.152632] [  338]  1000   338    39714     4302      80       3        0             0 openbox
[ 2400.152634] [  349]  1000   349    43472     3280      88       3        0             0 tint2
[ 2400.152636] [  355]  1000   355    34168     5754      58       3        0             0 urxvt
[ 2400.152638] [  356]  1000   356     4533     1248      15       3        0             0 bash
[ 2400.152640] [  435]     0   435     3691     2168      10       3        0             0 dhclient
[ 2400.152642] [  451]  1000   451     4445     1111      14       4        0             0 bash
[ 2400.152644] [  459]  1000   459    45577     6121      59       3        0             0 urxvt
[ 2400.152646] [  460]  1000   460     4445     1070      15       3        0             0 bash
[ 2400.152648] [  463]  1000   463     5207      728      16       3        0             0 tmux
[ 2400.152650] [  465]  1000   465     6276     1299      18       3        0             0 tmux
[ 2400.152653] [  466]  1000   466     4445     1113      14       3        0             0 bash
[ 2400.152655] [  473]  1000   473     4445     1087      15       3        0             0 bash
[ 2400.152657] [  476]  1000   476     5207      760      15       3        0             0 tmux
[ 2400.152659] [  477]  1000   477     4445     1080      14       3        0             0 bash
[ 2400.152661] [  484]  1000   484     4445     1076      14       3        0             0 bash
[ 2400.152663] [  487]  1000   487     4445     1129      14       3        0             0 bash
[ 2400.152665] [  490]  1000   490     4445     1115      14       3        0             0 bash
[ 2400.152667] [  493]  1000   493    10206     1135      24       3        0             0 top
[ 2400.152669] [  495]  1000   495     4445     1146      15       3        0             0 bash
[ 2400.152671] [  502]  1000   502     3745      814      13       3        0             0 coretemp-sensor
[ 2400.152673] [  536]  1000   536    27937     4429      53       3        0             0 urxvt
[ 2400.152675] [  537]  1000   537     4445     1092      14       3        0             0 bash
[ 2400.152677] [  543]  1000   543    29981     4138      53       3        0             0 urxvt
[ 2400.152680] [  544]  1000   544     4445     1095      14       3        0             0 bash
[ 2400.152682] [  549]  1000   549    29981     4132      53       3        0             0 urxvt
[ 2400.152684] [  550]  1000   550     4445     1121      13       3        0             0 bash
[ 2400.152686] [  555]  1000   555    45194     5728      62       3        0             0 urxvt
[ 2400.152688] [  556]  1000   556     4445     1116      14       3        0             0 bash
[ 2400.152690] [  561]  1000   561    30173     4317      51       3        0             0 urxvt
[ 2400.152692] [  562]  1000   562     4445     1075      14       3        0             0 bash
[ 2400.152694] [  586]  1000   586    57178     7499      65       4        0             0 urxvt
[ 2400.152696] [  587]  1000   587     4478     1156      14       3        0             0 bash
[ 2400.152698] [  593]     0   593    17836     1213      39       3        0             0 sudo
[ 2400.152700] [  594]     0   594   136671     1794     188       4        0             0 journalctl
[ 2400.152702] [  616]  1000   616    29981     4140      54       3        0             0 urxvt
[ 2400.152705] [  617]  1000   617     4445     1122      14       3        0             0 bash
[ 2400.152707] [  622]  1000   622    34169     8473      60       3        0             0 urxvt
[ 2400.152709] [  623]  1000   623     4445     1116      14       3        0             0 bash
[ 2400.152711] [  646]  1000   646     4445     1124      15       3        0             0 bash
[ 2400.152713] [  668]  1000   668     4445     1090      15       3        0             0 bash
[ 2400.152715] [  671]  1000   671     4445     1090      13       3        0             0 bash
[ 2400.152717] [  674]  1000   674     4445     1083      13       3        0             0 bash
[ 2400.152719] [  677]  1000   677     4445     1124      15       3        0             0 bash
[ 2400.152721] [  720]  1000   720     3717      707      12       3        0             0 build99
[ 2400.152723] [  721]  1000   721     9107     1244      21       3        0             0 ssh
[ 2400.152725] [  768]     0   768    17827     1292      40       3        0             0 sudo
[ 2400.152727] [  771]     0   771     4640      622      14       3        0             0 screen
[ 2400.152729] [  772]     0   772     4673      505      11       3        0             0 screen
[ 2400.152731] [  775]  1000   775     4445     1120      14       3        0             0 bash
[ 2400.152733] [  778]  1000   778     4445     1097      14       3        0             0 bash
[ 2400.152735] [  781]  1000   781     4445     1088      13       3        0             0 bash
[ 2400.152737] [  784]  1000   784     4445     1109      13       3        0             0 bash
[ 2400.152740] [  845]  1000   845     8144      799      20       3        0             0 dbus-daemon
[ 2400.152742] [  852]  1000   852    83828     1216      31       4        0             0 at-spi-bus-laun
[ 2400.152744] [ 9064]  1000  9064     4478     1154      13       3        0             0 bash
[ 2400.152746] [ 9068]  1000  9068     4478     1135      15       3        0             0 bash
[ 2400.152748] [ 9460]  1000  9460    11128      767      26       3        0             0 su
[ 2400.152750] [ 9463]     0  9463     4474     1188      14       4        0             0 bash
[ 2400.152752] [ 9517]     0  9517     3783      832      13       3        0             0 zram-test.sh
[ 2400.152754] [ 9917]  1000  9917     4444     1124      14       3        0             0 bash
[ 2400.152757] [14052]  1000 14052     1764      162       9       3        0             0 sleep
[ 2400.152758] Out of memory: Kill process 336 (Xorg) score 6 or sacrifice child
[ 2400.152767] Killed process 336 (Xorg) total-vm:190852kB, anon-rss:58728kB, file-rss:17684kB, shmem-rss:0kB
[ 2400.161723] oom_reaper: reaped process 336 (Xorg), now anon-rss:0kB, file-rss:0kB, shmem-rss:0lB
$ free
              total        used        free      shared  buff/cache   available
Mem:        3967928     1563132      310548      116936     2094248     2207584
Swap:       8388604         332     8388272
	-ss

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#57 lore raw
Quoted 329 lines · Hugh Dickins
Quoted 18 lines · Michal Hocko
Quoted 4 lines

Hi,
this thread went mostly quite. Are all the main concerns clarified? Are there any new concerns? Are there any objections to targeting this for the next merge window?

Sorry to say at this late date, but I do have one concern: hopefully you can tweak something somewhere, or point me to some tunable that I can adjust (I've not studied the patches, sorry).

This rework makes it impossible to run my tmpfs swapping loads:
they're soon OOM-killed when they ran forever before, so swapping does not get the exercise on mmotm that it used to. (But I'm not so arrogant as to expect you to optimize for my load!)

Maybe it's just that I'm using tmpfs, and there's code that's conscious of file and anon, but doesn't cope properly with the awkward shmem case.

Quoted 8 lines

(Of course, tmpfs is and always has been a problem for OOM-killing, given that it takes up memory, but none is freed by killing processes:
but although that is a tiresome problem, it's not what either of us is attacking here.)

Taking many of the irrelevancies out of my load, here's something you could try, first on v4.5-rc5 and then on mmotm.

FWIW,

I have recently noticed the same change while testing zram-zsmalloc. next/mmots are much more likely to OOM-kill apps now. and, unlike before, I don't see a lot of shrinker->zsmalloc->zs_shrinker_scan() calls or swapouts, the kernel just oom-kills Xorg, etc.

the test script just creates a zram device (ext4 fs, lzo compression) and fills it with some data, nothing special.

OOM example:

[ 2392.663170] zram-test.sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2,
oom_score_adj=0
[ 2392.663175] CPU: 1 PID: 9517 Comm: zram-test.sh Not tainted 4.5.0-rc5-next-20160225-dbg-00009-g334f687-dirty #190
[ 2392.663178]  0000000000000000 ffff88000b4efb88 ffffffff81237bac 0000000000000000
[ 2392.663181]  ffff88000b4efd28 ffff88000b4efbf8 ffffffff8113a077 ffff88000b4efba8
[ 2392.663184]  ffffffff81080e24 ffff88000b4efbc8 ffffffff8151584e ffffffff81a48460
[ 2392.663187] Call Trace:
[ 2392.663191]  [<ffffffff81237bac>] dump_stack+0x67/0x90
[ 2392.663195]  [<ffffffff8113a077>] dump_header.isra.5+0x54/0x351
[ 2392.663197]  [<ffffffff81080e24>] ? trace_hardirqs_on+0xd/0xf
[ 2392.663201]  [<ffffffff8151584e>] ? _raw_spin_unlock_irqrestore+0x4b/0x60
[ 2392.663204]  [<ffffffff810f7ae7>] oom_kill_process+0x89/0x4ff
[ 2392.663206]  [<ffffffff810f8319>] out_of_memory+0x36c/0x387
[ 2392.663208]  [<ffffffff810fc9c2>] __alloc_pages_nodemask+0x9ba/0xaa8
[ 2392.663211]  [<ffffffff810fcca8>] alloc_kmem_pages_node+0x1b/0x1d
[ 2392.663213]  [<ffffffff81040216>] copy_process.part.9+0xfe/0x183f
[ 2392.663216]  [<ffffffff81041aea>] _do_fork+0xbd/0x5f1
[ 2392.663218]  [<ffffffff81117402>] ? __might_fault+0x40/0x8d
[ 2392.663220]  [<ffffffff81515f52>] ? entry_SYSCALL_64_fastpath+0x5/0xa8
[ 2392.663223]  [<ffffffff81001844>] ? do_syscall_64+0x18/0xe6
[ 2392.663224]  [<ffffffff810420a4>] SyS_clone+0x19/0x1b
[ 2392.663226]  [<ffffffff81001886>] do_syscall_64+0x5a/0xe6
[ 2392.663228]  [<ffffffff8151601a>] entry_SYSCALL64_slow_path+0x25/0x25
[ 2392.663230] Mem-Info:
[ 2392.663233] active_anon:87788 inactive_anon:69289 isolated_anon:0
                active_file:161111 inactive_file:320022 isolated_file:0
                unevictable:0 dirty:51 writeback:0 unstable:0
                slab_reclaimable:80335 slab_unreclaimable:5920
                mapped:30115 shmem:29235 pagetables:2589 bounce:0
                free:10949 free_pcp:189 free_cma:0
[ 2392.663239] DMA free:15096kB min:28kB low:40kB high:52kB active_anon:0kB inactive_anon:0kB active_file:32kB
inactive_file:120kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15984kB managed:15900kB mlocked:0kB dirty:0kB
writeback:0kB mapped:136kB shmem:0kB slab_reclaimable:48kB slab_unreclaimable:92kB kernel_stack:0kB pagetables:0kB
unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 2392.663240] lowmem_reserve[]: 0 3031 3855 3855
[ 2392.663247] DMA32 free:22876kB min:6232kB low:9332kB high:12432kB active_anon:316384kB inactive_anon:172076kB

active_file:512592kB inactive_file:1011992kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3194880kB managed:3107516kB mlocked:0kB dirty:148kB writeback:0kB mapped:93284kB shmem:90904kB slab_reclaimable:248836kB slab_unreclaimable:14620kB kernel_stack:2208kB pagetables:7796kB unstable:0kB bounce:0kB free_pcp:628kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:256 all_unreclaimable? no

[ 2392.663249] lowmem_reserve[]: 0 0 824 824
[ 2392.663256] Normal free:5824kB min:1696kB low:2540kB high:3384kB active_anon:34768kB inactive_anon:105080kB

active_file:131820kB inactive_file:267720kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:917504kB managed:844512kB mlocked:0kB dirty:56kB writeback:0kB mapped:27040kB shmem:26036kB slab_reclaimable:72456kB slab_unreclaimable:8968kB kernel_stack:1296kB pagetables:2560kB unstable:0kB bounce:0kB free_pcp:128kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:128 all_unreclaimable? no

[ 2392.663257] lowmem_reserve[]: 0 0 0 0
[ 2392.663260] DMA: 4*4kB (M) 1*8kB (M) 4*16kB (ME) 1*32kB (M) 2*64kB (UE) 2*128kB (UE) 3*256kB (UME) 3*512kB (UME)
2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 15096kB
[ 2392.663284] DMA32: 5809*4kB (UME) 3*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB =
23260kB
[ 2392.663293] Normal: 1515*4kB (UME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB =
6060kB
[ 2392.663302] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 2392.663303] 510384 total pagecache pages
[ 2392.663305] 31 pages in swap cache
[ 2392.663306] Swap cache stats: add 113, delete 82, find 47/62
[ 2392.663307] Free swap  = 8388268kB
[ 2392.663308] Total swap = 8388604kB
[ 2392.663308] 1032092 pages RAM
[ 2392.663309] 0 pages HighMem/MovableOnly
[ 2392.663310] 40110 pages reserved
[ 2392.663311] 0 pages hwpoisoned
[ 2392.663312] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[ 2392.663316] [  149]     0   149     9683     1612      20       3        4             0 systemd-journal
[ 2392.663319] [  183]     0   183     8598     1103      19       3       18         -1000 systemd-udevd
[ 2392.663321] [  285]    81   285     8183      911      20       3        0          -900 dbus-daemon
[ 2392.663323] [  288]     0   288     3569      653      13       3        0             0 crond
[ 2392.663326] [  289]     0   289     3855      649      12       3        0             0 systemd-logind
[ 2392.663328] [  291]     0   291    22469      967      48       3        0             0 login
[ 2392.663330] [  299]  1000   299     8493     1140      21       3        0             0 systemd
[ 2392.663332] [  301]  1000   301    24226      416      47       3       20             0 (sd-pam)
[ 2392.663334] [  306]  1000   306     4471     1126      14       3        0             0 bash
[ 2392.663336] [  313]  1000   313     3717      739      13       3        0             0 startx
[ 2392.663339] [  335]  1000   335     3981      236      14       3        0             0 xinit
[ 2392.663341] [  336]  1000   336    47841    19104      94       3        0             0 Xorg
[ 2392.663343] [  338]  1000   338    39714     4302      80       3        0             0 openbox
[ 2392.663345] [  349]  1000   349    43472     3280      88       3        0             0 tint2
[ 2392.663347] [  355]  1000   355    34168     5710      57       3        0             0 urxvt
[ 2392.663349] [  356]  1000   356     4533     1248      15       3        0             0 bash
[ 2392.663351] [  435]     0   435     3691     2168      10       3        0             0 dhclient
[ 2392.663353] [  451]  1000   451     4445     1111      14       4        0             0 bash
[ 2392.663355] [  459]  1000   459    45577     6121      59       3        0             0 urxvt
[ 2392.663357] [  460]  1000   460     4445     1070      15       3        0             0 bash
[ 2392.663359] [  463]  1000   463     5207      728      16       3        0             0 tmux
[ 2392.663362] [  465]  1000   465     6276     1299      18       3        0             0 tmux
[ 2392.663364] [  466]  1000   466     4445     1113      14       3        0             0 bash
[ 2392.663366] [  473]  1000   473     4445     1087      15       3        0             0 bash
[ 2392.663368] [  476]  1000   476     5207      760      15       3        0             0 tmux
[ 2392.663370] [  477]  1000   477     4445     1080      14       3        0             0 bash
[ 2392.663372] [  484]  1000   484     4445     1076      14       3        0             0 bash
[ 2392.663374] [  487]  1000   487     4445     1129      14       3        0             0 bash
[ 2392.663376] [  490]  1000   490     4445     1115      14       3        0             0 bash
[ 2392.663378] [  493]  1000   493    10206     1135      24       3        0             0 top
[ 2392.663380] [  495]  1000   495     4445     1146      15       3        0             0 bash
[ 2392.663382] [  502]  1000   502     3745      814      13       3        0             0 coretemp-sensor
[ 2392.663385] [  536]  1000   536    27937     4429      53       3        0             0 urxvt
[ 2392.663387] [  537]  1000   537     4445     1092      14       3        0             0 bash
[ 2392.663389] [  543]  1000   543    29981     4138      53       3        0             0 urxvt
[ 2392.663391] [  544]  1000   544     4445     1095      14       3        0             0 bash
[ 2392.663393] [  549]  1000   549    29981     4132      53       3        0             0 urxvt
[ 2392.663395] [  550]  1000   550     4445     1121      13       3        0             0 bash
[ 2392.663397] [  555]  1000   555    45194     5728      62       3        0             0 urxvt
[ 2392.663399] [  556]  1000   556     4445     1116      14       3        0             0 bash
[ 2392.663401] [  561]  1000   561    30173     4317      51       3        0             0 urxvt
[ 2392.663403] [  562]  1000   562     4445     1075      14       3        0             0 bash
[ 2392.663405] [  586]  1000   586    57178     7499      65       4        0             0 urxvt
[ 2392.663408] [  587]  1000   587     4478     1156      14       3        0             0 bash
[ 2392.663410] [  593]     0   593    17836     1213      39       3        0             0 sudo
[ 2392.663412] [  594]     0   594   136671     1794     188       4        0             0 journalctl
[ 2392.663414] [  616]  1000   616    29981     4140      54       3        0             0 urxvt
[ 2392.663416] [  617]  1000   617     4445     1122      14       3        0             0 bash
[ 2392.663418] [  622]  1000   622    34169     8473      60       3        0             0 urxvt
[ 2392.663420] [  623]  1000   623     4445     1116      14       3        0             0 bash
[ 2392.663422] [  646]  1000   646     4445     1124      15       3        0             0 bash
[ 2392.663424] [  668]  1000   668     4445     1090      15       3        0             0 bash
[ 2392.663426] [  671]  1000   671     4445     1090      13       3        0             0 bash
[ 2392.663429] [  674]  1000   674     4445     1083      13       3        0             0 bash
[ 2392.663431] [  677]  1000   677     4445     1124      15       3        0             0 bash
[ 2392.663433] [  720]  1000   720     3717      707      12       3        0             0 build99
[ 2392.663435] [  721]  1000   721     9107     1244      21       3        0             0 ssh
[ 2392.663437] [  768]     0   768    17827     1292      40       3        0             0 sudo
[ 2392.663439] [  771]     0   771     4640      622      14       3        0             0 screen
[ 2392.663441] [  772]     0   772     4673      505      11       3        0             0 screen
[ 2392.663443] [  775]  1000   775     4445     1120      14       3        0             0 bash
[ 2392.663445] [  778]  1000   778     4445     1097      14       3        0             0 bash
[ 2392.663447] [  781]  1000   781     4445     1088      13       3        0             0 bash
[ 2392.663449] [  784]  1000   784     4445     1109      13       3        0             0 bash
[ 2392.663451] [  808]  1000   808   341606    79367     532       5        0             0 firefox
[ 2392.663454] [  845]  1000   845     8144      799      20       3        0             0 dbus-daemon
[ 2392.663456] [  852]  1000   852    83828     1216      31       4        0             0 at-spi-bus-laun
[ 2392.663458] [ 9064]  1000  9064     4478     1154      13       3        0             0 bash
[ 2392.663460] [ 9068]  1000  9068     4478     1135      15       3        0             0 bash
[ 2392.663462] [ 9460]  1000  9460    11128      767      26       3        0             0 su
[ 2392.663464] [ 9463]     0  9463     4474     1188      14       4        0             0 bash
[ 2392.663482] [ 9517]     0  9517     3750      830      13       3        0             0 zram-test.sh
[ 2392.663485] [ 9917]  1000  9917     4444     1124      14       3        0             0 bash
[ 2392.663487] [13623]  1000 13623     1764      186       9       3        0             0 sleep
[ 2392.663489] Out of memory: Kill process 808 (firefox) score 25 or sacrifice child
[ 2392.663769] Killed process 808 (firefox) total-vm:1366424kB, anon-rss:235572kB, file-rss:82320kB, shmem-rss:8kB
[ 2400.152464] zram-test.sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2,
oom_score_adj=0
[ 2400.152470] CPU: 1 PID: 9517 Comm: zram-test.sh Not tainted 4.5.0-rc5-next-20160225-dbg-00009-g334f687-dirty #190
[ 2400.152473]  0000000000000000 ffff88000b4efb88 ffffffff81237bac 0000000000000000
[ 2400.152476]  ffff88000b4efd28 ffff88000b4efbf8 ffffffff8113a077 ffff88000b4efba8
[ 2400.152479]  ffffffff81080e24 ffff88000b4efbc8 ffffffff8151584e ffffffff81a48460
[ 2400.152481] Call Trace:
[ 2400.152487]  [<ffffffff81237bac>] dump_stack+0x67/0x90
[ 2400.152490]  [<ffffffff8113a077>] dump_header.isra.5+0x54/0x351
[ 2400.152493]  [<ffffffff81080e24>] ? trace_hardirqs_on+0xd/0xf
[ 2400.152496]  [<ffffffff8151584e>] ? _raw_spin_unlock_irqrestore+0x4b/0x60
[ 2400.152500]  [<ffffffff810f7ae7>] oom_kill_process+0x89/0x4ff
[ 2400.152502]  [<ffffffff810f8319>] out_of_memory+0x36c/0x387
[ 2400.152504]  [<ffffffff810fc9c2>] __alloc_pages_nodemask+0x9ba/0xaa8
[ 2400.152506]  [<ffffffff810fcca8>] alloc_kmem_pages_node+0x1b/0x1d
[ 2400.152509]  [<ffffffff81040216>] copy_process.part.9+0xfe/0x183f
[ 2400.152511]  [<ffffffff81083178>] ? lock_acquire+0x11f/0x1c7
[ 2400.152513]  [<ffffffff81041aea>] _do_fork+0xbd/0x5f1
[ 2400.152515]  [<ffffffff81117402>] ? __might_fault+0x40/0x8d
[ 2400.152517]  [<ffffffff81515f52>] ? entry_SYSCALL_64_fastpath+0x5/0xa8
[ 2400.152520]  [<ffffffff81001844>] ? do_syscall_64+0x18/0xe6
[ 2400.152522]  [<ffffffff810420a4>] SyS_clone+0x19/0x1b
[ 2400.152524]  [<ffffffff81001886>] do_syscall_64+0x5a/0xe6
[ 2400.152526]  [<ffffffff8151601a>] entry_SYSCALL64_slow_path+0x25/0x25
[ 2400.152527] Mem-Info:
[ 2400.152531] active_anon:37648 inactive_anon:59709 isolated_anon:0
                active_file:160072 inactive_file:275086 isolated_file:0
                unevictable:0 dirty:49 writeback:0 unstable:0
                slab_reclaimable:54096 slab_unreclaimable:5978
                mapped:13650 shmem:29234 pagetables:2058 bounce:0
                free:13017 free_pcp:134 free_cma:0
[ 2400.152536] DMA free:15096kB min:28kB low:40kB high:52kB active_anon:0kB inactive_anon:0kB active_file:32kB
inactive_file:120kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15984kB managed:15900kB mlocked:0kB dirty:0kB
writeback:0kB mapped:136kB shmem:0kB slab_reclaimable:48kB slab_unreclaimable:92kB kernel_stack:0kB pagetables:0kB
unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[ 2400.152537] lowmem_reserve[]: 0 3031 3855 3855
[ 2400.152545] DMA32 free:31504kB min:6232kB low:9332kB high:12432kB active_anon:129548kB inactive_anon:172076kB

active_file:508480kB inactive_file:872492kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3194880kB managed:3107516kB mlocked:0kB dirty:132kB writeback:0kB mapped:42296kB shmem:90900kB slab_reclaimable:165548kB slab_unreclaimable:14964kB kernel_stack:1712kB pagetables:6176kB unstable:0kB bounce:0kB free_pcp:428kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:424 all_unreclaimable? no

[ 2400.152546] lowmem_reserve[]: 0 0 824 824
[ 2400.152553] Normal free:5468kB min:1696kB low:2540kB high:3384kB active_anon:21044kB inactive_anon:66760kB

active_file:131776kB inactive_file:227732kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:917504kB managed:844512kB mlocked:0kB dirty:64kB writeback:0kB mapped:12168kB shmem:26036kB slab_reclaimable:50788kB slab_unreclaimable:8856kB kernel_stack:912kB pagetables:2056kB unstable:0kB bounce:0kB free_pcp:108kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:160 all_unreclaimable? no

[ 2400.152555] lowmem_reserve[]: 0 0 0 0
[ 2400.152558] DMA: 4*4kB (M) 1*8kB (M) 4*16kB (ME) 1*32kB (M) 2*64kB (UE) 2*128kB (UE) 3*256kB (UME) 3*512kB (UME)
2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 15096kB
[ 2400.152573] DMA32: 7835*4kB (UME) 55*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB =
31780kB
[ 2400.152582] Normal: 1383*4kB (UM) 22*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB =
5708kB
[ 2400.152592] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 2400.152593] 464295 total pagecache pages
[ 2400.152594] 31 pages in swap cache
[ 2400.152595] Swap cache stats: add 113, delete 82, find 47/62
[ 2400.152596] Free swap  = 8388268kB
[ 2400.152597] Total swap = 8388604kB
[ 2400.152598] 1032092 pages RAM
[ 2400.152599] 0 pages HighMem/MovableOnly
[ 2400.152600] 40110 pages reserved
[ 2400.152600] 0 pages hwpoisoned
[ 2400.152601] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[ 2400.152605] [  149]     0   149     9683     1990      20       3        4             0 systemd-journal
[ 2400.152608] [  183]     0   183     8598     1103      19       3       18         -1000 systemd-udevd
[ 2400.152610] [  285]    81   285     8183      911      20       3        0          -900 dbus-daemon
[ 2400.152613] [  288]     0   288     3569      653      13       3        0             0 crond
[ 2400.152615] [  289]     0   289     3855      649      12       3        0             0 systemd-logind
[ 2400.152617] [  291]     0   291    22469      967      48       3        0             0 login
[ 2400.152619] [  299]  1000   299     8493     1140      21       3        0             0 systemd
[ 2400.152621] [  301]  1000   301    24226      416      47       3       20             0 (sd-pam)
[ 2400.152623] [  306]  1000   306     4471     1126      14       3        0             0 bash
[ 2400.152626] [  313]  1000   313     3717      739      13       3        0             0 startx
[ 2400.152628] [  335]  1000   335     3981      236      14       3        0             0 xinit
[ 2400.152630] [  336]  1000   336    47713    19103      93       3        0             0 Xorg
[ 2400.152632] [  338]  1000   338    39714     4302      80       3        0             0 openbox
[ 2400.152634] [  349]  1000   349    43472     3280      88       3        0             0 tint2
[ 2400.152636] [  355]  1000   355    34168     5754      58       3        0             0 urxvt
[ 2400.152638] [  356]  1000   356     4533     1248      15       3        0             0 bash
[ 2400.152640] [  435]     0   435     3691     2168      10       3        0             0 dhclient
[ 2400.152642] [  451]  1000   451     4445     1111      14       4        0             0 bash
[ 2400.152644] [  459]  1000   459    45577     6121      59       3        0             0 urxvt
[ 2400.152646] [  460]  1000   460     4445     1070      15       3        0             0 bash
[ 2400.152648] [  463]  1000   463     5207      728      16       3        0             0 tmux
[ 2400.152650] [  465]  1000   465     6276     1299      18       3        0             0 tmux
[ 2400.152653] [  466]  1000   466     4445     1113      14       3        0             0 bash
[ 2400.152655] [  473]  1000   473     4445     1087      15       3        0             0 bash
[ 2400.152657] [  476]  1000   476     5207      760      15       3        0             0 tmux
[ 2400.152659] [  477]  1000   477     4445     1080      14       3        0             0 bash
[ 2400.152661] [  484]  1000   484     4445     1076      14       3        0             0 bash
[ 2400.152663] [  487]  1000   487     4445     1129      14       3        0             0 bash
[ 2400.152665] [  490]  1000   490     4445     1115      14       3        0             0 bash
[ 2400.152667] [  493]  1000   493    10206     1135      24       3        0             0 top
[ 2400.152669] [  495]  1000   495     4445     1146      15       3        0             0 bash
[ 2400.152671] [  502]  1000   502     3745      814      13       3        0             0 coretemp-sensor
[ 2400.152673] [  536]  1000   536    27937     4429      53       3        0             0 urxvt
[ 2400.152675] [  537]  1000   537     4445     1092      14       3        0             0 bash
[ 2400.152677] [  543]  1000   543    29981     4138      53       3        0             0 urxvt
[ 2400.152680] [  544]  1000   544     4445     1095      14       3        0             0 bash
[ 2400.152682] [  549]  1000   549    29981     4132      53       3        0             0 urxvt
[ 2400.152684] [  550]  1000   550     4445     1121      13       3        0             0 bash
[ 2400.152686] [  555]  1000   555    45194     5728      62       3        0             0 urxvt
[ 2400.152688] [  556]  1000   556     4445     1116      14       3        0             0 bash
[ 2400.152690] [  561]  1000   561    30173     4317      51       3        0             0 urxvt
[ 2400.152692] [  562]  1000   562     4445     1075      14       3        0             0 bash
[ 2400.152694] [  586]  1000   586    57178     7499      65       4        0             0 urxvt
[ 2400.152696] [  587]  1000   587     4478     1156      14       3        0             0 bash
[ 2400.152698] [  593]     0   593    17836     1213      39       3        0             0 sudo
[ 2400.152700] [  594]     0   594   136671     1794     188       4        0             0 journalctl
[ 2400.152702] [  616]  1000   616    29981     4140      54       3        0             0 urxvt
[ 2400.152705] [  617]  1000   617     4445     1122      14       3        0             0 bash
[ 2400.152707] [  622]  1000   622    34169     8473      60       3        0             0 urxvt
[ 2400.152709] [  623]  1000   623     4445     1116      14       3        0             0 bash
[ 2400.152711] [  646]  1000   646     4445     1124      15       3        0             0 bash
[ 2400.152713] [  668]  1000   668     4445     1090      15       3        0             0 bash
[ 2400.152715] [  671]  1000   671     4445     1090      13       3        0             0 bash
[ 2400.152717] [  674]  1000   674     4445     1083      13       3        0             0 bash
[ 2400.152719] [  677]  1000   677     4445     1124      15       3        0             0 bash
[ 2400.152721] [  720]  1000   720     3717      707      12       3        0             0 build99
[ 2400.152723] [  721]  1000   721     9107     1244      21       3        0             0 ssh
[ 2400.152725] [  768]     0   768    17827     1292      40       3        0             0 sudo
[ 2400.152727] [  771]     0   771     4640      622      14       3        0             0 screen
[ 2400.152729] [  772]     0   772     4673      505      11       3        0             0 screen
[ 2400.152731] [  775]  1000   775     4445     1120      14       3        0             0 bash
[ 2400.152733] [  778]  1000   778     4445     1097      14       3        0             0 bash
[ 2400.152735] [  781]  1000   781     4445     1088      13       3        0             0 bash
[ 2400.152737] [  784]  1000   784     4445     1109      13       3        0             0 bash
[ 2400.152740] [  845]  1000   845     8144      799      20       3        0             0 dbus-daemon
[ 2400.152742] [  852]  1000   852    83828     1216      31       4        0             0 at-spi-bus-laun
[ 2400.152744] [ 9064]  1000  9064     4478     1154      13       3        0             0 bash
[ 2400.152746] [ 9068]  1000  9068     4478     1135      15       3        0             0 bash
[ 2400.152748] [ 9460]  1000  9460    11128      767      26       3        0             0 su
[ 2400.152750] [ 9463]     0  9463     4474     1188      14       4        0             0 bash
[ 2400.152752] [ 9517]     0  9517     3783      832      13       3        0             0 zram-test.sh
[ 2400.152754] [ 9917]  1000  9917     4444     1124      14       3        0             0 bash
[ 2400.152757] [14052]  1000 14052     1764      162       9       3        0             0 sleep
[ 2400.152758] Out of memory: Kill process 336 (Xorg) score 6 or sacrifice child
[ 2400.152767] Killed process 336 (Xorg) total-vm:190852kB, anon-rss:58728kB, file-rss:17684kB, shmem-rss:0kB
[ 2400.161723] oom_reaper: reaped process 336 (Xorg), now anon-rss:0kB, file-rss:0kB, shmem-rss:0lB
Quoted 5 lines
$ free
              total        used        free      shared  buff/cache   available
Mem:        3967928     1563132      310548      116936     2094248     2207584
Swap:       8388604         332     8388272

Hi Sergey

Thanks for your info.

Can you please schedule a run for the diff attached, in which non-expensive allocators are allowed to burn more CPU cycles.

thanks
Hillf

mm/page_alloc.c +2 -0
@@ -3113,6 +3113,8 @@ should_reclaim_retry(gfp_t gfp_mask, uns
 	struct zone *zone;
 	struct zoneref *z;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		no_progress_loops /= 2;
 	/*
 	 * Make sure we converge to OOM if we cannot make any progress
 	 * several times in the row.

--

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#58 lore raw

On Thu 25-02-16 17:17:45, Hillf Danton wrote:
[...]

Quoted 3 lines
Quoted 3 lines

OOM example:

[ 2392.663170] zram-test.sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2,  oom_score_adj=0

[...]

Quoted 3 lines
Quoted 3 lines
[ 2392.663260] DMA: 4*4kB (M) 1*8kB (M) 4*16kB (ME) 1*32kB (M) 2*64kB (UE) 2*128kB (UE) 3*256kB (UME) 3*512kB (UME) 2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 15096kB
[ 2392.663284] DMA32: 5809*4kB (UME) 3*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 23260kB
[ 2392.663293] Normal: 1515*4kB (UME) 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 6060kB

[...]

Quoted 1 line
Quoted 1 line
[ 2400.152464] zram-test.sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0

[...]

Quoted 3 lines
Quoted 3 lines
[ 2400.152558] DMA: 4*4kB (M) 1*8kB (M) 4*16kB (ME) 1*32kB (M) 2*64kB (UE) 2*128kB (UE) 3*256kB (UME) 3*512kB (UME)  2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 15096kB
[ 2400.152573] DMA32: 7835*4kB (UME) 55*8kB (M) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 31780kB
[ 2400.152582] Normal: 1383*4kB (UM) 22*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB =  5708kB

[...]

Quoted 4 lines

Thanks for your info.

Can you please schedule a run for the diff attached, in which non-expensive allocators are allowed to burn more CPU cycles.

I do not think your patch will help. As you can see, both OOMs were for order-2 and there simply are no order-2+ free blocks usable for the allocation request so the watermark check will fail for all eligible zones and no_progress_loops is simply ignored. This is what I've tried to address by patch I have just posted as a reply to Hugh's email
http://lkml.kernel.org/r/[email protected]

Quoted 11 lines
mm/page_alloc.c +2 -0
@@ -3113,6 +3113,8 @@ should_reclaim_retry(gfp_t gfp_mask, uns
 	struct zone *zone;
 	struct zoneref *z;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		no_progress_loops /= 2;
 	/*
 	 * Make sure we converge to OOM if we cannot make any progress
 	 * several times in the row.
Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#59 lore raw
Quoted 11 lines
Quoted 3 lines

Can you please schedule a run for the diff attached, in which non-expensive allocators are allowed to burn more CPU cycles.

I do not think your patch will help. As you can see, both OOMs were for order-2 and there simply are no order-2+ free blocks usable for the allocation request so the watermark check will fail for all eligible zones and no_progress_loops is simply ignored. This is what I've tried to address by patch I have just posted as a reply to Hugh's email
http://lkml.kernel.org/r/[email protected]

Hm, Mr. Swap can tell us more.

Hillf

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#60 lore raw
Quoted 11 lines · Hillf Danton
Quoted 10 lines
Quoted 2 lines

Can you please schedule a run for the diff attached, in which non-expensive allocators are allowed to burn more CPU cycles.

I do not think your patch will help. As you can see, both OOMs were for order-2 and there simply are no order-2+ free blocks usable for the allocation request so the watermark check will fail for all eligible zones and no_progress_loops is simply ignored. This is what I've tried to address by patch I have just posted as a reply to Hugh's email
http://lkml.kernel.org/r/[email protected]

Hm, Mr. Swap can tell us more.

Hi,

after *preliminary testing* both patches seem to work. at least I don't see oom-kills and there are some swapouts.

Michal Hocko's
total used free shared buff/cache available

Mem:        3836880     2458020       35992      115984     1342868     1181484
Swap:       8388604        2008     8386596

total used free shared buff/cache available

Mem:        3836880     2459516       39616      115880     1337748     1180156
Swap:       8388604        2052     8386552

total used free shared buff/cache available

Mem:        3836880     2460584       33944      115880     1342352     1179004
Swap:       8388604        2132     8386472

...

Hillf Danton's
total used free shared buff/cache available

Mem:        3836880     1661000      554236      116448     1621644     1978872
Swap:       8388604        1548     8387056

total used free shared buff/cache available

Mem:        3836880     1660500      554740      116448     1621640     1979376
Swap:       8388604        1548     8387056

...

I'll do more tests tomorrow.

	-ss

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#61 lore raw
In reply to Hugh Dickins

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 10 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Quoted 8 lines

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

This OOM detection rework significantly lowers the number of jobs which can be run in parallel without being OOM-killed.

This all smells like pre mature OOM because of a high order allocation (order-2 for fork) which Tetuo has seen already. Sergey Senozhatsky is reporting order-2 OOMs as well. It is true that what we have in the mmomt right now is quite fragile if all order-N+ are completely depleted. That was the case for both Tetsuo and Sergey. I have tried to mitigate this at least to some degree by
http://lkml.kernel.org/r/[email protected] (below with the full changelog) but I haven't heard back whether it helped so I haven't posted the official patch yet.

I also suspect that something is not quite right with the compaction and it gives up too early even though we have quite a lot reclaimable pages. I do not have any numbers for that because I didn't have a load to reproduce this problem yet. I will try your setup and see what I can do about that. It would be great if you could give the patch below a try and see if it helps.
---

#62 lore raw
Quoted 18 lines · Michal Hocko

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 16 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

Quoted 5 lines
Quoted 2 lines

This OOM detection rework significantly lowers the number of jobs which can be run in parallel without being OOM-killed.

This all smells like pre mature OOM because of a high order allocation (order-2 for fork) which Tetuo has seen already. Sergey Senozhatsky is

You're absolutely right, and I'm ashamed not to have noticed that, nor your comments and patch earlier in this thread, before bothering you. Order 2 they are.

Quoted 8 lines

reporting order-2 OOMs as well. It is true that what we have in the mmomt right now is quite fragile if all order-N+ are completely depleted. That was the case for both Tetsuo and Sergey. I have tried to mitigate this at least to some degree by
http://lkml.kernel.org/r/[email protected] (below with the full changelog) but I haven't heard back whether it helped so I haven't posted the official patch yet.

Quoted 4 lines

I also suspect that something is not quite right with the compaction and it gives up too early even though we have quite a lot reclaimable pages. I do not have any numbers for that because I didn't have a load to reproduce this problem yet. I will try your setup and see what I can do

Thanks a lot.

Quoted 21 lines

about that. It would be great if you could give the patch below a try and see if it helps.
---
From d09de26cee148b4d8c486943b4e8f3bd7ad6f4be Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Thu, 4 Feb 2016 14:56:59 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and the OOM killer is just a matter of time but let's stick to our MAX_RECLAIM_RETRIES for the high order and not costly requests to make sure we do not fail prematurely.

Quoted 8 lines

This also means that we do not reset no_progress_loops at the __alloc_pages_slowpath for high order allocations to guarantee a bounded number of retries.

Longterm it would be much better to communicate with the compaction and retry only if the compaction considers it meaningfull.

It didn't really help, I'm afraid: it reduces the actual number of OOM kills which occur before the job is terminated, but doesn't stop the job from being terminated very soon.

I also tried Hillf's patch (separately) too, but as you expected, it didn't seem to make any difference.

(I haven't tried on the PowerMac G5 yet, since that's busy with other testing; but expect that to tell the same story.)

Hugh

Quoted 49 lines
---
 mm/page_alloc.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)
mm/page_alloc.c +16 -4
index 269a04f20927..f05aca36469b 100644
@@ -3106,6 +3106,18 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		}
 	}
 
+	/*
+	 * OK, so the watermak check has failed. Make sure we do all the
+	 * retries for !costly high order requests and hope that multiple
+	 * runs of compaction will generate some high order ones for us.
+	 *
+	 * XXX: ideally we should teach the compaction to try _really_ hard
+	 * if we are in the retry path - something like priority 0 for the
+	 * reclaim
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
+
 	return false;
 }
 
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		goto noretry;
 
 	/*
-	 * Costly allocations might have made a progress but this doesn't mean
-	 * their order will become available due to high fragmentation so do
-	 * not reset the no progress counter for them
+	 * High order allocations might have made a progress but this doesn't
+	 * mean their order will become available due to high fragmentation so
+	 * do not reset the no progress counter for them
 	 */
-	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+	if (did_some_progress && !order)
 		no_progress_loops = 0;
 	else
 		no_progress_loops++;
Signature · 5 lines
2.7.0

-- 
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#63 lore raw
Quoted 8 lines

It didn't really help, I'm afraid: it reduces the actual number of OOM kills which occur before the job is terminated, but doesn't stop the job from being terminated very soon.

I also tried Hillf's patch (separately) too, but as you expected, it didn't seem to make any difference.

Perhaps non-costly means NOFAIL as shown by folding the two patches into one. Can it make any sense?

thanks
Hillf

mm/page_alloc.c +2 -0
@@ -3113,6 +3113,8 @@ should_reclaim_retry(gfp_t gfp_mask, uns
 	struct zone *zone;
 	struct zoneref *z;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
 	/*
 	 * Make sure we converge to OOM if we cannot make any progress
 	 * several times in the row.

--

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#64 lore raw
Quoted 9 lines · Hillf Danton
Quoted 8 lines

It didn't really help, I'm afraid: it reduces the actual number of OOM kills which occur before the job is terminated, but doesn't stop the job from being terminated very soon.

I also tried Hillf's patch (separately) too, but as you expected, it didn't seem to make any difference.

Perhaps non-costly means NOFAIL as shown by folding the two

nofail only means that the page allocator doesn't return with NULL. OOM killer is still not put aside...

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, uns · line 3117

Quoted 12 lines

patches into one. Can it make any sense?

thanks
Hillf

mm/page_alloc.c +2 -0
@@ -3113,6 +3113,8 @@ should_reclaim_retry(gfp_t gfp_mask, uns
 	struct zone *zone;
 	struct zoneref *z;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;

This is defeating the whole purpose of the rework - to behave deterministically. You have just disabled the oom killer completely. This is not the way to go

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#65 lore raw
Quoted 7 lines
Quoted 7 lines
mm/page_alloc.c +0 -0
@@ -3113,6 +3113,8 @@ should_reclaim_retry(gfp_t gfp_mask, uns
 	struct zone *zone;
 	struct zoneref *z;
 
Quoted 7 lines

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, uns · line 3117

Quoted 2 lines
mm/page_alloc.c +2 -0
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;

This is defeating the whole purpose of the rework - to behave deterministically. You have just disabled the oom killer completely. This is not the way to go

Then in another direction, below is what I can do.

thanks
Hillf

mm/page_alloc.c +4 -1
@@ -3366,8 +3366,11 @@ retry:
 		no_progress_loops++;
 
 	if (should_reclaim_retry(gfp_mask, order, ac, alloc_flags,
-				 did_some_progress > 0, no_progress_loops))
+				 did_some_progress > 0, no_progress_loops)) {
+		/* Burn more cycles if any zone seems to satisfy our request */
+		no_progress_loops /= 2;
 		goto retry;
+	}
 
 	/* Reclaim has failed us, start killing things */
 	page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);

--

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#66 lore raw
Quoted 21 lines · Hillf Danton
Quoted 7 lines
Quoted 7 lines
mm/page_alloc.c +0 -0
@@ -3113,6 +3113,8 @@ should_reclaim_retry(gfp_t gfp_mask, uns
 	struct zone *zone;
 	struct zoneref *z;
 
Quoted 7 lines

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, uns · line 3117

Quoted 2 lines
mm/page_alloc.c +2 -0
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;

This is defeating the whole purpose of the rework - to behave deterministically. You have just disabled the oom killer completely. This is not the way to go

Then in another direction, below is what I can do.

thanks
Hillf

mm/page_alloc.c retry: · line 3371

Quoted 7 lines
mm/page_alloc.c +3 -1
@@ -3366,8 +3366,11 @@ retry:
 		no_progress_loops++;
 
 	if (should_reclaim_retry(gfp_mask, order, ac, alloc_flags,
-				 did_some_progress > 0, no_progress_loops))
+				 did_some_progress > 0, no_progress_loops)) {
+		/* Burn more cycles if any zone seems to satisfy our request */
+		no_progress_loops /= 2;

No, I do not think this makes any sense. If we need more retry loops then we can do it by increasing MAX_RECLAIM_RETRIES.

Quoted 5 lines
mm/page_alloc.c +1 -0
 		goto retry;
+	}
 
 	/* Reclaim has failed us, start killing things */
 	page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);
Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#67 lore raw
In reply to Hugh Dickins
Quoted 1 line · Hugh Dickins

On Thu, 25 Feb 2016, Michal Hocko wrote:

[...]

Quoted 22 lines
Quoted 22 lines

From d09de26cee148b4d8c486943b4e8f3bd7ad6f4be Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Thu, 4 Feb 2016 14:56:59 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and the OOM killer is just a matter of time but let's stick to our MAX_RECLAIM_RETRIES for the high order and not costly requests to make sure we do not fail prematurely.

This also means that we do not reset no_progress_loops at the __alloc_pages_slowpath for high order allocations to guarantee a bounded number of retries.

Quoted 8 lines
Quoted 4 lines

Longterm it would be much better to communicate with the compaction and retry only if the compaction considers it meaningfull.

It didn't really help, I'm afraid: it reduces the actual number of OOM kills which occur before the job is terminated, but doesn't stop the job from being terminated very soon.

Yeah this is not a magic bullet. I am happy to hear that the patch actually helped to reduce the number of OOM kills, though, because that is what it aims to do. I also believe that supports (at least partially) my suspicious that it is compaction which doesn't try enough. order-0 reclaim, even when done repeatedly, doesn't have a great chances to form higher order pages. Especially when there is a lot of migrateable memory. I have already talked about this with Vlastimil and he said that compaction can indeed back off too early because it doesn't care about !costly request much at all. We will have a look into this more next week.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#68 lore raw
In reply to Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 75 lines · Michal Hocko

From d09de26cee148b4d8c486943b4e8f3bd7ad6f4be Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Thu, 4 Feb 2016 14:56:59 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and the OOM killer is just a matter of time but let's stick to our MAX_RECLAIM_RETRIES for the high order and not costly requests to make sure we do not fail prematurely.

This also means that we do not reset no_progress_loops at the __alloc_pages_slowpath for high order allocations to guarantee a bounded number of retries.

Longterm it would be much better to communicate with the compaction and retry only if the compaction considers it meaningfull.

---
 mm/page_alloc.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)
mm/page_alloc.c +16 -4
index 269a04f20927..f05aca36469b 100644
@@ -3106,6 +3106,18 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		}
 	}
 
+	/*
+	 * OK, so the watermak check has failed. Make sure we do all the
+	 * retries for !costly high order requests and hope that multiple
+	 * runs of compaction will generate some high order ones for us.
+	 *
+	 * XXX: ideally we should teach the compaction to try _really_ hard
+	 * if we are in the retry path - something like priority 0 for the
+	 * reclaim
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
+
 	return false;
 }
 
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		goto noretry;
 
 	/*
-	 * Costly allocations might have made a progress but this doesn't mean
-	 * their order will become available due to high fragmentation so do
-	 * not reset the no progress counter for them
+	 * High order allocations might have made a progress but this doesn't
+	 * mean their order will become available due to high fragmentation so
+	 * do not reset the no progress counter for them
 	 */
-	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+	if (did_some_progress && !order)
 		no_progress_loops = 0;
 	else
 		no_progress_loops++;
Signature · 5 lines
2.7.0

-- 
Michal Hocko
SUSE Labs
Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#69 lore raw
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

I'm not sure that this is a way to go. See below.

Quoted 40 lines · Michal Hocko
Quoted 38 lines

From d09de26cee148b4d8c486943b4e8f3bd7ad6f4be Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Thu, 4 Feb 2016 14:56:59 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and the OOM killer is just a matter of time but let's stick to our MAX_RECLAIM_RETRIES for the high order and not costly requests to make sure we do not fail prematurely.

This also means that we do not reset no_progress_loops at the __alloc_pages_slowpath for high order allocations to guarantee a bounded number of retries.

Longterm it would be much better to communicate with the compaction and retry only if the compaction considers it meaningfull.

---
 mm/page_alloc.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)
mm/page_alloc.c +0 -0
index 269a04f20927..f05aca36469b 100644
@@ -3106,6 +3106,18 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
 		}
 	}
 
Quoted 13 lines

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3121

Quoted 13 lines
mm/page_alloc.c +12 -0
+	/*
+	 * OK, so the watermak check has failed. Make sure we do all the
+	 * retries for !costly high order requests and hope that multiple
+	 * runs of compaction will generate some high order ones for us.
+	 *
+	 * XXX: ideally we should teach the compaction to try _really_ hard
+	 * if we are in the retry path - something like priority 0 for the
+	 * reclaim
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
+
 	return false;

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here. So, following fix is needed.

'if (order)' check isn't needed. It is used to clarify the meaning of this fix. You can remove it.

mm/page_alloc.c +4 -0
index 1993894..8c80375 100644
@@ -3125,6 +3125,10 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
        if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
                return false;
 
+       /* To check whether compaction is available or not */
+       if (order)
+               order = 0;
+
        /*
         * Keep reclaiming pages while there is a chance this will lead
         * somewhere.  If none of the target zones can satisfy our allocation
Quoted 5 lines
Quoted 5 lines
mm/page_alloc.c +0 -0
 }
 
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		goto noretry;
 
Quoted 13 lines

mm/page_alloc.c __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3303

Quoted 13 lines
mm/page_alloc.c +4 -4
 	/*
-	 * Costly allocations might have made a progress but this doesn't mean
-	 * their order will become available due to high fragmentation so do
-	 * not reset the no progress counter for them
+	 * High order allocations might have made a progress but this doesn't
+	 * mean their order will become available due to high fragmentation so
+	 * do not reset the no progress counter for them
 	 */
-	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+	if (did_some_progress && !order)
 		no_progress_loops = 0;
 	else
 		no_progress_loops++;

This unconditionally increases no_progress_loops for high order allocation, so, after 16 iterations, it will fail. If compaction isn't enabled in Kconfig, 16 times reclaim attempt would not be sufficient to make high order page. Should we consider this case also?

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#70 lore raw
Quoted 1 line · Joonsoo Kim

On Mon, Feb 29, 2016 at 10:02:13PM +0100, Michal Hocko wrote:

[...]

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
(quoted diff) +9 -0
+	/*
+	 * OK, so the watermak check has failed. Make sure we do all the
+	 * retries for !costly high order requests and hope that multiple
+	 * runs of compaction will generate some high order ones for us.
+	 *
+	 * XXX: ideally we should teach the compaction to try _really_ hard
+	 * if we are in the retry path - something like priority 0 for the
+	 * reclaim
+	 */
Quoted 8 lines
Quoted 4 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
+
 	return false;

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 10 lines

So, following fix is needed.

'if (order)' check isn't needed. It is used to clarify the meaning of this fix. You can remove it.

mm/page_alloc.c +0 -0
index 1993894..8c80375 100644

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3131

Quoted 7 lines
mm/page_alloc.c +4 -0
@@ -3125,6 +3125,10 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
        if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
                return false;
 
+       /* To check whether compaction is available or not */
+       if (order)
+               order = 0;
+

This would enforce the order 0 wmark check which is IMHO not correct as per above.

Quoted 19 lines
mm/page_alloc.c +0 -0
        /*
         * Keep reclaiming pages while there is a chance this will lead
         * somewhere.  If none of the target zones can satisfy our allocation
Quoted 16 lines
Quoted 16 lines

}

(quoted diff) +4 -4
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		goto noretry;
 
 	/*
-	 * Costly allocations might have made a progress but this doesn't mean
-	 * their order will become available due to high fragmentation so do
-	 * not reset the no progress counter for them
+	 * High order allocations might have made a progress but this doesn't
+	 * mean their order will become available due to high fragmentation so
+	 * do not reset the no progress counter for them
 	 */
-	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+	if (did_some_progress && !order)
Quoted 8 lines
Quoted 3 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3303

Quoted 3 lines
(quoted diff) +0 -0
 		no_progress_loops = 0;
 	else
 		no_progress_loops++;

This unconditionally increases no_progress_loops for high order allocation, so, after 16 iterations, it will fail. If compaction isn't enabled in Kconfig, 16 times reclaim attempt would not be sufficient to make high order page. Should we consider this case also?

How many retries would help? I do not think any number will work reliably. Configurations without compaction enabled are asking for problems by definition IMHO. Relying on order-0 reclaim for high order allocations simply cannot work.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#71 lore raw

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 21 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Quoted 6 lines

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

If we have >= order page available, we would not reach here. We would just allocate it.

And, should_reclaim_retry() is not just for reclaim. It is also for retrying compaction.

That watermark check is to check further reclaim/compaction is meaningful. And, for high order case, if there is enough freepage, compaction could make high order page even if there is no high order page now.

Adding freeable memory and checking watermark with it doesn't help in this case because number of high order page isn't changed with it.

I just did quick review to your patches so maybe I am wrong. Am I missing something?

Quoted 49 lines
Quoted 10 lines

So, following fix is needed.

'if (order)' check isn't needed. It is used to clarify the meaning of this fix. You can remove it.

mm/page_alloc.c +0 -0
index 1993894..8c80375 100644

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3131

Quoted 7 lines
mm/page_alloc.c +4 -0
@@ -3125,6 +3125,10 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
        if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
                return false;
 
+       /* To check whether compaction is available or not */
+       if (order)
+               order = 0;
+

This would enforce the order 0 wmark check which is IMHO not correct as per above.

Quoted 19 lines
mm/page_alloc.c +0 -0
        /*
         * Keep reclaiming pages while there is a chance this will lead
         * somewhere.  If none of the target zones can satisfy our allocation
Quoted 16 lines
Quoted 16 lines

}

(quoted diff) +4 -4
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
          goto noretry;
 
  /*
-  * Costly allocations might have made a progress but this doesn't mean
-  * their order will become available due to high fragmentation so do
-  * not reset the no progress counter for them
+  * High order allocations might have made a progress but this doesn't
+  * mean their order will become available due to high fragmentation so
+  * do not reset the no progress counter for them
   */
- if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+ if (did_some_progress && !order)
Quoted 8 lines
Quoted 3 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3303

Quoted 3 lines
(quoted diff) +0 -0
          no_progress_loops = 0;
  else
          no_progress_loops++;

This unconditionally increases no_progress_loops for high order allocation, so, after 16 iterations, it will fail. If compaction isn't enabled in Kconfig, 16 times reclaim attempt would not be sufficient to make high order page. Should we consider this case also?

Quoted 4 lines

How many retries would help? I do not think any number will work reliably. Configurations without compaction enabled are asking for problems by definition IMHO. Relying on order-0 reclaim for high order allocations simply cannot work.

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success. Why do we limit retry as 16 times with no evidence of potential impossibility of making high order page?

And, 16 retry looks not good to me because compaction could defer actual doing up to 64 times.

Thanks.

Quoted 10 lines

--
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#72 lore raw
Quoted 29 lines · Joonsoo Kim

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 21 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Quoted 6 lines

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 2 lines

If we have >= order page available, we would not reach here. We would just allocate it.

not really, we can still be under the low watermark. Note that the target for the should_reclaim_retry watermark check includes also the reclaimable memory.

Quoted 8 lines

And, should_reclaim_retry() is not just for reclaim. It is also for retrying compaction.

That watermark check is to check further reclaim/compaction is meaningful. And, for high order case, if there is enough freepage, compaction could make high order page even if there is no high order page now.

Quoted 5 lines

Adding freeable memory and checking watermark with it doesn't help in this case because number of high order page isn't changed with it.

I just did quick review to your patches so maybe I am wrong. Am I missing something?

The core idea behind should_reclaim_retry is to check whether the reclaiming all the pages would help to get over the watermark and there is at least one >= order page. Then it really makes sense to retry. As the compaction has already was performed before this is called we should have created some high order pages already. The decay guarantees that we eventually trigger the OOM killer after some attempts.

If the compaction can backoff and ignore our requests then we are screwed of course and that should be addressed imho at the compaction layer. Maybe we can tell the compaction to try harder but I would like to understand why this shouldn't be a default behavior for !costly orders.

[...]

Quoted 27 lines
Quoted 22 lines
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
(quoted diff) +4 -4
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
          goto noretry;
 
  /*
-  * Costly allocations might have made a progress but this doesn't mean
-  * their order will become available due to high fragmentation so do
-  * not reset the no progress counter for them
+  * High order allocations might have made a progress but this doesn't
+  * mean their order will become available due to high fragmentation so
+  * do not reset the no progress counter for them
   */
- if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+ if (did_some_progress && !order)
Quoted 8 lines
Quoted 3 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3303

Quoted 3 lines
(quoted diff) +0 -0
          no_progress_loops = 0;
  else
          no_progress_loops++;

This unconditionally increases no_progress_loops for high order allocation, so, after 16 iterations, it will fail. If compaction isn't enabled in Kconfig, 16 times reclaim attempt would not be sufficient to make high order page. Should we consider this case also?

Quoted 4 lines

How many retries would help? I do not think any number will work reliably. Configurations without compaction enabled are asking for problems by definition IMHO. Relying on order-0 reclaim for high order allocations simply cannot work.

Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

Quoted 2 lines

Why do we limit retry as 16 times with no evidence of potential impossibility of making high order page?

If we tried to compact 16 times without any progress then this sounds like a sufficient evidence to me. Well, this number is somehow arbitrary but the main point is to limit it to _some_ number, if we can show that a larger value would work better then we can update it of course.

Quoted 2 lines

And, 16 retry looks not good to me because compaction could defer actual doing up to 64 times.

OK, this is something that needs to be handled in a better way. The primary question would be why to defer the compaction for <= PAGE_ALLOC_COSTLY_ORDER requests in the first place. I guess I do see why it makes sense it for the best effort mode of operation but !costly orders should be trying much harder as they are nofail, no?

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#73 lore raw

2016-03-02 23:06 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 30 lines · Joonsoo Kim
Quoted 29 lines

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 27 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 4 lines
Quoted 2 lines

If we have >= order page available, we would not reach here. We would just allocate it.

not really, we can still be under the low watermark. Note that the

you mean min watermark?

Quoted 2 lines

target for the should_reclaim_retry watermark check includes also the reclaimable memory.

I guess that usual case for high order allocation failure has enough freepage.

Quoted 11 lines
Quoted 11 lines

And, should_reclaim_retry() is not just for reclaim. It is also for retrying compaction.

That watermark check is to check further reclaim/compaction is meaningful. And, for high order case, if there is enough freepage, compaction could make high order page even if there is no high order page now.

Adding freeable memory and checking watermark with it doesn't help in this case because number of high order page isn't changed with it.

Quoted 6 lines
Quoted 2 lines

I just did quick review to your patches so maybe I am wrong. Am I missing something?

The core idea behind should_reclaim_retry is to check whether the reclaiming all the pages would help to get over the watermark and there is at least one >= order page. Then it really makes sense to retry. As

How you can judge that reclaiming all the pages would help to check there is at least one >= order page?

Quoted 2 lines

the compaction has already was performed before this is called we should have created some high order pages already. The decay guarantees that we

Not really. Compaction could fail.

Quoted 1 line

eventually trigger the OOM killer after some attempts.

Yep.

Quoted 5 lines

If the compaction can backoff and ignore our requests then we are screwed of course and that should be addressed imho at the compaction layer. Maybe we can tell the compaction to try harder but I would like to understand why this shouldn't be a default behavior for !costly orders.

Yes, I agree that.

Quoted 28 lines

[...]

Quoted 27 lines
Quoted 26 lines
Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
(quoted diff) +4 -4
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
          goto noretry;
 
  /*
-  * Costly allocations might have made a progress but this doesn't mean
-  * their order will become available due to high fragmentation so do
-  * not reset the no progress counter for them
+  * High order allocations might have made a progress but this doesn't
+  * mean their order will become available due to high fragmentation so
+  * do not reset the no progress counter for them
   */
- if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+ if (did_some_progress && !order)
Quoted 8 lines
Quoted 3 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3303

Quoted 3 lines
(quoted diff) +0 -0
          no_progress_loops = 0;
  else
          no_progress_loops++;

This unconditionally increases no_progress_loops for high order allocation, so, after 16 iterations, it will fail. If compaction isn't enabled in Kconfig, 16 times reclaim attempt would not be sufficient to make high order page. Should we consider this case also?

How many retries would help? I do not think any number will work reliably. Configurations without compaction enabled are asking for problems by definition IMHO. Relying on order-0 reclaim for high order allocations simply cannot work.

Quoted 7 lines
Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

I just can tell simple version. Please check the link from me on another reply. Compaction could scan more range of memory if we have more freepage. This is due to algorithm limitation. Anyway, so, reclaiming random order-0 pages helps compaction.

Quoted 7 lines
Quoted 2 lines

Why do we limit retry as 16 times with no evidence of potential impossibility of making high order page?

If we tried to compact 16 times without any progress then this sounds like a sufficient evidence to me. Well, this number is somehow arbitrary but the main point is to limit it to _some_ number, if we can show that a larger value would work better then we can update it of course.

My arguing is for your band aid patch.
My point is that why retry count for order-0 is reset if there is some progress, but, retry counter for order up to costly isn't reset even if there is some progress

Quoted 8 lines
Quoted 2 lines

And, 16 retry looks not good to me because compaction could defer actual doing up to 64 times.

OK, this is something that needs to be handled in a better way. The primary question would be why to defer the compaction for <= PAGE_ALLOC_COSTLY_ORDER requests in the first place. I guess I do see why it makes sense it for the best effort mode of operation but !costly orders should be trying much harder as they are nofail, no?

Make sense.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#74 lore raw
Quoted 31 lines · Joonsoo Kim

2016-03-02 23:06 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 30 lines · Joonsoo Kim
Quoted 29 lines

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 27 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

If we have >= order page available, we would not reach here. We would just allocate it.

not really, we can still be under the low watermark. Note that the

you mean min watermark?

ohh, right...

Quoted 4 lines
Quoted 2 lines

target for the should_reclaim_retry watermark check includes also the reclaimable memory.

I guess that usual case for high order allocation failure has enough freepage.

Not sure I understand you mean here but I wouldn't be surprised if high order failed even with enough free pages. And that is exactly why I am claiming that reclaiming more pages is no free ticket to high order pages.

[...]

Quoted 9 lines
Quoted 6 lines
Quoted 2 lines

I just did quick review to your patches so maybe I am wrong. Am I missing something?

The core idea behind should_reclaim_retry is to check whether the reclaiming all the pages would help to get over the watermark and there is at least one >= order page. Then it really makes sense to retry. As

How you can judge that reclaiming all the pages would help to check there is at least one >= order page?

Again, not sure I understand you here. __zone_watermark_ok checks both wmark and an available page of the sufficient order. While increased free_pages (which includes reclaimable pages as well) will tell us whether we have a chance to get over the min wmark, the order check will tell us we have something to allocate from after we reach the min wmark.

Quoted 4 lines
Quoted 2 lines

the compaction has already was performed before this is called we should have created some high order pages already. The decay guarantees that we

Not really. Compaction could fail.

Yes it could have failed. But what is the point to retry endlessly then?

[...]

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

Quoted 7 lines
Quoted 2 lines

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

I just can tell simple version. Please check the link from me on another reply. Compaction could scan more range of memory if we have more freepage. This is due to algorithm limitation. Anyway, so, reclaiming random order-0 pages helps compaction.

I will have a look at that code but this just doesn't make any sense. The compaction should be reshuffling pages, this shouldn't be a function of free memory.

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

Why do we limit retry as 16 times with no evidence of potential impossibility of making high order page?

If we tried to compact 16 times without any progress then this sounds like a sufficient evidence to me. Well, this number is somehow arbitrary but the main point is to limit it to _some_ number, if we can show that a larger value would work better then we can update it of course.

Quoted 4 lines

My arguing is for your band aid patch.
My point is that why retry count for order-0 is reset if there is some progress, but, retry counter for order up to costly isn't reset even if there is some progress

Because we know that order-0 requests have chance to proceed if we keep reclaiming order-0 pages while this is not true for order > 0. If we did reset the no_progress_loops for order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER then we would be back to the zone_reclaimable heuristic. Why? Because order-0 reclaim progress will keep !costly in the reclaim loop while compaction still might not make any progress. So we either have to fail when __zone_watermark_ok fails for the order (which turned out to be too easy to trigger) or have the fixed amount of retries regardless the watermark check result. We cannot relax both unless we have other measures in place.

Sure we can be more intelligent and reset the counter if the feedback from compaction is optimistic and we are making some progress. This would be less hackish and the XXX comment points into that direction. For now I would like this to catch most loads reasonably and build better heuristics on top. I would like to do as much as possible to close the obvious regressions but I guess we have to expect there will be cases where the OOM fires and hasn't before and vice versa.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#75 lore raw
Quoted 8 lines · Michal Hocko

Sure we can be more intelligent and reset the counter if the feedback from compaction is optimistic and we are making some progress. This would be less hackish and the XXX comment points into that direction. For now I would like this to catch most loads reasonably and build better heuristics on top. I would like to do as much as possible to close the obvious regressions but I guess we have to expect there will be cases where the OOM fires and hasn't before and vice versa.

Aren't you forgetting that some people use panic_on_oom > 0 which means that premature OOM killer invocation is fatal for them?

#76 lore raw
In reply to Michal Hocko

2016-03-03 18:26 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 44 lines · Joonsoo Kim
Quoted 31 lines

2016-03-02 23:06 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 30 lines · Joonsoo Kim
Quoted 29 lines

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 27 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

If we have >= order page available, we would not reach here. We would just allocate it.

not really, we can still be under the low watermark. Note that the

you mean min watermark?

ohh, right...

Quoted 3 lines
Quoted 2 lines

target for the should_reclaim_retry watermark check includes also the reclaimable memory.

Quoted 6 lines
Quoted 1 line

I guess that usual case for high order allocation failure has enough freepage.

Not sure I understand you mean here but I wouldn't be surprised if high order failed even with enough free pages. And that is exactly why I am claiming that reclaiming more pages is no free ticket to high order pages.

I didn't say that it's free ticket. OOM kill would be the most expensive ticket that we have. Why do you want to kill something? It also doesn't guarantee to make high order pages. It is just another way of reclaiming memory. What is the difference between plain reclaim and OOM kill? Why do we use OOM kill in this case?

Quoted 8 lines

[...]

Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

I just did quick review to your patches so maybe I am wrong. Am I missing something?

The core idea behind should_reclaim_retry is to check whether the reclaiming all the pages would help to get over the watermark and there is at least one >= order page. Then it really makes sense to retry. As

Quoted 8 lines
Quoted 2 lines

How you can judge that reclaiming all the pages would help to check there is at least one >= order page?

Again, not sure I understand you here. __zone_watermark_ok checks both wmark and an available page of the sufficient order. While increased free_pages (which includes reclaimable pages as well) will tell us whether we have a chance to get over the min wmark, the order check will tell us we have something to allocate from after we reach the min wmark.

Again, your assumption would be different with mine. My assumption is that high order allocation problem happens due to fragmentation rather than low free memory. In this case, there is no high order page. Even if you can reclaim 1TB and add this counter to freepage counter, high order page counter will not be changed and watermark check would fail. So, high order allocation will not go through retry logic. This is what you want?

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

the compaction has already was performed before this is called we should have created some high order pages already. The decay guarantees that we

Not really. Compaction could fail.

Yes it could have failed. But what is the point to retry endlessly then?

I didn't say we should retry endlessly.

Quoted 9 lines

[...]

Quoted 8 lines
Quoted 7 lines
Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

Quoted 8 lines
Quoted 4 lines

I just can tell simple version. Please check the link from me on another reply. Compaction could scan more range of memory if we have more freepage. This is due to algorithm limitation. Anyway, so, reclaiming random order-0 pages helps compaction.

I will have a look at that code but this just doesn't make any sense. The compaction should be reshuffling pages, this shouldn't be a function of free memory.

Please refer the link I mentioned before. There is a reason why more free memory would help compaction success. Compaction doesn't work like as random reshuffling. It has an algorithm to reduce system overall fragmentation so there is limitation.

Quoted 13 lines
Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

Why do we limit retry as 16 times with no evidence of potential impossibility of making high order page?

If we tried to compact 16 times without any progress then this sounds like a sufficient evidence to me. Well, this number is somehow arbitrary but the main point is to limit it to _some_ number, if we can show that a larger value would work better then we can update it of course.

Quoted 4 lines

My arguing is for your band aid patch.
My point is that why retry count for order-0 is reset if there is some progress, but, retry counter for order up to costly isn't reset even if there is some progress

Quoted 10 lines

Because we know that order-0 requests have chance to proceed if we keep reclaiming order-0 pages while this is not true for order > 0. If we did reset the no_progress_loops for order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER then we would be back to the zone_reclaimable heuristic. Why? Because order-0 reclaim progress will keep !costly in the reclaim loop while compaction still might not make any progress. So we either have to fail when __zone_watermark_ok fails for the order (which turned out to be too easy to trigger) or have the fixed amount of retries regardless the watermark check result. We cannot relax both unless we have other measures in place.

As mentioned before, OOM kill also doesn't guarantee to make high order page. Reclaim more memory as much as possible makes more sense to me. Timing of OOM kill for order-0 is reasonable because there is not enough freeable page. But, it's not reasonable to kill something when we have much reclaimable memory like as your current implementation.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#77 lore raw
Quoted 47 lines · Joonsoo Kim

2016-03-03 18:26 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 46 lines · Joonsoo Kim
Quoted 31 lines

2016-03-02 23:06 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 30 lines · Joonsoo Kim
Quoted 29 lines

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 27 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

If we have >= order page available, we would not reach here. We would just allocate it.

not really, we can still be under the low watermark. Note that the

you mean min watermark?

ohh, right...

Quoted 4 lines
Quoted 2 lines

target for the should_reclaim_retry watermark check includes also the reclaimable memory.

I guess that usual case for high order allocation failure has enough freepage.

Quoted 7 lines
Quoted 4 lines

Not sure I understand you mean here but I wouldn't be surprised if high order failed even with enough free pages. And that is exactly why I am claiming that reclaiming more pages is no free ticket to high order pages.

I didn't say that it's free ticket. OOM kill would be the most expensive ticket that we have. Why do you want to kill something?

Because all the attempts so far have failed and we should rather not retry endlessly. With the band-aid we know we will retry MAX_RECLAIM_RETRIES at most. So compaction had that many attempts to resolve the situation along with the same amount of reclaim rounds to help and get over watermarks.

Quoted 3 lines

It also doesn't guarantee to make high order pages. It is just another way of reclaiming memory. What is the difference between plain reclaim and OOM kill? Why do we use OOM kill in this case?

What is our alternative other than keep looping endlessly?

Quoted 17 lines
Quoted 8 lines

[...]

Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

I just did quick review to your patches so maybe I am wrong. Am I missing something?

The core idea behind should_reclaim_retry is to check whether the reclaiming all the pages would help to get over the watermark and there is at least one >= order page. Then it really makes sense to retry. As

Quoted 8 lines
Quoted 2 lines

How you can judge that reclaiming all the pages would help to check there is at least one >= order page?

Again, not sure I understand you here. __zone_watermark_ok checks both wmark and an available page of the sufficient order. While increased free_pages (which includes reclaimable pages as well) will tell us whether we have a chance to get over the min wmark, the order check will tell us we have something to allocate from after we reach the min wmark.

Quoted 6 lines

Again, your assumption would be different with mine. My assumption is that high order allocation problem happens due to fragmentation rather than low free memory. In this case, there is no high order page. Even if you can reclaim 1TB and add this counter to freepage counter, high order page counter will not be changed and watermark check would fail. So, high order allocation will not go through retry logic. This is what you want?

I really want to base the decision on something measurable rather than a good hope. This is what all the zone_reclaimable() is about. I understand your concerns that compaction doesn't guarantee anything but I am quite convinced that we really need an upper bound for retries (unlike now when zone_reclaimable is basically unbounded assuming order-0 reclaim makes some progress). What is the best bound is harder to tell, of course.

[...]

Quoted 16 lines
Quoted 5 lines
Quoted 4 lines

My arguing is for your band aid patch.
My point is that why retry count for order-0 is reset if there is some progress, but, retry counter for order up to costly isn't reset even if there is some progress

Quoted 10 lines

Because we know that order-0 requests have chance to proceed if we keep reclaiming order-0 pages while this is not true for order > 0. If we did reset the no_progress_loops for order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER then we would be back to the zone_reclaimable heuristic. Why? Because order-0 reclaim progress will keep !costly in the reclaim loop while compaction still might not make any progress. So we either have to fail when __zone_watermark_ok fails for the order (which turned out to be too easy to trigger) or have the fixed amount of retries regardless the watermark check result. We cannot relax both unless we have other measures in place.

Quoted 1 line

As mentioned before, OOM kill also doesn't guarantee to make high order page.

Yes, of course, apart from the kernel stack which is high order there is no guarantee.

Quoted 1 line

Reclaim more memory as much as possible makes more sense to me.

But then we are back to square one. How much and how to decide when it makes sense to give up. Do you have any suggestions on what should be the criteria? Is there any feedback mechanism from the compaction which would tell us to keep retrying? Something like did_some_progress from the order-0 reclaim? Is any of deferred_compaction resp. contended_compaction usable? Or is there any per-zone flag we can check and prefer over wmark order check?

Thanks

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#78 lore raw
Quoted 62 lines · Michal Hocko
Quoted 47 lines · Joonsoo Kim

2016-03-03 18:26 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 46 lines · Joonsoo Kim
Quoted 31 lines

2016-03-02 23:06 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 30 lines · Joonsoo Kim
Quoted 29 lines

2016-03-02 18:50 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 27 lines · Joonsoo Kim
Quoted 1 line · Michal Hocko

[...]

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

+ /*

(quoted diff) +4 -0
+  * OK, so the watermak check has failed. Make sure we do all the
+  * retries for !costly high order requests and hope that multiple
+  * runs of compaction will generate some high order ones for us.
+  *
Quoted 8 lines

Review

Quoted 8 lines
(quoted diff) +7 -0
+  * XXX: ideally we should teach the compaction to try _really_ hard
+  * if we are in the retry path - something like priority 0 for the
+  * reclaim
+  */
+ if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+         return true;
+
  return false;
Quoted 3 lines

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

If we have >= order page available, we would not reach here. We would just allocate it.

not really, we can still be under the low watermark. Note that the

you mean min watermark?

ohh, right...

Quoted 4 lines
Quoted 2 lines

target for the should_reclaim_retry watermark check includes also the reclaimable memory.

I guess that usual case for high order allocation failure has enough freepage.

Quoted 7 lines
Quoted 4 lines

Not sure I understand you mean here but I wouldn't be surprised if high order failed even with enough free pages. And that is exactly why I am claiming that reclaiming more pages is no free ticket to high order pages.

I didn't say that it's free ticket. OOM kill would be the most expensive ticket that we have. Why do you want to kill something?

Because all the attempts so far have failed and we should rather not retry endlessly. With the band-aid we know we will retry MAX_RECLAIM_RETRIES at most. So compaction had that many attempts to resolve the situation along with the same amount of reclaim rounds to help and get over watermarks.

Quoted 5 lines
Quoted 3 lines

It also doesn't guarantee to make high order pages. It is just another way of reclaiming memory. What is the difference between plain reclaim and OOM kill? Why do we use OOM kill in this case?

What is our alternative other than keep looping endlessly?

Loop as long as free memory or estimated available memory (free + reclaimable) increases. This means that we did some progress. And, they will not grow forever because we have just limited reclaimable memory and limited memory. You can reset no_progress_loops = 0 when those metric increases than before.

With this bound, we can do our best to try to solve this unpleasant situation before OOM.

Unconditional 16 looping and then OOM kill really doesn't make any sense, because it doesn't mean that we already do our best. OOM should not be called prematurely and AFAIK it is one of goals on your patches.

If above suggestion doesn't make sense to you, please try to find another way rather than suggesting work-around that could cause OOM prematurely in high order allocation case.

Thanks.

Quoted 65 lines
Quoted 18 lines
Quoted 9 lines

[...]

Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

I just did quick review to your patches so maybe I am wrong. Am I missing something?

The core idea behind should_reclaim_retry is to check whether the reclaiming all the pages would help to get over the watermark and there is at least one >= order page. Then it really makes sense to retry. As

Quoted 8 lines
Quoted 2 lines

How you can judge that reclaiming all the pages would help to check there is at least one >= order page?

Again, not sure I understand you here. __zone_watermark_ok checks both wmark and an available page of the sufficient order. While increased free_pages (which includes reclaimable pages as well) will tell us whether we have a chance to get over the min wmark, the order check will tell us we have something to allocate from after we reach the min wmark.

Quoted 6 lines

Again, your assumption would be different with mine. My assumption is that high order allocation problem happens due to fragmentation rather than low free memory. In this case, there is no high order page. Even if you can reclaim 1TB and add this counter to freepage counter, high order page counter will not be changed and watermark check would fail. So, high order allocation will not go through retry logic. This is what you want?

I really want to base the decision on something measurable rather than a good hope. This is what all the zone_reclaimable() is about. I understand your concerns that compaction doesn't guarantee anything but I am quite convinced that we really need an upper bound for retries (unlike now when zone_reclaimable is basically unbounded assuming order-0 reclaim makes some progress). What is the best bound is harder to tell, of course.

[...]

Quoted 16 lines
Quoted 5 lines
Quoted 4 lines

My arguing is for your band aid patch.
My point is that why retry count for order-0 is reset if there is some progress, but, retry counter for order up to costly isn't reset even if there is some progress

Quoted 10 lines

Because we know that order-0 requests have chance to proceed if we keep reclaiming order-0 pages while this is not true for order > 0. If we did reset the no_progress_loops for order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER then we would be back to the zone_reclaimable heuristic. Why? Because order-0 reclaim progress will keep !costly in the reclaim loop while compaction still might not make any progress. So we either have to fail when __zone_watermark_ok fails for the order (which turned out to be too easy to trigger) or have the fixed amount of retries regardless the watermark check result. We cannot relax both unless we have other measures in place.

Quoted 1 line

As mentioned before, OOM kill also doesn't guarantee to make high order page.

Yes, of course, apart from the kernel stack which is high order there is no guarantee.

Quoted 1 line

Reclaim more memory as much as possible makes more sense to me.

But then we are back to square one. How much and how to decide when it makes sense to give up. Do you have any suggestions on what should be the criteria? Is there any feedback mechanism from the compaction which would tell us to keep retrying? Something like did_some_progress from the order-0 reclaim? Is any of deferred_compaction resp. contended_compaction usable? Or is there any per-zone flag we can check and prefer over wmark order check?

Quoted 10 lines

Thanks

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#79 lore raw
Quoted 3 lines · Joonsoo Kim
Quoted 2 lines · Michal Hocko
Quoted 1 line · Joonsoo Kim

2016-03-03 18:26 GMT+09:00 Michal Hocko <[email protected]>:

[...]

Quoted 20 lines
Quoted 20 lines
Quoted 9 lines
Quoted 6 lines
Quoted 1 line

I guess that usual case for high order allocation failure has enough freepage.

Not sure I understand you mean here but I wouldn't be surprised if high order failed even with enough free pages. And that is exactly why I am claiming that reclaiming more pages is no free ticket to high order pages.

I didn't say that it's free ticket. OOM kill would be the most expensive ticket that we have. Why do you want to kill something?

Because all the attempts so far have failed and we should rather not retry endlessly. With the band-aid we know we will retry MAX_RECLAIM_RETRIES at most. So compaction had that many attempts to resolve the situation along with the same amount of reclaim rounds to help and get over watermarks.

Quoted 3 lines

It also doesn't guarantee to make high order pages. It is just another way of reclaiming memory. What is the difference between plain reclaim and OOM kill? Why do we use OOM kill in this case?

Quoted 7 lines
Quoted 1 line

What is our alternative other than keep looping endlessly?

Loop as long as free memory or estimated available memory (free + reclaimable) increases. This means that we did some progress. And, they will not grow forever because we have just limited reclaimable memory and limited memory. You can reset no_progress_loops = 0 when those metric increases than before.

Hmm, why is this any better than taking the feedback from the reclaim (did_some_progress)?

Quoted 5 lines

With this bound, we can do our best to try to solve this unpleasant situation before OOM.

Unconditional 16 looping and then OOM kill really doesn't make any sense, because it doesn't mean that we already do our best.

16 is not really that important. We can change that if that doesn't sounds sufficient. But please note that each reclaim round means that we have scanned all eligible LRUs to find and reclaim something and asked direct compaction to prepare a high order page. This sounds like "do our best" to me.

Now it seems that we need more changes at least in the compaction area because the code doesn't seem to fit the nature of !costly allocation requests. I am also not satisfied with the fixed MAX_RECLAIM_RETRIES for high order pages, I would much rather see some feedback mechanism which would measurable and evaluated in some way but is this really necessary for the initial version?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#80 lore raw
Quoted 1 line · Michal Hocko

On Fri 04-03-16 14:23:27, Joonsoo Kim wrote:

[...]

Quoted 6 lines
Quoted 2 lines

Unconditional 16 looping and then OOM kill really doesn't make any sense, because it doesn't mean that we already do our best.

16 is not really that important. We can change that if that doesn't sounds sufficient. But please note that each reclaim round means that we have scanned all eligible LRUs to find and reclaim something

this should read "scanned potentially all eligible LRUs..."

Quoted 2 lines

and asked direct compaction to prepare a high order page. This sounds like "do our best" to me.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#81 lore raw
In reply to Michal Hocko
Quoted 27 lines · Michal Hocko
Quoted 3 lines · Joonsoo Kim
Quoted 2 lines · Michal Hocko
Quoted 1 line · Joonsoo Kim

2016-03-03 18:26 GMT+09:00 Michal Hocko <[email protected]>:

[...]

Quoted 22 lines
Quoted 21 lines
Quoted 9 lines
Quoted 6 lines
Quoted 1 line

I guess that usual case for high order allocation failure has enough freepage.

Not sure I understand you mean here but I wouldn't be surprised if high order failed even with enough free pages. And that is exactly why I am claiming that reclaiming more pages is no free ticket to high order pages.

I didn't say that it's free ticket. OOM kill would be the most expensive ticket that we have. Why do you want to kill something?

Because all the attempts so far have failed and we should rather not retry endlessly. With the band-aid we know we will retry MAX_RECLAIM_RETRIES at most. So compaction had that many attempts to resolve the situation along with the same amount of reclaim rounds to help and get over watermarks.

Quoted 3 lines

It also doesn't guarantee to make high order pages. It is just another way of reclaiming memory. What is the difference between plain reclaim and OOM kill? Why do we use OOM kill in this case?

What is our alternative other than keep looping endlessly?

Quoted 8 lines
Quoted 5 lines

Loop as long as free memory or estimated available memory (free + reclaimable) increases. This means that we did some progress. And, they will not grow forever because we have just limited reclaimable memory and limited memory. You can reset no_progress_loops = 0 when those metric increases than before.

Hmm, why is this any better than taking the feedback from the reclaim (did_some_progress)?

My suggestion could be only applied to high order case. In this case, free page and reclaimable page is already sufficient and parallel free page consumer would re-generate reclaimable page endlessly so positive did_some_progress will be returned endlessy. We need to stop retry at some point so we need some metric that ensures finite retry in any case.

Quoted 4 lines
Quoted 4 lines

With this bound, we can do our best to try to solve this unpleasant situation before OOM.

Quoted 8 lines
Quoted 2 lines

Unconditional 16 looping and then OOM kill really doesn't make any sense, because it doesn't mean that we already do our best.

16 is not really that important. We can change that if that doesn't sounds sufficient. But please note that each reclaim round means that we have scanned all eligible LRUs to find and reclaim something and asked direct compaction to prepare a high order page. This sounds like "do our best" to me.

AFAIK, each reclaim round doesn't reclaim all reclaimable page. It has a limit to reclaim. It looks not our best to me and N retry only multipies that limit N times. It also doesn't look like our best to me and will lead to premature OOM kill.

Quoted 6 lines

Now it seems that we need more changes at least in the compaction area because the code doesn't seem to fit the nature of !costly allocation requests. I am also not satisfied with the fixed MAX_RECLAIM_RETRIES for high order pages, I would much rather see some feedback mechanism which would measurable and evaluated in some way but is this really necessary for the initial version?

I don't know. My analysis is just based on my guess and background knowledge, not practical usecase, so I'm not sure it is necessary for the initial version or not. It's up to you.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#82 lore raw
In reply to Joonsoo Kim
Quoted 15 lines · Joonsoo Kim
Quoted 15 lines

[...]

Quoted 12 lines
Quoted 7 lines
Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

I just can tell simple version. Please check the link from me on another reply. Compaction could scan more range of memory if we have more freepage. This is due to algorithm limitation. Anyway, so, reclaiming random order-0 pages helps compaction.

Quoted 8 lines
Quoted 3 lines

I will have a look at that code but this just doesn't make any sense. The compaction should be reshuffling pages, this shouldn't be a function of free memory.

Please refer the link I mentioned before. There is a reason why more free memory would help compaction success. Compaction doesn't work like as random reshuffling. It has an algorithm to reduce system overall fragmentation so there is limitation.

I proposed another way to get better results from direct compaction - don't scan for free pages but get them directly from freelists:

https://lkml.org/lkml/2015/12/3/60

But your redesign would be useful too for kcompactd/khugepaged keeping overall fragmentation low.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#83 lore raw
Quoted 25 lines · Vlastimil Babka
Quoted 15 lines · Joonsoo Kim
Quoted 15 lines

[...]

Quoted 12 lines
Quoted 7 lines
Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

I just can tell simple version. Please check the link from me on another reply. Compaction could scan more range of memory if we have more freepage. This is due to algorithm limitation. Anyway, so, reclaiming random order-0 pages helps compaction.

Quoted 8 lines
Quoted 3 lines

I will have a look at that code but this just doesn't make any sense. The compaction should be reshuffling pages, this shouldn't be a function of free memory.

Please refer the link I mentioned before. There is a reason why more free memory would help compaction success. Compaction doesn't work like as random reshuffling. It has an algorithm to reduce system overall fragmentation so there is limitation.

Quoted 4 lines

I proposed another way to get better results from direct compaction - don't scan for free pages but get them directly from freelists:

https://lkml.org/lkml/2015/12/3/60

Yes this makes perfect sense to me (with my limited experience in this area so I might be missing some obvious problems this would introduce). The direct compaction for !costly orders is something we should better satisfy immediately. I would just object that this shouldn't be reduced to ASYNC compaction requests only. SYNC* modes are even a more desperate call (at least that is my understanding) for the page and we should treat them the appropriately.

Quoted 2 lines

But your redesign would be useful too for kcompactd/khugepaged keeping overall fragmentation low.

kcompactd can handle and should focus on the long term goals.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#84 lore raw
In reply to Vlastimil Babka
Quoted 25 lines · Vlastimil Babka
Quoted 15 lines · Joonsoo Kim
Quoted 15 lines

[...]

Quoted 12 lines
Quoted 7 lines
Quoted 4 lines

At least, reset no_progress_loops when did_some_progress. High order allocation up to PAGE_ALLOC_COSTLY_ORDER is as important as order 0. And, reclaim something would increase probability of compaction success.

This is something I still do not understand. Why would reclaiming random order-0 pages help compaction? Could you clarify this please?

I just can tell simple version. Please check the link from me on another reply. Compaction could scan more range of memory if we have more freepage. This is due to algorithm limitation. Anyway, so, reclaiming random order-0 pages helps compaction.

Quoted 8 lines
Quoted 3 lines

I will have a look at that code but this just doesn't make any sense. The compaction should be reshuffling pages, this shouldn't be a function of free memory.

Please refer the link I mentioned before. There is a reason why more free memory would help compaction success. Compaction doesn't work like as random reshuffling. It has an algorithm to reduce system overall fragmentation so there is limitation.

Quoted 5 lines

I proposed another way to get better results from direct compaction - don't scan for free pages but get them directly from freelists:

https://lkml.org/lkml/2015/12/3/60

I think that major problem of this approach is that there is no way to prevent other parallel compacting thread from taking freepage on targetted aligned block. So, if there are parallel compaction requestors, they would disturb each others. However, it would not be a problem for order up to PAGE_ALLOC_COSTLY_ORDER which would be finished so soon.

In fact, for quick allocation, migration scanner is also unnecessary. There would be a lot of pageblock we cannot do migration. Scanning all of them in this situation is unnecessary and costly. Moreover, scanning only half of zone due to limitation of compaction algorithm also looks not good. Instead, we can get base page on lru list and migrate neighborhood pages. I named this idea as "lumpy compaction" but didn't try it. If we only focus on quick allocation, this would be a better way. Any thought?

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#85 lore raw
In reply to Michal Hocko
Quoted 77 lines · Michal Hocko
Quoted 1 line · Joonsoo Kim

On Mon, Feb 29, 2016 at 10:02:13PM +0100, Michal Hocko wrote:

[...]

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
(quoted diff) +9 -0
+	/*
+	 * OK, so the watermak check has failed. Make sure we do all the
+	 * retries for !costly high order requests and hope that multiple
+	 * runs of compaction will generate some high order ones for us.
+	 *
+	 * XXX: ideally we should teach the compaction to try _really_ hard
+	 * if we are in the retry path - something like priority 0 for the
+	 * reclaim
+	 */
Quoted 8 lines
Quoted 4 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER)
+		return true;
+
 	return false;

This seems not a proper fix. Checking watermark with high order has another meaning that there is high order page or not. This isn't what we want here.

Why not? Why should we retry the reclaim if we do not have >=order page available? Reclaim itself doesn't guarantee any of the freed pages will form the requested order. The ordering on the LRU lists is pretty much random wrt. pfn ordering. On the other hand if we have a page available which is just hidden by watermarks then it makes perfect sense to retry and free even order-0 pages.

Quoted 10 lines

So, following fix is needed.

'if (order)' check isn't needed. It is used to clarify the meaning of this fix. You can remove it.

mm/page_alloc.c +0 -0
index 1993894..8c80375 100644

mm/page_alloc.c should_reclaim_retry(gfp_t gfp_mask, unsigned order, · line 3131

Quoted 7 lines
mm/page_alloc.c +4 -0
@@ -3125,6 +3125,10 @@ should_reclaim_retry(gfp_t gfp_mask, unsigned order,
        if (order > PAGE_ALLOC_COSTLY_ORDER && !(gfp_mask & __GFP_REPEAT))
                return false;
 
+       /* To check whether compaction is available or not */
+       if (order)
+               order = 0;
+

This would enforce the order 0 wmark check which is IMHO not correct as per above.

Quoted 19 lines
mm/page_alloc.c +0 -0
        /*
         * Keep reclaiming pages while there is a chance this will lead
         * somewhere.  If none of the target zones can satisfy our allocation
Quoted 16 lines
Quoted 16 lines

}

(quoted diff) +4 -4
@@ -3281,11 +3293,11 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		goto noretry;
 
 	/*
-	 * Costly allocations might have made a progress but this doesn't mean
-	 * their order will become available due to high fragmentation so do
-	 * not reset the no progress counter for them
+	 * High order allocations might have made a progress but this doesn't
+	 * mean their order will become available due to high fragmentation so
+	 * do not reset the no progress counter for them
 	 */
-	if (did_some_progress && order <= PAGE_ALLOC_COSTLY_ORDER)
+	if (did_some_progress && !order)
Quoted 8 lines
Quoted 3 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3303

Quoted 3 lines
(quoted diff) +0 -0
 		no_progress_loops = 0;
 	else
 		no_progress_loops++;

This unconditionally increases no_progress_loops for high order allocation, so, after 16 iterations, it will fail. If compaction isn't enabled in Kconfig, 16 times reclaim attempt would not be sufficient to make high order page. Should we consider this case also?

Quoted 4 lines

How many retries would help? I do not think any number will work reliably. Configurations without compaction enabled are asking for problems by definition IMHO. Relying on order-0 reclaim for high order allocations simply cannot work.

I left compaction code for a long time so a super hero might make it perfect now but I don't think the dream come true yet and I believe any algorithm has a drawback so we end up relying on a fallback approach in case of not working compaction correctly.

My suggestion is to reintroduce *lumpy reclaim* and kicks in only when compaction gave up by some reasons. It would be better to rely on random number retrial of reclaim.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#86 lore raw

PATCH 1/1 mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

In reply to Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Other changes in the compaction are still needed but I would like to not depend on them right now.
---

#87 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Hello Michal,

Quoted 11 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

(next-20160307 + "[PATCH] mm, oom: protect !costly allocations some more")

seems it's significantly less likely to oom-kill now, but I still can see something like this

[  501.942745] coretemp-sensor invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[  501.942796] CPU: 3 PID: 409 Comm: coretemp-sensor Not tainted 4.5.0-rc6-next-20160307-dbg-00015-g8a56edd-dirty #250
[  501.942801]  0000000000000000 ffff88013114fb88 ffffffff812364e9 0000000000000000
[  501.942804]  ffff88013114fd28 ffff88013114fbf8 ffffffff8113b11c ffff88013114fba8
[  501.942807]  ffffffff810835c1 ffff88013114fbc8 0000000000000206 ffffffff81a46de0
[  501.942808] Call Trace:
[  501.942813]  [<ffffffff812364e9>] dump_stack+0x67/0x90
[  501.942817]  [<ffffffff8113b11c>] dump_header.isra.5+0x54/0x359
[  501.942820]  [<ffffffff810835c1>] ? trace_hardirqs_on+0xd/0xf
[  501.942823]  [<ffffffff810f97c2>] oom_kill_process+0x89/0x503
[  501.942825]  [<ffffffff810f9ffe>] out_of_memory+0x372/0x38d
[  501.942827]  [<ffffffff810fe5ae>] __alloc_pages_nodemask+0x9b6/0xa92
[  501.942830]  [<ffffffff810fe882>] alloc_kmem_pages_node+0x1b/0x1d
[  501.942833]  [<ffffffff81041f86>] copy_process.part.9+0xfe/0x17f4
[  501.942835]  [<ffffffff810858f6>] ? lock_acquire+0x10f/0x1a3
[  501.942837]  [<ffffffff8104380f>] _do_fork+0xbd/0x5da
[  501.942838]  [<ffffffff81083598>] ? trace_hardirqs_on_caller+0x16c/0x188
[  501.942842]  [<ffffffff81001a79>] ? do_syscall_64+0x18/0xe6
[  501.942844]  [<ffffffff81043db2>] SyS_clone+0x19/0x1b
[  501.942845]  [<ffffffff81001abb>] do_syscall_64+0x5a/0xe6
[  501.942848]  [<ffffffff8151245a>] entry_SYSCALL64_slow_path+0x25/0x25
[  501.942850] Mem-Info:
[  501.942853] active_anon:151312 inactive_anon:54791 isolated_anon:0
                active_file:31213 inactive_file:302048 isolated_file:0
                unevictable:0 dirty:44 writeback:221 unstable:0
                slab_reclaimable:43570 slab_unreclaimable:5651
                mapped:16660 shmem:29495 pagetables:2542 bounce:0
                free:10884 free_pcp:214 free_cma:0
[  501.942859] DMA free:14896kB min:28kB low:40kB high:52kB active_anon:0kB inactive_anon:0kB active_file:96kB inactive_file:104kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15984kB managed:15900kB mlocked:0kB dirty:0kB writeback:0kB mapped:124kB shmem:0kB slab_reclaimable:28kB slab_unreclaimable:108kB kernel_stack:16kB pagetables:0kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  501.942862] lowmem_reserve[]: 0 3031 3855 3855
[  501.942867] DMA32 free:23664kB min:6232kB low:9332kB high:12432kB active_anon:516228kB inactive_anon:129136kB active_file:96508kB inactive_file:954780kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3194880kB managed:3107512kB mlocked:0kB dirty:136kB writeback:440kB mapped:51816kB shmem:91488kB slab_reclaimable:129856kB slab_unreclaimable:13876kB kernel_stack:2160kB pagetables:7888kB unstable:0kB bounce:0kB free_pcp:724kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:128 all_unreclaimable? no
[  501.942870] lowmem_reserve[]: 0 0 824 824
[  501.942876] Normal free:4784kB min:1696kB low:2540kB high:3384kB active_anon:89020kB inactive_anon:90028kB active_file:28248kB inactive_file:253308kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:917504kB managed:844512kB mlocked:0kB dirty:40kB writeback:444kB mapped:14700kB shmem:26492kB slab_reclaimable:44396kB slab_unreclaimable:8620kB kernel_stack:1328kB pagetables:2280kB unstable:0kB bounce:0kB free_pcp:244kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:60 all_unreclaimable? no
[  501.942879] lowmem_reserve[]: 0 0 0 0
[  501.942902] DMA: 6*4kB (UME) 3*8kB (M) 2*16kB (UM) 3*32kB (ME) 2*64kB (ME) 2*128kB (ME) 2*256kB (UE) 3*512kB (UME) 2*1024kB (ME) 1*2048kB (E) 2*4096kB (M) = 14896kB
[  501.942912] DMA32: 564*4kB (UME) 2700*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 23856kB
[  501.942921] Normal: 959*4kB (ME) 128*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 4860kB
[  501.942922] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[  501.942923] 362670 total pagecache pages
[  501.942924] 0 pages in swap cache
[  501.942926] Swap cache stats: add 150, delete 150, find 0/0
[  501.942926] Free swap  = 8388504kB
[  501.942927] Total swap = 8388604kB
[  501.942928] 1032092 pages RAM
[  501.942928] 0 pages HighMem/MovableOnly
[  501.942929] 40111 pages reserved
[  501.942930] 0 pages hwpoisoned
[  501.942930] [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name
[  501.942935] [  162]     0   162    15823     1065      35       3        0             0 systemd-journal
[  501.942991] [  186]     0   186     8586     1054      19       3        0         -1000 systemd-udevd
[  501.942993] [  287]     0   287     3557      651      12       3        0             0 crond
[  501.942995] [  288]    81   288     8159      775      20       3        0          -900 dbus-daemon
[  501.942997] [  289]     0   289     3843      518      13       3        0             0 systemd-logind
[  501.942999] [  294]     0   294    22455      856      47       3        0             0 login
[  501.943001] [  302]  1000   302     8481     1029      20       3        0             0 systemd
[  501.943003] [  304]  1000   304    24212      438      47       3        0             0 (sd-pam)
[  501.943005] [  309]  1000   309     4431     1123      14       3        0             0 bash
[  501.943007] [  316]  1000   316     3712      764      13       3        0             0 startx
[  501.943009] [  338]  1000   338     3976      255      14       3        0             0 xinit
[  501.943012] [  339]  1000   339    44397    11311      90       3        0             0 Xorg
[  501.943014] [  341]  1000   341    39703     4045      78       3        0             0 openbox
[  501.943016] [  352]  1000   352    43465     2997      86       4        0             0 tint2
[  501.943018] [  355]  1000   355    33962     4351      57       3        0             0 urxvt
[  501.943020] [  356]  1000   356     4466     1155      13       3        0             0 bash
[  501.943022] [  359]  1000   359     4433     1116      13       3        0             0 bash
[  501.943024] [  364]  1000   364    49365     6236      62       3        0             0 urxvt
[  501.943026] [  365]  1000   365     4433     1093      15       3        0             0 bash
[  501.943028] [  368]  1000   368     5203      745      15       3        0             0 tmux
[  501.943030] [  370]  1000   370     6336     1374      17       3        0             0 tmux
[  501.943046] [  371]  1000   371     4433     1100      14       3        0             0 bash
[  501.943049] [  378]  1000   378     4433     1115      13       3        0             0 bash
[  501.943051] [  381]  1000   381     5203      763      16       3        0             0 tmux
[  501.943053] [  382]  1000   382     4433     1089      15       3        0             0 bash
[  501.943055] [  389]  1000   389     4433     1078      15       3        0             0 bash
[  501.943057] [  392]  1000   392     4433     1078      15       3        0             0 bash
[  501.943058] [  395]  1000   395     4433     1090      14       3        0             0 bash
[  501.943060] [  398]  1000   398     4433     1111      14       3        0             0 bash
[  501.943062] [  401]  1000   401    10126     1010      25       3        0             0 top
[  501.943064] [  403]  1000   403     4433     1129      14       3        0             0 bash
[  501.943066] [  409]  1000   409     3740      786      13       3        0             0 coretemp-sensor
[  501.943069] [  443]  1000   443    25873     3141      51       3        0             0 urxvt
[  501.943071] [  444]  1000   444     4433     1110      13       3        0             0 bash
[  501.943073] [  447]  1000   447    68144    55547     138       3        0             0 mutt
[  501.943075] [  450]  1000   450    29966     3825      51       3        0             0 urxvt
[  501.943077] [  451]  1000   451     4433     1117      14       3        0             0 bash
[  501.943079] [  456]  1000   456    29967     3793      53       3        0             0 urxvt
[  501.943081] [  457]  1000   457     4433     1085      14       3        0             0 bash
[  501.943083] [  462]  1000   462    29967     3845      51       4        0             0 urxvt
[  501.943085] [  463]  1000   463     4433     1093      14       3        0             0 bash
[  501.943087] [  468]  1000   468    29967     3793      50       3        0             0 urxvt
[  501.943089] [  469]  1000   469     4433     1086      15       3        0             0 bash
[  501.943091] [  493]  1000   493    52976     6416      69       3        0             0 urxvt
[  501.943093] [  494]  1000   494     4433     1106      14       3        0             0 bash
[  501.943095] [  499]  1000   499    29966     3792      54       3        0             0 urxvt
[  501.943097] [  500]  1000   500     4433     1078      14       3        0             0 bash
[  501.943099] [  525]     0   525    17802     1108      38       3        0             0 sudo
[  501.943101] [  528]     0   528   186583      768     207       4        0             0 journalctl
[  501.943103] [  550]  1000   550    42144     9259      66       4        0             0 urxvt
[  501.943105] [  551]  1000   551     4433     1067      14       4        0             0 bash
[  501.943107] [  557]  1000   557    11115      768      27       3        0             0 su
[  501.943109] [  579]     0   579     4462     1148      13       3        0             0 bash
[  501.943111] [  963]  1000   963     4433     1075      14       3        0             0 bash
[  501.943113] [  981]  1000   981     4433     1114      13       3        0             0 bash
[  501.943115] [  993]  1000   993     4432     1118      14       3        0             0 bash
[  501.943117] [ 1062]  1000  1062     5203      734      15       3        0             0 tmux
[  501.943119] [ 1063]  1000  1063    13805    10479      32       3        0             0 bash
[  501.943121] [ 1145]  1000  1145     4466     1144      14       3        0             0 bash
[  501.943123] [ 4331]  1000  4331   287422    64040     429       4        0             0 firefox
[  501.943125] [ 4440]  1000  4440     8132      761      20       3        0             0 dbus-daemon
[  501.943127] [ 4470]  1000  4470    83823      934      31       4        0             0 at-spi-bus-laun
[  501.943129] [17875]  1000 17875     7549     1926      20       3        0             0 vim
[  501.943131] [27066]  1000 27066     4432     1120      15       3        0             0 bash
[  501.943133] [27073]  1000 27073     4432     1071      13       3        0             0 bash
[  501.943135] [27079]  1000 27079     4432     1077      15       3        0             0 bash
[  501.943137] [27085]  1000 27085     4432     1080      14       3        0             0 bash
[  501.943139] [27091]  1000 27091     4432     1091      14       3        0             0 bash
[  501.943141] [27097]  1000 27097     4432     1096      15       3        0             0 bash
[  501.943143] [ 1235]     0  1235     3745      809      11       3        0             0 zram-test.sh
[  501.943145] [ 2316]  1000  2316     1759      166       9       3        0             0 sleep
[  501.943147] [ 2323]     0  2323     3302     1946      12       3        0             0 dd
[  501.943148] Out of memory: Kill process 4331 (firefox) score 20 or sacrifice child
[  501.943352] Killed process 4331 (firefox) total-vm:1149688kB, anon-rss:207844kB, file-rss:48172kB, shmem-rss:516kB
	-ss

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#88 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Quoted 15 lines · Sergey Senozhatsky

Hello Michal,

Quoted 11 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Quoted 4 lines

(next-20160307 + "[PATCH] mm, oom: protect !costly allocations some more")

seems it's significantly less likely to oom-kill now, but I still can see something like this

Thanks for the testing. This is highly appreciated. If you are able to reproduce this then collecting compaction related tracepoints might be really helpful.

Quoted 1 line
[  501.942745] coretemp-sensor invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0

[...]

Quoted 6 lines
[  501.942853] active_anon:151312 inactive_anon:54791 isolated_anon:0
                active_file:31213 inactive_file:302048 isolated_file:0
                unevictable:0 dirty:44 writeback:221 unstable:0
                slab_reclaimable:43570 slab_unreclaimable:5651
                mapped:16660 shmem:29495 pagetables:2542 bounce:0
                free:10884 free_pcp:214 free_cma:0

[...]

Quoted 3 lines
[  501.942867] DMA32 free:23664kB min:6232kB low:9332kB high:12432kB active_anon:516228kB inactive_anon:129136kB active_file:96508kB inactive_file:954780kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3194880kB managed:3107512kB mlocked:0kB dirty:136kB writeback:440kB mapped:51816kB shmem:91488kB slab_reclaimable:129856kB slab_unreclaimable:13876kB kernel_stack:2160kB pagetables:7888kB unstable:0kB bounce:0kB free_pcp:724kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:128 all_unreclaimable? no
[  501.942870] lowmem_reserve[]: 0 0 824 824
[  501.942876] Normal free:4784kB min:1696kB low:2540kB high:3384kB active_anon:89020kB inactive_anon:90028kB active_file:28248kB inactive_file:253308kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:917504kB managed:844512kB mlocked:0kB dirty:40kB writeback:444kB mapped:14700kB shmem:26492kB slab_reclaimable:44396kB slab_unreclaimable:8620kB kernel_stack:1328kB pagetables:2280kB unstable:0kB bounce:0kB free_pcp:244kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:60 all_unreclaimable? no

Both DMA32 and Normal zones are over high watermarks so this OOM is due to the memory fragmentation.

Quoted 2 lines
[  501.942912] DMA32: 564*4kB (UME) 2700*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 23856kB
[  501.942921] Normal: 959*4kB (ME) 128*8kB (UM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 4860kB

There are no order-2+ pages usable even after we know that the compaction was active and didn't back out early. I might be missing something of course and the patch might still be tweaked to be more conservative. Tracepoints should tell us more though.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#89 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Quoted 16 lines · Michal Hocko
Quoted 15 lines · Sergey Senozhatsky

Hello Michal,

Quoted 11 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Quoted 9 lines
Quoted 4 lines

(next-20160307 + "[PATCH] mm, oom: protect !costly allocations some more")

seems it's significantly less likely to oom-kill now, but I still can see something like this

Thanks for the testing. This is highly appreciated. If you are able to reproduce this then collecting compaction related tracepoints might be really helpful.

oh, wow... compaction is disabled, somehow.

  $ zcat /proc/config.gz | grep -i CONFIG_COMPACTION
  # CONFIG_COMPACTION is not set

I should have checked that, sorry!

will enable and re-test.

	-ss

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#90 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

In reply to Michal Hocko
Quoted 80 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Other changes in the compaction are still needed but I would like to not depend on them right now.
---
From 0974f127e8eb7fe53e65f3a8b398db57effe9755 Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Mon, 7 Mar 2016 15:30:37 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or it told us to continue.

Additionally define COMPACT_NONE which reflects cases where the compaction is completely disabled.

---
 include/linux/compaction.h |  2 ++
 mm/page_alloc.c            | 41 ++++++++++++++++++++++++-----------------
 2 files changed, 26 insertions(+), 17 deletions(-)
include/linux/compaction.h +2 -0
index 4cd4ddf64cc7..a4cec4a03f7d 100644
@@ -1,6 +1,8 @@
 #ifndef _LINUX_COMPACTION_H
 #define _LINUX_COMPACTION_H
 
+/* compaction disabled */
+#define COMPACT_NONE		-1
 /* Return values for compact_zone() and try_to_compact_pages() */
 /* compaction didn't start as it was deferred due to past failures */
 #define COMPACT_DEFERRED	0
mm/page_alloc.c +4 -4
index 269a04f20927..f89e3cbfdf90 100644
@@ -2819,28 +2819,22 @@ static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		unsigned long *compact_result)
 {
-	unsigned long compact_result;
 	struct page *page;
 
-	if (!order)
+	if (!order) {
+		*compact_result = COMPACT_NONE;
 		return NULL;
+	}
 
 	current->flags |= PF_MEMALLOC;
-	compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,

mm/page_alloc.c static struct page * · line 2836

Quoted 10 lines
mm/page_alloc.c +2 -5
+	*compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
-	switch (compact_result) {
-	case COMPACT_DEFERRED:
-		*deferred_compaction = true;
-		/* fall-through */
-	case COMPACT_SKIPPED:
+	if (*compact_result <= COMPACT_SKIPPED)

COMPACT_NONE is -1 and compact_result is unsigned long, so this won't work as expected.

Quoted 46 lines
mm/page_alloc.c +5 -7
 		return NULL;
-	default:
-		break;
-	}
 
 	/*
 	 * At least in one zone compaction wasn't deferred or skipped, so let's
@@ -2875,8 +2869,9 @@ static inline struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		unsigned long *compact_result)
 {
+	*compact_result = COMPACT_NONE;
 	return NULL;
 }
 #endif /* CONFIG_COMPACTION */
@@ -3118,7 +3113,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	int alloc_flags;
 	unsigned long did_some_progress;
 	enum migrate_mode migration_mode = MIGRATE_ASYNC;
-	bool deferred_compaction = false;
+	unsigned long compact_result;
 	int contended_compaction = COMPACT_CONTENDED_NONE;
 	int no_progress_loops = 0;
 
@@ -3227,7 +3222,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags, ac,
 					migration_mode,
 					&contended_compaction,
-					&deferred_compaction);
+					&compact_result);
 	if (page)
 		goto got_pg;
 
@@ -3240,7 +3235,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		 * to heavily disrupt the system, so we fail the allocation
 		 * instead of entering direct reclaim.
 		 */
-		if (deferred_compaction)
+		if (compact_result == COMPACT_DEFERRED)
 			goto nopage;
 
 		/*

mm/page_alloc.c __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3298

Quoted 10 lines
mm/page_alloc.c +7 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

Quoted 10 lines
mm/page_alloc.c +5 -0
+			goto retry;
+		if (contended_compaction > COMPACT_CONTENDED_NONE)
+			goto retry;
+	}
+
 	/* Reclaim has failed us, start killing things */
 	page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);
 	if (page)

mm/page_alloc.c __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3327

Quoted 8 lines
mm/page_alloc.c +1 -1
@@ -3314,7 +3321,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags,
 					    ac, migration_mode,
 					    &contended_compaction,
-					    &deferred_compaction);
+					    &compact_result);
 	if (page)
 		goto got_pg;
 nopage:

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#91 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

On (03/08/16 10:24), Vlastimil Babka wrote:
[..]

Quoted 4 lines
Quoted 4 lines
(quoted diff) +0 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
Quoted 11 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3298

Quoted 7 lines
(quoted diff) +7 -0
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

I'm sorry, my test is not correct. I have disabled compaction last weeked on purpose - to provoke more OOM-kills and OOM conditions for reworked printk() patch set testing (http://marc.info/?l=linux-kernel&m=145734549308803); and I forgot to re-enable it.

	-ss

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#92 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

In reply to Vlastimil Babka

On Tue 08-03-16 10:24:56, Vlastimil Babka wrote:
[...]

Quoted 22 lines
Quoted 22 lines
(quoted diff) +5 -4
@@ -2819,28 +2819,22 @@ static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		unsigned long *compact_result)
 {
-	unsigned long compact_result;
 	struct page *page;
 
-	if (!order)
+	if (!order) {
+		*compact_result = COMPACT_NONE;
 		return NULL;
+	}
 
 	current->flags |= PF_MEMALLOC;
-	compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
+	*compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
Quoted 9 lines

Review static struct page * · line 2836

Quoted 6 lines
(quoted diff) +1 -5
-	switch (compact_result) {
-	case COMPACT_DEFERRED:
-		*deferred_compaction = true;
-		/* fall-through */
-	case COMPACT_SKIPPED:
+	if (*compact_result <= COMPACT_SKIPPED)

COMPACT_NONE is -1 and compact_result is unsigned long, so this won't work as expected.

Well, COMPACT_NONE is documented as /* compaction disabled */ so we should never get it from try_to_compact_pages.

[...]

Quoted 4 lines
Quoted 4 lines
(quoted diff) +0 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
Quoted 11 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3298

Quoted 7 lines
(quoted diff) +7 -0
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

This should work as expected because compact_result is unsigned long and so this is the unsigned arithmetic. I can make
#define COMPACT_NONE -1UL

to make the intention more obvious if you prefer, though.

Thanks for the review.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#93 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

Quoted 34 lines · Michal Hocko

On Tue 08-03-16 10:24:56, Vlastimil Babka wrote:
[...]

Quoted 22 lines
Quoted 22 lines
(quoted diff) +5 -4
@@ -2819,28 +2819,22 @@ static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		unsigned long *compact_result)
 {
-	unsigned long compact_result;
 	struct page *page;
 
-	if (!order)
+	if (!order) {
+		*compact_result = COMPACT_NONE;
 		return NULL;
+	}
 
 	current->flags |= PF_MEMALLOC;
-	compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
+	*compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
Quoted 9 lines

Review static struct page * · line 2836

Quoted 6 lines
(quoted diff) +1 -5
-	switch (compact_result) {
-	case COMPACT_DEFERRED:
-		*deferred_compaction = true;
-		/* fall-through */
-	case COMPACT_SKIPPED:
+	if (*compact_result <= COMPACT_SKIPPED)

COMPACT_NONE is -1 and compact_result is unsigned long, so this won't work as expected.

Quoted 2 lines

Well, COMPACT_NONE is documented as /* compaction disabled */ so we should never get it from try_to_compact_pages.

Right.

Quoted 18 lines

[...]

Quoted 4 lines
Quoted 4 lines
(quoted diff) +0 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
Quoted 11 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3298

Quoted 7 lines
(quoted diff) +7 -0
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

Quoted 5 lines

This should work as expected because compact_result is unsigned long and so this is the unsigned arithmetic. I can make
#define COMPACT_NONE -1UL

to make the intention more obvious if you prefer, though.

Well, what wasn't obvious to me is actually that here (unlike in the test above) it was actually intended that COMPACT_NONE doesn't result in a retry. But it makes sense, otherwise we would retry endlessly if reclaim couldn't form a higher-order page, right.

Quoted 2 lines

Thanks for the review.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#94 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

Quoted 1 line · Vlastimil Babka

On 03/08/2016 10:46 AM, Michal Hocko wrote:

[...]

Quoted 20 lines
Quoted 20 lines
Quoted 4 lines
Quoted 4 lines
(quoted diff) +0 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
Quoted 11 lines

Review __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3298

Quoted 7 lines
(quoted diff) +7 -0
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

This should work as expected because compact_result is unsigned long and so this is the unsigned arithmetic. I can make
#define COMPACT_NONE -1UL

Quoted 6 lines
Quoted 1 line

to make the intention more obvious if you prefer, though.

Well, what wasn't obvious to me is actually that here (unlike in the test above) it was actually intended that COMPACT_NONE doesn't result in a retry. But it makes sense, otherwise we would retry endlessly if reclaim couldn't form a higher-order page, right.

Yeah, that was the whole point. An alternative would be moving the test into should_compact_retry(order, compact_result, contended_compaction) which would be CONFIG_COMPACTION specific so we can get rid of the COMPACT_NONE altogether. Something like the following. We would lose the always initialized compact_result but this would matter only for order==0 and we check for that. Even gcc doesn't complain.

A more important question is whether the criteria I have chosen are reasonable and reasonably independent on the particular implementation of the compaction. I still cannot convince myself about the convergence here. Is it possible that the compaction would keep returning compact_result <= COMPACT_CONTINUE while not making any progress at all?

Sure we can see a case where somebody is stealing the compacted blocks but that is very same with the order-0 where parallel mem eaters will piggy back on the reclaimer and there is no upper boundary as well well.

---

include/linux/compaction.h +0 -2
index a4cec4a03f7d..4cd4ddf64cc7 100644
@@ -1,8 +1,6 @@
 #ifndef _LINUX_COMPACTION_H
 #define _LINUX_COMPACTION_H
 
-/* compaction disabled */
-#define COMPACT_NONE		-1
 /* Return values for compact_zone() and try_to_compact_pages() */
 /* compaction didn't start as it was deferred due to past failures */
 #define COMPACT_DEFERRED	0
mm/page_alloc.c +29 -15

71 lines not on this page · raw message

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#95 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

Quoted 30 lines · Michal Hocko
Quoted 1 line · Vlastimil Babka

On 03/08/2016 10:46 AM, Michal Hocko wrote:

[...]

Quoted 20 lines
Quoted 20 lines
Quoted 4 lines
Quoted 4 lines
mm/page_alloc.c +0 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
Quoted 11 lines

mm/page_alloc.c __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order, · line 2876

Quoted 7 lines
mm/page_alloc.c +7 -0
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

This should work as expected because compact_result is unsigned long and so this is the unsigned arithmetic. I can make
#define COMPACT_NONE -1UL

Quoted 6 lines
Quoted 1 line

to make the intention more obvious if you prefer, though.

Well, what wasn't obvious to me is actually that here (unlike in the test above) it was actually intended that COMPACT_NONE doesn't result in a retry. But it makes sense, otherwise we would retry endlessly if reclaim couldn't form a higher-order page, right.

Quoted 6 lines

Yeah, that was the whole point. An alternative would be moving the test into should_compact_retry(order, compact_result, contended_compaction) which would be CONFIG_COMPACTION specific so we can get rid of the COMPACT_NONE altogether. Something like the following. We would lose the always initialized compact_result but this would matter only for order==0 and we check for that. Even gcc doesn't complain.

Yeah I like this version better, you can add my Acked-By.

Thanks.

Quoted 5 lines

A more important question is whether the criteria I have chosen are reasonable and reasonably independent on the particular implementation of the compaction. I still cannot convince myself about the convergence here. Is it possible that the compaction would keep returning compact_result <= COMPACT_CONTINUE while not making any progress at all?

Theoretically, if reclaim/compaction suitability decisions and allocation attempts didn't match the watermark checks, including the alloc_flags and classzone_idx parameters. Possible scenarios:

- reclaim thinks compaction has enough to proceed, but compaction thinks otherwise and returns COMPACT_SKIPPED
- compaction thinks it succeeded and returns COMPACT_PARTIAL, but allocation attempt fails
- and perhaps some other combinations

Quoted 3 lines

Sure we can see a case where somebody is stealing the compacted blocks but that is very same with the order-0 where parallel mem eaters will piggy back on the reclaimer and there is no upper boundary as well well.

Yep.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#96 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

Quoted 31 lines · Vlastimil Babka
Quoted 30 lines · Michal Hocko
Quoted 1 line · Vlastimil Babka

On 03/08/2016 10:46 AM, Michal Hocko wrote:

[...]

Quoted 26 lines
Quoted 16 lines
Quoted 4 lines
Quoted 4 lines
mm/page_alloc.c +0 -0
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
Quoted 11 lines

mm/page_alloc.c __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order, · line 2876

Quoted 7 lines
mm/page_alloc.c +7 -0
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_CONTINUE)

Same here.
I was going to say that this didn't have effect on Sergey's test, but turns out it did :)

Quoted 5 lines

This should work as expected because compact_result is unsigned long and so this is the unsigned arithmetic. I can make
#define COMPACT_NONE -1UL

to make the intention more obvious if you prefer, though.

Well, what wasn't obvious to me is actually that here (unlike in the test above) it was actually intended that COMPACT_NONE doesn't result in a retry. But it makes sense, otherwise we would retry endlessly if reclaim couldn't form a higher-order page, right.

Quoted 8 lines
Quoted 6 lines

Yeah, that was the whole point. An alternative would be moving the test into should_compact_retry(order, compact_result, contended_compaction) which would be CONFIG_COMPACTION specific so we can get rid of the COMPACT_NONE altogether. Something like the following. We would lose the always initialized compact_result but this would matter only for order==0 and we check for that. Even gcc doesn't complain.

Yeah I like this version better, you can add my Acked-By.

OK, patch updated and I will post it as a reply to the original email.

Quoted 12 lines

Thanks.

Quoted 5 lines

A more important question is whether the criteria I have chosen are reasonable and reasonably independent on the particular implementation of the compaction. I still cannot convince myself about the convergence here. Is it possible that the compaction would keep returning compact_result <= COMPACT_CONTINUE while not making any progress at all?

Theoretically, if reclaim/compaction suitability decisions and allocation attempts didn't match the watermark checks, including the alloc_flags and classzone_idx parameters. Possible scenarios:

Quoted 5 lines

- reclaim thinks compaction has enough to proceed, but compaction thinks otherwise and returns COMPACT_SKIPPED
- compaction thinks it succeeded and returns COMPACT_PARTIAL, but allocation attempt fails
- and perhaps some other combinations

But that might happen right now as well so it wouldn't be a regression, right?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#97 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more

Quoted 12 lines · Michal Hocko
Quoted 12 lines

Thanks.

Quoted 5 lines

A more important question is whether the criteria I have chosen are reasonable and reasonably independent on the particular implementation of the compaction. I still cannot convince myself about the convergence here. Is it possible that the compaction would keep returning compact_result <= COMPACT_CONTINUE while not making any progress at all?

Theoretically, if reclaim/compaction suitability decisions and allocation attempts didn't match the watermark checks, including the alloc_flags and classzone_idx parameters. Possible scenarios:

Quoted 8 lines
Quoted 5 lines

- reclaim thinks compaction has enough to proceed, but compaction thinks otherwise and returns COMPACT_SKIPPED
- compaction thinks it succeeded and returns COMPACT_PARTIAL, but allocation attempt fails
- and perhaps some other combinations

But that might happen right now as well so it wouldn't be a regression, right?

Maybe, somehow, I didn't study closely how the retry decisions work. Your patch adds another way to retry so it's theoretically more dangerous. Just hinting at what to possibly check (the watermark checks) :)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Tested-by on v1 1/1
#98 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

In reply to Michal Hocko
Quoted 6 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 8 lines

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Other changes in the compaction are still needed but I would like to not depend on them right now.

works fine for me.

$  cat /proc/vmstat | egrep -e "compact|swap"
pgsteal_kswapd_dma 7
pgsteal_kswapd_dma32 6457075
pgsteal_kswapd_normal 1462767
pgsteal_kswapd_movable 0
pgscan_kswapd_dma 18
pgscan_kswapd_dma32 6544126
pgscan_kswapd_normal 1495604
pgscan_kswapd_movable 0
kswapd_inodesteal 29
kswapd_low_wmark_hit_quickly 1168
kswapd_high_wmark_hit_quickly 1627
compact_migrate_scanned 5762793
compact_free_scanned 54090239
compact_isolated 1303895
compact_stall 1542
compact_fail 1117
compact_success 425
compact_kcompatd_wake 0

no OOM-kills after 6 rounds of tests.

thanks!

	-ss

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#99 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Quoted 38 lines · Sergey Senozhatsky
Quoted 6 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 8 lines

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Other changes in the compaction are still needed but I would like to not depend on them right now.

works fine for me.

$  cat /proc/vmstat | egrep -e "compact|swap"
pgsteal_kswapd_dma 7
pgsteal_kswapd_dma32 6457075
pgsteal_kswapd_normal 1462767
pgsteal_kswapd_movable 0
pgscan_kswapd_dma 18
pgscan_kswapd_dma32 6544126
pgscan_kswapd_normal 1495604
pgscan_kswapd_movable 0
kswapd_inodesteal 29
kswapd_low_wmark_hit_quickly 1168
kswapd_high_wmark_hit_quickly 1627
compact_migrate_scanned 5762793
compact_free_scanned 54090239
compact_isolated 1303895
compact_stall 1542
compact_fail 1117
compact_success 425
compact_kcompatd_wake 0
Quoted 3 lines

no OOM-kills after 6 rounds of tests.

Thanks for retesting!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#100 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

In reply to Michal Hocko
Quoted 11 lines · Michal Hocko
Quoted 4 lines · Michal Hocko

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Did not help with my load at all, I'm afraid: quite the reverse, OOMed very much sooner (as usual on order=2), and with much more noise (multiple OOMs) than your previous patch.

vmstats.xz attached; sorry, I don't have tracing built in, and must move on to the powerpc issue before going back to bed.

I do hate replying without having something constructive to say, but have very little time to think about this, and no bright ideas so far.

I do not understand why it's so easy for me to reproduce, yet impossible for you - unless it's that you are still doing all your testing in a VM? Is Sergey the only other to see this issue?

Hugh

author reviewer
#101 lore raw

cover oom rework: high order enahncements

In reply to Michal Hocko

The first two patches are cleanups for the compaction and the second patch is updated as per Vlastimil's feedback. I didn't add his Acked-by because I have added COMPACT_SHOULD_RETRY to make the retry logic in the page allocator more robust for future changes.

Hugh has still reported this is not sufficient but I would prefer to handle the issue he is seeing in a separate patch once we understand what is going on there. The second patch sounds like a reasonable starting point to me.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#102 lore raw

PATCH 1/3 mm, compaction: change COMPACT_ constants into enum

From: Michal Hocko <[email protected]>

compaction code is doing weird dances between
COMPACT_FOO -> int -> unsigned long

but there doesn't seem to be any reason for that. All functions which return/use one of those constants are not expecting any other value so it really makes sense to define an enum for them and make it clear that no other values are expected.

This is a pure cleanup and shouldn't introduce any functional changes.

---
 include/linux/compaction.h | 45 +++++++++++++++++++++++++++------------------
 mm/compaction.c            | 27 ++++++++++++++-------------
 mm/page_alloc.c            |  2 +-
 3 files changed, 42 insertions(+), 32 deletions(-)
include/linux/compaction.h +27 -18
index 4cd4ddf64cc7..b167801187e7 100644
@@ -2,21 +2,29 @@
 #define _LINUX_COMPACTION_H
 
 /* Return values for compact_zone() and try_to_compact_pages() */
-/* compaction didn't start as it was deferred due to past failures */
-#define COMPACT_DEFERRED	0
-/* compaction didn't start as it was not possible or direct reclaim was more suitable */
-#define COMPACT_SKIPPED		1
-/* compaction should continue to another pageblock */
-#define COMPACT_CONTINUE	2
-/* direct compaction partially compacted a zone and there are suitable pages */
-#define COMPACT_PARTIAL		3
-/* The full zone was compacted */
-#define COMPACT_COMPLETE	4
-/* For more detailed tracepoint output */
-#define COMPACT_NO_SUITABLE_PAGE	5
-#define COMPACT_NOT_SUITABLE_ZONE	6
-#define COMPACT_CONTENDED		7
 /* When adding new states, please adjust include/trace/events/compaction.h */
+enum compact_result {
+	/* compaction didn't start as it was deferred due to past failures */
+	COMPACT_DEFERRED,
+	/*
+	 * compaction didn't start as it was not possible or direct reclaim
+	 * was more suitable
+	 */
+	COMPACT_SKIPPED,
+	/* compaction should continue to another pageblock */
+	COMPACT_CONTINUE,
+	/*
+	 * direct compaction partially compacted a zone and there are suitable
+	 * pages
+	 */
+	COMPACT_PARTIAL,
+	/* The full zone was compacted */
+	COMPACT_COMPLETE,
+	/* For more detailed tracepoint output */
+	COMPACT_NO_SUITABLE_PAGE,
+	COMPACT_NOT_SUITABLE_ZONE,
+	COMPACT_CONTENDED,
+};
 
 /* Used to signal whether compaction detected need_sched() or lock contention */
 /* No contention detected */
@@ -38,12 +46,13 @@ extern int sysctl_extfrag_handler(struct ctl_table *table, int write,
 extern int sysctl_compact_unevictable_allowed;
 
 extern int fragmentation_index(struct zone *zone, unsigned int order);
-extern unsigned long try_to_compact_pages(gfp_t gfp_mask, unsigned int order,
+extern enum compact_result try_to_compact_pages(gfp_t gfp_mask,
+			unsigned int order,
 			int alloc_flags, const struct alloc_context *ac,
 			enum migrate_mode mode, int *contended);
 extern void compact_pgdat(pg_data_t *pgdat, int order);
 extern void reset_isolation_suitable(pg_data_t *pgdat);
-extern unsigned long compaction_suitable(struct zone *zone, int order,
+extern enum compact_result compaction_suitable(struct zone *zone, int order,
 					int alloc_flags, int classzone_idx);
 
 extern void defer_compaction(struct zone *zone, int order);
@@ -53,7 +62,7 @@ extern void compaction_defer_reset(struct zone *zone, int order,
 extern bool compaction_restarting(struct zone *zone, int order);
 
 #else
-static inline unsigned long try_to_compact_pages(gfp_t gfp_mask,
+static inline enum compact_result try_to_compact_pages(gfp_t gfp_mask,
 			unsigned int order, int alloc_flags,
 			const struct alloc_context *ac,
 			enum migrate_mode mode, int *contended)
@@ -69,7 +78,7 @@ static inline void reset_isolation_suitable(pg_data_t *pgdat)
 {
 }
 
-static inline unsigned long compaction_suitable(struct zone *zone, int order,
+static inline enum compact_result compaction_suitable(struct zone *zone, int order,
 					int alloc_flags, int classzone_idx)
 {
 	return COMPACT_SKIPPED;
mm/compaction.c +14 -13
index 585de54dbe8c..0f61f12d82b6 100644
@@ -1195,7 +1195,7 @@ static inline bool is_via_compact_memory(int order)
 	return order == -1;
 }
 
-static int __compact_finished(struct zone *zone, struct compact_control *cc,
+static enum compact_result __compact_finished(struct zone *zone, struct compact_control *cc,
 			    const int migratetype)
 {
 	unsigned int order;
@@ -1258,8 +1258,9 @@ static int __compact_finished(struct zone *zone, struct compact_control *cc,
 	return COMPACT_NO_SUITABLE_PAGE;
 }
 
-static int compact_finished(struct zone *zone, struct compact_control *cc,
-			    const int migratetype)
+static enum compact_result compact_finished(struct zone *zone,
+			struct compact_control *cc,
+			const int migratetype)
 {
 	int ret;
 
@@ -1278,7 +1279,7 @@ static int compact_finished(struct zone *zone, struct compact_control *cc,
  *   COMPACT_PARTIAL  - If the allocation would succeed without compaction
  *   COMPACT_CONTINUE - If compaction should run now
  */
-static unsigned long __compaction_suitable(struct zone *zone, int order,
+static enum compact_result __compaction_suitable(struct zone *zone, int order,
 					int alloc_flags, int classzone_idx)
 {
 	int fragindex;
@@ -1323,10 +1324,10 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 	return COMPACT_CONTINUE;
 }
 
-unsigned long compaction_suitable(struct zone *zone, int order,
+enum compact_result compaction_suitable(struct zone *zone, int order,
 					int alloc_flags, int classzone_idx)
 {
-	unsigned long ret;
+	enum compact_result ret;
 
 	ret = __compaction_suitable(zone, order, alloc_flags, classzone_idx);
 	trace_mm_compaction_suitable(zone, order, ret);
@@ -1336,9 +1337,9 @@ unsigned long compaction_suitable(struct zone *zone, int order,
 	return ret;
 }
 
-static int compact_zone(struct zone *zone, struct compact_control *cc)
+static enum compact_result compact_zone(struct zone *zone, struct compact_control *cc)
 {
-	int ret;
+	enum compact_result ret;
 	unsigned long start_pfn = zone->zone_start_pfn;
 	unsigned long end_pfn = zone_end_pfn(zone);
 	const int migratetype = gfpflags_to_migratetype(cc->gfp_mask);
@@ -1483,11 +1484,11 @@ static int compact_zone(struct zone *zone, struct compact_control *cc)
 	return ret;
 }
 
-static unsigned long compact_zone_order(struct zone *zone, int order,
+static enum compact_result compact_zone_order(struct zone *zone, int order,
 		gfp_t gfp_mask, enum migrate_mode mode, int *contended,
 		int alloc_flags, int classzone_idx)
 {
-	unsigned long ret;
+	enum compact_result ret;
 	struct compact_control cc = {
 		.nr_freepages = 0,
 		.nr_migratepages = 0,
@@ -1524,7 +1525,7 @@ int sysctl_extfrag_threshold = 500;
  *
  * This is the main entry point for direct page compaction.
  */
-unsigned long try_to_compact_pages(gfp_t gfp_mask, unsigned int order,
+enum compact_result try_to_compact_pages(gfp_t gfp_mask, unsigned int order,
 			int alloc_flags, const struct alloc_context *ac,
 			enum migrate_mode mode, int *contended)
 {
@@ -1532,7 +1533,7 @@ unsigned long try_to_compact_pages(gfp_t gfp_mask, unsigned int order,
 	int may_perform_io = gfp_mask & __GFP_IO;
 	struct zoneref *z;
 	struct zone *zone;
-	int rc = COMPACT_DEFERRED;
+	enum compact_result rc = COMPACT_DEFERRED;
 	int all_zones_contended = COMPACT_CONTENDED_LOCK; /* init for &= op */
 
 	*contended = COMPACT_CONTENDED_NONE;
@@ -1546,7 +1547,7 @@ unsigned long try_to_compact_pages(gfp_t gfp_mask, unsigned int order,
 	/* Compact each zone in the list */
 	for_each_zone_zonelist_nodemask(zone, z, ac->zonelist, ac->high_zoneidx,
 								ac->nodemask) {
-		int status;
+		enum compact_result status;
 		int zone_contended;
 
 		if (compaction_deferred(zone, order))
mm/page_alloc.c +1 -1
index 269a04f20927..4acc0aa1aee0 100644
@@ -2821,7 +2821,7 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		enum migrate_mode mode, int *contended_compaction,
 		bool *deferred_compaction)
 {
-	unsigned long compact_result;
+	enum compact_result compact_result;
 	struct page *page;
 
 	if (!order)
Signature · 7 lines
2.7.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Acked-by on v1 1/3
#103 lore raw

Re: [PATCH 1/3] mm, compaction: change COMPACT_ constants into enum

Quoted 6 lines · Michal Hocko

From: Michal Hocko <[email protected]>

compaction code is doing weird dances between
COMPACT_FOO -> int -> unsigned long

but there doesn't seem to be any reason for that. All functions which

I vaguely recall trying this once and running into header dependency hell. But maybe it was something a bit different and involved storing a value in struct compact_control.

Quoted 7 lines

return/use one of those constants are not expecting any other value so it really makes sense to define an enum for them and make it clear that no other values are expected.

This is a pure cleanup and shouldn't introduce any functional changes.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Acked-by on v1 1/3
#104 lore raw

Re: [PATCH 1/3] mm, compaction: change COMPACT_ constants into enum

In reply to Michal Hocko
Quoted 11 lines

From: Michal Hocko <[email protected]>

compaction code is doing weird dances between
COMPACT_FOO -> int -> unsigned long

but there doesn't seem to be any reason for that. All functions which return/use one of those constants are not expecting any other value so it really makes sense to define an enum for them and make it clear that no other values are expected.

Quoted 4 lines

This is a pure cleanup and shouldn't introduce any functional changes.

---

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#105 lore raw

PATCH 2/3 mm, compaction: cover all compaction mode in compact_zone

In reply to Michal Hocko

From: Michal Hocko <[email protected]>

the compiler is complaining after "mm, compaction: change COMPACT_ constants into enum"

mm/compaction.c: In function a??compact_zonea??:
mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_DEFERREDa?? not handled in switch [-Wswitch]
switch (ret) {
^
mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_COMPLETEa?? not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_NO_SUITABLE_PAGEa?? not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_NOT_SUITABLE_ZONEa?? not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_CONTENDEDa?? not handled in switch [-Wswitch]

compaction_suitable is allowed to return only COMPACT_PARTIAL, COMPACT_SKIPPED and COMPACT_CONTINUE so other cases are simply impossible. Put a VM_BUG_ON to catch an impossible return value.

---
 mm/compaction.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)
mm/compaction.c +5 -8
index 0f61f12d82b6..86968d3a04e6 100644
@@ -1347,15 +1347,12 @@ static enum compact_result compact_zone(struct zone *zone, struct compact_contro
 
 	ret = compaction_suitable(zone, cc->order, cc->alloc_flags,
 							cc->classzone_idx);
-	switch (ret) {
-	case COMPACT_PARTIAL:
-	case COMPACT_SKIPPED:
-		/* Compaction is likely to fail */
+	/* Compaction is likely to fail */
+	if (ret == COMPACT_PARTIAL || ret == COMPACT_SKIPPED)
 		return ret;
-	case COMPACT_CONTINUE:
-		/* Fall through to compaction */
-		;
-	}
+
+	/* huh, compaction_suitable is returning something unexpected */
+	VM_BUG_ON(ret != COMPACT_CONTINUE);
 
 	/*
 	 * Clear pageblock skip if there were failures recently and compaction
Signature · 7 lines
2.7.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Acked-by on v1 2/3
#106 lore raw

Re: [PATCH 2/3] mm, compaction: cover all compaction mode in compact_zone

Quoted 4 lines · Michal Hocko

From: Michal Hocko <[email protected]>

the compiler is complaining after "mm, compaction: change COMPACT_ constants into enum"

Potentially a squash into that patch then?

Quoted 9 lines

mm/compaction.c: In function a??compact_zonea??:
mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_DEFERREDa?? not handled in switch [-Wswitch]
switch (ret) {
^
mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_COMPLETEa?? not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_NO_SUITABLE_PAGEa?? not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_NOT_SUITABLE_ZONEa?? not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value a??COMPACT_CONTENDEDa?? not handled in switch [-Wswitch]

Quoted 5 lines

compaction_suitable is allowed to return only COMPACT_PARTIAL, COMPACT_SKIPPED and COMPACT_CONTINUE so other cases are simply impossible. Put a VM_BUG_ON to catch an impossible return value.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Acked-by on v1 2/3
#107 lore raw

Re: [PATCH 2/3] mm, compaction: cover all compaction mode in compact_zone

In reply to Michal Hocko
Quoted 15 lines

From: Michal Hocko <[email protected]>

the compiler is complaining after "mm, compaction: change COMPACT_ constants into enum"

mm/compaction.c: In function ‘compact_zone’:
mm/compaction.c:1350:2: warning: enumeration value ‘COMPACT_DEFERRED’ not handled in switch [-Wswitch]
switch (ret) {
^
mm/compaction.c:1350:2: warning: enumeration value ‘COMPACT_COMPLETE’ not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value ‘COMPACT_NO_SUITABLE_PAGE’ not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value ‘COMPACT_NOT_SUITABLE_ZONE’ not handled in switch [-Wswitch] mm/compaction.c:1350:2: warning: enumeration value ‘COMPACT_CONTENDED’ not handled in switch [-Wswitch]

Quoted 6 lines

compaction_suitable is allowed to return only COMPACT_PARTIAL, COMPACT_SKIPPED and COMPACT_CONTINUE so other cases are simply impossible. Put a VM_BUG_ON to catch an impossible return value.

---

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#108 lore raw

PATCH 3/3 mm, oom: protect !costly allocations some more

In reply to Michal Hocko

From: Michal Hocko <[email protected]>

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or it told us to continue.

---
 include/linux/compaction.h |  5 +++++
 mm/page_alloc.c            | 53 ++++++++++++++++++++++++++++++++--------------
 2 files changed, 42 insertions(+), 16 deletions(-)
include/linux/compaction.h +5 -0
index b167801187e7..49e04326dcb8 100644
@@ -14,6 +14,11 @@ enum compact_result {
 	/* compaction should continue to another pageblock */
 	COMPACT_CONTINUE,
 	/*
+	 * whoever is calling compaction should retry because it was either
+	 * not active or it tells us there is more work to be done.
+	 */
+	COMPACT_SHOULD_RETRY = COMPACT_CONTINUE,
+	/*
 	 * direct compaction partially compacted a zone and there are suitable
 	 * pages
 	 */
mm/page_alloc.c +37 -16

114 lines not on this page · raw message

Signature · 7 lines
2.7.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#109 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 29 lines · Michal Hocko

From: Michal Hocko <[email protected]>

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or it told us to continue.

---
 include/linux/compaction.h |  5 +++++
 mm/page_alloc.c            | 53 ++++++++++++++++++++++++++++++++--------------
 2 files changed, 42 insertions(+), 16 deletions(-)
include/linux/compaction.h +0 -0
index b167801187e7..49e04326dcb8 100644

include/linux/compaction.h enum compact_result { · line 20

Quoted 7 lines
include/linux/compaction.h +4 -0
@@ -14,6 +14,11 @@ enum compact_result {
 	/* compaction should continue to another pageblock */
 	COMPACT_CONTINUE,
 	/*
+	 * whoever is calling compaction should retry because it was either
+	 * not active or it tells us there is more work to be done.
+	 */
+	COMPACT_SHOULD_RETRY = COMPACT_CONTINUE,

Hmm, I'm not sure about this. AFAIK compact_zone() doesn't ever return COMPACT_CONTINUE, and thus try_to_compact_pages() also doesn't. This overloading of CONTINUE only applies to compaction_suitable(). But the value that should_compact_retry() is testing comes only from try_to_compact_pages(). So this is not wrong, but perhaps a bit misleading?

Quoted 122 lines
include/linux/compaction.h +1 -0
+	/*
 	 * direct compaction partially compacted a zone and there are suitable
 	 * pages
 	 */
mm/page_alloc.c +36 -15
index 4acc0aa1aee0..041aeb1dc3b4 100644
@@ -2819,28 +2819,20 @@ static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		enum compact_result *compact_result)
 {
-	enum compact_result compact_result;
 	struct page *page;
 
 	if (!order)
 		return NULL;
 
 	current->flags |= PF_MEMALLOC;
-	compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
+	*compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
-	switch (compact_result) {
-	case COMPACT_DEFERRED:
-		*deferred_compaction = true;
-		/* fall-through */
-	case COMPACT_SKIPPED:
+	if (*compact_result <= COMPACT_SKIPPED)
 		return NULL;
-	default:
-		break;
-	}
 
 	/*
 	 * At least in one zone compaction wasn't deferred or skipped, so let's
@@ -2870,15 +2862,41 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 
 	return NULL;
 }
+
+static inline bool
+ (unsigned int order, enum compact_result compact_result,
+		     int contended_compaction)
+{
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_SHOULD_RETRY)
+			return true;
+		if (contended_compaction > COMPACT_CONTENDED_NONE)
+			return true;
+	}
+
+	return false;
+}
 #else
 static inline struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		enum compact_result *compact_result)
 {
 	return NULL;
 }
+
+static inline bool
+should_compact_retry(unsigned int order, enum compact_result compact_result,
+		     int contended_compaction)
+{
+	return false;
+}
 #endif /* CONFIG_COMPACTION */
 
 /* Perform direct synchronous page reclaim */
@@ -3118,7 +3136,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	int alloc_flags;
 	unsigned long did_some_progress;
 	enum migrate_mode migration_mode = MIGRATE_ASYNC;
-	bool deferred_compaction = false;
+	enum compact_result compact_result;
 	int contended_compaction = COMPACT_CONTENDED_NONE;
 	int no_progress_loops = 0;
 
@@ -3227,7 +3245,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags, ac,
 					migration_mode,
 					&contended_compaction,
-					&deferred_compaction);
+					&compact_result);
 	if (page)
 		goto got_pg;
 
@@ -3240,7 +3258,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		 * to heavily disrupt the system, so we fail the allocation
 		 * instead of entering direct reclaim.
 		 */
-		if (deferred_compaction)
+		if (compact_result == COMPACT_DEFERRED)
 			goto nopage;
 
 		/*
@@ -3294,6 +3312,9 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
+	if (should_compact_retry(order, compact_result, contended_compaction))
+		goto retry;
+
 	/* Reclaim has failed us, start killing things */
 	page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);
 	if (page)

mm/page_alloc.c __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3341

Quoted 8 lines
mm/page_alloc.c +1 -1
@@ -3314,7 +3335,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags,
 					    ac, migration_mode,
 					    &contended_compaction,
-					    &deferred_compaction);
+					    &compact_result);
 	if (page)
 		goto got_pg;
 nopage:

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#110 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 35 lines · Vlastimil Babka
Quoted 34 lines · Michal Hocko

From: Michal Hocko <[email protected]>

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or it told us to continue.

---
 include/linux/compaction.h |  5 +++++
 mm/page_alloc.c            | 53 ++++++++++++++++++++++++++++++++--------------
 2 files changed, 42 insertions(+), 16 deletions(-)
include/linux/compaction.h +2 -0
index b167801187e7..49e04326dcb8 100644
@@ -14,6 +14,11 @@ enum compact_result {
 	/* compaction should continue to another pageblock */
 	COMPACT_CONTINUE,
 	/*
+	 * whoever is calling compaction should retry because it was either
+	 * not active or it tells us there is more work to be done.
Quoted 8 lines

include/linux/compaction.h enum compact_result { · line 20

Quoted 2 lines
include/linux/compaction.h +2 -0
+	 */
+	COMPACT_SHOULD_RETRY = COMPACT_CONTINUE,

Hmm, I'm not sure about this. AFAIK compact_zone() doesn't ever return COMPACT_CONTINUE, and thus try_to_compact_pages() also doesn't. This overloading of CONTINUE only applies to compaction_suitable(). But the value that should_compact_retry() is testing comes only from try_to_compact_pages(). So this is not wrong, but perhaps a bit misleading?

Well the idea was that I wanted to cover all the _possible_ cases where compaction might want to tell us "please try again even when the last round wasn't really successful". COMPACT_CONTINUE might not be returned right now but we can come up with that in the future. It sounds like a sensible feedback to me. But maybe there would be a better name for such a feedback. I confess this is a bit oom-rework centric name...

Also I find it better to hide details behind a more generic name.

I am open to suggestions here, of course.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#111 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 18 lines · Michal Hocko
Quoted 8 lines · Vlastimil Babka
Quoted 8 lines
include/linux/compaction.h +2 -0
@@ -14,6 +14,11 @@ enum compact_result {
 	/* compaction should continue to another pageblock */
 	COMPACT_CONTINUE,
 	/*
+	 * whoever is calling compaction should retry because it was either
+	 * not active or it tells us there is more work to be done.
Quoted 8 lines

include/linux/compaction.h enum compact_result { · line 20

Quoted 2 lines
include/linux/compaction.h +2 -0
+	 */
+	COMPACT_SHOULD_RETRY = COMPACT_CONTINUE,

Hmm, I'm not sure about this. AFAIK compact_zone() doesn't ever return COMPACT_CONTINUE, and thus try_to_compact_pages() also doesn't. This overloading of CONTINUE only applies to compaction_suitable(). But the value that should_compact_retry() is testing comes only from try_to_compact_pages(). So this is not wrong, but perhaps a bit misleading?

Quoted 6 lines

Well the idea was that I wanted to cover all the _possible_ cases where compaction might want to tell us "please try again even when the last round wasn't really successful". COMPACT_CONTINUE might not be returned right now but we can come up with that in the future. It sounds like a sensible feedback to me. But maybe there would be a better name for such a feedback. I confess this is a bit oom-rework centric name...

Hmm, I see. But it doesn't really tell use to please try again. That interpretation is indeed oom-specific. What it's actually telling us is either a) reclaim and then try again (COMPACT_SKIPPED), b) try again just to overcome the deferred state (COMPACT_DEFERRED). COMPACT_CONTINUE says "go ahead", but only from compaction_suitable().
So the attempt a generic name doesn't really work here I'm afraid :/

Quoted 4 lines

Also I find it better to hide details behind a more generic name.

I am open to suggestions here, of course.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#112 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

In reply to Michal Hocko

Joonsoo has pointed out that this attempt is still not sufficient becasuse we might have invoked only a single compaction round which is might be not enough. I fully agree with that. Here is my take on that. It is again based on the number of retries loop.

I was also playing with an idea of doing something similar to the reclaim retry logic:

	if (order) {
		if (compaction_made_progress(compact_result)
			no_compact_progress = 0;
		else if (compaction_failed(compact_result)
			no_compact_progress++;
	}

but it is compaction_failed() part which is not really straightforward to define. Is it COMPACT_NO_SUITABLE_PAGE resp. COMPACT_NOT_SUITABLE_ZONE sufficient? compact_finished and compaction_suitable however hide this from compaction users so it seems like we can never see it.

Maybe we can update the feedback mechanism from the compaction but retries count seems reasonably easy to understand and pragmatic. If we cannot form a order page after we tried for N times then it really doesn't make much sense to continue and we are oom for this order. I am holding my breath to hear from Hugh on this, though. In case it doesn't then I would be really interested whether changing MAX_COMPACT_RETRIES makes any difference.

I haven't preserved Tested-by from Sergey to be on the safe side even though strictly speaking this should be less prone to high order OOMs because we clearly retry more times.
---

#113 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 9 lines · Michal Hocko

Joonsoo has pointed out that this attempt is still not sufficient becasuse we might have invoked only a single compaction round which is might be not enough. I fully agree with that. Here is my take on that. It is again based on the number of retries loop.

I was also playing with an idea of doing something similar to the reclaim retry logic:

	if (order) {
		if (compaction_made_progress(compact_result)

Progress for compaction would probably mean counting successful migrations. This would converge towards a definitive false (without parallel activity) in the current implementation, but probably not for the proposed redesigns where migration and free scanner initial positions are not fixed.

Quoted 9 lines
			no_compact_progress = 0;
		else if (compaction_failed(compact_result)
			no_compact_progress++;
	}

but it is compaction_failed() part which is not really straightforward to define. Is it COMPACT_NO_SUITABLE_PAGE resp. COMPACT_NOT_SUITABLE_ZONE sufficient? compact_finished and compaction_suitable however hide this from compaction users so it seems like we can never see it.

Anything other than COMPACT_PARTIAL is "failed" :) But it doesn't itself hint at whether retrying makes sense or not. Reclaim is simpler in this sense...

Quoted 42 lines

Maybe we can update the feedback mechanism from the compaction but retries count seems reasonably easy to understand and pragmatic. If we cannot form a order page after we tried for N times then it really doesn't make much sense to continue and we are oom for this order. I am holding my breath to hear from Hugh on this, though. In case it doesn't then I would be really interested whether changing MAX_COMPACT_RETRIES makes any difference.

I haven't preserved Tested-by from Sergey to be on the safe side even though strictly speaking this should be less prone to high order OOMs because we clearly retry more times.
---
From 33f08d6eeb0f5eaf1c73c292f070102ddec5878a Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Wed, 9 Mar 2016 10:57:42 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if
- the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or
- we haven't completed at least MAX_COMPACT_RETRIES successful (either COMPACT_PARTIAL or COMPACT_COMPLETE) compaction rounds.

The first rule ensures that the very last attempt for compaction was ignored while the second guarantees that the compaction has done some work. Multiple retries might be needed to prevent occasional pigggy packing of other contexts to steal the compacted pages while the current context manages to retry to allocate them.

Quoted 6 lines

If the given number of successful retries is not sufficient for a reasonable workloads we should focus on the collected compaction tracepoints data and try to address the issue in the compaction code. If this is not feasible we can increase the retries limit.

Yeah, this could work.

Quoted 175 lines
---
 include/linux/compaction.h | 10 +++++++
 mm/page_alloc.c            | 68 +++++++++++++++++++++++++++++++++++-----------
 2 files changed, 62 insertions(+), 16 deletions(-)
include/linux/compaction.h +10 -0
index b167801187e7..7d028ccf440a 100644
@@ -61,6 +61,12 @@ extern void compaction_defer_reset(struct zone *zone, int order,
 				bool alloc_success);
 extern bool compaction_restarting(struct zone *zone, int order);
 
+static inline bool compaction_made_progress(enum compact_result result)
+{
+	return (compact_result > COMPACT_SKIPPED &&
+				compact_result < COMPACT_NO_SUITABLE_PAGE)
+}
+
 #else
 static inline enum compact_result try_to_compact_pages(gfp_t gfp_mask,
 			unsigned int order, int alloc_flags,
@@ -93,6 +99,10 @@ static inline bool compaction_deferred(struct zone *zone, int order)
 	return true;
 }
 
+static inline bool compaction_made_progress(enum compact_result result)
+{
+	return false;
+}
 #endif /* CONFIG_COMPACTION */
 
 #if defined(CONFIG_COMPACTION) && defined(CONFIG_SYSFS) && defined(CONFIG_NUMA)
mm/page_alloc.c +51 -15
index 4acc0aa1aee0..5f1fc3793836 100644
@@ -2813,34 +2813,33 @@ __alloc_pages_may_oom(gfp_t gfp_mask, unsigned int order,
 	return page;
 }
 
+
+/*
+ * Maximum number of compaction retries wit a progress before OOM
+ * killer is consider as the only way to move forward.
+ */
+#define MAX_COMPACT_RETRIES 16
+
 #ifdef CONFIG_COMPACTION
 /* Try memory compaction for high-order allocations before reclaim */
 static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		enum compact_result *compact_result)
 {
-	enum compact_result compact_result;
 	struct page *page;
 
 	if (!order)
 		return NULL;
 
 	current->flags |= PF_MEMALLOC;
-	compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
+	*compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
-	switch (compact_result) {
-	case COMPACT_DEFERRED:
-		*deferred_compaction = true;
-		/* fall-through */
-	case COMPACT_SKIPPED:
+	if (*compact_result <= COMPACT_SKIPPED)
 		return NULL;
-	default:
-		break;
-	}
 
 	/*
 	 * At least in one zone compaction wasn't deferred or skipped, so let's
@@ -2870,15 +2869,44 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 
 	return NULL;
 }
+
+static inline bool
+should_compact_retry(unsigned int order, enum compact_result compact_result,
+		     int contended_compaction, int compaction_retries)
+{
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM. Still cap the reclaim retry loops with
+	 * progress to prevent from looping forever and potential trashing.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_SKIPPED)
+			return true;
+		if (contended_compaction > COMPACT_CONTENDED_NONE)
+			return true;
+		if (compaction_retries <= MAX_COMPACT_RETRIES)
+			return true;
+	}
+
+	return false;
+}
 #else
 static inline struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		enum compact_result *compact_result)
 {
 	return NULL;
 }
+
+static inline bool
+should_compact_retry(unsigned int order, enum compact_result compact_result,
+		     int contended_compaction, int compaction_retries)
+{
+	return false;
+}
 #endif /* CONFIG_COMPACTION */
 
 /* Perform direct synchronous page reclaim */
@@ -3118,7 +3146,8 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	int alloc_flags;
 	unsigned long did_some_progress;
 	enum migrate_mode migration_mode = MIGRATE_ASYNC;
-	bool deferred_compaction = false;
+	enum compact_result compact_result;
+	int compaction_retries = 0;
 	int contended_compaction = COMPACT_CONTENDED_NONE;
 	int no_progress_loops = 0;
 
@@ -3227,10 +3256,13 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags, ac,
 					migration_mode,
 					&contended_compaction,
-					&deferred_compaction);
+					&compact_result);
 	if (page)
 		goto got_pg;
 
+	if (order && compaction_made_progress(compact_result))
+		compaction_retries++;
+
 	/* Checks for THP-specific high-order allocations */
 	if (is_thp_gfp_mask(gfp_mask)) {
 		/*
@@ -3240,7 +3272,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		 * to heavily disrupt the system, so we fail the allocation
 		 * instead of entering direct reclaim.
 		 */
-		if (deferred_compaction)
+		if (compact_result == COMPACT_DEFERRED)
 			goto nopage;
 
 		/*
@@ -3294,6 +3326,10 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
+	if (should_compact_retry(order, compact_result, contended_compaction,
+				 compaction_retries))
+		goto retry;
+
 	/* Reclaim has failed us, start killing things */
 	page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);
 	if (page)

mm/page_alloc.c __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, · line 3356

Quoted 8 lines
mm/page_alloc.c +1 -1
@@ -3314,7 +3350,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags,
 					    ac, migration_mode,
 					    &contended_compaction,
-					    &deferred_compaction);
+					    &compact_result);
 	if (page)
 		goto got_pg;
 nopage:

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#114 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

In reply to Michal Hocko
Quoted 19 lines · Michal Hocko

Joonsoo has pointed out that this attempt is still not sufficient becasuse we might have invoked only a single compaction round which is might be not enough. I fully agree with that. Here is my take on that. It is again based on the number of retries loop.

I was also playing with an idea of doing something similar to the reclaim retry logic:

	if (order) {
		if (compaction_made_progress(compact_result)
			no_compact_progress = 0;
		else if (compaction_failed(compact_result)
			no_compact_progress++;
	}

but it is compaction_failed() part which is not really straightforward to define. Is it COMPACT_NO_SUITABLE_PAGE resp. COMPACT_NOT_SUITABLE_ZONE sufficient? compact_finished and compaction_suitable however hide this from compaction users so it seems like we can never see it.

Quoted 5 lines

Maybe we can update the feedback mechanism from the compaction but retries count seems reasonably easy to understand and pragmatic. If we cannot form a order page after we tried for N times then it really doesn't make much sense to continue and we are oom for this order. I am holding my breath to hear from Hugh on this, though.

Never a wise strategy. But I just got around to it tonight.

I do believe you've nailed it with this patch! Thank you!

I've applied 1/3, 2/3 and this (ah, it became the missing 3/3 later on) on top of 4.5.0-rc5-mm1 (I think there have been a couple of mmotms since, but I've not got to them yet): so far it is looking good on all machines.

After a quick go with the simple make -j20 in tmpfs, which survived a cycle on the laptop, I've switched back to my original tougher load, and that's going well so far: no sign of any OOMs. But I've interrupted on the laptop to report back to you now, then I'll leave it running overnight.

Quoted 54 lines

In case it doesn't
then I would be really interested whether changing MAX_COMPACT_RETRIES makes any difference.

I haven't preserved Tested-by from Sergey to be on the safe side even though strictly speaking this should be less prone to high order OOMs because we clearly retry more times.
---
From 33f08d6eeb0f5eaf1c73c292f070102ddec5878a Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Wed, 9 Mar 2016 10:57:42 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if
- the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or
- we haven't completed at least MAX_COMPACT_RETRIES successful (either COMPACT_PARTIAL or COMPACT_COMPLETE) compaction rounds.

The first rule ensures that the very last attempt for compaction was ignored while the second guarantees that the compaction has done some work. Multiple retries might be needed to prevent occasional pigggy packing of other contexts to steal the compacted pages while the current context manages to retry to allocate them.

If the given number of successful retries is not sufficient for a reasonable workloads we should focus on the collected compaction tracepoints data and try to address the issue in the compaction code. If this is not feasible we can increase the retries limit.

---
 include/linux/compaction.h | 10 +++++++
 mm/page_alloc.c            | 68 +++++++++++++++++++++++++++++++++++-----------
 2 files changed, 62 insertions(+), 16 deletions(-)
include/linux/compaction.h +0 -0
index b167801187e7..7d028ccf440a 100644

include/linux/compaction.h extern void compaction_defer_reset(struct zone *zone, int order, · line 67

Quoted 7 lines
include/linux/compaction.h +4 -0
@@ -61,6 +61,12 @@ extern void compaction_defer_reset(struct zone *zone, int order,
 				bool alloc_success);
 extern bool compaction_restarting(struct zone *zone, int order);
 
+static inline bool compaction_made_progress(enum compact_result result)
+{
+	return (compact_result > COMPACT_SKIPPED &&
+				compact_result < COMPACT_NO_SUITABLE_PAGE)

That line didn't build at all:

        return result > COMPACT_SKIPPED && result < COMPACT_NO_SUITABLE_PAGE;
Quoted 171 lines
include/linux/compaction.h +6 -0
+}
+
 #else
 static inline enum compact_result try_to_compact_pages(gfp_t gfp_mask,
 			unsigned int order, int alloc_flags,
@@ -93,6 +99,10 @@ static inline bool compaction_deferred(struct zone *zone, int order)
 	return true;
 }
 
+static inline bool compaction_made_progress(enum compact_result result)
+{
+	return false;
+}
 #endif /* CONFIG_COMPACTION */
 
 #if defined(CONFIG_COMPACTION) && defined(CONFIG_SYSFS) && defined(CONFIG_NUMA)
mm/page_alloc.c +52 -16
index 4acc0aa1aee0..5f1fc3793836 100644
@@ -2813,34 +2813,33 @@ __alloc_pages_may_oom(gfp_t gfp_mask, unsigned int order,
 	return page;
 }
 
+
+/*
+ * Maximum number of compaction retries wit a progress before OOM
+ * killer is consider as the only way to move forward.
+ */
+#define MAX_COMPACT_RETRIES 16
+
 #ifdef CONFIG_COMPACTION
 /* Try memory compaction for high-order allocations before reclaim */
 static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		enum compact_result *compact_result)
 {
-	enum compact_result compact_result;
 	struct page *page;
 
 	if (!order)
 		return NULL;
 
 	current->flags |= PF_MEMALLOC;
-	compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
+	*compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
-	switch (compact_result) {
-	case COMPACT_DEFERRED:
-		*deferred_compaction = true;
-		/* fall-through */
-	case COMPACT_SKIPPED:
+	if (*compact_result <= COMPACT_SKIPPED)
 		return NULL;
-	default:
-		break;
-	}
 
 	/*
 	 * At least in one zone compaction wasn't deferred or skipped, so let's
@@ -2870,15 +2869,44 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 
 	return NULL;
 }
+
+static inline bool
+should_compact_retry(unsigned int order, enum compact_result compact_result,
+		     int contended_compaction, int compaction_retries)
+{
+	/*
+	 * !costly allocations are really important and we have to make sure
+	 * the compaction wasn't deferred or didn't bail out early due to locks
+	 * contention before we go OOM. Still cap the reclaim retry loops with
+	 * progress to prevent from looping forever and potential trashing.
+	 */
+	if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+		if (compact_result <= COMPACT_SKIPPED)
+			return true;
+		if (contended_compaction > COMPACT_CONTENDED_NONE)
+			return true;
+		if (compaction_retries <= MAX_COMPACT_RETRIES)
+			return true;
+	}
+
+	return false;
+}
 #else
 static inline struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 		int alloc_flags, const struct alloc_context *ac,
 		enum migrate_mode mode, int *contended_compaction,
-		bool *deferred_compaction)
+		enum compact_result *compact_result)
 {
 	return NULL;
 }
+
+static inline bool
+should_compact_retry(unsigned int order, enum compact_result compact_result,
+		     int contended_compaction, int compaction_retries)
+{
+	return false;
+}
 #endif /* CONFIG_COMPACTION */
 
 /* Perform direct synchronous page reclaim */
@@ -3118,7 +3146,8 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	int alloc_flags;
 	unsigned long did_some_progress;
 	enum migrate_mode migration_mode = MIGRATE_ASYNC;
-	bool deferred_compaction = false;
+	enum compact_result compact_result;
+	int compaction_retries = 0;
 	int contended_compaction = COMPACT_CONTENDED_NONE;
 	int no_progress_loops = 0;
 
@@ -3227,10 +3256,13 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags, ac,
 					migration_mode,
 					&contended_compaction,
-					&deferred_compaction);
+					&compact_result);
 	if (page)
 		goto got_pg;
 
+	if (order && compaction_made_progress(compact_result))
+		compaction_retries++;
+
 	/* Checks for THP-specific high-order allocations */
 	if (is_thp_gfp_mask(gfp_mask)) {
 		/*
@@ -3240,7 +3272,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 		 * to heavily disrupt the system, so we fail the allocation
 		 * instead of entering direct reclaim.
 		 */
-		if (deferred_compaction)
+		if (compact_result == COMPACT_DEFERRED)
 			goto nopage;
 
 		/*
@@ -3294,6 +3326,10 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 				 did_some_progress > 0, no_progress_loops))
 		goto retry;
 
+	if (should_compact_retry(order, compact_result, contended_compaction,
+				 compaction_retries))
+		goto retry;
+
 	/* Reclaim has failed us, start killing things */
 	page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);
 	if (page)
@@ -3314,7 +3350,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
 	page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags,
 					    ac, migration_mode,
 					    &contended_compaction,
-					    &deferred_compaction);
+					    &compact_result);
 	if (page)
 		goto got_pg;
 nopage:
Signature · 5 lines
2.7.0

-- 
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#115 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 26 lines · Hugh Dickins
Quoted 19 lines · Michal Hocko

Joonsoo has pointed out that this attempt is still not sufficient becasuse we might have invoked only a single compaction round which is might be not enough. I fully agree with that. Here is my take on that. It is again based on the number of retries loop.

I was also playing with an idea of doing something similar to the reclaim retry logic:

	if (order) {
		if (compaction_made_progress(compact_result)
			no_compact_progress = 0;
		else if (compaction_failed(compact_result)
			no_compact_progress++;
	}

but it is compaction_failed() part which is not really straightforward to define. Is it COMPACT_NO_SUITABLE_PAGE resp. COMPACT_NOT_SUITABLE_ZONE sufficient? compact_finished and compaction_suitable however hide this from compaction users so it seems like we can never see it.

Quoted 5 lines

Maybe we can update the feedback mechanism from the compaction but retries count seems reasonably easy to understand and pragmatic. If we cannot form a order page after we tried for N times then it really doesn't make much sense to continue and we are oom for this order. I am holding my breath to hear from Hugh on this, though.

Quoted 3 lines

Never a wise strategy. But I just got around to it tonight.

I do believe you've nailed it with this patch! Thank you!

That's a great news! Thanks for testing.

Quoted 9 lines

I've applied 1/3, 2/3 and this (ah, it became the missing 3/3 later on) on top of 4.5.0-rc5-mm1 (I think there have been a couple of mmotms since, but I've not got to them yet): so far it is looking good on all machines.

After a quick go with the simple make -j20 in tmpfs, which survived a cycle on the laptop, I've switched back to my original tougher load, and that's going well so far: no sign of any OOMs. But I've interrupted on the laptop to report back to you now, then I'll leave it running overnight.

OK, let's wait for the rest of the tests but I find it really optimistic considering how easily you could trigger the issue previously. Anyway I hope for your Tested-by after you are reasonably confident your loads are behaving well.

[...]

Quoted 8 lines
Quoted 8 lines
include/linux/compaction.h +0 -0
index b167801187e7..7d028ccf440a 100644
@@ -61,6 +61,12 @@ extern void compaction_defer_reset(struct zone *zone, int order,
 				bool alloc_success);
 extern bool compaction_restarting(struct zone *zone, int order);
 
Quoted 8 lines

include/linux/compaction.h extern void compaction_defer_reset(struct zone *zone, int order, · line 67

Quoted 4 lines
include/linux/compaction.h +4 -0
+static inline bool compaction_made_progress(enum compact_result result)
+{
+	return (compact_result > COMPACT_SKIPPED &&
+				compact_result < COMPACT_NO_SUITABLE_PAGE)

That line didn't build at all:

        return result > COMPACT_SKIPPED && result < COMPACT_NO_SUITABLE_PAGE;

those last minute changes... Sorry about that. Fixed.

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Tested-by
#116 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 43 lines · Michal Hocko
Quoted 26 lines · Hugh Dickins
Quoted 19 lines · Michal Hocko

Joonsoo has pointed out that this attempt is still not sufficient becasuse we might have invoked only a single compaction round which is might be not enough. I fully agree with that. Here is my take on that. It is again based on the number of retries loop.

I was also playing with an idea of doing something similar to the reclaim retry logic:

	if (order) {
		if (compaction_made_progress(compact_result)
			no_compact_progress = 0;
		else if (compaction_failed(compact_result)
			no_compact_progress++;
	}

but it is compaction_failed() part which is not really straightforward to define. Is it COMPACT_NO_SUITABLE_PAGE resp. COMPACT_NOT_SUITABLE_ZONE sufficient? compact_finished and compaction_suitable however hide this from compaction users so it seems like we can never see it.

Quoted 5 lines

Maybe we can update the feedback mechanism from the compaction but retries count seems reasonably easy to understand and pragmatic. If we cannot form a order page after we tried for N times then it really doesn't make much sense to continue and we are oom for this order. I am holding my breath to hear from Hugh on this, though.

Quoted 3 lines

Never a wise strategy. But I just got around to it tonight.

I do believe you've nailed it with this patch! Thank you!

That's a great news! Thanks for testing.

Quoted 9 lines

I've applied 1/3, 2/3 and this (ah, it became the missing 3/3 later on) on top of 4.5.0-rc5-mm1 (I think there have been a couple of mmotms since, but I've not got to them yet): so far it is looking good on all machines.

After a quick go with the simple make -j20 in tmpfs, which survived a cycle on the laptop, I've switched back to my original tougher load, and that's going well so far: no sign of any OOMs. But I've interrupted on the laptop to report back to you now, then I'll leave it running overnight.

Quoted 4 lines

OK, let's wait for the rest of the tests but I find it really optimistic considering how easily you could trigger the issue previously. Anyway I hope for your Tested-by after you are reasonably confident your loads are behaving well.

Three have been stably running load for between 6 and 7 hours now, no problems, looking very good:

I'll be interested to see how my huge tmpfs loads fare with the rework, but I'm not quite ready to try that today; and any issue there (I've no reason to suppose that there will be) can be a separate investigation for me to make at some future date. It was this order=2 regression that was holding me back, and I've now no objection to your patches (though nobody should imagine that I've actually studied them).

Thank you, Michal.

Hugh

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#117 lore raw

Re: [PATCH 3/3] mm, oom: protect !costly allocations some more

Quoted 49 lines · Hugh Dickins
Quoted 43 lines · Michal Hocko
Quoted 26 lines · Hugh Dickins
Quoted 19 lines · Michal Hocko

Joonsoo has pointed out that this attempt is still not sufficient becasuse we might have invoked only a single compaction round which is might be not enough. I fully agree with that. Here is my take on that. It is again based on the number of retries loop.

I was also playing with an idea of doing something similar to the reclaim retry logic:

	if (order) {
		if (compaction_made_progress(compact_result)
			no_compact_progress = 0;
		else if (compaction_failed(compact_result)
			no_compact_progress++;
	}

but it is compaction_failed() part which is not really straightforward to define. Is it COMPACT_NO_SUITABLE_PAGE resp. COMPACT_NOT_SUITABLE_ZONE sufficient? compact_finished and compaction_suitable however hide this from compaction users so it seems like we can never see it.

Quoted 5 lines

Maybe we can update the feedback mechanism from the compaction but retries count seems reasonably easy to understand and pragmatic. If we cannot form a order page after we tried for N times then it really doesn't make much sense to continue and we are oom for this order. I am holding my breath to hear from Hugh on this, though.

Quoted 3 lines

Never a wise strategy. But I just got around to it tonight.

I do believe you've nailed it with this patch! Thank you!

That's a great news! Thanks for testing.

Quoted 9 lines

I've applied 1/3, 2/3 and this (ah, it became the missing 3/3 later on) on top of 4.5.0-rc5-mm1 (I think there have been a couple of mmotms since, but I've not got to them yet): so far it is looking good on all machines.

After a quick go with the simple make -j20 in tmpfs, which survived a cycle on the laptop, I've switched back to my original tougher load, and that's going well so far: no sign of any OOMs. But I've interrupted on the laptop to report back to you now, then I'll leave it running overnight.

Quoted 4 lines

OK, let's wait for the rest of the tests but I find it really optimistic considering how easily you could trigger the issue previously. Anyway I hope for your Tested-by after you are reasonably confident your loads are behaving well.

Quoted 4 lines

Three have been stably running load for between 6 and 7 hours now, no problems, looking very good:

Thanks!

Quoted 6 lines

I'll be interested to see how my huge tmpfs loads fare with the rework, but I'm not quite ready to try that today; and any issue there (I've no reason to suppose that there will be) can be a separate investigation for me to make at some future date. It was this order=2 regression that was holding me back, and I've now no objection to your patches (though nobody should imagine that I've actually studied them).

I still have some work on top pending and I do not want to rush these changes and target this for 4.7. 4.6 is just too close and I would hate to push some last minute changes. I think oom_reaper would be large enough for 4.6 in this area.

I will post the full series after rc1. Andrew feel free to drop it from the mmotm tree for now. I would prefer they got all reviewed together rather than a larger number of fixups.

Thanks Hugh for your testing. I wish I could depend on it less but I've not been able to reproduce not matter how much I tried.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#118 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

In reply to Michal Hocko

2016-03-08 1:08 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 8 lines · Michal Hocko
Quoted 4 lines

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help

I'm still not happy. Just ensuring one compaction run doesn't mean our best. What's your purpose of OOM rework? From my understanding, you'd like to trigger OOM kill deterministic and *not prematurely*. This makes sense.

But, what you did in case of high order allocation is completely different with original purpose. It may be deterministic but *completely premature*. There is no way to prevent premature OOM kill. So, I want to ask one more time. Why OOM kill is better than retry reclaiming when there is reclaimable page? Deterministic is for what? It ensures something more?

Please see Hugh's latest vmstat. There are plenty of anon pages when OOM kill happens and it may have enough swap space. Even if compaction runs and fails, why do we need to kill something in this case? OOM kill should be a last resort.

Please see Hugh's previous report and OOM dump.

[  796.540791] Mem-Info:
[  796.557378] active_anon:150198 inactive_anon:46022 isolated_anon:32
 active_file:5107 inactive_file:1664 isolated_file:57
 unevictable:3067 dirty:4 writeback:75 unstable:0
 slab_reclaimable:13907 slab_unreclaimable:23236
 mapped:8889 shmem:3171 pagetables:2176 bounce:0
 free:1637 free_pcp:54 free_cma:0
[  796.630465] Node 0 DMA32 free:13904kB min:3940kB low:4944kB

high:5948kB active_anon:588776kB inactive_anon:188816kB active_file:20432kB inactive_file:6928kB unevictable:12268kB isolated(anon):128kB isolated(file):8kB present:1046128kB managed:1004892kB mlocked:12268kB dirty:16kB writeback:1400kB mapped:35556kB shmem:12684kB slab_reclaimable:55628kB slab_unreclaimable:92944kB kernel_stack:4448kB pagetables:8604kB unstable:0kB bounce:0kB free_pcp:296kB local_pcp:164kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no

[  796.685815] lowmem_reserve[]: 0 0 0
[  796.687390] Node 0 DMA32: 969*4kB (UE) 184*8kB (UME) 167*16kB (UM)
19*32kB (UM) 3*64kB (UM) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB
0*4096kB = 8820kB
[  796.729696] Node 0 hugepages_total=0 hugepages_free=0

hugepages_surp=0 hugepages_size=2048kB

See [ 796.557378] and [ 796.630465].
In this 100 ms time interval, freepage increase a lot and there are enough high order pages. OOM kill happen later so freepage would come from reclaim. This shows
that your previous implementation which uses static retry number causes premature OOM.

This attempt using compaction result looks not different to me. It would also cause premature OOM kill.

I don't insist endless retry. I just want a more scientific criteria that prevents
premature OOM kill. I'm really tire to say same thing again and again. Am I missing something? This is the situation that I totally misunderstand something? Please let me know.

Note: your current implementation doesn't consider which zone is compacted. If DMA zone which easily fail to make high order page is compacted, your implementation will not do retry. It also looks not our best.

Thanks.

Quoted 158 lines

for your load? I have tested it with the Hugh's load and there was no major difference from the previous testing so at least nothing has blown up as I am not able to reproduce the issue here.

Other changes in the compaction are still needed but I would like to not depend on them right now.
---
From 0974f127e8eb7fe53e65f3a8b398db57effe9755 Mon Sep 17 00:00:00 2001
From: Michal Hocko <[email protected]>
Date: Mon, 7 Mar 2016 15:30:37 +0100
Subject: [PATCH] mm, oom: protect !costly allocations some more

should_reclaim_retry will give up retries for higher order allocations if none of the eligible zones has any requested or higher order pages available even if we pass the watermak check for order-0. This is done because there is no guarantee that the reclaimable and currently free pages will form the required order.

This can, however, lead to situations were the high-order request (e.g. order-2 required for the stack allocation during fork) will trigger OOM too early - e.g. after the first reclaim/compaction round. Such a system would have to be highly fragmented and there is no guarantee further reclaim/compaction attempts would help but at least make sure that the compaction was active before we go OOM and keep retrying even if should_reclaim_retry tells us to oom if the last compaction round was either inactive (deferred, skipped or bailed out early due to contention) or it told us to continue.

Additionally define COMPACT_NONE which reflects cases where the compaction is completely disabled.

---
 include/linux/compaction.h |  2 ++
 mm/page_alloc.c            | 41 ++++++++++++++++++++++++-----------------
 2 files changed, 26 insertions(+), 17 deletions(-)
include/linux/compaction.h +2 -0
index 4cd4ddf64cc7..a4cec4a03f7d 100644
@@ -1,6 +1,8 @@
 #ifndef _LINUX_COMPACTION_H
 #define _LINUX_COMPACTION_H
 
+/* compaction disabled */
+#define COMPACT_NONE           -1
 /* Return values for compact_zone() and try_to_compact_pages() */
 /* compaction didn't start as it was deferred due to past failures */
 #define COMPACT_DEFERRED       0
mm/page_alloc.c +24 -18
index 269a04f20927..f89e3cbfdf90 100644
@@ -2819,28 +2819,22 @@ static struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
                int alloc_flags, const struct alloc_context *ac,
                enum migrate_mode mode, int *contended_compaction,
-               bool *deferred_compaction)
+               unsigned long *compact_result)
 {
-       unsigned long compact_result;
        struct page *page;
 
-       if (!order)
+       if (!order) {
+               *compact_result = COMPACT_NONE;
                return NULL;
+       }
 
        current->flags |= PF_MEMALLOC;
-       compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
+       *compact_result = try_to_compact_pages(gfp_mask, order, alloc_flags, ac,
                                                mode, contended_compaction);
        current->flags &= ~PF_MEMALLOC;
 
-       switch (compact_result) {
-       case COMPACT_DEFERRED:
-               *deferred_compaction = true;
-               /* fall-through */
-       case COMPACT_SKIPPED:
+       if (*compact_result <= COMPACT_SKIPPED)
                return NULL;
-       default:
-               break;
-       }
 
        /*
         * At least in one zone compaction wasn't deferred or skipped, so let's
@@ -2875,8 +2869,9 @@ static inline struct page *
 __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
                int alloc_flags, const struct alloc_context *ac,
                enum migrate_mode mode, int *contended_compaction,
-               bool *deferred_compaction)
+               unsigned long *compact_result)
 {
+       *compact_result = COMPACT_NONE;
        return NULL;
 }
 #endif /* CONFIG_COMPACTION */
@@ -3118,7 +3113,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
        int alloc_flags;
        unsigned long did_some_progress;
        enum migrate_mode migration_mode = MIGRATE_ASYNC;
-       bool deferred_compaction = false;
+       unsigned long compact_result;
        int contended_compaction = COMPACT_CONTENDED_NONE;
        int no_progress_loops = 0;
 
@@ -3227,7 +3222,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
        page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags, ac,
                                        migration_mode,
                                        &contended_compaction,
-                                       &deferred_compaction);
+                                       &compact_result);
        if (page)
                goto got_pg;
 
@@ -3240,7 +3235,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
                 * to heavily disrupt the system, so we fail the allocation
                 * instead of entering direct reclaim.
                 */
-               if (deferred_compaction)
+               if (compact_result == COMPACT_DEFERRED)
                        goto nopage;
 
                /*
@@ -3294,6 +3289,18 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
                                 did_some_progress > 0, no_progress_loops))
                goto retry;
 
+       /*
+        * !costly allocations are really important and we have to make sure
+        * the compaction wasn't deferred or didn't bail out early due to locks
+        * contention before we go OOM.
+        */
+       if (order && order <= PAGE_ALLOC_COSTLY_ORDER) {
+               if (compact_result <= COMPACT_CONTINUE)
+                       goto retry;
+               if (contended_compaction > COMPACT_CONTENDED_NONE)
+                       goto retry;
+       }
+
        /* Reclaim has failed us, start killing things */
        page = __alloc_pages_may_oom(gfp_mask, order, ac, &did_some_progress);
        if (page)
@@ -3314,7 +3321,7 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
        page = __alloc_pages_direct_compact(gfp_mask, order, alloc_flags,
                                            ac, migration_mode,
                                            &contended_compaction,
-                                           &deferred_compaction);
+                                           &compact_result);
        if (page)
                goto got_pg;
 nopage:
--

2.7.0

Quoted 3 lines

--
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#119 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Quoted 7 lines · Joonsoo Kim

2016-03-08 1:08 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 6 lines · Michal Hocko
Quoted 4 lines

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 5 lines
Quoted 2 lines

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help

I'm still not happy. Just ensuring one compaction run doesn't mean our best.

OK, let me think about it some more.

Quoted 3 lines

What's your purpose of OOM rework? From my understanding, you'd like to trigger OOM kill deterministic and *not prematurely*. This makes sense.

Well this is a bit awkward because we do not have any proper definition of what prematurely actually means. We do not know whether something changes and decides to free some memory right after we made the decision. We also do not know whether reclaiming some more memory would help because we might be trashing over few remaining pages so there would be still some progress, albeit small, progress. The system would be basically unusable and the OOM killer would be a large relief. What I want to achieve is to have a clear definition of _when_ we fire and do not fire _often_ to be impractical. There are loads where the new implementation behaved slightly better (see the cover for my tests) and there surely be some where this will be worse. I want this to be reasonably good. I am not claiming we are there yet and the interaction with the compaction seems like it needs some work, no question about that.

Quoted 5 lines

But, what you did in case of high order allocation is completely different with original purpose. It may be deterministic but *completely premature*. There is no way to prevent premature OOM kill. So, I want to ask one more time. Why OOM kill is better than retry reclaiming when there is reclaimable page? Deterministic is for what? It ensures something more?

yes, If we keep reclaiming we can soon start trashing or over reclaim too much which would hurt more processes. If you invoke the OOM killer instead then chances are that you will release a lot of memory at once and that would help to reconcile the memory pressure as well as free some page blocks which couldn't have been compacted before and not affect potentially many processes. The effect would be reduced to a single process. If we had a proper trashing detection feedback we could do much more clever decisions of course.

But back to the !costly OOMs. Once your system is fragmented so heavily that there are no free blocks that would satisfy !costly request then something has gone terribly wrong and we should fix it. To me it sounds like we do not care about those requests early enough and only start carying after we hit the wall. Maybe kcompactd can help us in this regards.

Quoted 4 lines

Please see Hugh's latest vmstat. There are plenty of anon pages when OOM kill happens and it may have enough swap space. Even if compaction runs and fails, why do we need to kill something in this case? OOM kill should be a last resort.

Well this would be the case even if we were trashing over swap. Refaulting the swapped out memory all over again...

Quoted 23 lines

Please see Hugh's previous report and OOM dump.

[  796.540791] Mem-Info:
[  796.557378] active_anon:150198 inactive_anon:46022 isolated_anon:32
 active_file:5107 inactive_file:1664 isolated_file:57
 unevictable:3067 dirty:4 writeback:75 unstable:0
 slab_reclaimable:13907 slab_unreclaimable:23236
 mapped:8889 shmem:3171 pagetables:2176 bounce:0
 free:1637 free_pcp:54 free_cma:0
[  796.630465] Node 0 DMA32 free:13904kB min:3940kB low:4944kB

high:5948kB active_anon:588776kB inactive_anon:188816kB active_file:20432kB inactive_file:6928kB unevictable:12268kB isolated(anon):128kB isolated(file):8kB present:1046128kB managed:1004892kB mlocked:12268kB dirty:16kB writeback:1400kB mapped:35556kB shmem:12684kB slab_reclaimable:55628kB slab_unreclaimable:92944kB kernel_stack:4448kB pagetables:8604kB unstable:0kB bounce:0kB free_pcp:296kB local_pcp:164kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no

[  796.685815] lowmem_reserve[]: 0 0 0
[  796.687390] Node 0 DMA32: 969*4kB (UE) 184*8kB (UME) 167*16kB (UM)
19*32kB (UM) 3*64kB (UM) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB
0*4096kB = 8820kB
[  796.729696] Node 0 hugepages_total=0 hugepages_free=0
Quoted 8 lines

hugepages_surp=0 hugepages_size=2048kB

See [ 796.557378] and [ 796.630465].
In this 100 ms time interval, freepage increase a lot and there are enough high order pages. OOM kill happen later so freepage would come from reclaim. This shows
that your previous implementation which uses static retry number causes premature OOM.

Or simply one of the gcc simply exitted and freed up a memory which is more likely. As I've tried to explain in other email, we cannot prevent from those races. We simply do not have a crystal ball. All we know is that at the time we checked the watermarks the last time there were simply no eligible high order pages available.

Quoted 5 lines

This attempt using compaction result looks not different to me. It would also cause premature OOM kill.

I don't insist endless retry. I just want a more scientific criteria that prevents premature OOM kill.

That is exactly what I try to achive here. Right now we are relying on zone_reclaimable heuristic. That relies that some pages are freed (and reset NR_PAGES_SCANNED) while we are scanning. With a stream of order-0 pages this is basically unbounded. What I am trying to achieve here is to base the decision on the feedback. The first attempt was to use the reclaim feedback. This turned out to be not sufficient for higher orders because compaction can deffer and skip if we are close to watermarks which is really surprising to me. So now I've tried to make sure that we do not hit this path. I agree we can do better but there always will be a moment to simply give up. Whatever that moment will be we can still find loads which could theoretically go on for little more and survive.

Quoted 7 lines

I'm really tire to say same thing again and again.
Am I missing something? This is the situation that I totally misunderstand something? Please let me know.

Note: your current implementation doesn't consider which zone is compacted. If DMA zone which easily fail to make high order page is compacted, your implementation will not do retry. It also looks not our best.

Why are we even consider DMA zone when we cannot ever allocate from this zone?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#120 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

2016-03-09 1:05 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 14 lines · Joonsoo Kim
Quoted 7 lines

2016-03-08 1:08 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 6 lines · Michal Hocko
Quoted 4 lines

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 5 lines
Quoted 2 lines

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help

I'm still not happy. Just ensuring one compaction run doesn't mean our best.

Quoted 8 lines

OK, let me think about it some more.

Quoted 3 lines

What's your purpose of OOM rework? From my understanding, you'd like to trigger OOM kill deterministic and *not prematurely*. This makes sense.

Well this is a bit awkward because we do not have any proper definition of what prematurely actually means. We do not know whether something

If we don't have proper definition to it, please define it first. We need to improve
the situation toward the clear goal. Just certain number of retry which has no base doesn't make any sense.

Quoted 6 lines

changes and decides to free some memory right after we made the decision. We also do not know whether reclaiming some more memory would help because we might be trashing over few remaining pages so there would be still some progress, albeit small, progress. The system would be basically unusable and the OOM killer would be a large relief. What I want to achieve is to have a clear definition of _when_ we fire and do

If we have no clear definition about premature, what's the meaning of a clear definition of _when_? It would just mean random time.

Quoted 13 lines

not fire _often_ to be impractical. There are loads where the new implementation behaved slightly better (see the cover for my tests) and there surely be some where this will be worse. I want this to be reasonably good. I am not claiming we are there yet and the interaction with the compaction seems like it needs some work, no question about that.

Quoted 5 lines

But, what you did in case of high order allocation is completely different with original purpose. It may be deterministic but *completely premature*. There is no way to prevent premature OOM kill. So, I want to ask one more time. Why OOM kill is better than retry reclaiming when there is reclaimable page? Deterministic is for what? It ensures something more?

Quoted 8 lines

yes, If we keep reclaiming we can soon start trashing or over reclaim too much which would hurt more processes. If you invoke the OOM killer instead then chances are that you will release a lot of memory at once and that would help to reconcile the memory pressure as well as free some page blocks which couldn't have been compacted before and not affect potentially many processes. The effect would be reduced to a single process. If we had a proper trashing detection feedback we could do much more clever decisions of course.

It looks like you did it for performance reason. You'd better think again about effect of OOM kill. We don't have enough knowledge about user space program architecture and killing one important process could lead to whole system unusable. Moreover, OOM kill could cause important data loss so should be avoided as much as possible. Performance reason cannot justify OOM kill.

Quoted 7 lines

But back to the !costly OOMs. Once your system is fragmented so heavily that there are no free blocks that would satisfy !costly request then something has gone terribly wrong and we should fix it. To me it sounds like we do not care about those requests early enough and only start carying after we hit the wall. Maybe kcompactd can help us in this regards.

Yes, but, it's another issue. In any situation, !costly OOM should not happen prematurely.

Quoted 7 lines
Quoted 4 lines

Please see Hugh's latest vmstat. There are plenty of anon pages when OOM kill happens and it may have enough swap space. Even if compaction runs and fails, why do we need to kill something in this case? OOM kill should be a last resort.

Well this would be the case even if we were trashing over swap. Refaulting the swapped out memory all over again...

If thrashing is a main obstacle to decide proper OOM point, we need to invent a way to handle thrashing or invent reasonable metric which isn't affected by thrashing.

Quoted 25 lines
Quoted 25 lines

Please see Hugh's previous report and OOM dump.

[  796.540791] Mem-Info:
[  796.557378] active_anon:150198 inactive_anon:46022 isolated_anon:32
 active_file:5107 inactive_file:1664 isolated_file:57
 unevictable:3067 dirty:4 writeback:75 unstable:0
 slab_reclaimable:13907 slab_unreclaimable:23236
 mapped:8889 shmem:3171 pagetables:2176 bounce:0
 free:1637 free_pcp:54 free_cma:0
[  796.630465] Node 0 DMA32 free:13904kB min:3940kB low:4944kB

high:5948kB active_anon:588776kB inactive_anon:188816kB active_file:20432kB inactive_file:6928kB unevictable:12268kB isolated(anon):128kB isolated(file):8kB present:1046128kB managed:1004892kB mlocked:12268kB dirty:16kB writeback:1400kB mapped:35556kB shmem:12684kB slab_reclaimable:55628kB slab_unreclaimable:92944kB kernel_stack:4448kB pagetables:8604kB unstable:0kB bounce:0kB free_pcp:296kB local_pcp:164kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no

[  796.685815] lowmem_reserve[]: 0 0 0
[  796.687390] Node 0 DMA32: 969*4kB (UE) 184*8kB (UME) 167*16kB (UM)
19*32kB (UM) 3*64kB (UM) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB
0*4096kB = 8820kB
[  796.729696] Node 0 hugepages_total=0 hugepages_free=0

hugepages_surp=0 hugepages_size=2048kB

Quoted 8 lines
Quoted 6 lines

See [ 796.557378] and [ 796.630465].
In this 100 ms time interval, freepage increase a lot and there are enough high order pages. OOM kill happen later so freepage would come from reclaim. This shows
that your previous implementation which uses static retry number causes premature OOM.

Or simply one of the gcc simply exitted and freed up a memory which is

It doesn't matter where free memory comes from. If free memory increases due to gcc exit, it implies that we can reclaim some memory from it. There is no reason to trigger OOM in this case.

Quoted 11 lines

more likely. As I've tried to explain in other email, we cannot prevent from those races. We simply do not have a crystal ball. All we know is that at the time we checked the watermarks the last time there were simply no eligible high order pages available.

Quoted 5 lines

This attempt using compaction result looks not different to me. It would also cause premature OOM kill.

I don't insist endless retry. I just want a more scientific criteria that prevents premature OOM kill.

Quoted 12 lines

That is exactly what I try to achive here. Right now we are relying on zone_reclaimable heuristic. That relies that some pages are freed (and reset NR_PAGES_SCANNED) while we are scanning. With a stream of order-0 pages this is basically unbounded. What I am trying to achieve here is to base the decision on the feedback. The first attempt was to use the reclaim feedback. This turned out to be not sufficient for higher orders because compaction can deffer and skip if we are close to watermarks which is really surprising to me. So now I've tried to make sure that we do not hit this path. I agree we can do better but there always will be a moment to simply give up. Whatever that moment will be we can still find loads which could theoretically go on for little more and survive.

Problem is that, to me, current implementation looks really simple give up. Maybe, precise definition about premature would be helpful here. Without it, it would be just subjective.

Quoted 11 lines
Quoted 8 lines

I'm really tire to say same thing again and again.
Am I missing something? This is the situation that I totally misunderstand something? Please let me know.

Note: your current implementation doesn't consider which zone is compacted. If DMA zone which easily fail to make high order page is compacted, your implementation will not do retry. It also looks not our best.

Why are we even consider DMA zone when we cannot ever allocate from this zone?

This is just an example. It could be ZONE_NORMAL and something else. If we don't try all zones to compact, it's reasonable point to trigger OOM?

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#121 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Quoted 17 lines · Joonsoo Kim

2016-03-09 1:05 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 16 lines · Joonsoo Kim
Quoted 7 lines

2016-03-08 1:08 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 6 lines · Michal Hocko
Quoted 4 lines

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 5 lines
Quoted 2 lines

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help

I'm still not happy. Just ensuring one compaction run doesn't mean our best.

OK, let me think about it some more.

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

What's your purpose of OOM rework? From my understanding, you'd like to trigger OOM kill deterministic and *not prematurely*. This makes sense.

Well this is a bit awkward because we do not have any proper definition of what prematurely actually means. We do not know whether something

If we don't have proper definition to it, please define it first.

OK, I should have probably said that _there_is_no_proper_definition_... This will always be about heuristics as the clear cut can be pretty subjective and what some load might see as unreasonable retries others might see as insufficient. Our ultimate goal is to behave reasonable for reasonable workloads. I am somehow skeptical about formulating this into a single equation...

Quoted 2 lines

We need to improve the situation toward the clear goal. Just certain number of retry which has no base doesn't make any sense.

Certain number of retries is what we already have right now. And that certain number is hard to define even though it looks as simple as

NR_PAGES_SCANNED < 6*zone_reclaimable_pages && no_reclaimable_pages

because this is highly fragile when there are only few pages freed regularly but not sufficient to get us out of the loop... I am trying to formulate those retries somehow more deterministically considering the feedback _and_ an estimate about the feasibility of future reclaim/compaction. I admit that my attempts at compaction part have been far from ideal so far. Partially because I missed many aspects how it works.

[...]

Quoted 22 lines
Quoted 13 lines

not fire _often_ to be impractical. There are loads where the new implementation behaved slightly better (see the cover for my tests) and there surely be some where this will be worse. I want this to be reasonably good. I am not claiming we are there yet and the interaction with the compaction seems like it needs some work, no question about that.

Quoted 5 lines

But, what you did in case of high order allocation is completely different with original purpose. It may be deterministic but *completely premature*. There is no way to prevent premature OOM kill. So, I want to ask one more time. Why OOM kill is better than retry reclaiming when there is reclaimable page? Deterministic is for what? It ensures something more?

Quoted 8 lines

yes, If we keep reclaiming we can soon start trashing or over reclaim too much which would hurt more processes. If you invoke the OOM killer instead then chances are that you will release a lot of memory at once and that would help to reconcile the memory pressure as well as free some page blocks which couldn't have been compacted before and not affect potentially many processes. The effect would be reduced to a single process. If we had a proper trashing detection feedback we could do much more clever decisions of course.

Quoted 6 lines

It looks like you did it for performance reason. You'd better think again about effect of OOM kill. We don't have enough knowledge about user space program architecture and killing one important process could lead to whole system unusable. Moreover, OOM kill could cause important data loss so should be avoided as much as possible. Performance reason cannot justify OOM kill.

No I am not talking about performance. I am talking about the system healthiness as whole.

Quoted 9 lines
Quoted 6 lines

But back to the !costly OOMs. Once your system is fragmented so heavily that there are no free blocks that would satisfy !costly request then something has gone terribly wrong and we should fix it. To me it sounds like we do not care about those requests early enough and only start carying after we hit the wall. Maybe kcompactd can help us in this regards.

Yes, but, it's another issue. In any situation, !costly OOM should not happen prematurely.

I fully agree and I guess we also agree on the assumption that we shouldn't retry endlessly. So let's focus on what the OOM convergence criteria should look like. I have another proposal which I will send as a reply to the previous one.

Quoted 11 lines
Quoted 7 lines
Quoted 4 lines

Please see Hugh's latest vmstat. There are plenty of anon pages when OOM kill happens and it may have enough swap space. Even if compaction runs and fails, why do we need to kill something in this case? OOM kill should be a last resort.

Well this would be the case even if we were trashing over swap. Refaulting the swapped out memory all over again...

If thrashing is a main obstacle to decide proper OOM point, we need to invent a way to handle thrashing or invent reasonable metric which isn't affected by thrashing.

Great, you are welcome to come up with one. But more seriously, isn't the retries limiting a way to reduce the chances of threshing? It might be not the ideal one because it doesn't work 100% but can we simply come up with the one which works that reliable. This is a hard problem which we haven't been able to solve for ages.

[...]

Thanks!

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#122 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

2016-03-09 19:41 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 25 lines · Joonsoo Kim
Quoted 24 lines

2016-03-09 1:05 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 22 lines · Joonsoo Kim
Quoted 7 lines

2016-03-08 1:08 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 6 lines · Michal Hocko
Quoted 4 lines

Andrew,
could you queue this one as well, please? This is more a band aid than a real solution which I will be working on as soon as I am able to reproduce the issue but the patch should help to some degree at least.

Quoted 5 lines
Quoted 2 lines

Joonsoo wasn't very happy about this approach so let me try a different way. What do you think about the following? Hugh, Sergey does it help

I'm still not happy. Just ensuring one compaction run doesn't mean our best.

OK, let me think about it some more.

Quoted 3 lines

What's your purpose of OOM rework? From my understanding, you'd like to trigger OOM kill deterministic and *not prematurely*. This makes sense.

Well this is a bit awkward because we do not have any proper definition of what prematurely actually means. We do not know whether something

Quoted 8 lines
Quoted 1 line

If we don't have proper definition to it, please define it first.

OK, I should have probably said that _there_is_no_proper_definition_... This will always be about heuristics as the clear cut can be pretty subjective and what some load might see as unreasonable retries others might see as insufficient. Our ultimate goal is to behave reasonable for reasonable workloads. I am somehow skeptical about formulating this into a single equation...

I don't want a theoretically perfect definition. We need something that can be used for judging further changes. So, how can you judge that reasonable behave for reasonable workload? What's your criteria? If someone complains 16 retries is too small and the other complains 16 retries is too big, what's your decision in this case?

If you decide to increase number of retry in this case, when can we stop that increasing? If someone complains again that XX is too small then do you continue to increase it?

For me, for order 0 case, reasonable part is watermark checking with available (free + reclaimable) memory. It shows that we've done our best so it doesn't matter that how many times we retry.

But, for high order case, there is no *feasible* estimation. Watermark check as you did here isn't feasible because high order freepage problem usually happen when there are enough but fragmented freepages. It would be always failed. Without feasible estimation, N retry can't show anything.

Your logic here is just like below.

"We've tried N times reclaim/compaction and failed. It is proved that there is no possibility to make high order page. We should trigger OOM now."

Is it true that there is no possibility to make high order page in this case? Can you be sure?

If someone who get OOM complains regression, can you persuade him by above logic?

I don't think so. This is why I ask you to make proper definition on term *premature* here.

Quoted 45 lines
Quoted 2 lines

We need to improve the situation toward the clear goal. Just certain number of retry which has no base doesn't make any sense.

Certain number of retries is what we already have right now. And that certain number is hard to define even though it looks as simple as

NR_PAGES_SCANNED < 6*zone_reclaimable_pages && no_reclaimable_pages

because this is highly fragile when there are only few pages freed regularly but not sufficient to get us out of the loop... I am trying to formulate those retries somehow more deterministically considering the feedback _and_ an estimate about the feasibility of future reclaim/compaction. I admit that my attempts at compaction part have been far from ideal so far. Partially because I missed many aspects how it works.
[...]

Quoted 22 lines
Quoted 13 lines

not fire _often_ to be impractical. There are loads where the new implementation behaved slightly better (see the cover for my tests) and there surely be some where this will be worse. I want this to be reasonably good. I am not claiming we are there yet and the interaction with the compaction seems like it needs some work, no question about that.

Quoted 5 lines

But, what you did in case of high order allocation is completely different with original purpose. It may be deterministic but *completely premature*. There is no way to prevent premature OOM kill. So, I want to ask one more time. Why OOM kill is better than retry reclaiming when there is reclaimable page? Deterministic is for what? It ensures something more?

Quoted 8 lines

yes, If we keep reclaiming we can soon start trashing or over reclaim too much which would hurt more processes. If you invoke the OOM killer instead then chances are that you will release a lot of memory at once and that would help to reconcile the memory pressure as well as free some page blocks which couldn't have been compacted before and not affect potentially many processes. The effect would be reduced to a single process. If we had a proper trashing detection feedback we could do much more clever decisions of course.

Quoted 6 lines

It looks like you did it for performance reason. You'd better think again about effect of OOM kill. We don't have enough knowledge about user space program architecture and killing one important process could lead to whole system unusable. Moreover, OOM kill could cause important data loss so should be avoided as much as possible. Performance reason cannot justify OOM kill.

Quoted 2 lines

No I am not talking about performance. I am talking about the system healthiness as whole.

So, do you think that more frequent OOM kill is healthier than other ways?

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

But back to the !costly OOMs. Once your system is fragmented so heavily that there are no free blocks that would satisfy !costly request then something has gone terribly wrong and we should fix it. To me it sounds like we do not care about those requests early enough and only start carying after we hit the wall. Maybe kcompactd can help us in this regards.

Quoted 7 lines
Quoted 2 lines

Yes, but, it's another issue. In any situation, !costly OOM should not happen prematurely.

I fully agree and I guess we also agree on the assumption that we shouldn't retry endlessly. So let's focus on what the OOM convergence criteria should look like. I have another proposal which I will send as a reply to the previous one.

That's also insufficient to me. It just add one more brute force retry for compaction
without any reasonable estimation.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

Please see Hugh's latest vmstat. There are plenty of anon pages when OOM kill happens and it may have enough swap space. Even if compaction runs and fails, why do we need to kill something in this case? OOM kill should be a last resort.

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

Well this would be the case even if we were trashing over swap. Refaulting the swapped out memory all over again...

If thrashing is a main obstacle to decide proper OOM point, we need to invent a way to handle thrashing or invent reasonable metric which isn't affected by thrashing.

Great, you are welcome to come up with one. But more seriously, isn't

For example, we can collect how many pages reclaimed and compare it with reclaimable pages at start. If number of reclaimed pages exceed it, we can think that we've tried to reclaim all reclaimable pages at least once and can go next step such as OOM.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#123 lore raw

Re: [PATCH] mm, oom: protect !costly allocations some more (was: Re: [PATCH 0/3] OOM detection rework v4)

Quoted 4 lines · Joonsoo Kim

2016-03-09 19:41 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 3 lines · Joonsoo Kim
Quoted 2 lines

2016-03-09 1:05 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 1 line · Joonsoo Kim

[...]

Quoted 16 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines

What's your purpose of OOM rework? From my understanding, you'd like to trigger OOM kill deterministic and *not prematurely*. This makes sense.

Well this is a bit awkward because we do not have any proper definition of what prematurely actually means. We do not know whether something

Quoted 8 lines
Quoted 1 line

If we don't have proper definition to it, please define it first.

OK, I should have probably said that _there_is_no_proper_definition_... This will always be about heuristics as the clear cut can be pretty subjective and what some load might see as unreasonable retries others might see as insufficient. Our ultimate goal is to behave reasonable for reasonable workloads. I am somehow skeptical about formulating this into a single equation...

Quoted 5 lines

I don't want a theoretically perfect definition. We need something that can be used for judging further changes. So, how can you judge that reasonable behave for reasonable workload? What's your criteria? If someone complains 16 retries is too small and the other complains 16 retries is too big, what's your decision in this case?

The number of retries is the implementation detail. What matters, really, is whether we can argue about the particular load and why it should resp. shouldn't trigger the OOM killer. We can use our tracepoints to have a look and judge the overall progress or lack of it and see if we could do better. It is not the number of retries to tweak first. It is the reclaim/compaction to be made more reliable. Tweaking the retries would be just the very last resort. If we can see that compaction doesn't form the high order pages in a sufficient pace we should find out why.

Quoted 8 lines

If you decide to increase number of retry in this case, when can we stop that increasing? If someone complains again that XX is too small then do you continue to increase it?

For me, for order 0 case, reasonable part is watermark checking with available (free + reclaimable) memory. It shows that we've done our best so it doesn't matter that how many times we retry.

Quoted 5 lines

But, for high order case, there is no *feasible* estimation. Watermark check as you did here isn't feasible because high order freepage problem usually happen when there are enough but fragmented freepages. It would be always failed. Without feasible estimation, N retry can't show anything.

That's why I have done compaction retry loop independent on it in the last patch.

Quoted 4 lines

Your logic here is just like below.

"We've tried N times reclaim/compaction and failed. It is proved that there is no possibility to make high order page. We should trigger OOM now."

Have you seen the last patch where I make sure that the compaction had to report _success_ at least N times to declare the OOM? I think we can be reasonably sure that keep compacting again and again without any bound doesn't make much sense when that doesn't lead to a requested order page.

Quoted 2 lines

Is it true that there is no possibility to make high order page in this case? Can you be sure?

The thing I am trying to tell you, and I seem to fail here, is that you simply cannot be sure. Full stop. We might be staggering on the edge of the cliff and fall or be lucky and end up on the safe side.

Quoted 2 lines

If someone who get OOM complains regression, can you persuade him by above logic?

This really depends on the particular load of course.

Quoted 2 lines

I don't think so. This is why I ask you to make proper definition on term *premature* here.

Sigh. And what if that particular reporter doesn't agree with my "proper" definition because it doesn't suite the workload of the interest? I mean, anything we end up doing is highly subjective and it's been like that since ever OOM was introduced.

[...]

Quoted 11 lines
Quoted 9 lines
Quoted 6 lines

It looks like you did it for performance reason. You'd better think again about effect of OOM kill. We don't have enough knowledge about user space program architecture and killing one important process could lead to whole system unusable. Moreover, OOM kill could cause important data loss so should be avoided as much as possible. Performance reason cannot justify OOM kill.

No I am not talking about performance. I am talking about the system healthiness as whole.

So, do you think that more frequent OOM kill is healthier than other ways?

I didn't say so. And except for the Hugh's testcase I haven't seen the rework would cause that. As per the last testing result it seems that this particular case has been fixed. If you believe that you can see other cases than I am more than happy to look at them.

Quoted 10 lines
Quoted 10 lines
Quoted 9 lines
Quoted 6 lines

But back to the !costly OOMs. Once your system is fragmented so heavily that there are no free blocks that would satisfy !costly request then something has gone terribly wrong and we should fix it. To me it sounds like we do not care about those requests early enough and only start carying after we hit the wall. Maybe kcompactd can help us in this regards.

Yes, but, it's another issue. In any situation, !costly OOM should not happen prematurely.

Quoted 8 lines
Quoted 4 lines

I fully agree and I guess we also agree on the assumption that we shouldn't retry endlessly. So let's focus on what the OOM convergence criteria should look like. I have another proposal which I will send as a reply to the previous one.

That's also insufficient to me. It just add one more brute force retry for compaction
without any reasonable estimation.

The compaction absolutely lacks any useful feedback mechanism. If we ever grow one I am more than happy to make the estimate better.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#124 lore raw
In reply to Hugh Dickins

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 8 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

Quoted 7 lines

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

I have tried to reproduce and failed in a virtual on my laptop. I will try with another host with more CPUs (because my laptop has only two). Just for the record I did: boot 1G machine in kvm, I have 2G swap and reserve 800M for hugetlb pages (I got 445 of them). Then I extract the kernel source to tmpfs (-o size=2G), make defconfig and make -j20 (16, 10 no difference really). I was also collecting vmstat in the background. The compilation takes ages but the behavior seems consistent and stable.

If I try 900M for huge pages then I get OOMs but this happens with the mmotm without my oom rework patch set as well.

It would be great if you could retry and collect /proc/vmstat data around the OOM time to see what compaction did? (I was using the attached little program to reduce interference during OOM (no forks, the code locked in and the resulting file preallocated - e.g. read_vmstat 1s vmstat.log 10M and interrupt it by ctrl+c after the OOM hits).

Thanks!

Signature · 2 lines
Michal Hocko
SUSE Labs

#define _GNU_SOURCE
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <limits.h>
#include <unistd.h>
#include <time.h>

/*
* A simple /proc/vmstat collector into a file. It tries hard to guarantee
* that the content will get into the output file even under a strong memory
* pressure.
*
* Usage
* ./read_vmstat output_file timeout output_size
*
* output_file can be either a non-existing file or - for stdout
* timeout - time period between two snapshots. s - seconds, ms - miliseconds
* and m - minutes suffix is allowed
* output_file - size of the output file. The file is preallocated and pre-filled. *
* If the output reaches the end of the file it will start over overwriting the oldest
* data. Each snapshot is enclosed by header and footer.
* =S timestamp
* [...]
* E=
*
* Please note that your ulimit has to be sufficient to allow to mlock the code+
* all the buffers.
*
* This comes under GPL v2
* Copyright: Michal Hocko <[email protected]> 2015
*/

#define NS_PER_MS (1000*1000)
#define NS_PER_SEC (1000*NS_PER_MS)

int open_file(const char *str)

{
	int fd;

	fd = open(str, O_CREAT|O_EXCL|O_RDWR, 0755);
	if (fd == -1) {
		perror("open input");
		return 1;
	}

	return fd;
}

int read_timeout(const char *str, struct timespec *timeout)

{
	char *end;
	unsigned long val;

	val = strtoul(str, &end, 10);
	if (val == ULONG_MAX) {
		perror("Invalid number");
		return 1;
	}
	switch(*end) {
		case 's':
			timeout->tv_sec = val;
			break;
		case 'm':
			/* ms vs minute*/
			if (*(end+1) == 's') {
				timeout->tv_sec = (val * NS_PER_MS) / NS_PER_SEC;
				timeout->tv_nsec = (val * NS_PER_MS) % NS_PER_SEC;
			} else {
				timeout->tv_sec = val*60;
			}
			break;
		default:
			fprintf(stderr, "Uknown number %s\n", str);
			return 1;
	}

	return 0;
}

size_t read_size(const char *str)

{
	char *end;
	size_t val = strtoul(str, &end, 10);

	switch (*end) {
		case 'K':
			val <<= 10;
			break;
		case 'M':
			val <<= 20;
			break;
		case 'G':
			val <<= 30;
			break;
	}

	return val;
}

size_t dump_str(char *buffer, size_t buffer_size, size_t pos, const char *in, size_t size)

{
	size_t i;
	for (i = 0; i < size; i++) {
		buffer[pos] = in[i];
		pos = (pos + 1) % buffer_size;
	}

	return pos;
}

/* buffer == NULL -> stdout */
int __collect_logs(const struct timespec *timeout, char *buffer, size_t buffer_size)

{
	char buff[4096]; /* dump to the file automatically */
	time_t before, after;
	int in_fd = open("/proc/vmstat", O_RDONLY);
	size_t out_pos = 0;
	size_t in_pos = 0;
	size_t size = 0;
	int estimate = 0;

	if (in_fd == -1) {
		perror("open vmstat:");
		return 1;
	}

	/* lock everything in */
	if (mlockall(MCL_CURRENT) == -1) {
		perror("mlockall. Continuing anyway");
	}

	while (1) {
		before = time(NULL);

		size = snprintf(buff, sizeof(buff), "=S %lu\n", before);
		lseek(in_fd, 0, SEEK_SET);
		size += read(in_fd, buff + size, sizeof(buff) - size);
		size += snprintf(buff + size, sizeof(buff) - size, "E=\n");
		if (buffer && !estimate) {
			printf("Estimated %d entries fit to the buffer\n", buffer_size/size);
			estimate = 1;
		}

		/* Dump to stdout */
		if (!buffer) {
			printf("%s", buff);
		} else {
			size_t pos;
			pos = dump_str(buffer, buffer_size, out_pos, buff, size);
			if (pos < out_pos)
				fprintf(stderr, "%lu: Buffer wrapped\n", before);
			out_pos = pos;
		}

		after = time(NULL);

		if (after - before > 2) {
			fprintf(stderr, "%d: Snapshoting took %d!!!\n", before, after-before);
		}
		if (nanosleep(timeout, NULL) == -1)
			if (errno == EINTR)
				return 0;
		/* kick in the flushing */
		if (buffer)
			msync(buffer, buffer_size, MS_ASYNC);
	}
}

int collect_logs(int fd, const struct timespec *timeout, size_t buffer_size)

{
	unsigned char *buffer = NULL;

	if (fd != -1) {
		if (ftruncate(fd, buffer_size) == -1) {
			perror("ftruncate");
			return 1;
		}

		if (fallocate(fd, 0, 0, buffer_size) && errno != EOPNOTSUPP) {
			perror("fallocate");
			return 1;
		}

		/* commit it to the disk */
		sync();

		buffer = mmap(NULL, buffer_size, PROT_READ | PROT_WRITE,
				MAP_SHARED | MAP_POPULATE, fd, 0);
		if (buffer == MAP_FAILED) {
			perror("mmap");
			return 1;
		}
	}

	return __collect_logs(timeout, buffer, buffer_size);
}

int main(int argc, char **argv)

{
	struct timespec timeout = {.tv_sec = 1};
	int fd = -1;
	size_t buffer_size = 10UL<<20;

	if (argc > 1) {
		/* output file */
		if (strcmp(argv[1], "-")) {
			fd = open_file(argv[1]);
			if (fd == -1)
				return 1;
		}

		/* timeout */
		if (argc > 2) {
			if (read_timeout(argv[2], &timeout))
				return 1;

			/* buffer size */
			if (argc > 3) {
				buffer_size = read_size(argv[3]);
				if (buffer_size == -1UL)
					return 1;
			}
		}
	}
	printf("file:%s timeout:%lu.%lus buffer_size:%llu\n",
			(fd == -1)? "stdout" : argv[1],
			timeout.tv_sec, timeout.tv_nsec / NS_PER_MS,
			buffer_size);

	return collect_logs(fd, &timeout, buffer_size);
}
#125 lore raw
Quoted 18 lines · Michal Hocko

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 8 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

Quoted 7 lines

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

Quoted 8 lines

I have tried to reproduce and failed in a virtual on my laptop. I will try with another host with more CPUs (because my laptop has only two). Just for the record I did: boot 1G machine in kvm, I have 2G swap and reserve 800M for hugetlb pages (I got 445 of them). Then I extract the kernel source to tmpfs (-o size=2G), make defconfig and make -j20 (16, 10 no difference really). I was also collecting vmstat in the background. The compilation takes ages but the behavior seems consistent and stable.

Thanks a lot for giving it a go.

I'm puzzled. 445 hugetlb pages in 800M surprises me: some of them are less than 2M big?? But probably that's just a misunderstanding or typo somewhere.

Ignoring that, you're successfully doing a make -20 defconfig build in tmpfs, with only 224M of RAM available, plus 2G of swap? I'm not at all surprised that it takes ages, but I am very surprised that it does not OOM. I suppose by rights it ought not to OOM, the built tree occupies only a little more than 1G, so you do have enough swap; but I wouldn't get anywhere near that myself without OOMing - I give myself 1G of RAM (well, minus whatever the booted system takes up) to do that build in, four times your RAM, yet in my case it OOMs.

That source tree alone occupies more than 700M, so just copying it into your tmpfs would take a long time. I'd expect a build in 224M RAM plus 2G of swap to take so long, that I'd be very grateful to be OOM killed, even if there is technically enough space. Unless perhaps it's some superfast swap that you have?

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

It's true that my testing has been done on the physical machines, no virtualization involved: I expect that accounts for some difference between us, but as much difference as we're seeing? That's strange.

Quoted 3 lines

If I try 900M for huge pages then I get OOMs but this happens with the mmotm without my oom rework patch set as well.

Right, not at all surprising.

Quoted 9 lines

It would be great if you could retry and collect /proc/vmstat data around the OOM time to see what compaction did? (I was using the attached little program to reduce interference during OOM (no forks, the code locked in and the resulting file preallocated - e.g. read_vmstat 1s vmstat.log 10M and interrupt it by ctrl+c after the OOM hits).

Thanks!

I'll give it a try, thanks, but not tonight.

Hugh

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#126 lore raw

[Adding Vlastimil and Joonsoo for compaction related things - this was a large thread but the more interesting part starts with
http://lkml.kernel.org/r/[email protected]]

Quoted 28 lines · Hugh Dickins
Quoted 18 lines · Michal Hocko

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 8 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

Quoted 7 lines

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

Quoted 8 lines

I have tried to reproduce and failed in a virtual on my laptop. I will try with another host with more CPUs (because my laptop has only two). Just for the record I did: boot 1G machine in kvm, I have 2G swap and reserve 800M for hugetlb pages (I got 445 of them). Then I extract the kernel source to tmpfs (-o size=2G), make defconfig and make -j20 (16, 10 no difference really). I was also collecting vmstat in the background. The compilation takes ages but the behavior seems consistent and stable.

Quoted 5 lines

Thanks a lot for giving it a go.

I'm puzzled. 445 hugetlb pages in 800M surprises me: some of them are less than 2M big?? But probably that's just a misunderstanding or typo somewhere.

A typo. 445 was from 900M test which I was doing while writing the email. Sorry about the confusion.

Quoted 11 lines

Ignoring that, you're successfully doing a make -20 defconfig build in tmpfs, with only 224M of RAM available, plus 2G of swap? I'm not at all surprised that it takes ages, but I am very surprised that it does not OOM. I suppose by rights it ought not to OOM, the built tree occupies only a little more than 1G, so you do have enough swap; but I wouldn't get anywhere near that myself without OOMing - I give myself 1G of RAM (well, minus whatever the booted system takes up) to do that build in, four times your RAM, yet in my case it OOMs.

That source tree alone occupies more than 700M, so just copying it into your tmpfs would take a long time.

OK, I just found out that I was cheating a bit. I was building linux-3.7-rc5.tar.bz2 which is smaller:

$ du -sh /mnt/tmpfs/linux-3.7-rc5/
537M    /mnt/tmpfs/linux-3.7-rc5/

and after the defconfig build:

$ free
             total       used       free     shared    buffers     cached
Mem:       1008460     941904      66556          0       5092     806760
-/+ buffers/cache:     130052     878408
Swap:      2097148      42648    2054500
$ du -sh linux-3.7-rc5/
799M    linux-3.7-rc5/

Sorry about that but this is what my other tests were using and I forgot to check. Now let's try the same with the current linus tree:
host $ git archive v4.5-rc6 --prefix=linux-4.5-rc6/ | bzip2 > linux-4.5-rc6.tar.bz2

$ du -sh /mnt/tmpfs/linux-4.5-rc6/
707M    /mnt/tmpfs/linux-4.5-rc6/
$ free
             total       used       free     shared    buffers     cached
Mem:       1008460     962976      45484          0       7236     820064
-/+ buffers/cache:     135676     872784
Swap:      2097148         16    2097132
$ time make -j20 > /dev/null
drivers/acpi/property.c: In function a??acpi_data_prop_reada??:
drivers/acpi/property.c:745:8: warning: a??obja?? may be used uninitialized in this function [-Wmaybe-uninitialized]
real    8m36.621s
user    14m1.642s
sys     2m45.238s

so I wasn't cheating all that much...

Quoted 4 lines

I'd expect a build in 224M
RAM plus 2G of swap to take so long, that I'd be very grateful to be OOM killed, even if there is technically enough space. Unless perhaps it's some superfast swap that you have?

the swap partition is a standard qcow image stored on my SSD disk. So I guess the IO should be quite fast. This smells like a potential contributor because my reclaim seems to be much faster and that should lead to a more efficient reclaim (in the scanned/reclaimed sense). I realize I might be boring already when blaming compaction but let me try again ;)

$ grep compact /proc/vmstat 
compact_migrate_scanned 113983
compact_free_scanned 1433503
compact_isolated 134307
compact_stall 128
compact_fail 26
compact_success 102
compact_kcompatd_wake 0

So the whole load has done the direct compaction only 128 times during that test. This doesn't sound much to me

$ grep allocstall /proc/vmstat
allocstall 1061

we entered the direct reclaim much more but most of the load will be order-0 so this might be still ok. So I've tried the following:

mm/page_alloc.c +3 -0
index 1993894b4219..107d444afdb1 100644
@@ -2910,6 +2910,9 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
+	if (order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER)
+		trace_printk("order:%d gfp_mask:%pGg compact_result:%lu\n", order, &gfp_mask, compact_result);
+
 	switch (compact_result) {
 	case COMPACT_DEFERRED:
 		*deferred_compaction = true;

And the result was:

$ cat /debug/tracing/trace_pipe | tee ~/trace.log
             gcc-8707  [001] ....   137.946370: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1
             gcc-8726  [000] ....   138.528571: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1

this shows that order-2 memory pressure is not overly high in my setup. Both attempts ended up COMPACT_SKIPPED which is interesting.

So I went back to 800M of hugetlb pages and tried again. It took ages so I have interrupted that after one hour (there was still no OOM). The trace log is quite interesting regardless:

$ wc -l ~/trace.log
371 /root/trace.log
$ grep compact_stall /proc/vmstat 
compact_stall 190

so the compaction was still ignored more than actually invoked for
!costly allocations:
sed 's@.*order:\([[:digit:]]\).* compact_result:\([[:digit:]]\)@\1 \2@' ~/trace.log | sort | uniq -c

    190 2 1
    122 2 3
     59 2 4
#define COMPACT_SKIPPED         1               
#define COMPACT_PARTIAL         3
#define COMPACT_COMPLETE        4

that means that compaction is even not tried in half cases! This doesn't sounds right to me, especially when we are talking about <= PAGE_ALLOC_COSTLY_ORDER requests which are implicitly nofail, because then we simply rely on the order-0 reclaim to automagically form higher blocks. This might indeed work when we retry many times but I guess this is not a good approach. It leads to a excessive reclaim and the stall for allocation can be really large.

One of the suspicious places is __compaction_suitable which does order-0 watermark check (increased by 2<<order). I have put another trace_printk there and it clearly pointed out this was the case.

So I have tried the following:

mm/compaction.c +3 -0
index 4d99e1f5055c..7364e48cf69a 100644
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+
 	/*
 	 * Watermarks for order-0 must be met for compaction. Note the 2UL.
 	 * This is because during migration, copies of pages need to be

and retried the same test (without huge pages):

$ time make -j20 > /dev/null
real    8m46.626s
user    14m15.823s
sys     2m45.471s

the time increased but I haven't checked how stable the result is.

$ grep compact /proc/vmstat
compact_migrate_scanned 139822
compact_free_scanned 1661642
compact_isolated 139407
compact_stall 129
compact_fail 58
compact_success 71
compact_kcompatd_wake 1
$ grep allocstall /proc/vmstat
allocstall 1665

this is worse because we have scanned more pages for migration but the overall success rate was much smaller and the direct reclaim was invoked more. I do not have a good theory for that and will play with this some more. Maybe other changes are needed deeper in the compaction code.

I will play with this some more but I would be really interested to hear whether this helped Hugh with his setup. Vlastimi, Joonsoo does this even make sense to you?

Quoted 3 lines

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

Ohh, I see.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author reviewer
#127 lore raw

On Tue 01-03-16 14:38:46, Michal Hocko wrote:
[...]

Quoted 1 line

the time increased but I haven't checked how stable the result is.

And those results vary a lot (even when executed from the fresh boot) as per my further testing. Sure it might be related to the virtual environment but I do not think this particular test should be used for the performance regression comparision.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#128 lore raw
In reply to Michal Hocko
Quoted 67 lines · Michal Hocko
$ grep compact /proc/vmstat
compact_migrate_scanned 113983
compact_free_scanned 1433503
compact_isolated 134307
compact_stall 128
compact_fail 26
compact_success 102
compact_kcompatd_wake 0

So the whole load has done the direct compaction only 128 times during that test. This doesn't sound much to me

$ grep allocstall /proc/vmstat
allocstall 1061

we entered the direct reclaim much more but most of the load will be order-0 so this might be still ok. So I've tried the following:

mm/page_alloc.c +3 -0
index 1993894b4219..107d444afdb1 100644
@@ -2910,6 +2910,9 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
  						mode, contended_compaction);
  	current->flags &= ~PF_MEMALLOC;
 
+	if (order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER)
+		trace_printk("order:%d gfp_mask:%pGg compact_result:%lu\n", order, &gfp_mask, compact_result);
+
  	switch (compact_result) {
  	case COMPACT_DEFERRED:
  		*deferred_compaction = true;

And the result was:

$ cat /debug/tracing/trace_pipe | tee ~/trace.log
              gcc-8707  [001] ....   137.946370: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1
              gcc-8726  [000] ....   138.528571: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1

this shows that order-2 memory pressure is not overly high in my setup. Both attempts ended up COMPACT_SKIPPED which is interesting.

So I went back to 800M of hugetlb pages and tried again. It took ages so I have interrupted that after one hour (there was still no OOM). The trace log is quite interesting regardless:

$ wc -l ~/trace.log
371 /root/trace.log
$ grep compact_stall /proc/vmstat
compact_stall 190

so the compaction was still ignored more than actually invoked for
!costly allocations:
sed 's@.*order:\([[:digit:]]\).* compact_result:\([[:digit:]]\)@\1 \2@' ~/trace.log | sort | uniq -c

     190 2 1
     122 2 3
      59 2 4
#define COMPACT_SKIPPED         1
#define COMPACT_PARTIAL         3
#define COMPACT_COMPLETE        4

that means that compaction is even not tried in half cases! This doesn't sounds right to me, especially when we are talking about <= PAGE_ALLOC_COSTLY_ORDER requests which are implicitly nofail, because then we simply rely on the order-0 reclaim to automagically form higher blocks. This might indeed work when we retry many times but I guess this is not a good approach. It leads to a excessive reclaim and the stall for allocation can be really large.

Quoted 3 lines

One of the suspicious places is __compaction_suitable which does order-0 watermark check (increased by 2<<order). I have put another trace_printk there and it clearly pointed out this was the case.

Yes, compaction is historically quite careful to avoid making low memory conditions worse, and to prevent work if it doesn't look like it can ultimately succeed the allocation (so having not enough base pages means that compacting them is considered pointless). This aspect of preventing non-zero-order OOMs is somewhat unexpected :)

Quoted 34 lines

So I have tried the following:

mm/compaction.c +3 -0
index 4d99e1f5055c..7364e48cf69a 100644
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
  								alloc_flags))
  		return COMPACT_PARTIAL;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+
  	/*
  	 * Watermarks for order-0 must be met for compaction. Note the 2UL.
  	 * This is because during migration, copies of pages need to be

and retried the same test (without huge pages):

$ time make -j20 > /dev/null
real    8m46.626s
user    14m15.823s
sys     2m45.471s

the time increased but I haven't checked how stable the result is.

$ grep compact /proc/vmstat
compact_migrate_scanned 139822
compact_free_scanned 1661642
compact_isolated 139407
compact_stall 129
compact_fail 58
compact_success 71
compact_kcompatd_wake 1
Quoted 7 lines
$ grep allocstall /proc/vmstat
allocstall 1665

this is worse because we have scanned more pages for migration but the overall success rate was much smaller and the direct reclaim was invoked more. I do not have a good theory for that and will play with this some more. Maybe other changes are needed deeper in the compaction code.

I was under impression that similar checks to compaction_suitable() were done also in compact_finished(), to stop compacting if memory got low due to parallel activity. But I guess it was a patch from Joonsoo that didn't get merged.

My only other theory so far is that watermark checks fail in __isolate_free_page() when we want to grab page(s) as migration targets. I would suggest enabling all compaction tracepoint and the migration tracepoint. Looking at the trace could hopefully help faster than going one trace_printk() per attempt.

Once we learn all the relevant places/checks, we can think about how to communicate to them that this compaction attempt is "important" and should continue as long as possible even in low-memory conditions. Maybe not just a costly order check, but we also have alloc_flags or could add something to compact_control, etc.

Quoted 11 lines

I will play with this some more but I would be really interested to hear whether this helped Hugh with his setup. Vlastimi, Joonsoo does this even make sense to you?

Quoted 3 lines

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

Ohh, I see.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#129 lore raw
Quoted 68 lines · Vlastimil Babka
Quoted 67 lines · Michal Hocko
$ grep compact /proc/vmstat
compact_migrate_scanned 113983
compact_free_scanned 1433503
compact_isolated 134307
compact_stall 128
compact_fail 26
compact_success 102
compact_kcompatd_wake 0

So the whole load has done the direct compaction only 128 times during that test. This doesn't sound much to me

$ grep allocstall /proc/vmstat
allocstall 1061

we entered the direct reclaim much more but most of the load will be order-0 so this might be still ok. So I've tried the following:

mm/page_alloc.c +3 -0
index 1993894b4219..107d444afdb1 100644
@@ -2910,6 +2910,9 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
+	if (order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER)
+		trace_printk("order:%d gfp_mask:%pGg compact_result:%lu\n", order, &gfp_mask, compact_result);
+
 	switch (compact_result) {
 	case COMPACT_DEFERRED:
 		*deferred_compaction = true;

And the result was:

$ cat /debug/tracing/trace_pipe | tee ~/trace.log
             gcc-8707  [001] ....   137.946370: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1
             gcc-8726  [000] ....   138.528571: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1

this shows that order-2 memory pressure is not overly high in my setup. Both attempts ended up COMPACT_SKIPPED which is interesting.

So I went back to 800M of hugetlb pages and tried again. It took ages so I have interrupted that after one hour (there was still no OOM). The trace log is quite interesting regardless:

$ wc -l ~/trace.log
371 /root/trace.log
$ grep compact_stall /proc/vmstat
compact_stall 190

so the compaction was still ignored more than actually invoked for
!costly allocations:
sed 's@.*order:\([[:digit:]]\).* compact_result:\([[:digit:]]\)@\1 \2@' ~/trace.log | sort | uniq -c

    190 2 1
    122 2 3
     59 2 4
#define COMPACT_SKIPPED         1
#define COMPACT_PARTIAL         3
#define COMPACT_COMPLETE        4

that means that compaction is even not tried in half cases! This doesn't sounds right to me, especially when we are talking about <= PAGE_ALLOC_COSTLY_ORDER requests which are implicitly nofail, because then we simply rely on the order-0 reclaim to automagically form higher blocks. This might indeed work when we retry many times but I guess this is not a good approach. It leads to a excessive reclaim and the stall for allocation can be really large.

Quoted 9 lines
Quoted 3 lines

One of the suspicious places is __compaction_suitable which does order-0 watermark check (increased by 2<<order). I have put another trace_printk there and it clearly pointed out this was the case.

Yes, compaction is historically quite careful to avoid making low memory conditions worse, and to prevent work if it doesn't look like it can ultimately succeed the allocation (so having not enough base pages means that compacting them is considered pointless). This aspect of preventing non-zero-order OOMs is somewhat unexpected :)

It's better not to assume that compaction would succeed all the times. Compaction has some limitations so it sometimes fails.
For example, in lowmem situation, it only scans small parts of memory and if that part is fragmented by non-movable page, compaction would fail. And, compaction would defer requests 64 times at maximum if successive compaction failure happens before.

Depending on compaction heavily is right direction to go but I think that it's not ready for now. More reclaim would relieve problem.

I tried to fix this situation but not yet finished.

http://thread.gmane.org/gmane.linux.kernel.mm/142364
https://lkml.org/lkml/2015/8/23/182

Quoted 47 lines
Quoted 34 lines

So I have tried the following:

mm/compaction.c +3 -0
index 4d99e1f5055c..7364e48cf69a 100644
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+
 	/*
 	 * Watermarks for order-0 must be met for compaction. Note the 2UL.
 	 * This is because during migration, copies of pages need to be

and retried the same test (without huge pages):

$ time make -j20 > /dev/null
real    8m46.626s
user    14m15.823s
sys     2m45.471s

the time increased but I haven't checked how stable the result is.

$ grep compact /proc/vmstat
compact_migrate_scanned 139822
compact_free_scanned 1661642
compact_isolated 139407
compact_stall 129
compact_fail 58
compact_success 71
compact_kcompatd_wake 1
Quoted 7 lines
$ grep allocstall /proc/vmstat
allocstall 1665

this is worse because we have scanned more pages for migration but the overall success rate was much smaller and the direct reclaim was invoked more. I do not have a good theory for that and will play with this some more. Maybe other changes are needed deeper in the compaction code.

I was under impression that similar checks to compaction_suitable() were done also in compact_finished(), to stop compacting if memory got low due to parallel activity. But I guess it was a patch from Joonsoo that didn't get merged.

Quoted 5 lines

My only other theory so far is that watermark checks fail in __isolate_free_page() when we want to grab page(s) as migration targets. I would suggest enabling all compaction tracepoint and the migration tracepoint. Looking at the trace could hopefully help faster than going one trace_printk() per attempt.

Agreed. It's best thing to do now.

Thanks.

Quoted 19 lines

Once we learn all the relevant places/checks, we can think about how to communicate to them that this compaction attempt is "important" and should continue as long as possible even in low-memory conditions. Maybe not just a costly order check, but we also have alloc_flags or could add something to compact_control, etc.

Quoted 11 lines

I will play with this some more but I would be really interested to hear whether this helped Hugh with his setup. Vlastimi, Joonsoo does this even make sense to you?

Quoted 3 lines

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

Ohh, I see.

Quoted 5 lines

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#130 lore raw
Quoted 1 line · Joonsoo Kim

On Tue, Mar 01, 2016 at 07:14:08PM +0100, Vlastimil Babka wrote:

[...]

Quoted 13 lines
Quoted 5 lines

Yes, compaction is historically quite careful to avoid making low memory conditions worse, and to prevent work if it doesn't look like it can ultimately succeed the allocation (so having not enough base pages means that compacting them is considered pointless). This aspect of preventing non-zero-order OOMs is somewhat unexpected :)

It's better not to assume that compaction would succeed all the times. Compaction has some limitations so it sometimes fails.
For example, in lowmem situation, it only scans small parts of memory and if that part is fragmented by non-movable page, compaction would fail. And, compaction would defer requests 64 times at maximum if successive compaction failure happens before.

Quoted 2 lines

Depending on compaction heavily is right direction to go but I think that it's not ready for now. More reclaim would relieve problem.

I really fail to see why. The reclaimable memory can be migrated as well, no? Relying on the order-0 reclaim makes only sense to get over wmarks.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#131 lore raw

2016-03-02 21:37 GMT+09:00 Michal Hocko <[email protected]>:

Quoted 16 lines · Joonsoo Kim
Quoted 1 line · Vlastimil Babka

[...]

Quoted 13 lines
Quoted 5 lines

Yes, compaction is historically quite careful to avoid making low memory conditions worse, and to prevent work if it doesn't look like it can ultimately succeed the allocation (so having not enough base pages means that compacting them is considered pointless). This aspect of preventing non-zero-order OOMs is somewhat unexpected :)

It's better not to assume that compaction would succeed all the times. Compaction has some limitations so it sometimes fails.
For example, in lowmem situation, it only scans small parts of memory and if that part is fragmented by non-movable page, compaction would fail. And, compaction would defer requests 64 times at maximum if successive compaction failure happens before.

Quoted 6 lines
Quoted 2 lines

Depending on compaction heavily is right direction to go but I think that it's not ready for now. More reclaim would relieve problem.

I really fail to see why. The reclaimable memory can be migrated as well, no? Relying on the order-0 reclaim makes only sense to get over wmarks.

Attached link on previous reply mentioned limitation of current compaction implementation. Briefly speaking, It would not scan all range of memory due to algorithm limitation so even if there is reclaimable memory that can be also migrated, compaction could fail.

There is no such limitation on reclaim and that's why I think that compaction is not ready for now.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#132 lore raw
In reply to Vlastimil Babka
Quoted 1 line · Vlastimil Babka

On 03/01/2016 02:38 PM, Michal Hocko wrote:

[...]

Quoted 8 lines
Quoted 8 lines

that means that compaction is even not tried in half cases! This doesn't sounds right to me, especially when we are talking about <= PAGE_ALLOC_COSTLY_ORDER requests which are implicitly nofail, because then we simply rely on the order-0 reclaim to automagically form higher blocks. This might indeed work when we retry many times but I guess this is not a good approach. It leads to a excessive reclaim and the stall for allocation can be really large.

Quoted 8 lines
Quoted 3 lines

One of the suspicious places is __compaction_suitable which does order-0 watermark check (increased by 2<<order). I have put another trace_printk there and it clearly pointed out this was the case.

Yes, compaction is historically quite careful to avoid making low memory conditions worse, and to prevent work if it doesn't look like it can ultimately succeed the allocation (so having not enough base pages means that compacting them is considered pointless).

The compaction is running in PF_MEMALLOC context so it shouldn't fail the allocation. Moreover the additional memory is only temporal until the migration finishes. Or am I missing something?

Quoted 2 lines

This aspect of preventing non-zero-order OOMs is somewhat unexpected :)

I hope we can do something about it then...

[...]

Quoted 5 lines
Quoted 4 lines

this is worse because we have scanned more pages for migration but the overall success rate was much smaller and the direct reclaim was invoked more. I do not have a good theory for that and will play with this some more. Maybe other changes are needed deeper in the compaction code.

Quoted 7 lines

I was under impression that similar checks to compaction_suitable() were done also in compact_finished(), to stop compacting if memory got low due to parallel activity. But I guess it was a patch from Joonsoo that didn't get merged.

My only other theory so far is that watermark checks fail in __isolate_free_page() when we want to grab page(s) as migration targets.

yes this certainly contributes to the problem and triggered in my case a lot:

$ grep __isolate_free_page trace.log | wc -l
181
$ grep __alloc_pages_direct_compact: trace.log | wc -l
7
Quoted 3 lines

I would suggest enabling all compaction tracepoint and the migration tracepoint. Looking at the trace could hopefully help faster than going one trace_printk() per attempt.

OK, here we go with both watermarks checks removed and hopefully all the compaction related tracepoints enabled:
echo 1 > /debug/tracing/events/compaction/enable
echo 1 > /debug/tracing/events/migrate/mm_migrate_pages/enable

this was without the hugetlb handicap. See the trace log and vmstat after the run attached.

Thanks

Signature · 2 lines
Michal Hocko
SUSE Labs

nr_free_pages 151306
nr_alloc_batch 123
nr_inactive_anon 12815
nr_active_anon 44507
nr_inactive_file 1160
nr_active_file 5910
nr_unevictable 0
nr_mlock 0
nr_anon_pages 232
nr_mapped 1025
nr_file_pages 64246
nr_dirty 2
nr_writeback 0
nr_slab_reclaimable 12344
nr_slab_unreclaimable 21129
nr_page_table_pages 260
nr_kernel_stack 90
nr_unstable 0
nr_bounce 0
nr_vmscan_write 362270
nr_vmscan_immediate_reclaim 43
nr_writeback_temp 0
nr_isolated_anon 0
nr_isolated_file 0
nr_shmem 54592
nr_dirtied 5363
nr_written 364001
nr_pages_scanned 0
workingset_refault 16574
workingset_activate 9062
workingset_nodereclaim 640
nr_anon_transparent_hugepages 0
nr_free_cma 0
nr_dirty_threshold 31188
nr_dirty_background_threshold 15594
pgpgin 564127
pgpgout 1457932
pswpin 85569
pswpout 362180
pgalloc_dma 226916
pgalloc_dma32 21472873
pgalloc_normal 0
pgalloc_movable 0
pgfree 22057596
pgactivate 174766
pgdeactivate 919764
pgfault 23950701
pgmajfault 31819
pglazyfreed 0
pgrefill_dma 15589
pgrefill_dma32 999305
pgrefill_normal 0
pgrefill_movable 0
pgsteal_kswapd_dma 5339
pgsteal_kswapd_dma32 322951
pgsteal_kswapd_normal 0
pgsteal_kswapd_movable 0
pgsteal_direct_dma 334
pgsteal_direct_dma32 71877
pgsteal_direct_normal 0
pgsteal_direct_movable 0
pgscan_kswapd_dma 11213
pgscan_kswapd_dma32 653096
pgscan_kswapd_normal 0
pgscan_kswapd_movable 0
pgscan_direct_dma 670
pgscan_direct_dma32 137488
pgscan_direct_normal 0
pgscan_direct_movable 0
pgscan_direct_throttle 0
pginodesteal 0
slabs_scanned 1920
kswapd_inodesteal 0
kswapd_low_wmark_hit_quickly 351
kswapd_high_wmark_hit_quickly 13
pageoutrun 458
allocstall 1376
pgrotated 360480
drop_pagecache 0
drop_slab 0
pgmigrate_success 204875
pgmigrate_fail 169
compact_migrate_scanned 343087
compact_free_scanned 3597902
compact_isolated 412234
compact_stall 163
compact_fail 99
compact_success 64
compact_kcompatd_wake 2
htlb_buddy_alloc_success 0
htlb_buddy_alloc_fail 0
unevictable_pgs_culled 1089
unevictable_pgs_scanned 0
unevictable_pgs_rescued 1561
unevictable_pgs_mlocked 1561
unevictable_pgs_munlocked 1561
unevictable_pgs_cleared 0
unevictable_pgs_stranded 0
thp_fault_alloc 152
thp_fault_fallback 39
thp_collapse_alloc 69
thp_collapse_alloc_failed 11
thp_split_page 1
thp_split_page_failed 0
thp_deferred_split_page 212
thp_split_pmd 10
thp_zero_page_alloc 2
thp_zero_page_alloc_failed 1

author reviewer
#133 lore raw
Quoted 1 line · Michal Hocko

On Tue 01-03-16 19:14:08, Vlastimil Babka wrote:

[...]

Quoted 11 lines
Quoted 3 lines

I would suggest enabling all compaction tracepoint and the migration tracepoint. Looking at the trace could hopefully help faster than going one trace_printk() per attempt.

OK, here we go with both watermarks checks removed and hopefully all the compaction related tracepoints enabled:
echo 1 > /debug/tracing/events/compaction/enable
echo 1 > /debug/tracing/events/migrate/mm_migrate_pages/enable

this was without the hugetlb handicap. See the trace log and vmstat after the run attached.

Just for the reference the above was with:

mm/compaction.c +3 -0
index 4d99e1f5055c..7364e48cf69a 100644
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+
 	/*
 	 * Watermarks for order-0 must be met for compaction. Note the 2UL.
 	 * This is because during migration, copies of pages need to be
mm/page_alloc.c +1 -8
index 1993894b4219..50954a9a4433 100644
@@ -2245,7 +2245,6 @@ EXPORT_SYMBOL_GPL(split_page);
 
 int __isolate_free_page(struct page *page, unsigned int order)
 {
-	unsigned long watermark;
 	struct zone *zone;
 	int mt;
 
@@ -2254,14 +2253,8 @@ int __isolate_free_page(struct page *page, unsigned int order)
 	zone = page_zone(page);
 	mt = get_pageblock_migratetype(page);
 
-	if (!is_migrate_isolate(mt)) {
-		/* Obey watermarks as if the page was being allocated */
-		watermark = low_wmark_pages(zone) + (1 << order);
-		if (!zone_watermark_ok(zone, 0, watermark, 0, 0))
-			return 0;
-
+	if (!is_migrate_isolate(mt))
 		__mod_zone_freepage_state(zone, -(1UL << order), mt);
-	}
 
 	/* Remove page from free list */
 	list_del(&page->lru);

And I rerun the same with the clean mmotm tree and the results are attached.

As we can see there was less scanning on dma32 both in direct and kswapd reclaim.

$ grep direct vmstat.*
vmstat.mmotm.log:pgsteal_direct_dma 420
vmstat.mmotm.log:pgsteal_direct_dma32 71234
vmstat.mmotm.log:pgsteal_direct_normal 0
vmstat.mmotm.log:pgsteal_direct_movable 0
vmstat.mmotm.log:pgscan_direct_dma 990
vmstat.mmotm.log:pgscan_direct_dma32 144376
vmstat.mmotm.log:pgscan_direct_normal 0
vmstat.mmotm.log:pgscan_direct_movable 0
vmstat.mmotm.log:pgscan_direct_throttle 0
vmstat.updated.log:pgsteal_direct_dma 334
vmstat.updated.log:pgsteal_direct_dma32 71877
vmstat.updated.log:pgsteal_direct_normal 0
vmstat.updated.log:pgsteal_direct_movable 0
vmstat.updated.log:pgscan_direct_dma 670
vmstat.updated.log:pgscan_direct_dma32 137488
vmstat.updated.log:pgscan_direct_normal 0
vmstat.updated.log:pgscan_direct_movable 0
vmstat.updated.log:pgscan_direct_throttle 0
$ grep kswapd vmstat.*
vmstat.mmotm.log:pgsteal_kswapd_dma 5602
vmstat.mmotm.log:pgsteal_kswapd_dma32 332336
vmstat.mmotm.log:pgsteal_kswapd_normal 0
vmstat.mmotm.log:pgsteal_kswapd_movable 0
vmstat.mmotm.log:pgscan_kswapd_dma 12187
vmstat.mmotm.log:pgscan_kswapd_dma32 679667
vmstat.mmotm.log:pgscan_kswapd_normal 0
vmstat.mmotm.log:pgscan_kswapd_movable 0
vmstat.mmotm.log:kswapd_inodesteal 0
vmstat.mmotm.log:kswapd_low_wmark_hit_quickly 339
vmstat.mmotm.log:kswapd_high_wmark_hit_quickly 10
vmstat.updated.log:pgsteal_kswapd_dma 5339
vmstat.updated.log:pgsteal_kswapd_dma32 322951
vmstat.updated.log:pgsteal_kswapd_normal 0
vmstat.updated.log:pgsteal_kswapd_movable 0
vmstat.updated.log:pgscan_kswapd_dma 11213
vmstat.updated.log:pgscan_kswapd_dma32 653096
vmstat.updated.log:pgscan_kswapd_normal 0
vmstat.updated.log:pgscan_kswapd_movable 0
vmstat.updated.log:kswapd_inodesteal 0
vmstat.updated.log:kswapd_low_wmark_hit_quickly 351
vmstat.updated.log:kswapd_high_wmark_hit_quickly 13
Signature · 2 lines
Michal Hocko
SUSE Labs

nr_free_pages 149226
nr_alloc_batch 114
nr_inactive_anon 13962
nr_active_anon 46754
nr_inactive_file 634
nr_active_file 5010
nr_unevictable 0
nr_mlock 0
nr_anon_pages 219
nr_mapped 793
nr_file_pages 66233
nr_dirty 0
nr_writeback 0
nr_slab_reclaimable 12355
nr_slab_unreclaimable 21208
nr_page_table_pages 320
nr_kernel_stack 92
nr_unstable 0
nr_bounce 0
nr_vmscan_write 358705
nr_vmscan_immediate_reclaim 111
nr_writeback_temp 0
nr_isolated_anon 0
nr_isolated_file 0
nr_shmem 58505
nr_dirtied 5516
nr_written 360677
nr_pages_scanned 0
workingset_refault 17291
workingset_activate 11908
workingset_nodereclaim 644
nr_anon_transparent_hugepages 0
nr_free_cma 0
nr_dirty_threshold 30487
nr_dirty_background_threshold 15243
pgpgin 525267
pgpgout 1444464
pswpin 75386
pswpout 358705
pgalloc_dma 241466
pgalloc_dma32 21491760
pgalloc_normal 0
pgalloc_movable 0
pgfree 22110844
pgactivate 204005
pgdeactivate 1033621
pgfault 23929641
pgmajfault 27748
pglazyfreed 0
pgrefill_dma 18759
pgrefill_dma32 1122090
pgrefill_normal 0
pgrefill_movable 0
pgsteal_kswapd_dma 5602
pgsteal_kswapd_dma32 332336
pgsteal_kswapd_normal 0
pgsteal_kswapd_movable 0
pgsteal_direct_dma 420
pgsteal_direct_dma32 71234
pgsteal_direct_normal 0
pgsteal_direct_movable 0
pgscan_kswapd_dma 12187
pgscan_kswapd_dma32 679667
pgscan_kswapd_normal 0
pgscan_kswapd_movable 0
pgscan_direct_dma 990
pgscan_direct_dma32 144376
pgscan_direct_normal 0
pgscan_direct_movable 0
pgscan_direct_throttle 0
pginodesteal 0
slabs_scanned 2052
kswapd_inodesteal 0
kswapd_low_wmark_hit_quickly 339
kswapd_high_wmark_hit_quickly 10
pageoutrun 448
allocstall 1376
pgrotated 357091
drop_pagecache 0
drop_slab 0
pgmigrate_success 227102
pgmigrate_fail 142
compact_migrate_scanned 374515
compact_free_scanned 4000566
compact_isolated 456131
compact_stall 133
compact_fail 73
compact_success 60
compact_kcompatd_wake 0
htlb_buddy_alloc_success 0
htlb_buddy_alloc_fail 0
unevictable_pgs_culled 1087
unevictable_pgs_scanned 0
unevictable_pgs_rescued 1530
unevictable_pgs_mlocked 1530
unevictable_pgs_munlocked 1529
unevictable_pgs_cleared 1
unevictable_pgs_stranded 0
thp_fault_alloc 164
thp_fault_fallback 26
thp_collapse_alloc 159
thp_collapse_alloc_failed 11
thp_split_page 0
thp_split_page_failed 0
thp_deferred_split_page 309
thp_split_pmd 7
thp_zero_page_alloc 3
thp_zero_page_alloc_failed 0

#134 lore raw
In reply to Michal Hocko
Quoted 17 lines · Michal Hocko
Quoted 8 lines · Vlastimil Babka

I was under impression that similar checks to compaction_suitable() were done also in compact_finished(), to stop compacting if memory got low due to parallel activity. But I guess it was a patch from Joonsoo that didn't get merged.

My only other theory so far is that watermark checks fail in __isolate_free_page() when we want to grab page(s) as migration targets.

yes this certainly contributes to the problem and triggered in my case a lot:

$ grep __isolate_free_page trace.log | wc -l
181
$ grep __alloc_pages_direct_compact: trace.log | wc -l
7
Quoted 8 lines
Quoted 3 lines

I would suggest enabling all compaction tracepoint and the migration tracepoint. Looking at the trace could hopefully help faster than going one trace_printk() per attempt.

OK, here we go with both watermarks checks removed and hopefully all the compaction related tracepoints enabled:
echo 1 > /debug/tracing/events/compaction/enable
echo 1 > /debug/tracing/events/migrate/mm_migrate_pages/enable

The trace shows only 4 direct compaction attempts with order=2. The rest is order=9, i.e. THP, which has little chances of success under such pressure, and thus those failures and defers. The few order=2 attempts appear all successful (defer_reset is called).

So it seems your system is mostly fine with just reclaim, and there's little need for order-2 compaction, and that's also why you can't reproduce the OOMs. So I'm afraid we'll learn nothing here, and looks like Hugh will have to try those watermark check adjustments/removals and/or provide the same kind of trace.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#135 lore raw
In reply to Michal Hocko
Quoted 9 lines · Michal Hocko
Quoted 4 lines

I'd expect a build in 224M
RAM plus 2G of swap to take so long, that I'd be very grateful to be OOM killed, even if there is technically enough space. Unless perhaps it's some superfast swap that you have?

the swap partition is a standard qcow image stored on my SSD disk. So I guess the IO should be quite fast. This smells like a potential contributor because my reclaim seems to be much faster and that should lead to a more efficient reclaim (in the scanned/reclaimed sense).

Hmm... This looks like one of potential culprit. If page is in writeback, it can't be migrated by compaction with MIGRATE_SYNC_LIGHT. In this case, this page works as pinned page and prevent compaction. It'd be better to check that changing 'migration_mode = MIGRATE_SYNC' at 'no_progress_loops > XXX' will help in this situation.

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#136 lore raw
Quoted 11 lines · Joonsoo Kim
Quoted 9 lines · Michal Hocko
Quoted 4 lines

I'd expect a build in 224M
RAM plus 2G of swap to take so long, that I'd be very grateful to be OOM killed, even if there is technically enough space. Unless perhaps it's some superfast swap that you have?

the swap partition is a standard qcow image stored on my SSD disk. So I guess the IO should be quite fast. This smells like a potential contributor because my reclaim seems to be much faster and that should lead to a more efficient reclaim (in the scanned/reclaimed sense).

Quoted 5 lines

Hmm... This looks like one of potential culprit. If page is in writeback, it can't be migrated by compaction with MIGRATE_SYNC_LIGHT. In this case, this page works as pinned page and prevent compaction. It'd be better to check that changing 'migration_mode = MIGRATE_SYNC' at 'no_progress_loops > XXX' will help in this situation.

Would it make sense to use MIGRATE_SYNC for !costly allocations by default?

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#137 lore raw
In reply to Michal Hocko
Quoted 4 lines · Michal Hocko

[Adding Vlastimil and Joonsoo for compaction related things - this was a large thread but the more interesting part starts with
http://lkml.kernel.org/r/[email protected]]

Quoted 23 lines
Quoted 22 lines · Hugh Dickins
Quoted 21 lines · Michal Hocko

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 8 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

Quoted 7 lines

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

I have tried to reproduce and failed in a virtual on my laptop. I will try with another host with more CPUs (because my laptop has only two). Just for the record I did: boot 1G machine in kvm, I have 2G swap

I've found that the number of CPUs makes quite a difference - I have 4.

And another difference between us may be in our configs: on this laptop I had lots of debug options on (including DEBUG_VM, DEBUG_SPINLOCK and PROVE_LOCKING, though not DEBUG_PAGEALLOC), which approximately doubles the size of each shmem_inode (and those of course are not swappable).

I found that I could avoid the OOM if I ran the "make -j20" on a kernel without all those debug options, and booted with nr_cpus=2. And currently I'm booting the kernel with the debug options in, but with nr_cpus=2, which does still OOM (whereas not if nr_cpus=1).

Maybe in the OOM rework, threads are cancelling each other's progress more destructively, where before they co-operated to some extent?

(All that is on the laptop. The G5 is still busy full-time bisecting a powerpc issue: I know it was OOMing with the rework, but I have not verified the effect of nr_cpus on it. My x86 workstation has not been OOMing with the rework - I think that means that I've not been exerting as much memory pressure on it as I'd thought, that it copes with the load better, and would only show the difference if I loaded it more heavily.)

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines

and reserve 800M for hugetlb pages (I got 445 of them). Then I extract the kernel source to tmpfs (-o size=2G), make defconfig and make -j20 (16, 10 no difference really). I was also collecting vmstat in the background. The compilation takes ages but the behavior seems consistent and stable.

Quoted 8 lines
Quoted 5 lines

Thanks a lot for giving it a go.

I'm puzzled. 445 hugetlb pages in 800M surprises me: some of them are less than 2M big?? But probably that's just a misunderstanding or typo somewhere.

A typo. 445 was from 900M test which I was doing while writing the email. Sorry about the confusion.

That makes more sense! Though I'm still amazed that you got anywhere, taking so much of the usable memory out.

Quoted 10 lines
Quoted 10 lines

Ignoring that, you're successfully doing a make -20 defconfig build in tmpfs, with only 224M of RAM available, plus 2G of swap? I'm not at all surprised that it takes ages, but I am very surprised that it does not OOM. I suppose by rights it ought not to OOM, the built tree occupies only a little more than 1G, so you do have enough swap; but I wouldn't get anywhere near that myself without OOMing - I give myself 1G of RAM (well, minus whatever the booted system takes up) to do that build in, four times your RAM, yet in my case it OOMs.

Quoted 7 lines
Quoted 2 lines

That source tree alone occupies more than 700M, so just copying it into your tmpfs would take a long time.

OK, I just found out that I was cheating a bit. I was building linux-3.7-rc5.tar.bz2 which is smaller:

$ du -sh /mnt/tmpfs/linux-3.7-rc5/
537M    /mnt/tmpfs/linux-3.7-rc5/

Right, I have a habit like that too; but my habitual testing still uses the 2.6.24 source tree, which is rather too old to ask others to reproduce with - but we both find that the kernel source tree keeps growing, and prefer to stick with something of a fixed size.

Quoted 10 lines

and after the defconfig build:

$ free
             total       used       free     shared    buffers     cached
Mem:       1008460     941904      66556          0       5092     806760
-/+ buffers/cache:     130052     878408
Swap:      2097148      42648    2054500
$ du -sh linux-3.7-rc5/
799M    linux-3.7-rc5/
Quoted 8 lines

Sorry about that but this is what my other tests were using and I forgot to check. Now let's try the same with the current linus tree:
host $ git archive v4.5-rc6 --prefix=linux-4.5-rc6/ | bzip2 > linux-4.5-rc6.tar.bz2

$ du -sh /mnt/tmpfs/linux-4.5-rc6/
707M    /mnt/tmpfs/linux-4.5-rc6/
$ free
             total       used       free     shared    buffers     cached
Mem:       1008460     962976      45484          0       7236     820064

I guess we have different versions of "free": mine shows Shmem as shared, but yours appears to be an older version, just showing 0.

Quoted 100 lines

-/+ buffers/cache: 135676 872784
Swap: 2097148 16 2097132

$ time make -j20 > /dev/null
drivers/acpi/property.c: In function ‘acpi_data_prop_read’:
drivers/acpi/property.c:745:8: warning: ‘obj’ may be used uninitialized in this function [-Wmaybe-uninitialized]
real    8m36.621s
user    14m1.642s
sys     2m45.238s

so I wasn't cheating all that much...

Quoted 4 lines

I'd expect a build in 224M
RAM plus 2G of swap to take so long, that I'd be very grateful to be OOM killed, even if there is technically enough space. Unless perhaps it's some superfast swap that you have?

the swap partition is a standard qcow image stored on my SSD disk. So I guess the IO should be quite fast. This smells like a potential contributor because my reclaim seems to be much faster and that should lead to a more efficient reclaim (in the scanned/reclaimed sense). I realize I might be boring already when blaming compaction but let me try again ;)

$ grep compact /proc/vmstat 
compact_migrate_scanned 113983
compact_free_scanned 1433503
compact_isolated 134307
compact_stall 128
compact_fail 26
compact_success 102
compact_kcompatd_wake 0

So the whole load has done the direct compaction only 128 times during that test. This doesn't sound much to me

$ grep allocstall /proc/vmstat
allocstall 1061

we entered the direct reclaim much more but most of the load will be order-0 so this might be still ok. So I've tried the following:

mm/page_alloc.c +3 -0
index 1993894b4219..107d444afdb1 100644
@@ -2910,6 +2910,9 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
+	if (order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER)
+		trace_printk("order:%d gfp_mask:%pGg compact_result:%lu\n", order, &gfp_mask, compact_result);
+
 	switch (compact_result) {
 	case COMPACT_DEFERRED:
 		*deferred_compaction = true;

And the result was:

$ cat /debug/tracing/trace_pipe | tee ~/trace.log
             gcc-8707  [001] ....   137.946370: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1
             gcc-8726  [000] ....   138.528571: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1

this shows that order-2 memory pressure is not overly high in my setup. Both attempts ended up COMPACT_SKIPPED which is interesting.

So I went back to 800M of hugetlb pages and tried again. It took ages so I have interrupted that after one hour (there was still no OOM). The trace log is quite interesting regardless:

$ wc -l ~/trace.log
371 /root/trace.log
$ grep compact_stall /proc/vmstat 
compact_stall 190

so the compaction was still ignored more than actually invoked for
!costly allocations:
sed 's@.*order:\([[:digit:]]\).* compact_result:\([[:digit:]]\)@\1 \2@' ~/trace.log | sort | uniq -c

    190 2 1
    122 2 3
     59 2 4
#define COMPACT_SKIPPED         1               
#define COMPACT_PARTIAL         3
#define COMPACT_COMPLETE        4

that means that compaction is even not tried in half cases! This doesn't sounds right to me, especially when we are talking about <= PAGE_ALLOC_COSTLY_ORDER requests which are implicitly nofail, because then we simply rely on the order-0 reclaim to automagically form higher blocks. This might indeed work when we retry many times but I guess this is not a good approach. It leads to a excessive reclaim and the stall for allocation can be really large.

One of the suspicious places is __compaction_suitable which does order-0 watermark check (increased by 2<<order). I have put another trace_printk there and it clearly pointed out this was the case.

So I have tried the following:

mm/compaction.c +0 -0
index 4d99e1f5055c..7364e48cf69a 100644

mm/compaction.c static unsigned long __compaction_suitable(struct zone *zone, int order, · line 1281

Quoted 6 lines
mm/compaction.c +3 -0
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+

I gave that a try just now, but it didn't help me: OOMed much sooner, after doing half as much work. (FWIW, I have been including your other patch, the "Andrew, could you queue this one as well, please" patch.)

I do agree that compaction appears to have closed down when we OOM:
taking that along with my nr_cpus remark (and the make -jNumber), are parallel compactions interfering with each other destructively, in a way that they did not before the rework?

Quoted 4 lines
mm/compaction.c +0 -0
 	/*
 	 * Watermarks for order-0 must be met for compaction. Note the 2UL.
 	 * This is because during migration, copies of pages need to be
Quoted 8 lines

and retried the same test (without huge pages):

$ time make -j20 > /dev/null
real    8m46.626s
user    14m15.823s
sys     2m45.471s

the time increased but I haven't checked how stable the result is.

But I didn't investigate its stability either, may have judged against it too soon.

Quoted 9 lines
$ grep compact /proc/vmstat
compact_migrate_scanned 139822
compact_free_scanned 1661642
compact_isolated 139407
compact_stall 129
compact_fail 58
compact_success 71
compact_kcompatd_wake 1

I have not seen any compact_kcompatd_wakes at all:
perhaps we're too busy compacting directly.

(Vlastimil, there's a "c" missing from that name, it should be "compact_kcompactd_wake" - though "compact_daemon_wake" might be nicer.)

Quoted 4 lines
$ grep allocstall /proc/vmstat
allocstall 1665
Quoted 8 lines

this is worse because we have scanned more pages for migration but the overall success rate was much smaller and the direct reclaim was invoked more. I do not have a good theory for that and will play with this some more. Maybe other changes are needed deeper in the compaction code.

I will play with this some more but I would be really interested to hear whether this helped Hugh with his setup. Vlastimi, Joonsoo does this even make sense to you?

It didn't help me; but I do suspect you're right to be worrying about the treatment of compaction of 0 < order <= PAGE_ALLOC_COSTLY_ORDER.

Quoted 6 lines
Quoted 4 lines

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

Ohh, I see.

I've attached vmstats.xz, output from your read_vmstat proggy; together with oom.xz, the dmesg for the OOM in question.

I hacked out_of_memory() to count_vm_event(BALLOON_DEFLATE), that being a count that's always 0 for me: so when you see "balloon_deflate 1" towards the end, that's where the OOM kill came in, and shortly after I Ctrl-C'ed.

I hope you can get more out of it than I have - thanks!

Hugh

author reviewer
#138 lore raw
Quoted 1 line · Hugh Dickins

On Tue, 1 Mar 2016, Michal Hocko wrote:

[...]

Quoted 9 lines
Quoted 9 lines

So I have tried the following:

mm/compaction.c +0 -0
index 4d99e1f5055c..7364e48cf69a 100644
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
Quoted 6 lines

mm/compaction.c static unsigned long __compaction_suitable(struct zone *zone, int order, · line 1281

Quoted 3 lines
mm/compaction.c +3 -0
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+

I gave that a try just now, but it didn't help me: OOMed much sooner, after doing half as much work.

I do not have an explanation why it would cause oom sooner but this turned out to be incomplete. There is another wmaark check deeper in the compaction path. Could you try the one from
http://lkml.kernel.org/r/[email protected]

I will try to find a machine with more CPUs and try to reproduce this in the mean time.

I will also have a look at the data you have collected.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#139 lore raw
Quoted 12 lines · Michal Hocko
Quoted 1 line · Hugh Dickins

On Tue, 1 Mar 2016, Michal Hocko wrote:

[...]

Quoted 9 lines
Quoted 9 lines

So I have tried the following:

mm/compaction.c +0 -0
index 4d99e1f5055c..7364e48cf69a 100644
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
Quoted 6 lines
Quoted 6 lines

mm/compaction.c static unsigned long __compaction_suitable(struct zone *zone, int order, · line 1281

Quoted 3 lines
mm/compaction.c +3 -0
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+

I gave that a try just now, but it didn't help me: OOMed much sooner, after doing half as much work.

I think I exaggerated: sooner, but not _much_ sooner; and I cannot see now what I based that estimate of "half as much work" on.

Quoted 5 lines

I do not have an explanation why it would cause oom sooner but this turned out to be incomplete. There is another wmaark check deeper in the compaction path. Could you try the one from
http://lkml.kernel.org/r/[email protected]

I've now added that in: it corrects the "sooner", but does not make any difference to the fact of OOMing for me.

Hugh

Quoted 8 lines

I will try to find a machine with more CPUs and try to reproduce this in the mean time.

I will also have a look at the data you have collected.

Signature · 2 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#140 lore raw
Quoted 9 lines · Hugh Dickins
Quoted 6 lines

I do not have an explanation why it would cause oom sooner but this turned out to be incomplete. There is another wmaark check deeper in the compaction path. Could you try the one from
http://lkml.kernel.org/r/[email protected]

I've now added that in: it corrects the "sooner", but does not make any difference to the fact of OOMing for me.

Could you try producing a trace with
echo 1 > /debug/tracing/events/compaction/enable echo 1 > /debug/tracing/events/migrate/mm_migrate_pages/enable

Hopefully it will hint at what's wrong with:
compact_migrate_scanned 424920
compact_free_scanned 9278408
compact_isolated 469472
compact_stall 377
compact_fail 297
compact_success 80
compact_kcompatd_wake 0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#141 lore raw
In reply to Hugh Dickins
Quoted 279 lines · Hugh Dickins
Quoted 4 lines · Michal Hocko

[Adding Vlastimil and Joonsoo for compaction related things - this was a large thread but the more interesting part starts with
http://lkml.kernel.org/r/[email protected]]

Quoted 23 lines
Quoted 22 lines · Hugh Dickins
Quoted 21 lines · Michal Hocko

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 8 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

Quoted 7 lines

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

I have tried to reproduce and failed in a virtual on my laptop. I will try with another host with more CPUs (because my laptop has only two). Just for the record I did: boot 1G machine in kvm, I have 2G swap

I've found that the number of CPUs makes quite a difference - I have 4.

And another difference between us may be in our configs: on this laptop I had lots of debug options on (including DEBUG_VM, DEBUG_SPINLOCK and PROVE_LOCKING, though not DEBUG_PAGEALLOC), which approximately doubles the size of each shmem_inode (and those of course are not swappable).

I found that I could avoid the OOM if I ran the "make -j20" on a kernel without all those debug options, and booted with nr_cpus=2. And currently I'm booting the kernel with the debug options in, but with nr_cpus=2, which does still OOM (whereas not if nr_cpus=1).

Maybe in the OOM rework, threads are cancelling each other's progress more destructively, where before they co-operated to some extent?

(All that is on the laptop. The G5 is still busy full-time bisecting a powerpc issue: I know it was OOMing with the rework, but I have not verified the effect of nr_cpus on it. My x86 workstation has not been OOMing with the rework - I think that means that I've not been exerting as much memory pressure on it as I'd thought, that it copes with the load better, and would only show the difference if I loaded it more heavily.)

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines

and reserve 800M for hugetlb pages (I got 445 of them). Then I extract the kernel source to tmpfs (-o size=2G), make defconfig and make -j20 (16, 10 no difference really). I was also collecting vmstat in the background. The compilation takes ages but the behavior seems consistent and stable.

Quoted 8 lines
Quoted 5 lines

Thanks a lot for giving it a go.

I'm puzzled. 445 hugetlb pages in 800M surprises me: some of them are less than 2M big?? But probably that's just a misunderstanding or typo somewhere.

A typo. 445 was from 900M test which I was doing while writing the email. Sorry about the confusion.

That makes more sense! Though I'm still amazed that you got anywhere, taking so much of the usable memory out.

Quoted 10 lines
Quoted 10 lines

Ignoring that, you're successfully doing a make -20 defconfig build in tmpfs, with only 224M of RAM available, plus 2G of swap? I'm not at all surprised that it takes ages, but I am very surprised that it does not OOM. I suppose by rights it ought not to OOM, the built tree occupies only a little more than 1G, so you do have enough swap; but I wouldn't get anywhere near that myself without OOMing - I give myself 1G of RAM (well, minus whatever the booted system takes up) to do that build in, four times your RAM, yet in my case it OOMs.

Quoted 7 lines
Quoted 2 lines

That source tree alone occupies more than 700M, so just copying it into your tmpfs would take a long time.

OK, I just found out that I was cheating a bit. I was building linux-3.7-rc5.tar.bz2 which is smaller:

$ du -sh /mnt/tmpfs/linux-3.7-rc5/
537M    /mnt/tmpfs/linux-3.7-rc5/

Right, I have a habit like that too; but my habitual testing still uses the 2.6.24 source tree, which is rather too old to ask others to reproduce with - but we both find that the kernel source tree keeps growing, and prefer to stick with something of a fixed size.

Quoted 10 lines

and after the defconfig build:

$ free
             total       used       free     shared    buffers     cached
Mem:       1008460     941904      66556          0       5092     806760
-/+ buffers/cache:     130052     878408
Swap:      2097148      42648    2054500
$ du -sh linux-3.7-rc5/
799M    linux-3.7-rc5/
Quoted 8 lines

Sorry about that but this is what my other tests were using and I forgot to check. Now let's try the same with the current linus tree:
host $ git archive v4.5-rc6 --prefix=linux-4.5-rc6/ | bzip2 > linux-4.5-rc6.tar.bz2

$ du -sh /mnt/tmpfs/linux-4.5-rc6/
707M    /mnt/tmpfs/linux-4.5-rc6/
$ free
             total       used       free     shared    buffers     cached
Mem:       1008460     962976      45484          0       7236     820064

I guess we have different versions of "free": mine shows Shmem as shared, but yours appears to be an older version, just showing 0.

Quoted 100 lines

-/+ buffers/cache: 135676 872784
Swap: 2097148 16 2097132

$ time make -j20 > /dev/null
drivers/acpi/property.c: In function a??acpi_data_prop_reada??:
drivers/acpi/property.c:745:8: warning: a??obja?? may be used uninitialized in this function [-Wmaybe-uninitialized]
real    8m36.621s
user    14m1.642s
sys     2m45.238s

so I wasn't cheating all that much...

Quoted 4 lines

I'd expect a build in 224M
RAM plus 2G of swap to take so long, that I'd be very grateful to be OOM killed, even if there is technically enough space. Unless perhaps it's some superfast swap that you have?

the swap partition is a standard qcow image stored on my SSD disk. So I guess the IO should be quite fast. This smells like a potential contributor because my reclaim seems to be much faster and that should lead to a more efficient reclaim (in the scanned/reclaimed sense). I realize I might be boring already when blaming compaction but let me try again ;)

$ grep compact /proc/vmstat 
compact_migrate_scanned 113983
compact_free_scanned 1433503
compact_isolated 134307
compact_stall 128
compact_fail 26
compact_success 102
compact_kcompatd_wake 0

So the whole load has done the direct compaction only 128 times during that test. This doesn't sound much to me

$ grep allocstall /proc/vmstat
allocstall 1061

we entered the direct reclaim much more but most of the load will be order-0 so this might be still ok. So I've tried the following:

mm/page_alloc.c +3 -0
index 1993894b4219..107d444afdb1 100644
@@ -2910,6 +2910,9 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 						mode, contended_compaction);
 	current->flags &= ~PF_MEMALLOC;
 
+	if (order > 0 && order <= PAGE_ALLOC_COSTLY_ORDER)
+		trace_printk("order:%d gfp_mask:%pGg compact_result:%lu\n", order, &gfp_mask, compact_result);
+
 	switch (compact_result) {
 	case COMPACT_DEFERRED:
 		*deferred_compaction = true;

And the result was:

$ cat /debug/tracing/trace_pipe | tee ~/trace.log
             gcc-8707  [001] ....   137.946370: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1
             gcc-8726  [000] ....   138.528571: __alloc_pages_direct_compact: order:2 gfp_mask:GFP_KERNEL_ACCOUNT|__GFP_NOTRACK compact_result:1

this shows that order-2 memory pressure is not overly high in my setup. Both attempts ended up COMPACT_SKIPPED which is interesting.

So I went back to 800M of hugetlb pages and tried again. It took ages so I have interrupted that after one hour (there was still no OOM). The trace log is quite interesting regardless:

$ wc -l ~/trace.log
371 /root/trace.log
$ grep compact_stall /proc/vmstat 
compact_stall 190

so the compaction was still ignored more than actually invoked for
!costly allocations:
sed 's@.*order:\([[:digit:]]\).* compact_result:\([[:digit:]]\)@\1 \2@' ~/trace.log | sort | uniq -c

    190 2 1
    122 2 3
     59 2 4
#define COMPACT_SKIPPED         1               
#define COMPACT_PARTIAL         3
#define COMPACT_COMPLETE        4

that means that compaction is even not tried in half cases! This doesn't sounds right to me, especially when we are talking about <= PAGE_ALLOC_COSTLY_ORDER requests which are implicitly nofail, because then we simply rely on the order-0 reclaim to automagically form higher blocks. This might indeed work when we retry many times but I guess this is not a good approach. It leads to a excessive reclaim and the stall for allocation can be really large.

One of the suspicious places is __compaction_suitable which does order-0 watermark check (increased by 2<<order). I have put another trace_printk there and it clearly pointed out this was the case.

So I have tried the following:

mm/compaction.c +0 -0
index 4d99e1f5055c..7364e48cf69a 100644

mm/compaction.c static unsigned long __compaction_suitable(struct zone *zone, int order, · line 1281

Quoted 6 lines
mm/compaction.c +3 -0
@@ -1276,6 +1276,9 @@ static unsigned long __compaction_suitable(struct zone *zone, int order,
 								alloc_flags))
 		return COMPACT_PARTIAL;
 
+	if (order <= PAGE_ALLOC_COSTLY_ORDER)
+		return COMPACT_CONTINUE;
+

I gave that a try just now, but it didn't help me: OOMed much sooner, after doing half as much work. (FWIW, I have been including your other patch, the "Andrew, could you queue this one as well, please" patch.)

I do agree that compaction appears to have closed down when we OOM:
taking that along with my nr_cpus remark (and the make -jNumber), are parallel compactions interfering with each other destructively, in a way that they did not before the rework?

Quoted 4 lines
mm/compaction.c +0 -0
 	/*
 	 * Watermarks for order-0 must be met for compaction. Note the 2UL.
 	 * This is because during migration, copies of pages need to be
Quoted 8 lines

and retried the same test (without huge pages):

$ time make -j20 > /dev/null
real    8m46.626s
user    14m15.823s
sys     2m45.471s

the time increased but I haven't checked how stable the result is.

But I didn't investigate its stability either, may have judged against it too soon.

Quoted 9 lines
$ grep compact /proc/vmstat
compact_migrate_scanned 139822
compact_free_scanned 1661642
compact_isolated 139407
compact_stall 129
compact_fail 58
compact_success 71
compact_kcompatd_wake 1

I have not seen any compact_kcompatd_wakes at all:
perhaps we're too busy compacting directly.

(Vlastimil, there's a "c" missing from that name, it should be "compact_kcompactd_wake" - though "compact_daemon_wake" might be nicer.)

Quoted 4 lines
$ grep allocstall /proc/vmstat
allocstall 1665
Quoted 8 lines

this is worse because we have scanned more pages for migration but the overall success rate was much smaller and the direct reclaim was invoked more. I do not have a good theory for that and will play with this some more. Maybe other changes are needed deeper in the compaction code.

I will play with this some more but I would be really interested to hear whether this helped Hugh with his setup. Vlastimi, Joonsoo does this even make sense to you?

It didn't help me; but I do suspect you're right to be worrying about the treatment of compaction of 0 < order <= PAGE_ALLOC_COSTLY_ORDER.

Quoted 1 line
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

Ohh, I see.

I've attached vmstats.xz, output from your read_vmstat proggy; together with oom.xz, the dmesg for the OOM in question.

Hello, Hugh.

I guess following things from your vmstat.
it could be wrong so please be careful. :)

Before OOM happens,

pgmigrate_success 230007
pgmigrate_fail 94
compact_migrate_scanned 422734
compact_free_scanned 9277915
compact_isolated 469308
compact_stall 370
compact_fail 291
compact_success 79
...
balloon_deflate 0

After OOM happens,

pgmigrate_success 230007
pgmigrate_fail 94
compact_migrate_scanned 424920
compact_free_scanned 9278408
compact_isolated 469472
compact_stall 377
compact_fail 297
compact_success 80
...
balloon_deflate 1

This shows that we tried compaction (compaction stall increases). Increased compact_isolated tell us that we isolated something for migration. But, pgmigrate_xxx isn't changed and it means that we didn't do any actual migration. It could happen when we can't find freepage. compact_free_scanned changed a little so it seems that there are many pageblocks with skipbit set and compaction would skip almost range in this case. This skipbit could be reset when we try more and reach the reset threshold. How about do test
with MAX_RECLAIM_RETRIES 128 or something larger to see that makes some difference?

Thanks.

Quoted 9 lines

I hacked out_of_memory() to count_vm_event(BALLOON_DEFLATE), that being a count that's always 0 for me: so when you see "balloon_deflate 1" towards the end, that's where the OOM kill came in, and shortly after I Ctrl-C'ed.

I hope you can get more out of it than I have - thanks!

Hugh

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author reviewer
#142 lore raw
In reply to Hugh Dickins
Quoted 29 lines · Hugh Dickins
Quoted 4 lines · Michal Hocko

[Adding Vlastimil and Joonsoo for compaction related things - this was a large thread but the more interesting part starts with
http://lkml.kernel.org/r/[email protected]]

Quoted 23 lines
Quoted 22 lines · Hugh Dickins
Quoted 21 lines · Michal Hocko

On Wed 24-02-16 19:47:06, Hugh Dickins wrote:
[...]

Quoted 8 lines

Boot with mem=1G (or boot your usual way, and do something to occupy most of the memory: I think /proc/sys/vm/nr_hugepages provides a great way to gobble up most of the memory, though it's not how I've done it).

Make sure you have swap: 2G is more than enough. Copy the v4.5-rc5 kernel source tree into a tmpfs: size=2G is more than enough. make defconfig there, then make -j20.

Quoted 7 lines

On a v4.5-rc5 kernel that builds fine, on mmotm it is soon OOM-killed.

Except that you'll probably need to fiddle around with that j20, it's true for my laptop but not for my workstation. j20 just happens to be what I've had there for years, that I now see breaking down (I can lower to j6 to proceed, perhaps could go a bit higher, but it still doesn't exercise swap very much).

I have tried to reproduce and failed in a virtual on my laptop. I will try with another host with more CPUs (because my laptop has only two). Just for the record I did: boot 1G machine in kvm, I have 2G swap

Quoted 6 lines

I've found that the number of CPUs makes quite a difference - I have 4.

And another difference between us may be in our configs: on this laptop I had lots of debug options on (including DEBUG_VM, DEBUG_SPINLOCK and PROVE_LOCKING, though not DEBUG_PAGEALLOC), which approximately doubles the size of each shmem_inode (and those of course are not swappable).

I had everything but PROVE_LOCKING. Enabling this option doesn't change anything (except for the overal runtime which is longer of course) in my 2 cpus setup, though.

All the following is with the clean mmotm (mmotm-2016-02-24-16-18) without any additional change. I have moved my kvm setup to a larger machine. The storage is a standard spinning rust and I've made sure that the swap is not cached on the host and the swap IO is done directly by doing
-drive file=swap-2G.qcow,if=ide,index=2,cache=none

retested with 4CPUs and make -j20

real    8m42.263s
user    20m52.838s
sys     8m8.805s

with 16CPU and make -j20

real    3m34.806s
user    20m25.245s
sys     8m39.366s

and the same with -j60 which actually triggered the OOM

$ grep "invoked oom-killer:" oomrework.qcow_serial.log
[10064.286799] cc1 invoked oom-killer: gfp_mask=0x24201ca(GFP_HIGHUSER_MOVABLE|__GFP_COLD), order=0, oom_score_adj=0
[...]
[10064.394172] DMA32 free:3764kB min:3796kB low:4776kB high:5756kB active_anon:394184kB inactive_anon:394168kB active_file:1836kB inactive_file:2156kB unevictable:0kB isolated(anon):148kB isolated(file):0kB present:1032060kB managed:987556kB mlocked:0kB dirty:0kB writeback:96kB mapped:1308kB shmem:6704kB slab_reclaimable:51356kB slab_unreclaimable:100532kB kernel_stack:7328kB pagetables:15944kB unstable:0kB bounce:0kB free_pcp:1796kB local_pcp:120kB free_cma:0kB writeback_tmp:0kB pages_scanned:63244 all_unreclaimable? yes
[...]
[10560.926971] cc1 invoked oom-killer: gfp_mask=0x24200ca(GFP_HIGHUSER_MOVABLE), order=0, oom_score_adj=0
[...]
[10561.007362] DMA32 free:4800kB min:3796kB low:4776kB high:5756kB active_anon:393112kB inactive_anon:393508kB active_file:1560kB inactive_file:1428kB unevictable:0kB isolated(anon):2452kB isolated(file):212kB present:1032060kB managed:987556kB mlocked:0kB dirty:0kB writeback:564kB mapped:2552kB shmem:7664kB slab_reclaimable:51352kB slab_unreclaimable:100396kB kernel_stack:7392kB pagetables:16196kB unstable:0kB bounce:0kB free_pcp:812kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:1172 all_unreclaimable? no

but those are simple order-0 OOMs. So this cannot be a compaction related. The second oom is probably racing with the exiting task because we are over the low wmark. This would suggest we have exhausted all the attempts with no progress.

This was all after fresh boot so then I stayed with 16CPUs and did make -j20 > /dev/null
make clean

in the loop and left it run overnight. This should randomize the swap IO and also should have a better chance of longterm fragmentation. It survived 300 iterations.

I really have no idea what might be the difference with your setup. So I've tried to test linux-next (next-20160226) just to make sure that this is not something mmotm git tree (which I maintain) specific.

Quoted 8 lines

I found that I could avoid the OOM if I ran the "make -j20" on a kernel without all those debug options, and booted with nr_cpus=2. And currently I'm booting the kernel with the debug options in, but with nr_cpus=2, which does still OOM (whereas not if nr_cpus=1).

Maybe in the OOM rework, threads are cancelling each other's progress more destructively, where before they co-operated to some extent?

Quoted 6 lines

(All that is on the laptop. The G5 is still busy full-time bisecting a powerpc issue: I know it was OOMing with the rework, but I have not verified the effect of nr_cpus on it. My x86 workstation has not been OOMing with the rework - I think that means that I've not been exerting as much memory pressure on it as I'd thought, that it copes with the load better, and would only show the difference if I loaded it more heavily.)

I am currently testing with the swap backed on sshfs (with -o direct_io) which should emulate a really slow storage. But still not OOM, I only managed to hit:

INFO: task khugepaged:246 blocked for more than 120 seconds.
int the IO path
[  480.422500]  [<ffffffff812b0c9b>] get_request+0x440/0x55e
[  480.423444]  [<ffffffff81081148>] ? wait_woken+0x72/0x72
[  480.424447]  [<ffffffff812b3071>] blk_queue_bio+0x16d/0x302
[  480.425566]  [<ffffffff812b1607>] generic_make_request+0xc0/0x15e
[  480.426642]  [<ffffffff812b17ae>] submit_bio+0x109/0x114
[  480.427704]  [<ffffffff81147101>] __swap_writepage+0x1ea/0x1f9
[  480.430364]  [<ffffffff81149346>] ? page_swapcount+0x45/0x4c
[  480.432718]  [<ffffffff815a8aed>] ? _raw_spin_unlock+0x31/0x44
[  480.433722]  [<ffffffff81149346>] ? page_swapcount+0x45/0x4c
[  480.434697]  [<ffffffff8114714a>] swap_writepage+0x3a/0x3e
[  480.435718]  [<ffffffff81122bbe>] shmem_writepage+0x37b/0x3d1
[  480.436757]  [<ffffffff8111dbe8>] shrink_page_list+0x49c/0xd88

[...]

Quoted 8 lines
Quoted 3 lines

I will play with this some more but I would be really interested to hear whether this helped Hugh with his setup. Vlastimi, Joonsoo does this even make sense to you?

It didn't help me; but I do suspect you're right to be worrying about the treatment of compaction of 0 < order <= PAGE_ALLOC_COSTLY_ORDER.

Quoted 1 line
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

I was only suggesting to allocate hugetlb pages, if you preferred not to reboot with artificially reduced RAM. Not an issue if you're booting VMs.

Ohh, I see.

I've attached vmstats.xz, output from your read_vmstat proggy; together with oom.xz, the dmesg for the OOM in question.

[  796.225322] sh invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=0
[...]
[  796.630465] Node 0 DMA32 free:13904kB min:3940kB low:4944kB high:5948kB active_anon:588776kB inactive_anon:188816kB active_file:20432kB inactive_file:6928kB unevictable:12268kB isolated(anon):128kB isolated(file):8kB present:1046128kB managed:1004892kB mlocked:12268kB dirty:16kB writeback:1400kB mapped:35556kB shmem:12684kB slab_reclaimable:55628kB slab_unreclaimable:92944kB kernel_stack:4448kB pagetables:8604kB unstable:0kB bounce:0kB free_pcp:296kB local_pcp:164kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
[  796.687390] Node 0 DMA32: 969*4kB (UE) 184*8kB (UME) 167*16kB (UM) 19*32kB (UM) 3*64kB (UM) 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 8820kB

[...]

This is really interesting because there are some order-2+ pages available. Even more striking is that free is way above high watermark. This would suggest that declaring OOM must have raced with an exiting task. This is not that unexpected because gcc are quite shortlived and `make' spawns new as soon the last one terminated. This race is not new and we cannot do much better without a moving the wmark check closer to the actual do_send_sig_info. This is not the main problem though. The thing that you are able to trigger this consistently is what bothers me.

Signature · 8 lines
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>