mm/page_alloc: Add lockdep assertion for pageblock type change
Subsystems (3)
PATCH 1/1 mm/page_alloc: Add lockdep assertion for pageblock type change
Since the migratetype hygiene patches [0], the locking here is a bit more formalised, so write it down with an assert.
[0] https://lore.kernel.org/lkml/[email protected]/T/
--- include/linux/memory_hotplug.h | 1 + mm/memory_hotplug.c | 5 +++++ mm/page_alloc.c | 4 ++++ 3 files changed, 10 insertions(+)
include/linux/memory_hotplug.h +1 -0
@@ -175,6 +175,7 @@ void put_online_mems(void);void mem_hotplug_begin (void );void mem_hotplug_done (void ); +bool in_mem_hotplug (void );/* See kswapd_is_running() */ static inline void pgdat_kswapd_lock (pg_data_t *pgdat)
mm/memory_hotplug.c +5 -0
@@ -267,6 +267,11 @@ void mem_hotplug_done(void)cpus_read_unlock (); } +bool in_mem_hotplug (void ) +{ +return percpu_is_write_locked (&mem_hotplug_lock); +} +u64 max_mem_size =U64_MAX ;/* add this memory to iomem resource */
mm/page_alloc.c +4 -0
@@ -417,6 +417,10 @@ void set_pfnblock_flags_mask(struct page *page, unsigned long flags,void set_pageblock_migratetype (struct page *page,int migratetype) { +lockdep_assert_once (system_state ==SYSTEM_BOOTING || +in_mem_hotplug () || +lockdep_is_held (&page_zone (page)->lock)); +if (unlikely (page_group_by_mobility_disabled && migratetype <MIGRATE_PCPTYPES )) migratetype =MIGRATE_UNMOVABLE ;
---
base-commit: d58172d128acbafa2295aa17cc96e28260da9a86
change-id: 20250227-pageblock-lockdep-9628c48d7e08
Best regards,
Signature · 1 line
Brendan Jackman <[email protected]>
Andrew Morton
maintainerQuoted 9 lines · On Thu, 27 Feb 2025 16:15:47 +0000 Brendan Jackman <[email protected]> wrote:
Since the migratetype hygiene patches [0], the locking here is a bit more formalised, so write it down with an assert.
...
mm/page_alloc.c void set_pfnblock_flags_mask(struct page *page, unsigned long flags, · line 426
Quoted 10 lines
mm/page_alloc.c +4
-0@@ -417,6 +417,10 @@ void set_pfnblock_flags_mask(struct page *page, unsigned long flags,void set_pageblock_migratetype (struct page *page,int migratetype) { +lockdep_assert_once (system_state ==SYSTEM_BOOTING || +in_mem_hotplug () || +lockdep_is_held (&page_zone (page)->lock)); +if (unlikely (page_group_by_mobility_disabled && migratetype <MIGRATE_PCPTYPES )) migratetype =MIGRATE_UNMOVABLE ;
We could add such assertions all over the place. Why this place in particular?
Quoted 11 lines · Andrew Morton
Quoted 9 lines · On Thu, 27 Feb 2025 16:15:47 +0000 Brendan Jackman <[email protected]> wrote:
Since the migratetype hygiene patches [0], the locking here is a bit more formalised, so write it down with an assert.
...
mm/page_alloc.c +0
-0@@ -417,6 +417,10 @@ void set_pfnblock_flags_mask(struct page *page, unsigned long flags,
Quoted 13 lines
mm/page_alloc.c void set_pfnblock_flags_mask(struct page *page, unsigned long flags, · line 426
Quoted 10 lines
mm/page_alloc.c +4
-0void set_pageblock_migratetype (struct page *page,int migratetype) { +lockdep_assert_once (system_state ==SYSTEM_BOOTING || +in_mem_hotplug () || +lockdep_is_held (&page_zone (page)->lock)); +if (unlikely (page_group_by_mobility_disabled && migratetype <MIGRATE_PCPTYPES )) migratetype =MIGRATE_UNMOVABLE ;We could add such assertions all over the place. Why this place in particular?
For the other stuff, it's pretty obvious that it would be protected by the zone lock (or, I don't know about it!). But it didn't seem totally self-evident to me that it should protect the pageblock type. So it seems particularly helpful to have it written in the code.
I may be heavily biased about this though, because of the code I'm working on for [0]. I use the pageblock type to remember whether it's mapped in the ASI restricted address space.
Vlastimil Babka
Acked-byQuoted 26 lines · Brendan Jackman
Quoted 11 lines · Andrew Morton
Quoted 9 lines · On Thu, 27 Feb 2025 16:15:47 +0000 Brendan Jackman <[email protected]> wrote:
Since the migratetype hygiene patches [0], the locking here is a bit more formalised, so write it down with an assert.
...
mm/page_alloc.c +0
-0@@ -417,6 +417,10 @@ void set_pfnblock_flags_mask(struct page *page, unsigned long flags,Quoted 13 lines
mm/page_alloc.c void set_pfnblock_flags_mask(struct page *page, unsigned long flags, · line 426
Quoted 10 lines
mm/page_alloc.c +4
-0void set_pageblock_migratetype (struct page *page,int migratetype) { +lockdep_assert_once (system_state ==SYSTEM_BOOTING || +in_mem_hotplug () || +lockdep_is_held (&page_zone (page)->lock)); +if (unlikely (page_group_by_mobility_disabled && migratetype <MIGRATE_PCPTYPES )) migratetype =MIGRATE_UNMOVABLE ;We could add such assertions all over the place. Why this place in particular?
Quoted 4 lines
For the other stuff, it's pretty obvious that it would be protected by the zone lock (or, I don't know about it!). But it didn't seem totally self-evident to me that it should protect the pageblock type. So it seems particularly helpful to have it written in the code.
I guess that's a good addition to the changelog.
Quoted 7 lines
I may be heavily biased about this though, because of the code I'm working on for [0]. I use the pageblock type to remember whether it's mapped in the ASI restricted address space.
kernel test robot
build botkernel test robot noticed the following build errors:
Hi Brendan,
kernel test robot noticed the following build errors:
[auto build test ERROR on d58172d128acbafa2295aa17cc96e28260da9a86]
url: https://github.com/intel-lab-lkp/linux/commits/Brendan-Jackman/mm-page_alloc-Add-lockdep-assertion-for-pageblock-type-change/20250228-002107 base: d58172d128acbafa2295aa17cc96e28260da9a86 patch link: https://lore.kernel.org/r/20250227-pageblock-lockdep-v1-1-3701efb331bb%40google.com patch subject: [PATCH] mm/page_alloc: Add lockdep assertion for pageblock type change config: x86_64-buildonly-randconfig-002-20250228 (https://download.01.org/0day-ci/archive/20250301/[email protected]/config) compiler: clang version 19.1.7 (https://github.com/llvm/llvm-project cd708029e0b2869e80abe31ddb175f7c35361f90) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250301/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All errors (new ones prefixed by >>):
In file included from mm/page_alloc.c:19:
In file included from include/linux/mm.h:2302:
include/linux/vmstat.h:518:36: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion]
518 | return node_stat_name(NR_LRU_BASE + lru) + 3; // skip "nr_"
| ~~~~~~~~~~~ ^ ~~~
In file included from mm/page_alloc.c:44:
include/linux/mm_inline.h:47:41: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion]
47 | __mod_lruvec_state(lruvec, NR_LRU_BASE + lru, nr_pages);
| ~~~~~~~~~~~ ^ ~~~
include/linux/mm_inline.h:49:22: warning: arithmetic between different enumeration types ('enum zone_stat_item' and 'enum lru_list') [-Wenum-enum-conversion]
49 | NR_ZONE_LRU_BASE + lru, nr_pages);
| ~~~~~~~~~~~~~~~~ ^ ~~~
>> mm/page_alloc.c:421:3: error: call to undeclared function 'in_mem_hotplug'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
421 | in_mem_hotplug() ||
| ^
mm/page_alloc.c:2857:2: warning: arithmetic between different enumeration types ('enum vm_event_item' and 'enum zone_type') [-Wenum-enum-conversion]
2857 | __count_zid_vm_events(PGALLOC, page_zonenum(page), 1 << order);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/linux/vmstat.h:139:34: note: expanded from macro '__count_zid_vm_events'
139 | __count_vm_events(item##_NORMAL - ZONE_NORMAL + zid, delta)
| ~~~~~~~~~~~~~ ^ ~~~~~~~~~~~
mm/page_alloc.c:2974:3: warning: arithmetic between different enumeration types ('enum vm_event_item' and 'enum zone_type') [-Wenum-enum-conversion]
2974 | __count_zid_vm_events(PGALLOC, page_zonenum(page), 1 << order);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/linux/vmstat.h:139:34: note: expanded from macro '__count_zid_vm_events'
139 | __count_vm_events(item##_NORMAL - ZONE_NORMAL + zid, delta)
| ~~~~~~~~~~~~~ ^ ~~~~~~~~~~~
mm/page_alloc.c:4745:2: warning: arithmetic between different enumeration types ('enum vm_event_item' and 'enum zone_type') [-Wenum-enum-conversion]
4745 | __count_zid_vm_events(PGALLOC, zone_idx(zone), nr_account);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/vmstat.h:139:34: note: expanded from macro '__count_zid_vm_events'
139 | __count_vm_events(item##_NORMAL - ZONE_NORMAL + zid, delta)
| ~~~~~~~~~~~~~ ^ ~~~~~~~~~~~
6 warnings and 1 error generated.
vim +/in_mem_hotplug +421 mm/page_alloc.c
417
418 void set_pageblock_migratetype(struct page *page, int migratetype)
419 {
420 lockdep_assert_once(system_state == SYSTEM_BOOTING ||
> 421 in_mem_hotplug() ||
422 lockdep_is_held(&page_zone(page)->lock));
423
424 if (unlikely(page_group_by_mobility_disabled &&
425 migratetype < MIGRATE_PCPTYPES))
426 migratetype = MIGRATE_UNMOVABLE;
427
428 set_pfnblock_flags_mask(page, (unsigned long)migratetype,
429 page_to_pfn(page), MIGRATETYPE_MASK);
430 }
431
Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
Johannes Weiner
reviewerQuoted 33 lines · kernel test robot
Hi Brendan,
kernel test robot noticed the following build errors:
[auto build test ERROR on
d58172d128acbafa2295aa17cc96e28260da9a86]url: https://github.com/intel-lab-lkp/linux/commits/Brendan-Jackman/mm-page_alloc-Add-lockdep-assertion-for-pageblock-type-change/20250228-002107 base: d58172d128acbafa2295aa17cc96e28260da9a86 patch link: https://lore.kernel.org/r/20250227-pageblock-lockdep-v1-1-3701efb331bb%40google.com patch subject: [PATCH] mm/page_alloc: Add lockdep assertion for pageblock type change config: x86_64-buildonly-randconfig-002-20250228 (https://download.01.org/0day-ci/archive/20250301/[email protected]/config) compiler: clang version 19.1.7 (https://github.com/llvm/llvm-project cd708029e0b2869e80abe31ddb175f7c35361f90) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250301/[email protected]/reproduce)If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/All errors (new ones prefixed by >>):
In file included from
mm/page_alloc.c:19:In file included from include/linux/mm.h:2302: include/linux/vmstat.h:518:36: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion] 518 | return node_stat_name(NR_LRU_BASE + lru) + 3; // skip "nr_" | ~~~~~~~~~~~ ^ ~~~ In file included from mm/page_alloc.c:44: include/linux/mm_inline.h:47:41: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion] 47 | __mod_lruvec_state(lruvec, NR_LRU_BASE + lru, nr_pages); | ~~~~~~~~~~~ ^ ~~~ include/linux/mm_inline.h:49:22: warning: arithmetic between different enumeration types ('enum zone_stat_item' and 'enum lru_list') [-Wenum-enum-conversion] 49 | NR_ZONE_LRU_BASE + lru, nr_pages); | ~~~~~~~~~~~~~~~~ ^ ~~~
Quoted 2 lines
Quoted 1 line
Quoted 1 line
mm/page_alloc.c:421:3: error: call to undeclared function 'in_mem_hotplug'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]421 | in_mem_hotplug() ||
The patch is missing a dummy in_mem_hotplug() in the !CONFIG_MEMORY_HOTPLUG section of <linux/memory_hotplug.h>.
Quoted 4 lines · Johannes Weiner
On Sat, Mar 01, 2025 at 01:31:30AM +0800, kernel test robot wrote:
The patch is missing a dummy
in_mem_hotplug()in the !CONFIG_MEMORY_HOTPLUG section of <linux/memory_hotplug.h>.
+1, I just stumbled over and this is not fixed in today's Linux Next. I'm wondering how this was missed during merge into Linux Next. Stephen?
Signature · 2 lines
With Best Regards, Andy Shevchenko
Hi Andy,
Quoted 9 lines · On Mon, 3 Mar 2025 13:18:42 +0200 Andy Shevchenko…
Quoted 4 lines · Johannes Weiner
On Sat, Mar 01, 2025 at 01:31:30AM +0800, kernel test robot wrote:
The patch is missing a dummy
in_mem_hotplug()in the !CONFIG_MEMORY_HOTPLUG section of <linux/memory_hotplug.h>.+1, I just stumbled over and this is not fixed in today's Linux Next. I'm wondering how this was missed during merge into Linux Next. Stephen?
I just get what people put in their trees. There are no conflicts around this and none of my builds failed, so I didn't see the problem. Has someone sent a fix patch to Andrew? If so, if you forward it to me, I will add it to linux-next today.
Signature · 2 lines
Cheers, Stephen Rothwell
Quoted 10 lines · Stephen Rothwell
Hi Andy,
Quoted 7 lines · On Mon, 3 Mar 2025 13:18:42 +0200 Andy Shevchenko…
Quoted 4 lines · Johannes Weiner
On Sat, Mar 01, 2025 at 01:31:30AM +0800, kernel test robot wrote:
The patch is missing a dummy
in_mem_hotplug()in the !CONFIG_MEMORY_HOTPLUG section of <linux/memory_hotplug.h>.
Quoted 7 lines
Quoted 2 lines
+1, I just stumbled over and this is not fixed in today's Linux Next. I'm wondering how this was missed during merge into Linux Next. Stephen?
I just get what people put in their trees. There are no conflicts around this and none of my builds failed, so I didn't see the problem. Has someone sent a fix patch to Andrew? If so, if you forward it to me, I will add it to linux-next today.
Andrew has backed it out of mm-unstable now. There's a v2 [0] which still has runtime issues but AFAIK it's not in any tree yet.
[0] https://lore.kernel.org/linux-mm/[email protected]/
In case it helps calibrate expectations: I think this particular patch had been reviewed but in general some patches get into mm-unstable without any review being recorded at all. My understanding is that Andrew squints at it and goes "that looks like it will probably eventually get merged" and puts it in so that people get a view of likely upcoming changes.
So if an issue like this reaching linux-next is a big problem then I think the solution is not to merge mm-unstable. I'm not sure how high the bar is supposed to be for feeding into linux-next.
Quoted 31 lines · Brendan Jackman
Quoted 10 lines · Stephen Rothwell
Hi Andy,
Quoted 7 lines · On Mon, 3 Mar 2025 13:18:42 +0200 Andy Shevchenko…
Quoted 4 lines · Johannes Weiner
On Sat, Mar 01, 2025 at 01:31:30AM +0800, kernel test robot wrote:
The patch is missing a dummy
in_mem_hotplug()in the !CONFIG_MEMORY_HOTPLUG section of <linux/memory_hotplug.h>.Quoted 7 lines
Quoted 2 lines
+1, I just stumbled over and this is not fixed in today's Linux Next. I'm wondering how this was missed during merge into Linux Next. Stephen?
I just get what people put in their trees. There are no conflicts around this and none of my builds failed, so I didn't see the problem. Has someone sent a fix patch to Andrew? If so, if you forward it to me, I will add it to linux-next today.
Andrew has backed it out of mm-unstable now. There's a v2 [0] which still has runtime issues but AFAIK it's not in any tree yet.
[0] https://lore.kernel.org/linux-mm/[email protected]/
In case it helps calibrate expectations: I think this particular patch had been reviewed but in general some patches get into mm-unstable without any review being recorded at all. My understanding is that Andrew squints at it and goes "that looks like it will probably eventually get merged" and puts it in so that people get a view of likely upcoming changes.
Quoted 3 lines
So if an issue like this reaching linux-next is a big problem then I think the solution is not to merge mm-unstable. I'm not sure how high the bar is supposed to be for feeding into linux-next.
Last year at LSF/MM I raised that maybe we should have something in-between mm-unstable and mm-stable that would get merged into -next. ( "mm-almost-stable" / "mm-for-next" ;) )
Alternatively, we could add something like "mm-staging" before "mm-unstable" and "mm-stable", whereby only "mm-unstable" would get merged into -next.
Ideally, we'd have at least build-bots and some basic sanity checks going on on such a staging environment. After a couple of days (not more) of survival in such an environment, it could be moved to the tree that gets exposed to -next/
I guess the downside is that it's "yet another branch" and "yet more build+testing" effort; in particular, if build+testing doesn't happen it will be worthless. And likely, a lot of build+testing is happening on linux-next only.
I have pretty good results with build bots going crazy on my branches, so most stuff that I send (after letting is rest for a couple of days on a public branch) doesn't really result in build issues.
Signature · 3 lines
Cheers, David / dhildenb
kernel test robot
build botkernel test robot noticed the following build errors:
Hi Brendan,
kernel test robot noticed the following build errors:
[auto build test ERROR on d58172d128acbafa2295aa17cc96e28260da9a86]
url: https://github.com/intel-lab-lkp/linux/commits/Brendan-Jackman/mm-page_alloc-Add-lockdep-assertion-for-pageblock-type-change/20250228-002107 base: d58172d128acbafa2295aa17cc96e28260da9a86 patch link: https://lore.kernel.org/r/20250227-pageblock-lockdep-v1-1-3701efb331bb%40google.com patch subject: [PATCH] mm/page_alloc: Add lockdep assertion for pageblock type change config: i386-buildonly-randconfig-005-20250301 (https://download.01.org/0day-ci/archive/20250301/[email protected]/config) compiler: gcc-12 (Debian 12.2.0-14) 12.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250301/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All errors (new ones prefixed by >>):
In file included from arch/x86/include/asm/bug.h:99,
from include/linux/bug.h:5,
from include/linux/mmdebug.h:5,
from include/linux/mm.h:6,
from mm/page_alloc.c:19:
mm/page_alloc.c: In function 'set_pageblock_migratetype':
>> mm/page_alloc.c:421:17: error: implicit declaration of function 'in_mem_hotplug' [-Werror=implicit-function-declaration]
421 | in_mem_hotplug() ||
| ^~~~~~~~~~~~~~
include/asm-generic/bug.h:171:32: note: in definition of macro 'WARN_ON'
171 | int __ret_warn_on = !!(condition); \
| ^~~~~~~~~
include/linux/lockdep.h:282:14: note: in expansion of macro 'WARN_ON_ONCE'
282 | do { WARN_ON_ONCE(debug_locks && !(cond)); } while (0)
| ^~~~~~~~~~~~
mm/page_alloc.c:420:9: note: in expansion of macro 'lockdep_assert_once'
420 | lockdep_assert_once(system_state == SYSTEM_BOOTING ||
| ^~~~~~~~~~~~~~~~~~~cc1: some warnings being treated as errors
vim +/in_mem_hotplug +421 mm/page_alloc.c
417
418 void set_pageblock_migratetype(struct page *page, int migratetype)
419 {
420 lockdep_assert_once(system_state == SYSTEM_BOOTING ||
> 421 in_mem_hotplug() ||
422 lockdep_is_held(&page_zone(page)->lock));
423
424 if (unlikely(page_group_by_mobility_disabled &&
425 migratetype < MIGRATE_PCPTYPES))
426 migratetype = MIGRATE_UNMOVABLE;
427
428 set_pfnblock_flags_mask(page, (unsigned long)migratetype,
429 page_to_pfn(page), MIGRATETYPE_MASK);
430 }
431
Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki