swap_cluster_info lockdep splat

Hi Huang,

With changing from bit lock to spinlock of swap_cluster_info, my zram test failed with below message. It seems nested lock problem so need to play with lockdep.

Thanks.

=============================================
[ INFO: possible recursive locking detected ]
4.10.0-rc8-next-20170214-zram #24 Not tainted
---------------------------------------------
as/6557 is trying to acquire lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811ddd03>] cluster_list_add_tail.part.31+0x33/0x70

but task is already holding lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0
       ----

lock(&(&((cluster_info + ci)->lock))->rlock);
lock(&(&((cluster_info + ci)->lock))->rlock);

*** DEADLOCK ***

May be due to missing lock nesting notation

3 locks held by as/6557:

 #0:  (&(&cache->free_lock)->rlock){......}, at: [<ffffffff811c206b>] free_swap_slot+0x8b/0x110
 #1:  (&(&p->lock)->rlock){+.+.-.}, at: [<ffffffff811df295>] swapcache_free_entries+0x75/0x330
 #2:  (&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

stack backtrace:
CPU: 3 PID: 6557 Comm: as Not tainted 4.10.0-rc8-next-20170214-zram #24
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
Call Trace:
 dump_stack+0x85/0xc2
 __lock_acquire+0x15ea/0x1640
 lock_acquire+0x100/0x1f0
 ? cluster_list_add_tail.part.31+0x33/0x70
 _raw_spin_lock+0x38/0x50
 ? cluster_list_add_tail.part.31+0x33/0x70
 cluster_list_add_tail.part.31+0x33/0x70
 swapcache_free_entries+0x2f9/0x330
 free_swap_slot+0xf8/0x110
 swapcache_free+0x36/0x40
 delete_from_swap_cache+0x5f/0xa0
 try_to_free_swap+0x6e/0xa0
 free_pages_and_swap_cache+0x7d/0xb0
 tlb_flush_mmu_free+0x36/0x60
 tlb_finish_mmu+0x1c/0x50
 exit_mmap+0xc7/0x150
 mmput+0x51/0x110
 do_exit+0x2b2/0xc30
 ? trace_hardirqs_on_caller+0x129/0x1b0
 do_group_exit+0x50/0xd0
 SyS_exit_group+0x14/0x20
 entry_SYSCALL_64_fastpath+0x23/0xc6
RIP: 0033:0x2b9a2dbdf309
RSP: 002b:00007ffe71887528 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00002b9a2dbdf309
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 00002b9a2ded8858 R08: 000000000000003c R09: 00000000000000e7
R10: ffffffffffffff60 R11: 0000000000000246 R12: 00002b9a2ded8858
R13: 00002b9a2dedde80 R14: 000000000255f770 R15: 0000000000000001

Hi Huang,

With changing from bit lock to spinlock of swap_cluster_info, my zram test failed with below message. It seems nested lock problem so need to play with lockdep.

Hi, Minchan,

Quoted 5 lines · Minchan Kim

Hi Huang,

With changing from bit lock to spinlock of swap_cluster_info, my zram test failed with below message. It seems nested lock problem so need to play with lockdep.

Sorry, I could not reproduce the warning in my tests. Could you try the
patches as below? And could you share your test case?

Best Regards,
Huang, Ying

-------------------------------------------------------------> From 2b9e2f78a6e389442f308c4f9e8d5ac40fe6aa2f Mon Sep 17 00:00:00 2001
From: Huang Ying <[email protected]>
Date: Thu, 16 Feb 2017 16:38:17 +0800
Subject: [PATCH] mm, swap: Annotate nested locking for cluster lock

There is a nested locking in cluster_list_add_tail() for cluster lock, which caused lockdep to complain as below. The nested locking is safe because both cluster locks are only acquired when we held the swap_info_struct->lock. Annotated the nested locking via spin_lock_nested() to fix the complain of lockdep.

=============================================
[ INFO: possible recursive locking detected ]
4.10.0-rc8-next-20170214-zram #24 Not tainted
---------------------------------------------
as/6557 is trying to acquire lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811ddd03>] cluster_list_add_tail.part.31+0x33/0x70

but task is already holding lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0
       ----

lock(&(&((cluster_info + ci)->lock))->rlock);
lock(&(&((cluster_info + ci)->lock))->rlock);

*** DEADLOCK ***

May be due to missing lock nesting notation

3 locks held by as/6557:

 #0:  (&(&cache->free_lock)->rlock){......}, at: [<ffffffff811c206b>] free_swap_slot+0x8b/0x110
 #1:  (&(&p->lock)->rlock){+.+.-.}, at: [<ffffffff811df295>] swapcache_free_entries+0x75/0x330
 #2:  (&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

stack backtrace:
CPU: 3 PID: 6557 Comm: as Not tainted 4.10.0-rc8-next-20170214-zram #24
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
Call Trace:
 dump_stack+0x85/0xc2
 __lock_acquire+0x15ea/0x1640
 lock_acquire+0x100/0x1f0
 ? cluster_list_add_tail.part.31+0x33/0x70
 _raw_spin_lock+0x38/0x50
 ? cluster_list_add_tail.part.31+0x33/0x70
 cluster_list_add_tail.part.31+0x33/0x70
 swapcache_free_entries+0x2f9/0x330
 free_swap_slot+0xf8/0x110
 swapcache_free+0x36/0x40
 delete_from_swap_cache+0x5f/0xa0
 try_to_free_swap+0x6e/0xa0
 free_pages_and_swap_cache+0x7d/0xb0
 tlb_flush_mmu_free+0x36/0x60
 tlb_finish_mmu+0x1c/0x50
 exit_mmap+0xc7/0x150
 mmput+0x51/0x110
 do_exit+0x2b2/0xc30
 ? trace_hardirqs_on_caller+0x129/0x1b0
 do_group_exit+0x50/0xd0
 SyS_exit_group+0x14/0x20
 entry_SYSCALL_64_fastpath+0x23/0xc6
RIP: 0033:0x2b9a2dbdf309
RSP: 002b:00007ffe71887528 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00002b9a2dbdf309
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 00002b9a2ded8858 R08: 000000000000003c R09: 00000000000000e7
R10: ffffffffffffff60 R11: 0000000000000246 R12: 00002b9a2ded8858
R13: 00002b9a2dedde80 R14: 000000000255f770 R15: 0000000000000001

Reported-by Minchan Kim <[email protected]>

Signed-off-by "Huang, Ying" <[email protected]>

---
 include/linux/swap.h | 6 ++++++
 mm/swapfile.c        | 8 +++++++-
 2 files changed, 13 insertions(+), 1 deletion(-)

include/linux/swap.h +6 -0

index 4d12b381821f..ef044ea8fe79 100644
@@ -166,6 +166,12 @@ enum {
 #define COUNT_CONTINUED	0x80	/* See swap_map continuation for full count */
 #define SWAP_MAP_SHMEM	0xbf	/* Owned by shmem/tmpfs, in first swap_map */
 
+enum swap_cluster_lock_class
+{
+	SWAP_CLUSTER_LOCK_NORMAL,  /* implicitly used by plain spin_lock() APIs. */
+	SWAP_CLUSTER_LOCK_NESTED,
+};
+
 /*
  * We use this to track usage of a cluster. A cluster is a block of swap disk
  * space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All

mm/swapfile.c +7 -1

index 5ac2cb40dbd3..0a52e9b2f843 100644
@@ -263,6 +263,12 @@ static inline void __lock_cluster(struct swap_cluster_info *ci)
 	spin_lock(&ci->lock);
 }
 
+static inline void __lock_cluster_nested(struct swap_cluster_info *ci,
+					 unsigned subclass)
+{
+	spin_lock_nested(&ci->lock, subclass);
+}
+
 static inline struct swap_cluster_info *lock_cluster(struct swap_info_struct *si,
 						     unsigned long offset)
 {

@@ -336,7 +342,7 @@ static void cluster_list_add_tail(struct swap_cluster_list *list,
 		 * only acquired when we held swap_info_struct->lock
 		 */
 		ci_tail = ci + tail;
-		__lock_cluster(ci_tail);
+		__lock_cluster_nested(ci_tail, SWAP_CLUSTER_LOCK_NESTED);
 		cluster_set_next(ci_tail, idx);
 		unlock_cluster(ci_tail);
 		cluster_set_next_flag(&list->tail, idx, 0);

Signature · 1 line

2.11.0

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be.A Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

Quoted 6 lines

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be.  Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

Hugh, trying to duplicate your test case.  So you were doing swapping, then swap off, swap on the swap device and restart swapping?

Quoted 9 lines · Tim Chen

Quoted 6 lines

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be.  Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

Hugh, trying to duplicate your test case.  So you were doing swapping, then swap off, swap on the swap device and restart swapping?

Repeated pair of make -j20 kernel builds in 700M RAM, 1.5G swap on SSD, 8 cpus; one of the builds in tmpfs, other in ext4 on loop on tmpfs file; sizes tuned for plenty of swapping but no OOMing (it's an ancient 2.6.24 kernel I build, modern one needing a lot more space with a lot less in use).

How much of that is relevant I don't know: hopefully none of it, it's hard to get the tunings right from scratch. To answer your specific
question: yes, I'm not doing concurrent swapoffs in this test showing the leak, just waiting for each of the pair of builds to complete, then tearing down the trees, doing swapoff followed by swapon, and starting a new pair of builds.

Sometimes it's the swapoff that fails with ENOMEM, more often it's a fork during build that fails with ENOMEM: after 6 or 7 hours of load (but timings show it getting slower leading up to that). /proc/meminfo did not give me an immediate clue, Slab didn't look surprising but I may not have studied close enough.

I quilt-bisected it as far as the mm-swap series, good before, bad after, but didn't manage to narrow it down further because of hitting a presumably different issue inside the series, where swapoff ENOMEMed much sooner (after 25 mins one time, during first iteration the next).

Hugh

Hi, Hugh,

Quoted 23 lines · Hugh Dickins

Quoted 9 lines · Tim Chen

Quoted 6 lines

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be.A Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

Hugh, trying to duplicate your test case. A So you were doing swapping, then swap off, swap on the swap device and restart swapping?

Repeated pair of make -j20 kernel builds in 700M RAM, 1.5G swap on SSD, 8 cpus; one of the builds in tmpfs, other in ext4 on loop on tmpfs file; sizes tuned for plenty of swapping but no OOMing (it's an ancient 2.6.24 kernel I build, modern one needing a lot more space with a lot less in use).

How much of that is relevant I don't know: hopefully none of it, it's hard to get the tunings right from scratch. To answer your specific
question: yes, I'm not doing concurrent swapoffs in this test showing the leak, just waiting for each of the pair of builds to complete, then tearing down the trees, doing swapoff followed by swapon, and starting a new pair of builds.

Quoted 5 lines

Sometimes it's the swapoff that fails with ENOMEM, more often it's a fork during build that fails with ENOMEM: after 6 or 7 hours of load (but timings show it getting slower leading up to that). /proc/meminfo did not give me an immediate clue, Slab didn't look surprising but I may not have studied close enough.

Thanks for you information!

Memory newly allocated in the mm-swap series are allocated via vmalloc, could you find anything special for vmalloc in /proc/meminfo?

Quoted 9 lines · Tim Chen

Quoted 6 lines

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be.A Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

Hugh, trying to duplicate your test case. A So you were doing swapping, then swap off, swap on the swap device and restart swapping?

Repeated pair of make -j20 kernel builds in 700M RAM, 1.5G swap on SSD, 8 cpus; one of the builds in tmpfs, other in ext4 on loop on tmpfs file; sizes tuned for plenty of swapping but no OOMing (it's an ancient 2.6.24 kernel I build, modern one needing a lot more space with a lot less in use).

How much of that is relevant I don't know: hopefully none of it, it's hard to get the tunings right from scratch. To answer your specific
question: yes, I'm not doing concurrent swapoffs in this test showing the leak, just waiting for each of the pair of builds to complete, then tearing down the trees, doing swapoff followed by swapon, and starting a new pair of builds.

Sometimes it's the swapoff that fails with ENOMEM, more often it's a fork during build that fails with ENOMEM: after 6 or 7 hours of load (but timings show it getting slower leading up to that). /proc/meminfo did not give me an immediate clue, Slab didn't look surprising but I may not have studied close enough.

I quilt-bisected it as far as the mm-swap series, good before, bad after, but didn't manage to narrow it down further because of hitting a presumably different issue inside the series, where swapoff ENOMEMed much sooner (after 25 mins one time, during first iteration the next).

I found a memory leak in __read_swap_cache_async() introduced by mm-swap series, and confirmed it via testing. Could you verify whether it fixed your cases? Thanks a lot for reporting.

Best Regards,
Huang, Ying

-------------------------------------------------------------------------> From 4b96423796ab7435104eb2cb4dcf5d525b9e0800 Mon Sep 17 00:00:00 2001
From: Huang Ying <[email protected]>
Date: Fri, 17 Feb 2017 10:31:37 +0800
Subject: [PATCH] mm, swap: Fix memory leak in __read_swap_cache_async()

The memory may be leaked in __read_swap_cache_async(). For the cases as below,

CPU 0 CPU 1
----- -----

find_get_page() == NULL
__swp_swapcount() != 0
new_page = alloc_page_vma()
radix_tree_maybe_preload()
swap in swap slot
swapcache_prepare() == -EEXIST
cond_resched()
reclaim the swap slot
find_get_page() == NULL
__swp_swapcount() == 0
return NULL <- new_page leaked here !!!

The memory leak has been confirmed via checking the value of new_page when returning inside the loop in __read_swap_cache_async().

This is fixed via replacing return with break inside of loop in __read_swap_cache_async(), so that there is opportunity for the new_page to be checked and freed.

Reported-by Hugh Dickins <[email protected]>

Cc Tim Chen <[email protected]>

Signed-off-by "Huang, Ying" <[email protected]>

---
 mm/swap_state.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

mm/swap_state.c +1 -1

index 2126e9ba23b2..473b71e052a8 100644
@@ -333,7 +333,7 @@ struct page *__read_swap_cache_async(swp_entry_t entry, gfp_t gfp_mask,
 		 * else swap_off will be aborted if we return NULL.
 		 */
 		if (!__swp_swapcount(entry) && swap_slot_cache_enabled)
-			return NULL;
+			break;
 
 		/*
 		 * Get a new page to read into from swap.

Signature · 1 line

2.11.0

Quoted 4 lines · Ying

Hi, Minchan,

Minchan Kim <[email protected]> writes:

Quoted 8 lines

Quoted 5 lines

Hi Huang,

With changing from bit lock to spinlock of swap_cluster_info, my zram test failed with below message. It seems nested lock problem so need to play with lockdep.

Sorry, I could not reproduce the warning in my tests. Could you try the
patches as below? And could you share your test case?

Quoted 129 lines

Best Regards,
Huang, Ying

-------------------------------------------------------------> From 2b9e2f78a6e389442f308c4f9e8d5ac40fe6aa2f Mon Sep 17 00:00:00 2001
From: Huang Ying <[email protected]>
Date: Thu, 16 Feb 2017 16:38:17 +0800
Subject: [PATCH] mm, swap: Annotate nested locking for cluster lock

There is a nested locking in cluster_list_add_tail() for cluster lock, which caused lockdep to complain as below. The nested locking is safe because both cluster locks are only acquired when we held the swap_info_struct->lock. Annotated the nested locking via spin_lock_nested() to fix the complain of lockdep.

=============================================
[ INFO: possible recursive locking detected ]
4.10.0-rc8-next-20170214-zram #24 Not tainted
---------------------------------------------
as/6557 is trying to acquire lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811ddd03>] cluster_list_add_tail.part.31+0x33/0x70

but task is already holding lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0
       ----

lock(&(&((cluster_info + ci)->lock))->rlock);
lock(&(&((cluster_info + ci)->lock))->rlock);

*** DEADLOCK ***

May be due to missing lock nesting notation

3 locks held by as/6557:

 #0:  (&(&cache->free_lock)->rlock){......}, at: [<ffffffff811c206b>] free_swap_slot+0x8b/0x110
 #1:  (&(&p->lock)->rlock){+.+.-.}, at: [<ffffffff811df295>] swapcache_free_entries+0x75/0x330
 #2:  (&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

stack backtrace:
CPU: 3 PID: 6557 Comm: as Not tainted 4.10.0-rc8-next-20170214-zram #24
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
Call Trace:
 dump_stack+0x85/0xc2
 __lock_acquire+0x15ea/0x1640
 lock_acquire+0x100/0x1f0
 ? cluster_list_add_tail.part.31+0x33/0x70
 _raw_spin_lock+0x38/0x50
 ? cluster_list_add_tail.part.31+0x33/0x70
 cluster_list_add_tail.part.31+0x33/0x70
 swapcache_free_entries+0x2f9/0x330
 free_swap_slot+0xf8/0x110
 swapcache_free+0x36/0x40
 delete_from_swap_cache+0x5f/0xa0
 try_to_free_swap+0x6e/0xa0
 free_pages_and_swap_cache+0x7d/0xb0
 tlb_flush_mmu_free+0x36/0x60
 tlb_finish_mmu+0x1c/0x50
 exit_mmap+0xc7/0x150
 mmput+0x51/0x110
 do_exit+0x2b2/0xc30
 ? trace_hardirqs_on_caller+0x129/0x1b0
 do_group_exit+0x50/0xd0
 SyS_exit_group+0x14/0x20
 entry_SYSCALL_64_fastpath+0x23/0xc6
RIP: 0033:0x2b9a2dbdf309
RSP: 002b:00007ffe71887528 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00002b9a2dbdf309
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 00002b9a2ded8858 R08: 000000000000003c R09: 00000000000000e7
R10: ffffffffffffff60 R11: 0000000000000246 R12: 00002b9a2ded8858
R13: 00002b9a2dedde80 R14: 000000000255f770 R15: 0000000000000001

Reported-by Minchan Kim <[email protected]>

Signed-off-by "Huang, Ying" <[email protected]>

---
 include/linux/swap.h | 6 ++++++
 mm/swapfile.c        | 8 +++++++-
 2 files changed, 13 insertions(+), 1 deletion(-)

include/linux/swap.h +6 -0

index 4d12b381821f..ef044ea8fe79 100644
@@ -166,6 +166,12 @@ enum {
 #define COUNT_CONTINUED	0x80	/* See swap_map continuation for full count */
 #define SWAP_MAP_SHMEM	0xbf	/* Owned by shmem/tmpfs, in first swap_map */
 
+enum swap_cluster_lock_class
+{
+	SWAP_CLUSTER_LOCK_NORMAL,  /* implicitly used by plain spin_lock() APIs. */
+	SWAP_CLUSTER_LOCK_NESTED,
+};
+
 /*
  * We use this to track usage of a cluster. A cluster is a block of swap disk
  * space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All

mm/swapfile.c +7 -1

index 5ac2cb40dbd3..0a52e9b2f843 100644
@@ -263,6 +263,12 @@ static inline void __lock_cluster(struct swap_cluster_info *ci)
 	spin_lock(&ci->lock);
 }
 
+static inline void __lock_cluster_nested(struct swap_cluster_info *ci,
+					 unsigned subclass)
+{
+	spin_lock_nested(&ci->lock, subclass);
+}
+
 static inline struct swap_cluster_info *lock_cluster(struct swap_info_struct *si,
 						     unsigned long offset)
 {

@@ -336,7 +342,7 @@ static void cluster_list_add_tail(struct swap_cluster_list *list,
 		 * only acquired when we held swap_info_struct->lock
 		 */
 		ci_tail = ci + tail;
-		__lock_cluster(ci_tail);
+		__lock_cluster_nested(ci_tail, SWAP_CLUSTER_LOCK_NESTED);
 		cluster_set_next(ci_tail, idx);
 		unlock_cluster(ci_tail);
 		cluster_set_next_flag(&list->tail, idx, 0);

Signature · 1 line

2.11.0

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be. Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

[PATCH] mm, swap: Annotate nested locking for cluster lock

Fix swap cluster lockdep warnings.

Reported-by Minchan Kim <[email protected]>

Signed-off-by Hugh Dickins <[email protected]>

Quoted 138 lines · Ying

Hi, Minchan,

Quoted 5 lines · Minchan Kim

Hi Huang,

With changing from bit lock to spinlock of swap_cluster_info, my zram test failed with below message. It seems nested lock problem so need to play with lockdep.

Sorry, I could not reproduce the warning in my tests. Could you try the
patches as below? And could you share your test case?

Best Regards,
Huang, Ying

-------------------------------------------------------------> From 2b9e2f78a6e389442f308c4f9e8d5ac40fe6aa2f Mon Sep 17 00:00:00 2001
From: Huang Ying <[email protected]>
Date: Thu, 16 Feb 2017 16:38:17 +0800
Subject: [PATCH] mm, swap: Annotate nested locking for cluster lock

There is a nested locking in cluster_list_add_tail() for cluster lock, which caused lockdep to complain as below. The nested locking is safe because both cluster locks are only acquired when we held the swap_info_struct->lock. Annotated the nested locking via spin_lock_nested() to fix the complain of lockdep.

=============================================
[ INFO: possible recursive locking detected ]
4.10.0-rc8-next-20170214-zram #24 Not tainted
---------------------------------------------
as/6557 is trying to acquire lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811ddd03>] cluster_list_add_tail.part.31+0x33/0x70

but task is already holding lock:
(&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0
       ----

lock(&(&((cluster_info + ci)->lock))->rlock);
lock(&(&((cluster_info + ci)->lock))->rlock);

*** DEADLOCK ***

May be due to missing lock nesting notation

3 locks held by as/6557:

 #0:  (&(&cache->free_lock)->rlock){......}, at: [<ffffffff811c206b>] free_swap_slot+0x8b/0x110
 #1:  (&(&p->lock)->rlock){+.+.-.}, at: [<ffffffff811df295>] swapcache_free_entries+0x75/0x330
 #2:  (&(&((cluster_info + ci)->lock))->rlock){+.+.-.}, at: [<ffffffff811df2bb>] swapcache_free_entries+0x9b/0x330

stack backtrace:
CPU: 3 PID: 6557 Comm: as Not tainted 4.10.0-rc8-next-20170214-zram #24
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
Call Trace:
 dump_stack+0x85/0xc2
 __lock_acquire+0x15ea/0x1640
 lock_acquire+0x100/0x1f0
 ? cluster_list_add_tail.part.31+0x33/0x70
 _raw_spin_lock+0x38/0x50
 ? cluster_list_add_tail.part.31+0x33/0x70
 cluster_list_add_tail.part.31+0x33/0x70
 swapcache_free_entries+0x2f9/0x330
 free_swap_slot+0xf8/0x110
 swapcache_free+0x36/0x40
 delete_from_swap_cache+0x5f/0xa0
 try_to_free_swap+0x6e/0xa0
 free_pages_and_swap_cache+0x7d/0xb0
 tlb_flush_mmu_free+0x36/0x60
 tlb_finish_mmu+0x1c/0x50
 exit_mmap+0xc7/0x150
 mmput+0x51/0x110
 do_exit+0x2b2/0xc30
 ? trace_hardirqs_on_caller+0x129/0x1b0
 do_group_exit+0x50/0xd0
 SyS_exit_group+0x14/0x20
 entry_SYSCALL_64_fastpath+0x23/0xc6
RIP: 0033:0x2b9a2dbdf309
RSP: 002b:00007ffe71887528 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00002b9a2dbdf309
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 00002b9a2ded8858 R08: 000000000000003c R09: 00000000000000e7
R10: ffffffffffffff60 R11: 0000000000000246 R12: 00002b9a2ded8858
R13: 00002b9a2dedde80 R14: 000000000255f770 R15: 0000000000000001

Reported-by Minchan Kim <[email protected]>

Signed-off-by "Huang, Ying" <[email protected]>

---
 include/linux/swap.h | 6 ++++++
 mm/swapfile.c        | 8 +++++++-
 2 files changed, 13 insertions(+), 1 deletion(-)

include/linux/swap.h +6 -0

index 4d12b381821f..ef044ea8fe79 100644
@@ -166,6 +166,12 @@ enum {
 #define COUNT_CONTINUED	0x80	/* See swap_map continuation for full count */
 #define SWAP_MAP_SHMEM	0xbf	/* Owned by shmem/tmpfs, in first swap_map */
 
+enum swap_cluster_lock_class
+{
+	SWAP_CLUSTER_LOCK_NORMAL,  /* implicitly used by plain spin_lock() APIs. */
+	SWAP_CLUSTER_LOCK_NESTED,
+};
+
 /*
  * We use this to track usage of a cluster. A cluster is a block of swap disk
  * space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All

mm/swapfile.c +7 -1

index 5ac2cb40dbd3..0a52e9b2f843 100644
@@ -263,6 +263,12 @@ static inline void __lock_cluster(struct swap_cluster_info *ci)
 	spin_lock(&ci->lock);
 }
 
+static inline void __lock_cluster_nested(struct swap_cluster_info *ci,
+					 unsigned subclass)
+{
+	spin_lock_nested(&ci->lock, subclass);
+}
+
 static inline struct swap_cluster_info *lock_cluster(struct swap_info_struct *si,
 						     unsigned long offset)
 {

@@ -336,7 +342,7 @@ static void cluster_list_add_tail(struct swap_cluster_list *list,
 		 * only acquired when we held swap_info_struct->lock
 		 */
 		ci_tail = ci + tail;
-		__lock_cluster(ci_tail);
+		__lock_cluster_nested(ci_tail, SWAP_CLUSTER_LOCK_NESTED);
 		cluster_set_next(ci_tail, idx);
 		unlock_cluster(ci_tail);

Quoted 3 lines

mm/swapfile.c +0 -0

 		cluster_set_next_flag(&list->tail, idx, 0);

Signature · 1 line

2.11.0

I do not understand your zest for putting wrappers around every little thing, making it all harder to follow than it need be. Here's the patch I've been running with (but you have a leak somewhere, and I don't have time to search out and fix it: please try sustained swapping and swapoff).

[PATCH] mm, swap: Annotate nested locking for cluster lock

Fix swap cluster lockdep warnings.

Reported-by Minchan Kim <[email protected]>

Signed-off-by Hugh Dickins <[email protected]>

---

 mm/swapfile.c |    9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

mm/swapfile.c +1 -6

@@ -258,11 +258,6 @@ static inline void cluster_set_null(stru
 	info->data = 0;
 }
 
-static inline void __lock_cluster(struct swap_cluster_info *ci)
-{
-	spin_lock(&ci->lock);
-}
-
 static inline struct swap_cluster_info *lock_cluster(struct swap_info_struct *si,
 						     unsigned long offset)
 {

@@ -271,7 +266,7 @@ static inline struct swap_cluster_info *
 	ci = si->cluster_info;
 	if (ci) {
 		ci += offset / SWAPFILE_CLUSTER;
-		__lock_cluster(ci);
+		spin_lock(&ci->lock);
 	}
 	return ci;
 }