Patchwork kernel panic on null pointer on page->mem_cgroup

login
register
mail settings
Submitter Johannes Weiner
Date Aug. 9, 2017, 6:38 p.m.
Message ID <20170809183825.GA26387@cmpxchg.org>
Download mbox | patch
Permalink /patch/9891645/
State New
Headers show

Comments

Johannes Weiner - Aug. 9, 2017, 6:38 p.m.
On Tue, Aug 08, 2017 at 10:39:27PM -0400, Brad Bolen wrote:
> Yes, the BUG_ON(!page_count(page)) fired for me as well.

Brad, Jaegeuk, does the following patch address this problem?

---

From cf0060892eb70bccbc8cedeac0a5756c8f7b975e Mon Sep 17 00:00:00 2001
From: Johannes Weiner <hannes@cmpxchg.org>
Date: Wed, 9 Aug 2017 12:06:03 -0400
Subject: [PATCH] mm: memcontrol: fix NULL pointer crash in
 test_clear_page_writeback()

Jaegeuk and Brad report a NULL pointer crash when writeback ending
tries to update the memcg stats:

[] BUG: unable to handle kernel NULL pointer dereference at 00000000000003b0
[] IP: test_clear_page_writeback+0x12e/0x2c0
[...]
[] RIP: 0010:test_clear_page_writeback+0x12e/0x2c0
[] RSP: 0018:ffff8e3abfd03d78 EFLAGS: 00010046
[] RAX: 0000000000000000 RBX: ffffdb59c03f8900 RCX: ffffffffffffffe8
[] RDX: 0000000000000000 RSI: 0000000000000010 RDI: ffff8e3abffeb000
[] RBP: ffff8e3abfd03da8 R08: 0000000000020059 R09: 00000000fffffffc
[] R10: 0000000000000000 R11: 0000000000020048 R12: ffff8e3a8c39f668
[] R13: 0000000000000001 R14: ffff8e3a8c39f680 R15: 0000000000000000
[] FS:  0000000000000000(0000) GS:ffff8e3abfd00000(0000) knlGS:0000000000000000
[] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[] CR2: 00000000000003b0 CR3: 000000002c5e1000 CR4: 00000000000406e0
[] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[] Call Trace:
[]  <IRQ>
[]  end_page_writeback+0x47/0x70
[]  f2fs_write_end_io+0x76/0x180 [f2fs]
[]  bio_endio+0x9f/0x120
[]  blk_update_request+0xa8/0x2f0
[]  scsi_end_request+0x39/0x1d0
[]  scsi_io_completion+0x211/0x690
[]  scsi_finish_command+0xd9/0x120
[]  scsi_softirq_done+0x127/0x150
[]  __blk_mq_complete_request_remote+0x13/0x20
[]  flush_smp_call_function_queue+0x56/0x110
[]  generic_smp_call_function_single_interrupt+0x13/0x30
[]  smp_call_function_single_interrupt+0x27/0x40
[]  call_function_single_interrupt+0x89/0x90
[] RIP: 0010:native_safe_halt+0x6/0x10

(gdb) l *(test_clear_page_writeback+0x12e)
0xffffffff811bae3e is in test_clear_page_writeback (./include/linux/memcontrol.h:619).
614		mod_node_page_state(page_pgdat(page), idx, val);
615		if (mem_cgroup_disabled() || !page->mem_cgroup)
616			return;
617		mod_memcg_state(page->mem_cgroup, idx, val);
618		pn = page->mem_cgroup->nodeinfo[page_to_nid(page)];
619		this_cpu_add(pn->lruvec_stat->count[idx], val);
620	}
621
622	unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order,
623							gfp_t gfp_mask,

The issue is that writeback doesn't hold a page reference and the page
might get freed after PG_writeback is cleared (and the mapping is
unlocked) in test_clear_page_writeback(). The stat functions looking
up the page's node or zone are safe, as those attributes are static
across allocation and free cycles. But page->mem_cgroup is not, and it
will get cleared if we race with truncation or migration.

It appears this race window has been around for a while, but less
likely to trigger when the memcg stats were updated first thing after
PG_writeback is cleared. Recent changes reshuffled this code to update
the global node stats before the memcg ones, though, stretching the
race window out to an extent where people can reproduce the problem.

Update test_clear_page_writeback() to look up and pin page->mem_cgroup
before clearing PG_writeback, then not use that pointer afterward. It
is a partial revert of 62cccb8c8e7a ("mm: simplify lock_page_memcg()")
but leaves the pageref-holding callsites that aren't affected alone.

Fixes: 62cccb8c8e7a ("mm: simplify lock_page_memcg()")
Reported-by: Jaegeuk Kim <jaegeuk@kernel.org>
Reported-by: Bradley Bolen <bradleybolen@gmail.com>
Cc: <stable@vger.kernel.org> # 4.6+
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 include/linux/memcontrol.h | 10 ++++++++--
 mm/memcontrol.c            | 43 +++++++++++++++++++++++++++++++------------
 mm/page-writeback.c        | 15 ++++++++++++---
 3 files changed, 51 insertions(+), 17 deletions(-)
Michal Hocko - Aug. 10, 2017, 11:56 a.m.
On Wed 09-08-17 14:38:25, Johannes Weiner wrote:
> On Tue, Aug 08, 2017 at 10:39:27PM -0400, Brad Bolen wrote:
> > Yes, the BUG_ON(!page_count(page)) fired for me as well.
> 
> Brad, Jaegeuk, does the following patch address this problem?
> 
> ---
> 
> >From cf0060892eb70bccbc8cedeac0a5756c8f7b975e Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Wed, 9 Aug 2017 12:06:03 -0400
> Subject: [PATCH] mm: memcontrol: fix NULL pointer crash in
>  test_clear_page_writeback()
> 
> Jaegeuk and Brad report a NULL pointer crash when writeback ending
> tries to update the memcg stats:
> 
> [] BUG: unable to handle kernel NULL pointer dereference at 00000000000003b0
> [] IP: test_clear_page_writeback+0x12e/0x2c0
> [...]
> [] RIP: 0010:test_clear_page_writeback+0x12e/0x2c0
> [] RSP: 0018:ffff8e3abfd03d78 EFLAGS: 00010046
> [] RAX: 0000000000000000 RBX: ffffdb59c03f8900 RCX: ffffffffffffffe8
> [] RDX: 0000000000000000 RSI: 0000000000000010 RDI: ffff8e3abffeb000
> [] RBP: ffff8e3abfd03da8 R08: 0000000000020059 R09: 00000000fffffffc
> [] R10: 0000000000000000 R11: 0000000000020048 R12: ffff8e3a8c39f668
> [] R13: 0000000000000001 R14: ffff8e3a8c39f680 R15: 0000000000000000
> [] FS:  0000000000000000(0000) GS:ffff8e3abfd00000(0000) knlGS:0000000000000000
> [] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [] CR2: 00000000000003b0 CR3: 000000002c5e1000 CR4: 00000000000406e0
> [] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [] Call Trace:
> []  <IRQ>
> []  end_page_writeback+0x47/0x70
> []  f2fs_write_end_io+0x76/0x180 [f2fs]
> []  bio_endio+0x9f/0x120
> []  blk_update_request+0xa8/0x2f0
> []  scsi_end_request+0x39/0x1d0
> []  scsi_io_completion+0x211/0x690
> []  scsi_finish_command+0xd9/0x120
> []  scsi_softirq_done+0x127/0x150
> []  __blk_mq_complete_request_remote+0x13/0x20
> []  flush_smp_call_function_queue+0x56/0x110
> []  generic_smp_call_function_single_interrupt+0x13/0x30
> []  smp_call_function_single_interrupt+0x27/0x40
> []  call_function_single_interrupt+0x89/0x90
> [] RIP: 0010:native_safe_halt+0x6/0x10
> 
> (gdb) l *(test_clear_page_writeback+0x12e)
> 0xffffffff811bae3e is in test_clear_page_writeback (./include/linux/memcontrol.h:619).
> 614		mod_node_page_state(page_pgdat(page), idx, val);
> 615		if (mem_cgroup_disabled() || !page->mem_cgroup)
> 616			return;
> 617		mod_memcg_state(page->mem_cgroup, idx, val);
> 618		pn = page->mem_cgroup->nodeinfo[page_to_nid(page)];
> 619		this_cpu_add(pn->lruvec_stat->count[idx], val);
> 620	}
> 621
> 622	unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order,
> 623							gfp_t gfp_mask,
> 
> The issue is that writeback doesn't hold a page reference and the page
> might get freed after PG_writeback is cleared (and the mapping is
> unlocked) in test_clear_page_writeback(). The stat functions looking
> up the page's node or zone are safe, as those attributes are static
> across allocation and free cycles. But page->mem_cgroup is not, and it
> will get cleared if we race with truncation or migration.

Is there anything that prevents us from holding a reference on a page
under writeback?
Brad Bolen - Aug. 11, 2017, 1:30 a.m.
Johannes,

Yes, the patch (slightly modified to apply for 4.11) does make my problem
go away.  Thank you for driving this to a solution.

Brad

On Wed, Aug 9, 2017 at 2:38 PM, Johannes Weiner <hannes@cmpxchg.org> wrote:
> On Tue, Aug 08, 2017 at 10:39:27PM -0400, Brad Bolen wrote:
>> Yes, the BUG_ON(!page_count(page)) fired for me as well.
>
> Brad, Jaegeuk, does the following patch address this problem?
>
> ---
>
> From cf0060892eb70bccbc8cedeac0a5756c8f7b975e Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Wed, 9 Aug 2017 12:06:03 -0400
> Subject: [PATCH] mm: memcontrol: fix NULL pointer crash in
>  test_clear_page_writeback()
>
> Jaegeuk and Brad report a NULL pointer crash when writeback ending
> tries to update the memcg stats:
>
> [] BUG: unable to handle kernel NULL pointer dereference at 00000000000003b0
> [] IP: test_clear_page_writeback+0x12e/0x2c0
> [...]
> [] RIP: 0010:test_clear_page_writeback+0x12e/0x2c0
> [] RSP: 0018:ffff8e3abfd03d78 EFLAGS: 00010046
> [] RAX: 0000000000000000 RBX: ffffdb59c03f8900 RCX: ffffffffffffffe8
> [] RDX: 0000000000000000 RSI: 0000000000000010 RDI: ffff8e3abffeb000
> [] RBP: ffff8e3abfd03da8 R08: 0000000000020059 R09: 00000000fffffffc
> [] R10: 0000000000000000 R11: 0000000000020048 R12: ffff8e3a8c39f668
> [] R13: 0000000000000001 R14: ffff8e3a8c39f680 R15: 0000000000000000
> [] FS:  0000000000000000(0000) GS:ffff8e3abfd00000(0000) knlGS:0000000000000000
> [] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [] CR2: 00000000000003b0 CR3: 000000002c5e1000 CR4: 00000000000406e0
> [] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [] Call Trace:
> []  <IRQ>
> []  end_page_writeback+0x47/0x70
> []  f2fs_write_end_io+0x76/0x180 [f2fs]
> []  bio_endio+0x9f/0x120
> []  blk_update_request+0xa8/0x2f0
> []  scsi_end_request+0x39/0x1d0
> []  scsi_io_completion+0x211/0x690
> []  scsi_finish_command+0xd9/0x120
> []  scsi_softirq_done+0x127/0x150
> []  __blk_mq_complete_request_remote+0x13/0x20
> []  flush_smp_call_function_queue+0x56/0x110
> []  generic_smp_call_function_single_interrupt+0x13/0x30
> []  smp_call_function_single_interrupt+0x27/0x40
> []  call_function_single_interrupt+0x89/0x90
> [] RIP: 0010:native_safe_halt+0x6/0x10
>
> (gdb) l *(test_clear_page_writeback+0x12e)
> 0xffffffff811bae3e is in test_clear_page_writeback (./include/linux/memcontrol.h:619).
> 614             mod_node_page_state(page_pgdat(page), idx, val);
> 615             if (mem_cgroup_disabled() || !page->mem_cgroup)
> 616                     return;
> 617             mod_memcg_state(page->mem_cgroup, idx, val);
> 618             pn = page->mem_cgroup->nodeinfo[page_to_nid(page)];
> 619             this_cpu_add(pn->lruvec_stat->count[idx], val);
> 620     }
> 621
> 622     unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order,
> 623                                                     gfp_t gfp_mask,
>
> The issue is that writeback doesn't hold a page reference and the page
> might get freed after PG_writeback is cleared (and the mapping is
> unlocked) in test_clear_page_writeback(). The stat functions looking
> up the page's node or zone are safe, as those attributes are static
> across allocation and free cycles. But page->mem_cgroup is not, and it
> will get cleared if we race with truncation or migration.
>
> It appears this race window has been around for a while, but less
> likely to trigger when the memcg stats were updated first thing after
> PG_writeback is cleared. Recent changes reshuffled this code to update
> the global node stats before the memcg ones, though, stretching the
> race window out to an extent where people can reproduce the problem.
>
> Update test_clear_page_writeback() to look up and pin page->mem_cgroup
> before clearing PG_writeback, then not use that pointer afterward. It
> is a partial revert of 62cccb8c8e7a ("mm: simplify lock_page_memcg()")
> but leaves the pageref-holding callsites that aren't affected alone.
>
> Fixes: 62cccb8c8e7a ("mm: simplify lock_page_memcg()")
> Reported-by: Jaegeuk Kim <jaegeuk@kernel.org>
> Reported-by: Bradley Bolen <bradleybolen@gmail.com>
> Cc: <stable@vger.kernel.org> # 4.6+
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> ---
>  include/linux/memcontrol.h | 10 ++++++++--
>  mm/memcontrol.c            | 43 +++++++++++++++++++++++++++++++------------
>  mm/page-writeback.c        | 15 ++++++++++++---
>  3 files changed, 51 insertions(+), 17 deletions(-)
>
> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> index 3914e3dd6168..9b15a4bcfa77 100644
> --- a/include/linux/memcontrol.h
> +++ b/include/linux/memcontrol.h
> @@ -484,7 +484,8 @@ bool mem_cgroup_oom_synchronize(bool wait);
>  extern int do_swap_account;
>  #endif
>
> -void lock_page_memcg(struct page *page);
> +struct mem_cgroup *lock_page_memcg(struct page *page);
> +void __unlock_page_memcg(struct mem_cgroup *memcg);
>  void unlock_page_memcg(struct page *page);
>
>  static inline unsigned long memcg_page_state(struct mem_cgroup *memcg,
> @@ -809,7 +810,12 @@ mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p)
>  {
>  }
>
> -static inline void lock_page_memcg(struct page *page)
> +static inline struct mem_cgroup *lock_page_memcg(struct page *page)
> +{
> +       return NULL;
> +}
> +
> +static inline void __unlock_page_memcg(struct mem_cgroup *memcg)
>  {
>  }
>
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 3df3c04d73ab..e09741af816f 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -1611,9 +1611,13 @@ bool mem_cgroup_oom_synchronize(bool handle)
>   * @page: the page
>   *
>   * This function protects unlocked LRU pages from being moved to
> - * another cgroup and stabilizes their page->mem_cgroup binding.
> + * another cgroup.
> + *
> + * It ensures lifetime of the returned memcg. Caller is responsible
> + * for the lifetime of the page; __unlock_page_memcg() is available
> + * when @page might get freed inside the locked section.
>   */
> -void lock_page_memcg(struct page *page)
> +struct mem_cgroup *lock_page_memcg(struct page *page)
>  {
>         struct mem_cgroup *memcg;
>         unsigned long flags;
> @@ -1622,18 +1626,24 @@ void lock_page_memcg(struct page *page)
>          * The RCU lock is held throughout the transaction.  The fast
>          * path can get away without acquiring the memcg->move_lock
>          * because page moving starts with an RCU grace period.
> -        */
> +        *
> +        * The RCU lock also protects the memcg from being freed when
> +        * the page state that is going to change is the only thing
> +        * preventing the page itself from being freed. E.g. writeback
> +        * doesn't hold a page reference and relies on PG_writeback to
> +        * keep off truncation, migration and so forth.
> +         */
>         rcu_read_lock();
>
>         if (mem_cgroup_disabled())
> -               return;
> +               return NULL;
>  again:
>         memcg = page->mem_cgroup;
>         if (unlikely(!memcg))
> -               return;
> +               return NULL;
>
>         if (atomic_read(&memcg->moving_account) <= 0)
> -               return;
> +               return memcg;
>
>         spin_lock_irqsave(&memcg->move_lock, flags);
>         if (memcg != page->mem_cgroup) {
> @@ -1649,18 +1659,18 @@ void lock_page_memcg(struct page *page)
>         memcg->move_lock_task = current;
>         memcg->move_lock_flags = flags;
>
> -       return;
> +       return memcg;
>  }
>  EXPORT_SYMBOL(lock_page_memcg);
>
>  /**
> - * unlock_page_memcg - unlock a page->mem_cgroup binding
> - * @page: the page
> + * __unlock_page_memcg - unlock and unpin a memcg
> + * @memcg: the memcg
> + *
> + * Unlock and unpin a memcg returned by lock_page_memcg().
>   */
> -void unlock_page_memcg(struct page *page)
> +void __unlock_page_memcg(struct mem_cgroup *memcg)
>  {
> -       struct mem_cgroup *memcg = page->mem_cgroup;
> -
>         if (memcg && memcg->move_lock_task == current) {
>                 unsigned long flags = memcg->move_lock_flags;
>
> @@ -1672,6 +1682,15 @@ void unlock_page_memcg(struct page *page)
>
>         rcu_read_unlock();
>  }
> +
> +/**
> + * unlock_page_memcg - unlock a page->mem_cgroup binding
> + * @page: the page
> + */
> +void unlock_page_memcg(struct page *page)
> +{
> +       __unlock_page_memcg(page->mem_cgroup);
> +}
>  EXPORT_SYMBOL(unlock_page_memcg);
>
>  /*
> diff --git a/mm/page-writeback.c b/mm/page-writeback.c
> index 96e93b214d31..bf050ab025b7 100644
> --- a/mm/page-writeback.c
> +++ b/mm/page-writeback.c
> @@ -2724,9 +2724,12 @@ EXPORT_SYMBOL(clear_page_dirty_for_io);
>  int test_clear_page_writeback(struct page *page)
>  {
>         struct address_space *mapping = page_mapping(page);
> +       struct mem_cgroup *memcg;
> +       struct lruvec *lruvec;
>         int ret;
>
> -       lock_page_memcg(page);
> +       memcg = lock_page_memcg(page);
> +       lruvec = mem_cgroup_page_lruvec(page, page_pgdat(page));
>         if (mapping && mapping_use_writeback_tags(mapping)) {
>                 struct inode *inode = mapping->host;
>                 struct backing_dev_info *bdi = inode_to_bdi(inode);
> @@ -2754,12 +2757,18 @@ int test_clear_page_writeback(struct page *page)
>         } else {
>                 ret = TestClearPageWriteback(page);
>         }
> +       /*
> +        * NOTE: Page might be free now! Writeback doesn't hold a page
> +        * reference on its own, it relies on truncation to wait for
> +        * the clearing of PG_writeback. The below can only access
> +        * page state that is static across allocation cycles.
> +        */
>         if (ret) {
> -               dec_lruvec_page_state(page, NR_WRITEBACK);
> +               dec_lruvec_state(lruvec, NR_WRITEBACK);
>                 dec_zone_page_state(page, NR_ZONE_WRITE_PENDING);
>                 inc_node_page_state(page, NR_WRITTEN);
>         }
> -       unlock_page_memcg(page);
> +       __unlock_page_memcg(memcg);
>         return ret;
>  }
>
> --
> 2.13.3
>
Jaegeuk Kim - Aug. 11, 2017, 9:37 p.m.
Hi Johannes,

On 08/09, Johannes Weiner wrote:
> On Tue, Aug 08, 2017 at 10:39:27PM -0400, Brad Bolen wrote:
> > Yes, the BUG_ON(!page_count(page)) fired for me as well.
> 
> Brad, Jaegeuk, does the following patch address this problem?

I also confirmed that this patch addresses the problem.

Tested-by: Jaegeuk Kim <jaegeuk@kernel.org>

Thanks,

> 
> ---
> 
> >From cf0060892eb70bccbc8cedeac0a5756c8f7b975e Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Wed, 9 Aug 2017 12:06:03 -0400
> Subject: [PATCH] mm: memcontrol: fix NULL pointer crash in
>  test_clear_page_writeback()
> 
> Jaegeuk and Brad report a NULL pointer crash when writeback ending
> tries to update the memcg stats:
> 
> [] BUG: unable to handle kernel NULL pointer dereference at 00000000000003b0
> [] IP: test_clear_page_writeback+0x12e/0x2c0
> [...]
> [] RIP: 0010:test_clear_page_writeback+0x12e/0x2c0
> [] RSP: 0018:ffff8e3abfd03d78 EFLAGS: 00010046
> [] RAX: 0000000000000000 RBX: ffffdb59c03f8900 RCX: ffffffffffffffe8
> [] RDX: 0000000000000000 RSI: 0000000000000010 RDI: ffff8e3abffeb000
> [] RBP: ffff8e3abfd03da8 R08: 0000000000020059 R09: 00000000fffffffc
> [] R10: 0000000000000000 R11: 0000000000020048 R12: ffff8e3a8c39f668
> [] R13: 0000000000000001 R14: ffff8e3a8c39f680 R15: 0000000000000000
> [] FS:  0000000000000000(0000) GS:ffff8e3abfd00000(0000) knlGS:0000000000000000
> [] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [] CR2: 00000000000003b0 CR3: 000000002c5e1000 CR4: 00000000000406e0
> [] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [] Call Trace:
> []  <IRQ>
> []  end_page_writeback+0x47/0x70
> []  f2fs_write_end_io+0x76/0x180 [f2fs]
> []  bio_endio+0x9f/0x120
> []  blk_update_request+0xa8/0x2f0
> []  scsi_end_request+0x39/0x1d0
> []  scsi_io_completion+0x211/0x690
> []  scsi_finish_command+0xd9/0x120
> []  scsi_softirq_done+0x127/0x150
> []  __blk_mq_complete_request_remote+0x13/0x20
> []  flush_smp_call_function_queue+0x56/0x110
> []  generic_smp_call_function_single_interrupt+0x13/0x30
> []  smp_call_function_single_interrupt+0x27/0x40
> []  call_function_single_interrupt+0x89/0x90
> [] RIP: 0010:native_safe_halt+0x6/0x10
> 
> (gdb) l *(test_clear_page_writeback+0x12e)
> 0xffffffff811bae3e is in test_clear_page_writeback (./include/linux/memcontrol.h:619).
> 614		mod_node_page_state(page_pgdat(page), idx, val);
> 615		if (mem_cgroup_disabled() || !page->mem_cgroup)
> 616			return;
> 617		mod_memcg_state(page->mem_cgroup, idx, val);
> 618		pn = page->mem_cgroup->nodeinfo[page_to_nid(page)];
> 619		this_cpu_add(pn->lruvec_stat->count[idx], val);
> 620	}
> 621
> 622	unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order,
> 623							gfp_t gfp_mask,
> 
> The issue is that writeback doesn't hold a page reference and the page
> might get freed after PG_writeback is cleared (and the mapping is
> unlocked) in test_clear_page_writeback(). The stat functions looking
> up the page's node or zone are safe, as those attributes are static
> across allocation and free cycles. But page->mem_cgroup is not, and it
> will get cleared if we race with truncation or migration.
> 
> It appears this race window has been around for a while, but less
> likely to trigger when the memcg stats were updated first thing after
> PG_writeback is cleared. Recent changes reshuffled this code to update
> the global node stats before the memcg ones, though, stretching the
> race window out to an extent where people can reproduce the problem.
> 
> Update test_clear_page_writeback() to look up and pin page->mem_cgroup
> before clearing PG_writeback, then not use that pointer afterward. It
> is a partial revert of 62cccb8c8e7a ("mm: simplify lock_page_memcg()")
> but leaves the pageref-holding callsites that aren't affected alone.
> 
> Fixes: 62cccb8c8e7a ("mm: simplify lock_page_memcg()")
> Reported-by: Jaegeuk Kim <jaegeuk@kernel.org>
> Reported-by: Bradley Bolen <bradleybolen@gmail.com>
> Cc: <stable@vger.kernel.org> # 4.6+
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> ---
>  include/linux/memcontrol.h | 10 ++++++++--
>  mm/memcontrol.c            | 43 +++++++++++++++++++++++++++++++------------
>  mm/page-writeback.c        | 15 ++++++++++++---
>  3 files changed, 51 insertions(+), 17 deletions(-)
> 
> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> index 3914e3dd6168..9b15a4bcfa77 100644
> --- a/include/linux/memcontrol.h
> +++ b/include/linux/memcontrol.h
> @@ -484,7 +484,8 @@ bool mem_cgroup_oom_synchronize(bool wait);
>  extern int do_swap_account;
>  #endif
>  
> -void lock_page_memcg(struct page *page);
> +struct mem_cgroup *lock_page_memcg(struct page *page);
> +void __unlock_page_memcg(struct mem_cgroup *memcg);
>  void unlock_page_memcg(struct page *page);
>  
>  static inline unsigned long memcg_page_state(struct mem_cgroup *memcg,
> @@ -809,7 +810,12 @@ mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p)
>  {
>  }
>  
> -static inline void lock_page_memcg(struct page *page)
> +static inline struct mem_cgroup *lock_page_memcg(struct page *page)
> +{
> +	return NULL;
> +}
> +
> +static inline void __unlock_page_memcg(struct mem_cgroup *memcg)
>  {
>  }
>  
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 3df3c04d73ab..e09741af816f 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -1611,9 +1611,13 @@ bool mem_cgroup_oom_synchronize(bool handle)
>   * @page: the page
>   *
>   * This function protects unlocked LRU pages from being moved to
> - * another cgroup and stabilizes their page->mem_cgroup binding.
> + * another cgroup.
> + *
> + * It ensures lifetime of the returned memcg. Caller is responsible
> + * for the lifetime of the page; __unlock_page_memcg() is available
> + * when @page might get freed inside the locked section.
>   */
> -void lock_page_memcg(struct page *page)
> +struct mem_cgroup *lock_page_memcg(struct page *page)
>  {
>  	struct mem_cgroup *memcg;
>  	unsigned long flags;
> @@ -1622,18 +1626,24 @@ void lock_page_memcg(struct page *page)
>  	 * The RCU lock is held throughout the transaction.  The fast
>  	 * path can get away without acquiring the memcg->move_lock
>  	 * because page moving starts with an RCU grace period.
> -	 */
> +	 *
> +	 * The RCU lock also protects the memcg from being freed when
> +	 * the page state that is going to change is the only thing
> +	 * preventing the page itself from being freed. E.g. writeback
> +	 * doesn't hold a page reference and relies on PG_writeback to
> +	 * keep off truncation, migration and so forth.
> +         */
>  	rcu_read_lock();
>  
>  	if (mem_cgroup_disabled())
> -		return;
> +		return NULL;
>  again:
>  	memcg = page->mem_cgroup;
>  	if (unlikely(!memcg))
> -		return;
> +		return NULL;
>  
>  	if (atomic_read(&memcg->moving_account) <= 0)
> -		return;
> +		return memcg;
>  
>  	spin_lock_irqsave(&memcg->move_lock, flags);
>  	if (memcg != page->mem_cgroup) {
> @@ -1649,18 +1659,18 @@ void lock_page_memcg(struct page *page)
>  	memcg->move_lock_task = current;
>  	memcg->move_lock_flags = flags;
>  
> -	return;
> +	return memcg;
>  }
>  EXPORT_SYMBOL(lock_page_memcg);
>  
>  /**
> - * unlock_page_memcg - unlock a page->mem_cgroup binding
> - * @page: the page
> + * __unlock_page_memcg - unlock and unpin a memcg
> + * @memcg: the memcg
> + *
> + * Unlock and unpin a memcg returned by lock_page_memcg().
>   */
> -void unlock_page_memcg(struct page *page)
> +void __unlock_page_memcg(struct mem_cgroup *memcg)
>  {
> -	struct mem_cgroup *memcg = page->mem_cgroup;
> -
>  	if (memcg && memcg->move_lock_task == current) {
>  		unsigned long flags = memcg->move_lock_flags;
>  
> @@ -1672,6 +1682,15 @@ void unlock_page_memcg(struct page *page)
>  
>  	rcu_read_unlock();
>  }
> +
> +/**
> + * unlock_page_memcg - unlock a page->mem_cgroup binding
> + * @page: the page
> + */
> +void unlock_page_memcg(struct page *page)
> +{
> +	__unlock_page_memcg(page->mem_cgroup);
> +}
>  EXPORT_SYMBOL(unlock_page_memcg);
>  
>  /*
> diff --git a/mm/page-writeback.c b/mm/page-writeback.c
> index 96e93b214d31..bf050ab025b7 100644
> --- a/mm/page-writeback.c
> +++ b/mm/page-writeback.c
> @@ -2724,9 +2724,12 @@ EXPORT_SYMBOL(clear_page_dirty_for_io);
>  int test_clear_page_writeback(struct page *page)
>  {
>  	struct address_space *mapping = page_mapping(page);
> +	struct mem_cgroup *memcg;
> +	struct lruvec *lruvec;
>  	int ret;
>  
> -	lock_page_memcg(page);
> +	memcg = lock_page_memcg(page);
> +	lruvec = mem_cgroup_page_lruvec(page, page_pgdat(page));
>  	if (mapping && mapping_use_writeback_tags(mapping)) {
>  		struct inode *inode = mapping->host;
>  		struct backing_dev_info *bdi = inode_to_bdi(inode);
> @@ -2754,12 +2757,18 @@ int test_clear_page_writeback(struct page *page)
>  	} else {
>  		ret = TestClearPageWriteback(page);
>  	}
> +	/*
> +	 * NOTE: Page might be free now! Writeback doesn't hold a page
> +	 * reference on its own, it relies on truncation to wait for
> +	 * the clearing of PG_writeback. The below can only access
> +	 * page state that is static across allocation cycles.
> +	 */
>  	if (ret) {
> -		dec_lruvec_page_state(page, NR_WRITEBACK);
> +		dec_lruvec_state(lruvec, NR_WRITEBACK);
>  		dec_zone_page_state(page, NR_ZONE_WRITE_PENDING);
>  		inc_node_page_state(page, NR_WRITTEN);
>  	}
> -	unlock_page_memcg(page);
> +	__unlock_page_memcg(memcg);
>  	return ret;
>  }
>  
> -- 
> 2.13.3
Johannes Weiner - Aug. 21, 2017, 1:02 p.m.
On Thu, Aug 10, 2017 at 01:56:05PM +0200, Michal Hocko wrote:
> On Wed 09-08-17 14:38:25, Johannes Weiner wrote:
> > The issue is that writeback doesn't hold a page reference and the page
> > might get freed after PG_writeback is cleared (and the mapping is
> > unlocked) in test_clear_page_writeback(). The stat functions looking
> > up the page's node or zone are safe, as those attributes are static
> > across allocation and free cycles. But page->mem_cgroup is not, and it
> > will get cleared if we race with truncation or migration.
> 
> Is there anything that prevents us from holding a reference on a page
> under writeback?

Hm, I'm hesitant to add redundant life-time management to the page
there just for memcg, which is not always configured in.

Pinning the memcg instead is slightly more complex, but IMO has the
complexity in a preferrable place.

Would you agree?
Michal Hocko - Aug. 21, 2017, 1:23 p.m.
On Mon 21-08-17 09:02:18, Johannes Weiner wrote:
> On Thu, Aug 10, 2017 at 01:56:05PM +0200, Michal Hocko wrote:
> > On Wed 09-08-17 14:38:25, Johannes Weiner wrote:
> > > The issue is that writeback doesn't hold a page reference and the page
> > > might get freed after PG_writeback is cleared (and the mapping is
> > > unlocked) in test_clear_page_writeback(). The stat functions looking
> > > up the page's node or zone are safe, as those attributes are static
> > > across allocation and free cycles. But page->mem_cgroup is not, and it
> > > will get cleared if we race with truncation or migration.
> > 
> > Is there anything that prevents us from holding a reference on a page
> > under writeback?
> 
> Hm, I'm hesitant to add redundant life-time management to the page
> there just for memcg, which is not always configured in.
> 
> Pinning the memcg instead is slightly more complex, but IMO has the
> complexity in a preferrable place.

If that is the single place that needs such a special handling and it is
very likely to stay that way then the additional complexity is probably
justified. I am just worried that this is really subtle and history
tells us that such a code usually kicks us back later.
 
> Would you agree?

Well, I was not objecting to the patch. It seems correct I am just
worried a robust fix would be preferable. And a clear object life time
sounds like a more robust thing to do.

Patch

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 3914e3dd6168..9b15a4bcfa77 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -484,7 +484,8 @@  bool mem_cgroup_oom_synchronize(bool wait);
 extern int do_swap_account;
 #endif
 
-void lock_page_memcg(struct page *page);
+struct mem_cgroup *lock_page_memcg(struct page *page);
+void __unlock_page_memcg(struct mem_cgroup *memcg);
 void unlock_page_memcg(struct page *page);
 
 static inline unsigned long memcg_page_state(struct mem_cgroup *memcg,
@@ -809,7 +810,12 @@  mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p)
 {
 }
 
-static inline void lock_page_memcg(struct page *page)
+static inline struct mem_cgroup *lock_page_memcg(struct page *page)
+{
+	return NULL;
+}
+
+static inline void __unlock_page_memcg(struct mem_cgroup *memcg)
 {
 }
 
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 3df3c04d73ab..e09741af816f 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1611,9 +1611,13 @@  bool mem_cgroup_oom_synchronize(bool handle)
  * @page: the page
  *
  * This function protects unlocked LRU pages from being moved to
- * another cgroup and stabilizes their page->mem_cgroup binding.
+ * another cgroup.
+ *
+ * It ensures lifetime of the returned memcg. Caller is responsible
+ * for the lifetime of the page; __unlock_page_memcg() is available
+ * when @page might get freed inside the locked section.
  */
-void lock_page_memcg(struct page *page)
+struct mem_cgroup *lock_page_memcg(struct page *page)
 {
 	struct mem_cgroup *memcg;
 	unsigned long flags;
@@ -1622,18 +1626,24 @@  void lock_page_memcg(struct page *page)
 	 * The RCU lock is held throughout the transaction.  The fast
 	 * path can get away without acquiring the memcg->move_lock
 	 * because page moving starts with an RCU grace period.
-	 */
+	 *
+	 * The RCU lock also protects the memcg from being freed when
+	 * the page state that is going to change is the only thing
+	 * preventing the page itself from being freed. E.g. writeback
+	 * doesn't hold a page reference and relies on PG_writeback to
+	 * keep off truncation, migration and so forth.
+         */
 	rcu_read_lock();
 
 	if (mem_cgroup_disabled())
-		return;
+		return NULL;
 again:
 	memcg = page->mem_cgroup;
 	if (unlikely(!memcg))
-		return;
+		return NULL;
 
 	if (atomic_read(&memcg->moving_account) <= 0)
-		return;
+		return memcg;
 
 	spin_lock_irqsave(&memcg->move_lock, flags);
 	if (memcg != page->mem_cgroup) {
@@ -1649,18 +1659,18 @@  void lock_page_memcg(struct page *page)
 	memcg->move_lock_task = current;
 	memcg->move_lock_flags = flags;
 
-	return;
+	return memcg;
 }
 EXPORT_SYMBOL(lock_page_memcg);
 
 /**
- * unlock_page_memcg - unlock a page->mem_cgroup binding
- * @page: the page
+ * __unlock_page_memcg - unlock and unpin a memcg
+ * @memcg: the memcg
+ *
+ * Unlock and unpin a memcg returned by lock_page_memcg().
  */
-void unlock_page_memcg(struct page *page)
+void __unlock_page_memcg(struct mem_cgroup *memcg)
 {
-	struct mem_cgroup *memcg = page->mem_cgroup;
-
 	if (memcg && memcg->move_lock_task == current) {
 		unsigned long flags = memcg->move_lock_flags;
 
@@ -1672,6 +1682,15 @@  void unlock_page_memcg(struct page *page)
 
 	rcu_read_unlock();
 }
+
+/**
+ * unlock_page_memcg - unlock a page->mem_cgroup binding
+ * @page: the page
+ */
+void unlock_page_memcg(struct page *page)
+{
+	__unlock_page_memcg(page->mem_cgroup);
+}
 EXPORT_SYMBOL(unlock_page_memcg);
 
 /*
diff --git a/mm/page-writeback.c b/mm/page-writeback.c
index 96e93b214d31..bf050ab025b7 100644
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -2724,9 +2724,12 @@  EXPORT_SYMBOL(clear_page_dirty_for_io);
 int test_clear_page_writeback(struct page *page)
 {
 	struct address_space *mapping = page_mapping(page);
+	struct mem_cgroup *memcg;
+	struct lruvec *lruvec;
 	int ret;
 
-	lock_page_memcg(page);
+	memcg = lock_page_memcg(page);
+	lruvec = mem_cgroup_page_lruvec(page, page_pgdat(page));
 	if (mapping && mapping_use_writeback_tags(mapping)) {
 		struct inode *inode = mapping->host;
 		struct backing_dev_info *bdi = inode_to_bdi(inode);
@@ -2754,12 +2757,18 @@  int test_clear_page_writeback(struct page *page)
 	} else {
 		ret = TestClearPageWriteback(page);
 	}
+	/*
+	 * NOTE: Page might be free now! Writeback doesn't hold a page
+	 * reference on its own, it relies on truncation to wait for
+	 * the clearing of PG_writeback. The below can only access
+	 * page state that is static across allocation cycles.
+	 */
 	if (ret) {
-		dec_lruvec_page_state(page, NR_WRITEBACK);
+		dec_lruvec_state(lruvec, NR_WRITEBACK);
 		dec_zone_page_state(page, NR_ZONE_WRITE_PENDING);
 		inc_node_page_state(page, NR_WRITTEN);
 	}
-	unlock_page_memcg(page);
+	__unlock_page_memcg(memcg);
 	return ret;
 }