Re: [syzbot] possible deadlock in start_this_handle (3)

Re: [syzbot] possible deadlock in start_this_handle (3)

[Jan Kara]

Hello,

so this lockdep report looks real but is more related to OOM handling than
to ext4 as such. The immediate problem I can see is that
mem_cgroup_print_oom_meminfo() which is called under oom_lock calls
memory_stat_format() which does GFP_KERNEL allocations to allocate buffers
for dumping of MM statistics. This creates oom_lock -> fs reclaim
dependency and because OOM can be hit (and thus oom_lock acquired) in
practically any allocation (regardless of GFP_NOFS) this has a potential of
creating real deadlock cycles.

So should mem_cgroup_print_oom_meminfo() be using
memalloc_nofs_save/restore() to avoid such deadlocks? Or perhaps someone
sees another solution? Generally allocating memory to report OOM looks a
bit dangerous to me ;).

								Honza

On Thu 14-07-22 05:08:26, syzbot wrote:
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    5a29232d870d Merge tag 'for-5.19-rc6-tag' of git://git.ker..
> git tree:       upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=16619ce8080000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=525bc0635a2b942a
> dashboard link: https://syzkaller.appspot.com/bug?extid=2d2aeadc6ce1e1f11d45
> compiler:       gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> userspace arch: i386
> 
> Unfortunately, I don't have any reproducer for this issue yet.
> 
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+2d2aeadc6ce1e1f11d45@syzkaller.appspotmail.com
> 
> ======================================================
> WARNING: possible circular locking dependency detected
> 5.19.0-rc6-syzkaller-00026-g5a29232d870d #0 Not tainted
> ------------------------------------------------------
> khugepaged/48 is trying to acquire lock:
> ffff888044598990 (jbd2_handle){++++}-{0:0}, at: start_this_handle+0xfb4/0x14a0 fs/jbd2/transaction.c:461
> 
> but task is already holding lock:
> ffffffff8bebdb20 (fs_reclaim){+.+.}-{0:0}, at: __perform_reclaim mm/page_alloc.c:4638 [inline]
> ffffffff8bebdb20 (fs_reclaim){+.+.}-{0:0}, at: __alloc_pages_direct_reclaim mm/page_alloc.c:4663 [inline]
> ffffffff8bebdb20 (fs_reclaim){+.+.}-{0:0}, at: __alloc_pages_slowpath.constprop.0+0x9e1/0x2160 mm/page_alloc.c:5066
> 
> which lock already depends on the new lock.
> 
> 
> the existing dependency chain (in reverse order) is:
> 
> -> #2 (fs_reclaim){+.+.}-{0:0}:
>        __fs_reclaim_acquire mm/page_alloc.c:4589 [inline]
>        fs_reclaim_acquire+0x115/0x160 mm/page_alloc.c:4603
>        might_alloc include/linux/sched/mm.h:271 [inline]
>        slab_pre_alloc_hook mm/slab.h:723 [inline]
>        slab_alloc_node mm/slub.c:3157 [inline]
>        slab_alloc mm/slub.c:3251 [inline]
>        kmem_cache_alloc_trace+0x40/0x3f0 mm/slub.c:3282
>        kmalloc include/linux/slab.h:600 [inline]
>        memory_stat_format+0x95/0xae0 mm/memcontrol.c:1468
>        mem_cgroup_print_oom_meminfo.cold+0x50/0x7e mm/memcontrol.c:1594
>        dump_header+0x13f/0x7f9 mm/oom_kill.c:462
>        oom_kill_process.cold+0x10/0x15 mm/oom_kill.c:1037
>        out_of_memory+0x358/0x14b0 mm/oom_kill.c:1175
>        mem_cgroup_out_of_memory+0x206/0x270 mm/memcontrol.c:1650
>        memory_max_write+0x25c/0x3b0 mm/memcontrol.c:6299
>        cgroup_file_write+0x1de/0x770 kernel/cgroup/cgroup.c:3882
>        kernfs_fop_write_iter+0x3f8/0x610 fs/kernfs/file.c:290
>        call_write_iter include/linux/fs.h:2058 [inline]
>        new_sync_write+0x38a/0x560 fs/read_write.c:504
>        vfs_write+0x7c0/0xac0 fs/read_write.c:591
>        ksys_write+0x127/0x250 fs/read_write.c:644
>        do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
>        __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
>        do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
>        entry_SYSENTER_compat_after_hwframe+0x53/0x62
> 
> -> #1 (oom_lock){+.+.}-{3:3}:
>        __mutex_lock_common kernel/locking/mutex.c:603 [inline]
>        __mutex_lock+0x12f/0x1350 kernel/locking/mutex.c:747
>        mem_cgroup_out_of_memory+0x8d/0x270 mm/memcontrol.c:1640
>        mem_cgroup_oom mm/memcontrol.c:1880 [inline]
>        try_charge_memcg+0xef9/0x1380 mm/memcontrol.c:2670
>        obj_cgroup_charge_pages mm/memcontrol.c:2999 [inline]
>        obj_cgroup_charge+0x2ab/0x5e0 mm/memcontrol.c:3289
>        memcg_slab_pre_alloc_hook mm/slab.h:505 [inline]
>        slab_pre_alloc_hook mm/slab.h:728 [inline]
>        slab_alloc_node mm/slub.c:3157 [inline]
>        slab_alloc mm/slub.c:3251 [inline]
>        __kmem_cache_alloc_lru mm/slub.c:3258 [inline]
>        kmem_cache_alloc+0x92/0x3b0 mm/slub.c:3268
>        kmem_cache_zalloc include/linux/slab.h:723 [inline]
>        alloc_buffer_head+0x20/0x140 fs/buffer.c:3294
>        alloc_page_buffers+0x285/0x7a0 fs/buffer.c:829
>        grow_dev_page fs/buffer.c:965 [inline]
>        grow_buffers fs/buffer.c:1011 [inline]
>        __getblk_slow+0x525/0x1080 fs/buffer.c:1038
>        __getblk_gfp+0x6e/0x80 fs/buffer.c:1333
>        sb_getblk include/linux/buffer_head.h:326 [inline]
>        ext4_getblk+0x20d/0x7c0 fs/ext4/inode.c:866
>        ext4_bread+0x2a/0x1c0 fs/ext4/inode.c:912
>        ext4_append+0x177/0x3a0 fs/ext4/namei.c:67
>        ext4_init_new_dir+0x25e/0x4d0 fs/ext4/namei.c:2920
>        ext4_mkdir+0x3cf/0xb20 fs/ext4/namei.c:2966
>        vfs_mkdir+0x1c3/0x3b0 fs/namei.c:3975
>        do_mkdirat+0x285/0x300 fs/namei.c:4001
>        __do_sys_mkdirat fs/namei.c:4016 [inline]
>        __se_sys_mkdirat fs/namei.c:4014 [inline]
>        __ia32_sys_mkdirat+0x81/0xa0 fs/namei.c:4014
>        do_syscall_32_irqs_on arch/x86/entry/common.c:112 [inline]
>        __do_fast_syscall_32+0x65/0xf0 arch/x86/entry/common.c:178
>        do_fast_syscall_32+0x2f/0x70 arch/x86/entry/common.c:203
>        entry_SYSENTER_compat_after_hwframe+0x53/0x62
> 
> -> #0 (jbd2_handle){++++}-{0:0}:
>        check_prev_add kernel/locking/lockdep.c:3095 [inline]
>        check_prevs_add kernel/locking/lockdep.c:3214 [inline]
>        validate_chain kernel/locking/lockdep.c:3829 [inline]
>        __lock_acquire+0x2abe/0x5660 kernel/locking/lockdep.c:5053
>        lock_acquire kernel/locking/lockdep.c:5665 [inline]
>        lock_acquire+0x1ab/0x570 kernel/locking/lockdep.c:5630
>        start_this_handle+0xfe7/0x14a0 fs/jbd2/transaction.c:463
>        jbd2__journal_start+0x399/0x930 fs/jbd2/transaction.c:520
>        __ext4_journal_start_sb+0x3a8/0x4a0 fs/ext4/ext4_jbd2.c:105
>        __ext4_journal_start fs/ext4/ext4_jbd2.h:326 [inline]
>        ext4_dirty_inode+0x9d/0x110 fs/ext4/inode.c:5949
>        __mark_inode_dirty+0x495/0x1050 fs/fs-writeback.c:2381
>        mark_inode_dirty_sync include/linux/fs.h:2337 [inline]
>        iput.part.0+0x57/0x820 fs/inode.c:1767
>        iput+0x58/0x70 fs/inode.c:1760
>        dentry_unlink_inode+0x2b1/0x460 fs/dcache.c:401
>        __dentry_kill+0x3c0/0x640 fs/dcache.c:607
>        shrink_dentry_list+0x23c/0x800 fs/dcache.c:1201
>        prune_dcache_sb+0xe7/0x140 fs/dcache.c:1282
>        super_cache_scan+0x336/0x590 fs/super.c:104
>        do_shrink_slab+0x42d/0xbd0 mm/vmscan.c:770
>        shrink_slab+0x17c/0x6f0 mm/vmscan.c:930
>        shrink_node_memcgs mm/vmscan.c:3124 [inline]
>        shrink_node+0x8b3/0x1db0 mm/vmscan.c:3245
>        shrink_zones mm/vmscan.c:3482 [inline]
>        do_try_to_free_pages+0x3b5/0x1700 mm/vmscan.c:3540
>        try_to_free_pages+0x2ac/0x840 mm/vmscan.c:3775
>        __perform_reclaim mm/page_alloc.c:4641 [inline]
>        __alloc_pages_direct_reclaim mm/page_alloc.c:4663 [inline]
>        __alloc_pages_slowpath.constprop.0+0xa8a/0x2160 mm/page_alloc.c:5066
>        __alloc_pages+0x436/0x510 mm/page_alloc.c:5439
>        __alloc_pages_node include/linux/gfp.h:587 [inline]
>        khugepaged_alloc_page+0xa0/0x170 mm/khugepaged.c:859
>        collapse_huge_page mm/khugepaged.c:1062 [inline]
>        khugepaged_scan_pmd mm/khugepaged.c:1348 [inline]
>        khugepaged_scan_mm_slot mm/khugepaged.c:2170 [inline]
>        khugepaged_do_scan mm/khugepaged.c:2251 [inline]
>        khugepaged+0x3473/0x66a0 mm/khugepaged.c:2296
>        kthread+0x2e9/0x3a0 kernel/kthread.c:376
>        ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:302
> 
> other info that might help us debug this:
> 
> Chain exists of:
>   jbd2_handle --> oom_lock --> fs_reclaim
> 
>  Possible unsafe locking scenario:
> 
>        CPU0                    CPU1
>        ----                    ----
>   lock(fs_reclaim);
>                                lock(oom_lock);
>                                lock(fs_reclaim);
>   lock(jbd2_handle);
> 
>  *** DEADLOCK ***
> 
> 3 locks held by khugepaged/48:
>  #0: ffffffff8bebdb20 (fs_reclaim){+.+.}-{0:0}, at: __perform_reclaim mm/page_alloc.c:4638 [inline]
>  #0: ffffffff8bebdb20 (fs_reclaim){+.+.}-{0:0}, at: __alloc_pages_direct_reclaim mm/page_alloc.c:4663 [inline]
>  #0: ffffffff8bebdb20 (fs_reclaim){+.+.}-{0:0}, at: __alloc_pages_slowpath.constprop.0+0x9e1/0x2160 mm/page_alloc.c:5066
>  #1: ffffffff8be7d850 (shrinker_rwsem){++++}-{3:3}, at: shrink_slab+0xc9/0x6f0 mm/vmscan.c:920
>  #2: ffff8880445800e0 (&type->s_umount_key#33){++++}-{3:3}, at: trylock_super fs/super.c:415 [inline]
>  #2: ffff8880445800e0 (&type->s_umount_key#33){++++}-{3:3}, at: super_cache_scan+0x6c/0x590 fs/super.c:79
> 
> stack backtrace:
> CPU: 2 PID: 48 Comm: khugepaged Not tainted 5.19.0-rc6-syzkaller-00026-g5a29232d870d #0
> Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.14.0-2 04/01/2014
> Call Trace:
>  <TASK>
>  __dump_stack lib/dump_stack.c:88 [inline]
>  dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
>  check_noncircular+0x25f/0x2e0 kernel/locking/lockdep.c:2175
>  check_prev_add kernel/locking/lockdep.c:3095 [inline]
>  check_prevs_add kernel/locking/lockdep.c:3214 [inline]
>  validate_chain kernel/locking/lockdep.c:3829 [inline]
>  __lock_acquire+0x2abe/0x5660 kernel/locking/lockdep.c:5053
>  lock_acquire kernel/locking/lockdep.c:5665 [inline]
>  lock_acquire+0x1ab/0x570 kernel/locking/lockdep.c:5630
>  start_this_handle+0xfe7/0x14a0 fs/jbd2/transaction.c:463
>  jbd2__journal_start+0x399/0x930 fs/jbd2/transaction.c:520
>  __ext4_journal_start_sb+0x3a8/0x4a0 fs/ext4/ext4_jbd2.c:105
>  __ext4_journal_start fs/ext4/ext4_jbd2.h:326 [inline]
>  ext4_dirty_inode+0x9d/0x110 fs/ext4/inode.c:5949
>  __mark_inode_dirty+0x495/0x1050 fs/fs-writeback.c:2381
>  mark_inode_dirty_sync include/linux/fs.h:2337 [inline]
>  iput.part.0+0x57/0x820 fs/inode.c:1767
>  iput+0x58/0x70 fs/inode.c:1760
>  dentry_unlink_inode+0x2b1/0x460 fs/dcache.c:401
>  __dentry_kill+0x3c0/0x640 fs/dcache.c:607
>  shrink_dentry_list+0x23c/0x800 fs/dcache.c:1201
>  prune_dcache_sb+0xe7/0x140 fs/dcache.c:1282
>  super_cache_scan+0x336/0x590 fs/super.c:104
>  do_shrink_slab+0x42d/0xbd0 mm/vmscan.c:770
>  shrink_slab+0x17c/0x6f0 mm/vmscan.c:930
>  shrink_node_memcgs mm/vmscan.c:3124 [inline]
>  shrink_node+0x8b3/0x1db0 mm/vmscan.c:3245
>  shrink_zones mm/vmscan.c:3482 [inline]
>  do_try_to_free_pages+0x3b5/0x1700 mm/vmscan.c:3540
>  try_to_free_pages+0x2ac/0x840 mm/vmscan.c:3775
>  __perform_reclaim mm/page_alloc.c:4641 [inline]
>  __alloc_pages_direct_reclaim mm/page_alloc.c:4663 [inline]
>  __alloc_pages_slowpath.constprop.0+0xa8a/0x2160 mm/page_alloc.c:5066
>  __alloc_pages+0x436/0x510 mm/page_alloc.c:5439
>  __alloc_pages_node include/linux/gfp.h:587 [inline]
>  khugepaged_alloc_page+0xa0/0x170 mm/khugepaged.c:859
>  collapse_huge_page mm/khugepaged.c:1062 [inline]
>  khugepaged_scan_pmd mm/khugepaged.c:1348 [inline]
>  khugepaged_scan_mm_slot mm/khugepaged.c:2170 [inline]
>  khugepaged_do_scan mm/khugepaged.c:2251 [inline]
>  khugepaged+0x3473/0x66a0 mm/khugepaged.c:2296
>  kthread+0x2e9/0x3a0 kernel/kthread.c:376
>  ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:302
>  </TASK>
> 
> 
> ---
> This report is generated by a bot. It may contain errors.
> See https://goo.gl/tpsmEJ for more information about syzbot.
> syzbot engineers can be reached at syzkaller@googlegroups.com.
> 
> syzbot will keep track of this issue. See:
> https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [syzbot] possible deadlock in start_this_handle (3)

[Shakeel Butt]

On Fri, Jul 15, 2022 at 07:24:55AM +0900, Tetsuo Handa wrote:
> On 2022/07/14 23:18, Jan Kara wrote:
> > Hello,
> > 
> > so this lockdep report looks real but is more related to OOM handling than
> > to ext4 as such. The immediate problem I can see is that
> > mem_cgroup_print_oom_meminfo() which is called under oom_lock calls
> > memory_stat_format() which does GFP_KERNEL allocations to allocate buffers
> > for dumping of MM statistics. This creates oom_lock -> fs reclaim
> > dependency and because OOM can be hit (and thus oom_lock acquired) in
> > practically any allocation (regardless of GFP_NOFS) this has a potential of
> > creating real deadlock cycles.
> > 
> > So should mem_cgroup_print_oom_meminfo() be using
> > memalloc_nofs_save/restore() to avoid such deadlocks? Or perhaps someone
> > sees another solution? Generally allocating memory to report OOM looks a
> > bit dangerous to me ;).

mem_cgroup_print_oom_meminfo() is called only for memcg OOMs. So, the
situaion would be dangerous only if the system is also OOM at that time.

> > 
> > 								Honza
> 
> I think mem_cgroup_print_oom_meminfo() should use GFP_ATOMIC, for it will fall into
> infinite loop if kmalloc(GFP_NOFS) under oom_lock reached __alloc_pages_may_oom() path.

I would prefer GFP_NOWAIT. This is printing info for memcg OOMs and if
the system is low on memory then memcg OOMs has lower importance than
the system state.

[PATCH] mm: memcontrol: fix potential oom_lock recursion deadlock

[Tetsuo Handa]

syzbot is reporting fs_reclaim allocation with oom_lock held [1]. We
must make sure that such allocation won't hit __alloc_pages_may_oom()
path which will retry forever if oom_lock is already held.

I choose GFP_ATOMIC than GFP_NOWAIT, for since global OOM situation will
likely be avoided by killing some process in memcg, and memory will be
released after printk(), trying a little hard will be acceptable.

Link: https://syzkaller.appspot.com/bug?extid=2d2aeadc6ce1e1f11d45 [1]
Reported-by: syzbot <syzbot+2d2aeadc6ce1e1f11d45@syzkaller.appspotmail.com>
Fixes: c8713d0b23123759 ("mm: memcontrol: dump memory.stat during cgroup OOM")
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
---
 mm/memcontrol.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 618c366a2f07..11cd900729b9 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1465,7 +1465,8 @@ static char *memory_stat_format(struct mem_cgroup *memcg)
 	struct seq_buf s;
 	int i;
 
-	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_KERNEL), PAGE_SIZE);
+	/* Caller might be holding oom_lock. */
+	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_ATOMIC), PAGE_SIZE);
 	if (!s.buffer)
 		return NULL;
 
-- 
2.18.4

Re: [PATCH] mm: memcontrol: fix potential oom_lock recursion deadlock

[Michal Hocko]

On Thu 21-07-22 08:49:57, Tetsuo Handa wrote:
> syzbot is reporting fs_reclaim allocation with oom_lock held [1]. We
> must make sure that such allocation won't hit __alloc_pages_may_oom()
> path which will retry forever if oom_lock is already held.
> 
> I choose GFP_ATOMIC than GFP_NOWAIT, for since global OOM situation will
> likely be avoided by killing some process in memcg, and memory will be
> released after printk(), trying a little hard will be acceptable.

Nope, this is not a proper fix. You are making memory.stat more likely
to fail.

An uncoditional GFP_KERNEL allocation is certainly not good but is there
any reason to not use GFP_NOIO instead?

Or even better. In an ideal world we won't allocate from here at
all. Can we pre-allocate that single page and re-use it for the oom
report?
--- 
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index abec50f31fe6..13483cb278bb 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1460,14 +1460,12 @@ static inline unsigned long memcg_page_state_output(struct mem_cgroup *memcg,
 	return memcg_page_state(memcg, item) * memcg_page_state_unit(item);
 }
 
-static char *memory_stat_format(struct mem_cgroup *memcg)
+void memory_stat_format(struct mem_cgroup *memcg, char *buf)
 {
 	struct seq_buf s;
 	int i;
 
-	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_KERNEL), PAGE_SIZE);
-	if (!s.buffer)
-		return NULL;
+	seq_buf_init(&s, buf, PAGE_SIZE);
 
 	/*
 	 * Provide statistics on the state of the memory subsystem as
@@ -1533,8 +1531,6 @@ static char *memory_stat_format(struct mem_cgroup *memcg)
 
 	/* The above should easily fit into one page */
 	WARN_ON_ONCE(seq_buf_has_overflowed(&s));
-
-	return s.buffer;
 }
 
 #define K(x) ((x) << (PAGE_SHIFT-10))
@@ -1563,6 +1559,12 @@ void mem_cgroup_print_oom_context(struct mem_cgroup *memcg, struct task_struct *
 	rcu_read_unlock();
 }
 
+/*
+ * preallocated buffer to collect memory stats for the oom situation.
+ * Usage protected by oom_lock
+ */
+char oombuf[PAGE_SIZE];
+
 /**
  * mem_cgroup_print_oom_meminfo: Print OOM memory information relevant to
  * memory controller.
@@ -1570,7 +1572,7 @@ void mem_cgroup_print_oom_context(struct mem_cgroup *memcg, struct task_struct *
  */
 void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 {
-	char *buf;
+	lockdep_assert_held(&oom_lock);
 
 	pr_info("memory: usage %llukB, limit %llukB, failcnt %lu\n",
 		K((u64)page_counter_read(&memcg->memory)),
@@ -1591,11 +1593,8 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 	pr_info("Memory cgroup stats for ");
 	pr_cont_cgroup_path(memcg->css.cgroup);
 	pr_cont(":");
-	buf = memory_stat_format(memcg);
-	if (!buf)
-		return;
-	pr_info("%s", buf);
-	kfree(buf);
+	memory_stat_format(memcg, oombuf);
+	pr_info("%s", oombuf);
 }
 
 /*
@@ -6335,11 +6334,11 @@ static int memory_events_local_show(struct seq_file *m, void *v)
 static int memory_stat_show(struct seq_file *m, void *v)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_seq(m);
-	char *buf;
+	char *buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
 
-	buf = memory_stat_format(memcg);
 	if (!buf)
 		return -ENOMEM;
+	memory_stat_format(memcg, buf);
 	seq_puts(m, buf);
 	kfree(buf);
 	return 0;

-- 
Michal Hocko
SUSE Labs

[PATCH v2] mm: memcontrol: fix potential oom_lock recursion deadlock

[Tetsuo Handa]

syzbot is reporting GFP_KERNEL allocation with oom_lock held [1]. We
must make sure that such allocation won't hit __alloc_pages_may_oom()
path which will retry forever if oom_lock is already held. Use static
buffer when oom_lock is already held.

Link: https://syzkaller.appspot.com/bug?extid=2d2aeadc6ce1e1f11d45 [1]
Reported-by: syzbot <syzbot+2d2aeadc6ce1e1f11d45@syzkaller.appspotmail.com>
Suggested-by: Michal Hocko <mhocko@suse.com>
Fixes: c8713d0b23123759 ("mm: memcontrol: dump memory.stat during cgroup OOM")
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
---
Changes in v2:
  Use static buffer for OOM reporting, suggested by Michal Hocko <mhocko@suse.com>.

 mm/memcontrol.c | 22 +++++++++-------------
 1 file changed, 9 insertions(+), 13 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 618c366a2f07..8092be2fbb7c 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1460,14 +1460,12 @@ static inline unsigned long memcg_page_state_output(struct mem_cgroup *memcg,
 	return memcg_page_state(memcg, item) * memcg_page_state_unit(item);
 }
 
-static char *memory_stat_format(struct mem_cgroup *memcg)
+static void memory_stat_format(struct mem_cgroup *memcg, char *buf, int bufsize)
 {
 	struct seq_buf s;
 	int i;
 
-	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_KERNEL), PAGE_SIZE);
-	if (!s.buffer)
-		return NULL;
+	seq_buf_init(&s, buf, bufsize);
 
 	/*
 	 * Provide statistics on the state of the memory subsystem as
@@ -1533,8 +1531,6 @@ static char *memory_stat_format(struct mem_cgroup *memcg)
 
 	/* The above should easily fit into one page */
 	WARN_ON_ONCE(seq_buf_has_overflowed(&s));
-
-	return s.buffer;
 }
 
 #define K(x) ((x) << (PAGE_SHIFT-10))
@@ -1570,7 +1566,10 @@ void mem_cgroup_print_oom_context(struct mem_cgroup *memcg, struct task_struct *
  */
 void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 {
-	char *buf;
+	/* Use static buffer, for the caller is holding oom_lock. */
+	static char buf[PAGE_SIZE];
+
+	lockdep_assert_held(&oom_lock);
 
 	pr_info("memory: usage %llukB, limit %llukB, failcnt %lu\n",
 		K((u64)page_counter_read(&memcg->memory)),
@@ -1591,11 +1590,8 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 	pr_info("Memory cgroup stats for ");
 	pr_cont_cgroup_path(memcg->css.cgroup);
 	pr_cont(":");
-	buf = memory_stat_format(memcg);
-	if (!buf)
-		return;
+	memory_stat_format(memcg, buf, sizeof(buf));
 	pr_info("%s", buf);
-	kfree(buf);
 }
 
 /*
@@ -6335,11 +6331,11 @@ static int memory_events_local_show(struct seq_file *m, void *v)
 static int memory_stat_show(struct seq_file *m, void *v)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_seq(m);
-	char *buf;
+	char *buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
 
-	buf = memory_stat_format(memcg);
 	if (!buf)
 		return -ENOMEM;
+	memory_stat_format(memcg, buf, PAGE_SIZE);
 	seq_puts(m, buf);
 	kfree(buf);
 	return 0;
-- 
2.18.4

Re: [PATCH v2] mm: memcontrol: fix potential oom_lock recursion deadlock

[Michal Hocko]

On Fri 22-07-22 09:46:27, Tetsuo Handa wrote:
> syzbot is reporting GFP_KERNEL allocation with oom_lock held [1]. We
> must make sure that such allocation won't hit __alloc_pages_may_oom()
> path which will retry forever if oom_lock is already held. Use static
> buffer when oom_lock is already held.

The changelog is rather cryptic. Your previous one was more readable.
I would go with:
"
syzbot is reporting GFP_KERNEL allocation with oom_lock held [1]
when reporting memcg oom. This is problematic because this creates a
dependency between GFP_NOFS and GFP_KERNEL over oom_lock which could
dead lock the system.

Fix the problem by removing the allocation from memory_stat_format
completely. Use a statically preallocated buffer instead for this path.
OOM dumping is synchronized by the oom_lock so there is no exclusion
required here. memory_stat_show can use GFP_KERNEL allocation.
"
 
> Link: https://syzkaller.appspot.com/bug?extid=2d2aeadc6ce1e1f11d45 [1]
> Reported-by: syzbot <syzbot+2d2aeadc6ce1e1f11d45@syzkaller.appspotmail.com>
> Suggested-by: Michal Hocko <mhocko@suse.com>
> Fixes: c8713d0b23123759 ("mm: memcontrol: dump memory.stat during cgroup OOM")
> Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>

Acked-by: Michal Hocko <mhocko@suse.com>

Thanks!

> ---
> Changes in v2:
>   Use static buffer for OOM reporting, suggested by Michal Hocko <mhocko@suse.com>.
> 
>  mm/memcontrol.c | 22 +++++++++-------------
>  1 file changed, 9 insertions(+), 13 deletions(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 618c366a2f07..8092be2fbb7c 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -1460,14 +1460,12 @@ static inline unsigned long memcg_page_state_output(struct mem_cgroup *memcg,
>  	return memcg_page_state(memcg, item) * memcg_page_state_unit(item);
>  }
>  
> -static char *memory_stat_format(struct mem_cgroup *memcg)
> +static void memory_stat_format(struct mem_cgroup *memcg, char *buf, int bufsize)
>  {
>  	struct seq_buf s;
>  	int i;
>  
> -	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_KERNEL), PAGE_SIZE);
> -	if (!s.buffer)
> -		return NULL;
> +	seq_buf_init(&s, buf, bufsize);
>  
>  	/*
>  	 * Provide statistics on the state of the memory subsystem as
> @@ -1533,8 +1531,6 @@ static char *memory_stat_format(struct mem_cgroup *memcg)
>  
>  	/* The above should easily fit into one page */
>  	WARN_ON_ONCE(seq_buf_has_overflowed(&s));
> -
> -	return s.buffer;
>  }
>  
>  #define K(x) ((x) << (PAGE_SHIFT-10))
> @@ -1570,7 +1566,10 @@ void mem_cgroup_print_oom_context(struct mem_cgroup *memcg, struct task_struct *
>   */
>  void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
>  {
> -	char *buf;
> +	/* Use static buffer, for the caller is holding oom_lock. */
> +	static char buf[PAGE_SIZE];
> +
> +	lockdep_assert_held(&oom_lock);
>  
>  	pr_info("memory: usage %llukB, limit %llukB, failcnt %lu\n",
>  		K((u64)page_counter_read(&memcg->memory)),
> @@ -1591,11 +1590,8 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
>  	pr_info("Memory cgroup stats for ");
>  	pr_cont_cgroup_path(memcg->css.cgroup);
>  	pr_cont(":");
> -	buf = memory_stat_format(memcg);
> -	if (!buf)
> -		return;
> +	memory_stat_format(memcg, buf, sizeof(buf));
>  	pr_info("%s", buf);
> -	kfree(buf);
>  }
>  
>  /*
> @@ -6335,11 +6331,11 @@ static int memory_events_local_show(struct seq_file *m, void *v)
>  static int memory_stat_show(struct seq_file *m, void *v)
>  {
>  	struct mem_cgroup *memcg = mem_cgroup_from_seq(m);
> -	char *buf;
> +	char *buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>  
> -	buf = memory_stat_format(memcg);
>  	if (!buf)
>  		return -ENOMEM;
> +	memory_stat_format(memcg, buf, PAGE_SIZE);
>  	seq_puts(m, buf);
>  	kfree(buf);
>  	return 0;
> -- 
> 2.18.4
> 

-- 
Michal Hocko
SUSE Labs

[PATCH v3] mm: memcontrol: fix potential oom_lock recursion deadlock

[Tetsuo Handa]

syzbot is reporting GFP_KERNEL allocation with oom_lock held when reporting
memcg OOM [1]. Such allocation request might deadlock the system, for
__alloc_pages_may_oom() cannot invoke global OOM killer due to oom_lock
being already held by the caller.

Fix this problem by removing the allocation from memory_stat_format()
completely, and pass static buffer when calling from memcg OOM path.

Link: https://syzkaller.appspot.com/bug?extid=2d2aeadc6ce1e1f11d45 [1]
Reported-by: syzbot <syzbot+2d2aeadc6ce1e1f11d45@syzkaller.appspotmail.com>
Suggested-by: Michal Hocko <mhocko@suse.com>
Fixes: c8713d0b23123759 ("mm: memcontrol: dump memory.stat during cgroup OOM")
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Acked-by: Michal Hocko <mhocko@suse.com>
---
Changes in v3:
  Update patch description.

Changes in v2:
  Use static buffer for OOM reporting, suggested by Michal Hocko <mhocko@suse.com>.

 mm/memcontrol.c | 22 +++++++++-------------
 1 file changed, 9 insertions(+), 13 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 618c366a2f07..8092be2fbb7c 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1460,14 +1460,12 @@ static inline unsigned long memcg_page_state_output(struct mem_cgroup *memcg,
 	return memcg_page_state(memcg, item) * memcg_page_state_unit(item);
 }
 
-static char *memory_stat_format(struct mem_cgroup *memcg)
+static void memory_stat_format(struct mem_cgroup *memcg, char *buf, int bufsize)
 {
 	struct seq_buf s;
 	int i;
 
-	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_KERNEL), PAGE_SIZE);
-	if (!s.buffer)
-		return NULL;
+	seq_buf_init(&s, buf, bufsize);
 
 	/*
 	 * Provide statistics on the state of the memory subsystem as
@@ -1533,8 +1531,6 @@ static char *memory_stat_format(struct mem_cgroup *memcg)
 
 	/* The above should easily fit into one page */
 	WARN_ON_ONCE(seq_buf_has_overflowed(&s));
-
-	return s.buffer;
 }
 
 #define K(x) ((x) << (PAGE_SHIFT-10))
@@ -1570,7 +1566,10 @@ void mem_cgroup_print_oom_context(struct mem_cgroup *memcg, struct task_struct *
  */
 void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 {
-	char *buf;
+	/* Use static buffer, for the caller is holding oom_lock. */
+	static char buf[PAGE_SIZE];
+
+	lockdep_assert_held(&oom_lock);
 
 	pr_info("memory: usage %llukB, limit %llukB, failcnt %lu\n",
 		K((u64)page_counter_read(&memcg->memory)),
@@ -1591,11 +1590,8 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 	pr_info("Memory cgroup stats for ");
 	pr_cont_cgroup_path(memcg->css.cgroup);
 	pr_cont(":");
-	buf = memory_stat_format(memcg);
-	if (!buf)
-		return;
+	memory_stat_format(memcg, buf, sizeof(buf));
 	pr_info("%s", buf);
-	kfree(buf);
 }
 
 /*
@@ -6335,11 +6331,11 @@ static int memory_events_local_show(struct seq_file *m, void *v)
 static int memory_stat_show(struct seq_file *m, void *v)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_seq(m);
-	char *buf;
+	char *buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
 
-	buf = memory_stat_format(memcg);
 	if (!buf)
 		return -ENOMEM;
+	memory_stat_format(memcg, buf, PAGE_SIZE);
 	seq_puts(m, buf);
 	kfree(buf);
 	return 0;
-- 
2.18.4

Re: [PATCH v3] mm: memcontrol: fix potential oom_lock recursion deadlock

[Michal Hocko]

On Fri 22-07-22 19:45:39, Tetsuo Handa wrote:
> syzbot is reporting GFP_KERNEL allocation with oom_lock held when reporting
> memcg OOM [1]. Such allocation request might deadlock the system, for
> __alloc_pages_may_oom() cannot invoke global OOM killer due to oom_lock
> being already held by the caller.

OK, I have misunderstood your previous wording and now I have realized
that there are 2 issues here. One of them is a (less likely) dead lock on
the oom_lock not making a fwd progress (that would require global OOM
racing with memcg OOM)  and the other is the GFP_NOFS->GFP_KERNEL dependency
which can deadlock even without global the above race.
Correct?

Sorry I could have realized that sooner.

> Fix this problem by removing the allocation from memory_stat_format()
> completely, and pass static buffer when calling from memcg OOM path.
> 
> Link: https://syzkaller.appspot.com/bug?extid=2d2aeadc6ce1e1f11d45 [1]
> Reported-by: syzbot <syzbot+2d2aeadc6ce1e1f11d45@syzkaller.appspotmail.com>
> Suggested-by: Michal Hocko <mhocko@suse.com>
> Fixes: c8713d0b23123759 ("mm: memcontrol: dump memory.stat during cgroup OOM")
> Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> Acked-by: Michal Hocko <mhocko@suse.com>
> ---
> Changes in v3:
>   Update patch description.
> 
> Changes in v2:
>   Use static buffer for OOM reporting, suggested by Michal Hocko <mhocko@suse.com>.
> 
>  mm/memcontrol.c | 22 +++++++++-------------
>  1 file changed, 9 insertions(+), 13 deletions(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 618c366a2f07..8092be2fbb7c 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -1460,14 +1460,12 @@ static inline unsigned long memcg_page_state_output(struct mem_cgroup *memcg,
>  	return memcg_page_state(memcg, item) * memcg_page_state_unit(item);
>  }
>  
> -static char *memory_stat_format(struct mem_cgroup *memcg)
> +static void memory_stat_format(struct mem_cgroup *memcg, char *buf, int bufsize)
>  {
>  	struct seq_buf s;
>  	int i;
>  
> -	seq_buf_init(&s, kmalloc(PAGE_SIZE, GFP_KERNEL), PAGE_SIZE);
> -	if (!s.buffer)
> -		return NULL;
> +	seq_buf_init(&s, buf, bufsize);
>  
>  	/*
>  	 * Provide statistics on the state of the memory subsystem as
> @@ -1533,8 +1531,6 @@ static char *memory_stat_format(struct mem_cgroup *memcg)
>  
>  	/* The above should easily fit into one page */
>  	WARN_ON_ONCE(seq_buf_has_overflowed(&s));
> -
> -	return s.buffer;
>  }
>  
>  #define K(x) ((x) << (PAGE_SHIFT-10))
> @@ -1570,7 +1566,10 @@ void mem_cgroup_print_oom_context(struct mem_cgroup *memcg, struct task_struct *
>   */
>  void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
>  {
> -	char *buf;
> +	/* Use static buffer, for the caller is holding oom_lock. */
> +	static char buf[PAGE_SIZE];
> +
> +	lockdep_assert_held(&oom_lock);
>  
>  	pr_info("memory: usage %llukB, limit %llukB, failcnt %lu\n",
>  		K((u64)page_counter_read(&memcg->memory)),
> @@ -1591,11 +1590,8 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
>  	pr_info("Memory cgroup stats for ");
>  	pr_cont_cgroup_path(memcg->css.cgroup);
>  	pr_cont(":");
> -	buf = memory_stat_format(memcg);
> -	if (!buf)
> -		return;
> +	memory_stat_format(memcg, buf, sizeof(buf));
>  	pr_info("%s", buf);
> -	kfree(buf);
>  }
>  
>  /*
> @@ -6335,11 +6331,11 @@ static int memory_events_local_show(struct seq_file *m, void *v)
>  static int memory_stat_show(struct seq_file *m, void *v)
>  {
>  	struct mem_cgroup *memcg = mem_cgroup_from_seq(m);
> -	char *buf;
> +	char *buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>  
> -	buf = memory_stat_format(memcg);
>  	if (!buf)
>  		return -ENOMEM;
> +	memory_stat_format(memcg, buf, PAGE_SIZE);
>  	seq_puts(m, buf);
>  	kfree(buf);
>  	return 0;
> -- 
> 2.18.4

-- 
Michal Hocko
SUSE Labs

Re: [PATCH v3] mm: memcontrol: fix potential oom_lock recursion deadlock

[Tetsuo Handa]

On 2022/07/22 20:04, Michal Hocko wrote:
> On Fri 22-07-22 19:45:39, Tetsuo Handa wrote:
>> syzbot is reporting GFP_KERNEL allocation with oom_lock held when reporting
>> memcg OOM [1]. Such allocation request might deadlock the system, for
>> __alloc_pages_may_oom() cannot invoke global OOM killer due to oom_lock
>> being already held by the caller.
> 
> OK, I have misunderstood your previous wording and now I have realized
> that there are 2 issues here. One of them is a (less likely) dead lock on
> the oom_lock not making a fwd progress (that would require global OOM
> racing with memcg OOM)  and the other is the GFP_NOFS->GFP_KERNEL dependency
> which can deadlock even without global the above race.
> Correct?

Correct.

> 
> Sorry I could have realized that sooner.
> 

No problem. Thanks for suggestion.