Re: [PATCH] memcg: mem_cgroup_migrate() may be called with irq disabled

[Christian Borntraeger]

On 06/20/2016 08:41 PM, Tejun Heo wrote:
> Hello,
> 
> Christian, I *think* this should fix it.  Can you please verify?

I cannot reliably reproduce the bug :-/,but at least I have not seen
it with this patch and the patch makes sense and matches the traces.

> 
> Thanks!
> ------ 8< ------
> mem_cgroup_migrate() uses local_irq_disable/enable() but can be called
> with irq disabled from migrate_page_copy().  This ends up enabling irq
> while holding a irq context lock triggering the following lockdep
> warning.  Fix it by using irq_save/restore instead.
> 
>   =================================
>   [ INFO: inconsistent lock state ]
>   4.7.0-rc1+ #52 Tainted: G        W      
>   ---------------------------------
>   inconsistent {IN-SOFTIRQ-W} -> {SOFTIRQ-ON-W} usage.
>   kcompactd0/151 [HC0[0]:SC0[0]:HE1:SE1] takes:
>    (&(&ctx->completion_lock)->rlock){+.?.-.}, at: [<000000000038fd96>] aio_migratepage+0x156/0x1e8
>   {IN-SOFTIRQ-W} state was registered at:
>     [<00000000001a8366>] __lock_acquire+0x5b6/0x1930
>     [<00000000001a9b9e>] lock_acquire+0xee/0x270
>     [<0000000000951fee>] _raw_spin_lock_irqsave+0x66/0xb0
>     [<0000000000390108>] aio_complete+0x98/0x328
>     [<000000000037c7d4>] dio_complete+0xe4/0x1e0
>     [<0000000000650e64>] blk_update_request+0xd4/0x450
>     [<000000000072a1a8>] scsi_end_request+0x48/0x1c8
>     [<000000000072d7e2>] scsi_io_completion+0x272/0x698
>     [<000000000065adb2>] blk_done_softirq+0xca/0xe8
>     [<0000000000953f80>] __do_softirq+0xc8/0x518
>     [<00000000001495de>] irq_exit+0xee/0x110
>     [<000000000010ceba>] do_IRQ+0x6a/0x88
>     [<000000000095342e>] io_int_handler+0x11a/0x25c
>     [<000000000094fb5c>] __mutex_unlock_slowpath+0x144/0x1d8
>     [<000000000094fb58>] __mutex_unlock_slowpath+0x140/0x1d8
>     [<00000000003c6114>] kernfs_iop_permission+0x64/0x80
>     [<000000000033ba86>] __inode_permission+0x9e/0xf0
>     [<000000000033ea96>] link_path_walk+0x6e/0x510
>     [<000000000033f09c>] path_lookupat+0xc4/0x1a8
>     [<000000000034195c>] filename_lookup+0x9c/0x160
>     [<0000000000341b44>] user_path_at_empty+0x5c/0x70
>     [<0000000000335250>] SyS_readlinkat+0x68/0x140
>     [<0000000000952f8e>] system_call+0xd6/0x270
>   irq event stamp: 971410
>   hardirqs last  enabled at (971409): [<000000000030f982>] migrate_page_move_mapping+0x3ea/0x588
>   hardirqs last disabled at (971410): [<0000000000951fc4>] _raw_spin_lock_irqsave+0x3c/0xb0
>   softirqs last  enabled at (970526): [<0000000000954318>] __do_softirq+0x460/0x518
>   softirqs last disabled at (970519): [<00000000001495de>] irq_exit+0xee/0x110
> 
>   other info that might help us debug this:
>    Possible unsafe locking scenario:
> 
> 	 CPU0
> 	 ----
>     lock(&(&ctx->completion_lock)->rlock);
>     <Interrupt>
>       lock(&(&ctx->completion_lock)->rlock);
> 
>     *** DEADLOCK ***
> 
>   3 locks held by kcompactd0/151:
>    #0:  (&(&mapping->private_lock)->rlock){+.+.-.}, at: [<000000000038fc82>] aio_migratepage+0x42/0x1e8
>    #1:  (&ctx->ring_lock){+.+.+.}, at: [<000000000038fc9a>] aio_migratepage+0x5a/0x1e8
>    #2:  (&(&ctx->completion_lock)->rlock){+.?.-.}, at: [<000000000038fd96>] aio_migratepage+0x156/0x1e8
> 
>   stack backtrace:
>   CPU: 20 PID: 151 Comm: kcompactd0 Tainted: G        W       4.7.0-rc1+ #52
> 	 00000001c6cbb730 00000001c6cbb7c0 0000000000000002 0000000000000000 
> 	 00000001c6cbb860 00000001c6cbb7d8 00000001c6cbb7d8 0000000000114496 
> 	 0000000000000000 0000000000b517ec 0000000000b680b6 000000000000000b 
> 	 00000001c6cbb820 00000001c6cbb7c0 0000000000000000 0000000000000000 
> 	 040000000184ad18 0000000000114496 00000001c6cbb7c0 00000001c6cbb820 
>   Call Trace:
>   ([<00000000001143d2>] show_trace+0xea/0xf0)
>   ([<000000000011444a>] show_stack+0x72/0xf0)
>   ([<0000000000684522>] dump_stack+0x9a/0xd8)
>   ([<000000000028679c>] print_usage_bug.part.27+0x2d4/0x2e8)
>   ([<00000000001a71ce>] mark_lock+0x17e/0x758)
>   ([<00000000001a784a>] mark_held_locks+0xa2/0xd0)
>   ([<00000000001a79b8>] trace_hardirqs_on_caller+0x140/0x1c0)
>   ([<0000000000326026>] mem_cgroup_migrate+0x266/0x370)
>   ([<000000000038fdaa>] aio_migratepage+0x16a/0x1e8)
>   ([<0000000000310568>] move_to_new_page+0xb0/0x260)
>   ([<00000000003111b4>] migrate_pages+0x8f4/0x9f0)
>   ([<00000000002c507c>] compact_zone+0x4dc/0xdc8)
>   ([<00000000002c5e22>] kcompactd_do_work+0x1aa/0x358)
>   ([<00000000002c608a>] kcompactd+0xba/0x2c8)
>   ([<000000000016b09a>] kthread+0x10a/0x110)
>   ([<000000000095315a>] kernel_thread_starter+0x6/0xc)
>   ([<0000000000953154>] kernel_thread_starter+0x0/0xc)
>   INFO: lockdep is turned off.
> 
> Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
> Reported-by: Christian Borntraeger <borntraeger@xxxxxxxxxx>
> Link: http://lkml.kernel.org/g/5767CFE5.7080904@xxxxxxxxxx
> ---
>  mm/memcontrol.c |    5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 75e7440..d4a33e1 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -5544,6 +5544,7 @@ void mem_cgroup_migrate(struct page *oldpage, struct page *newpage)
>  	struct mem_cgroup *memcg;
>  	unsigned int nr_pages;
>  	bool compound;
> +	unsigned long flags;
> 
>  	VM_BUG_ON_PAGE(!PageLocked(oldpage), oldpage);
>  	VM_BUG_ON_PAGE(!PageLocked(newpage), newpage);
> @@ -5574,10 +5575,10 @@ void mem_cgroup_migrate(struct page *oldpage, struct page *newpage)
> 
>  	commit_charge(newpage, memcg, false);
> 
> -	local_irq_disable();
> +	local_irq_save(flags);
>  	mem_cgroup_charge_statistics(memcg, newpage, compound, nr_pages);
>  	memcg_check_events(memcg, newpage);
> -	local_irq_enable();
> +	local_irq_restore(flags);
>  }
> 
>  DEFINE_STATIC_KEY_FALSE(memcg_sockets_enabled_key);
>