Re: ext4 lockdep trace (3.1.0rc3)

[Tao Ma <tm@tao.ma>]

Hi Dave,
On 08/30/2011 04:54 AM, Dave Jones wrote:
> On Mon, Aug 29, 2011 at 04:48:30PM -0400, Dave Jones wrote:
>  > On Fri, Aug 26, 2011 at 05:49:30PM -0400, Dave Jones wrote:
>  >  > just hit this while building a kernel.  Laptop wedged for a few seconds
>  >  > during the final link, and this was in the log when it unwedged.
>  >  
>  > I still see this in rc4, and can reproduce it reliably every time I build.
>  > It only started happening in the last week.  I don't see any ext4 or vfs commits
>  > within a few days of that, so I'm not sure why it only just begun
>  > (I do daily builds, and the 26th was the first time I saw it appear)
>  > 
>  > Given the lack of obvious commits in that timeframe, I'm not sure a bisect is
>  > going to be particularly fruitful. It might just be that my IO patterns changed ?
> 
> also a second variant with a different trace.
Ted has sent out a fix in the ext4 mailist. Please search for the subject:
[URGENT PATCH] ext4: fix potential deadlock in ext4_evict_inode()

Thanks
Tao
> 
> 	Dave
> 
> 
>  =======================================================
>  [ INFO: possible circular locking dependency detected ]
>  3.1.0-rc3+ #151
>  -------------------------------------------------------
>  gnome-settings-/2037 is trying to acquire lock:
>   (&sb->s_type->i_mutex_key#14){+.+.+.}, at: [<ffffffff811cee83>] ext4_evict_inode+0x76/0x33c
>  
>  but task is already holding lock:
>   (&mm->mmap_sem){++++++}, at: [<ffffffff8111ba13>] sys_munmap+0x3b/0x60
>  
>  which lock already depends on the new lock.
>  
>  
>  the existing dependency chain (in reverse order) is:
>  
>  -> #1 (&mm->mmap_sem){++++++}:
>         [<ffffffff81091251>] lock_acquire+0xf3/0x13e
>         [<ffffffff8111499a>] might_fault+0x80/0xa3
>         [<ffffffff811570ef>] filldir+0x6f/0xc7
>         [<ffffffff811c5c3f>] call_filldir+0x96/0xc0
>         [<ffffffff811c5f78>] ext4_readdir+0x1bd/0x548
>         [<ffffffff81157344>] vfs_readdir+0x7b/0xb4
>         [<ffffffff81157463>] sys_getdents+0x7e/0xd1
>         [<ffffffff814f2342>] system_call_fastpath+0x16/0x1b
>  
>  -> #0 (&sb->s_type->i_mutex_key#14){+.+.+.}:
>         [<ffffffff81090a71>] __lock_acquire+0xa2f/0xd0c
>         [<ffffffff81091251>] lock_acquire+0xf3/0x13e
>         [<ffffffff814ea40d>] __mutex_lock_common+0x65/0x44a
>         [<ffffffff814ea8ed>] mutex_lock_nested+0x3b/0x40
>         [<ffffffff811cee83>] ext4_evict_inode+0x76/0x33c
>         [<ffffffff8115d2a5>] evict+0x98/0x152
>         [<ffffffff8115d4f0>] iput+0x191/0x199
>         [<ffffffff8115a0f2>] dentry_kill+0x123/0x145
>         [<ffffffff8115a4fc>] dput+0xf2/0x102
>         [<ffffffff8114948a>] fput+0x1d8/0x1f0
>         [<ffffffff8111a6bd>] remove_vma+0x51/0x82
>         [<ffffffff8111b8a0>] do_munmap+0x2f2/0x30b
>         [<ffffffff8111ba21>] sys_munmap+0x49/0x60
>         [<ffffffff814f2342>] system_call_fastpath+0x16/0x1b
>  
>  other info that might help us debug this:
>  
>   Possible unsafe locking scenario:
>  
>         CPU0                    CPU1
>         ----                    ----
>    lock(&mm->mmap_sem);
>                                 lock(&sb->s_type->i_mutex_key);
>                                 lock(&mm->mmap_sem);
>    lock(&sb->s_type->i_mutex_key);
>  
>   *** DEADLOCK ***
>  
>  1 lock held by gnome-settings-/2037:
>   #0:  (&mm->mmap_sem){++++++}, at: [<ffffffff8111ba13>] sys_munmap+0x3b/0x60
>  
>  stack backtrace:
>  Pid: 2037, comm: gnome-settings- Not tainted 3.1.0-rc3+ #151
>  Call Trace:
>   [<ffffffff810811d1>] ? up+0x39/0x3e
>   [<ffffffff814e1469>] print_circular_bug+0x1f8/0x209
>   [<ffffffff81090a71>] __lock_acquire+0xa2f/0xd0c
>   [<ffffffff810821ef>] ? local_clock+0x35/0x4c
>   [<ffffffff8108fe4f>] ? mark_lock+0x2d/0x220
>   [<ffffffff811cee83>] ? ext4_evict_inode+0x76/0x33c
>   [<ffffffff81091251>] lock_acquire+0xf3/0x13e
>   [<ffffffff811cee83>] ? ext4_evict_inode+0x76/0x33c
>   [<ffffffff814ea3e5>] ? __mutex_lock_common+0x3d/0x44a
>   [<ffffffff814ea8ed>] ? mutex_lock_nested+0x3b/0x40
>   [<ffffffff811cee83>] ? ext4_evict_inode+0x76/0x33c
>   [<ffffffff814ea40d>] __mutex_lock_common+0x65/0x44a
>   [<ffffffff811cee83>] ? ext4_evict_inode+0x76/0x33c
>   [<ffffffff810821ef>] ? local_clock+0x35/0x4c
>   [<ffffffff8115d297>] ? evict+0x8a/0x152
>   [<ffffffff8108d9ba>] ? put_lock_stats+0xe/0x29
>   [<ffffffff8108e03e>] ? lock_release_holdtime.part.10+0x59/0x62
>   [<ffffffff8115d297>] ? evict+0x8a/0x152
>   [<ffffffff814ea8ed>] mutex_lock_nested+0x3b/0x40
>   [<ffffffff811cee83>] ext4_evict_inode+0x76/0x33c
>   [<ffffffff8115d2a5>] evict+0x98/0x152
>   [<ffffffff8115d4f0>] iput+0x191/0x199
>   [<ffffffff8115a0f2>] dentry_kill+0x123/0x145
>   [<ffffffff8115a4fc>] dput+0xf2/0x102
>   [<ffffffff8114948a>] fput+0x1d8/0x1f0
>   [<ffffffff8111a6bd>] remove_vma+0x51/0x82
>   [<ffffffff8111b8a0>] do_munmap+0x2f2/0x30b
>   [<ffffffff8111ba21>] sys_munmap+0x49/0x60
>   [<ffffffff814f2342>] system_call_fastpath+0x16/0x1b
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/