Re: Regression on linux-next (next-20240712)

[David Hildenbrand <david@redhat.com>]

On 16.07.24 07:37, Borah, Chaitanya Kumar wrote:
> Hello Pei,
> 
> Hope you are doing well. I am Chaitanya from the linux graphics team in Intel.
> 
> This mail is regarding a regression we are seeing in our CI runs[1] on linux-next repository.
> 
> In version next-20240712[2], we saw the following regression (currently being masked by another regression)
> 
> `````````````````````````````````````````````````````````````````````````````````
> <4>[   14.530533] ============================================
> <4>[   14.530533] WARNING: possible recursive locking detected
> <4>[   14.530534] 6.10.0-rc7-next-20240712-next-20240712-g3fe121b62282+ #1 Not tainted
> <4>[   14.530535] --------------------------------------------
> <4>[   14.530535] (direxec)/171 is trying to acquire lock:
> <4>[   14.530536] ffff8881010725d8 (&mm->mmap_lock){++++}-{3:3}, at: unmap_single_vma+0xea/0x170
> <4>[   14.530541]
>                    but task is already holding lock:
> <4>[   14.530542] ffff8881010725d8 (&mm->mmap_lock){++++}-{3:3}, at: exit_mmap+0x6a/0x450
> <4>[   14.530545]
>                    other info that might help us debug this:
> <4>[   14.530545]  Possible unsafe locking scenario:
> `````````````````````````````````````````````````````````````````````````````````
> Details log can be found in [3].
> 
> After bisecting the tree, the following patch [4] seems to be the first "bad"
> commit
> 
> `````````````````````````````````````````````````````````````````````````````````````````````````````````
> commit a13252049629a8225f38a9be7d8d4fc4ff5350e8
> Author: Pei Li mailto:peili.dev@gmail.com
> Date:   Wed Jul 10 22:13:17 2024 -0700
> 
>      mm: fix mmap_assert_locked() in follow_pte()
> 
> `````````````````````````````````````````````````````````````````````````````````````````````````````````
> 
> We also verified that if we revert the patch the issue is not seen.
> 
> Could you please check why the patch causes this regression and provide a fix if necessary?

This is know.

There is a discussion along the original patch [1] on how to do it 
differently. But likely we'll tackle it differently [2]. So this patch 
should be dropped for -- which I think already happened because I cannot 
spot that patch in mm-unstable anymore.

[1] https://lore.kernel.org/all/20240710-bug12-v1-1-0e5440f9b8d3@gmail.com/

[2] https://lkml.kernel.org/r/20240712144244.3090089-1-peterx@redhat.com

-- 
Cheers,

David / dhildenb