Re: [bug report] Potential atomicity bug in fs/btrfs/compression.c, between btrfs_put_workspace() and free_workspace_manager()

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2026/4/23 14:25, Ginger 写道:
> Dear Linux kernel maintainers,
> 
> My research-based static analyzer found a potential atomicity bug
> within the 'fs/btrfs' subsystem, more specifically, in
> 'fs/btrfs/compression.c'.

I guess your "research-based static analyzer" is some pretty bad LLM.

> 
> Kernel version: long-term kernel v6.18.9
> 
> Potential concurrent triggering executions:
> T0:
> btrfs_put_workspace
>       --> ws_lock = fs_info->compr_wsm[type]->ws_lock;
>       --> spin_lock(ws_lock)
>       --> list_add(ws, gwsm->idl_ws)
>       --> spin_unlock(ws_lock)
> 
> T1:
> free_workspace_manager
>      --> gwsm = fs_info->compr_wsm[type];
>      -->  while(!list_empty(gwsm->idle_ws))

Go check when free_workspace_manager() is called.
At that time there should be no compression workload at all, because 
that's when the super block of the fs is being removed.

> 
> In T1, the accesses to the idle_ws are plain memory reads and writes,
> without holding the 'ws_lock' to protect against potential data races.
> 
> Please kindly check at your convenience. Thank you for your time and
> consideration.

Please kindly switch to a better LLM.

> 
> Best regards,
> Ginger
>