Re: [PATCH 5.15] fuse: Fix race condition in writethrough path A race

[Bernd Schubert]

On 10/14/25 18:10, Miklos Szeredi wrote:
> On Tue, 14 Oct 2025 at 15:57, Brian Foster <bfoster@xxxxxxxxxx> wrote:
> 
>> But TBH, if the writeback thing or something similarly simple works for
>> resolving the immediate bug, I wouldnt worry too much about it
>> until/unless there are userspace fs' explicitly looking for that sort of
>> behavior. Just my .02.
> 
> Agreed.
> 
> I just feel it unfortunate that this is default in libfuse and so many
> filesystems will have auto_inval_data enabled which don't even need
> it, and some mixed read-write workloads suffering badly as a
> consequence.

Maybe we should disable it in libfuse-3.18 and hope for reports, in case
there are regressions?


Thanks,
Bernd