Re: [PATCH] xfs: Remove i_rwsem lock in buffered read

[John Garry]

On 09/01/2025 10:07, Amir Goldstein wrote:
> > Please note that IOCB_ATOMIC is not supported for buffered IO, so we
> > can't do this - we only support direct IO today.
> Oops. I see now.
>
> > And supporting buffered IO has its challenges; how to handle overlapping
> > atomic writes of differing sizes sitting in the page cache is the main
> > issue which comes to mind.
> How about the combination of RWF_ATOMIC | RWF_UNCACHED [1]
> Would it be easier/possible to support this considering that the write of folio
> is started before the write system call returns?

I am not sure exactly what you are proposing. Is it that RWF_ATOMIC for 
buffered IO auto-sets RWF_UNCACHED? Or that RWF_ATOMIC requires 
RWF_UNCACHED to be set?

But that is not so important, as I just think that future users of 
RWF_ATOMIC may not want the behavior of RWF_UNCACHED always (for 
buffered IO).

And I don't think that RWF_UNCACHED even properly solves the issues of 
RWF_ATOMIC for buffered IO in terms of handling overlapping atomic 
writes in the page cache.

Thanks,
John

> Note that application that desires mutithreaded atomicity of writes vs. reads
> will only need to opt-in for RWF_ATOMIC | RWF_UNCACHED writes, so this
> is not expected to actually break its performance by killing the read caching.
>
> Thanks,
> Amir.
>
> [1]https://urldefense.com/v3/__https://lore.kernel.org/linux- 
> fsdevel/20241220154831.1086649-1-axboe@xxxxxxxxx/__;!!ACWV5N9M2RV99hQ! 
> J7_5N_kSixl5iSy8IX37Cup3uKTHAaC5Oy-RlvsJeTE2kr3iJ2IXNww_rApK7TwI_ocCBSE- 
> G2vZSKSRHqY$