Re: [PATCH RFC] block: enable RWF_DONTCACHE for block devices

[Tal Zussman]

On Thu, Feb 19, 2026 at 10:24 AM Jens Axboe <axboe@xxxxxxxxx> wrote:
> On 2/18/26 2:13 PM, Tal Zussman wrote:
> > Block device buffered reads and writes already pass through
> > filemap_read() and iomap_file_buffered_write() respectively, both of
> > which handle IOCB_DONTCACHE. Enable RWF_DONTCACHE for block device files
> > by setting FOP_DONTCACHE in def_blk_fops.
> >
> > For CONFIG_BUFFER_HEAD paths, thread the kiocb through
> > block_write_begin() so that buffer_head-based I/O can use DONTCACHE
> > behavior as well. Callers without a kiocb context (e.g. nilfs2 recovery)
> > pass NULL, which preserves the existing behavior.
> >
> > This support is useful for databases that operate on raw block devices,
> > among other userspace applications.
>
> OOO right now so I'll take a real look when I'm back, but when I
> originally did this work, it's not the issue side that's the issue. It's
> the pruning done from completion context, and you need to ensure that's
> sane context for that (non-irq).

Thanks for taking a look! That was very helpful feedback.
I sent out a v2 hopefully addressing that here:
https://lore.kernel.org/lkml/20260225-blk-dontcache-v2-0-70e7ac4f7108@xxxxxxxxxxxx/

> --
> Jens Axboe