* bcachefs-tools: O_DIRECT necessary?
@ 2022-10-31 1:18 Thomas Weißschuh
2022-10-31 1:41 ` Kent Overstreet
0 siblings, 1 reply; 3+ messages in thread
From: Thomas Weißschuh @ 2022-10-31 1:18 UTC (permalink / raw)
To: linux-bcache
Hi all,
I just tried to run the unittests of bcachefs-tools and they are failing for
me.
The culprit is that mkfs.bcachefs tries to open the disk image with O_DIRECT
which is not allowed on tmpfs.
Is O_DIRECT really necessary for mkfs? It does not seem necessary for other
filesystems.
Thomas
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: bcachefs-tools: O_DIRECT necessary?
2022-10-31 1:18 bcachefs-tools: O_DIRECT necessary? Thomas Weißschuh
@ 2022-10-31 1:41 ` Kent Overstreet
2022-10-31 3:10 ` Thomas Weißschuh
0 siblings, 1 reply; 3+ messages in thread
From: Kent Overstreet @ 2022-10-31 1:41 UTC (permalink / raw)
To: Thomas Weißschuh; +Cc: linux-bcache, linux-bcachefs
On Mon, Oct 31, 2022 at 02:18:40AM +0100, Thomas Weißschuh wrote:
> Hi all,
>
> I just tried to run the unittests of bcachefs-tools and they are failing for
> me.
> The culprit is that mkfs.bcachefs tries to open the disk image with O_DIRECT
> which is not allowed on tmpfs.
>
> Is O_DIRECT really necessary for mkfs? It does not seem necessary for other
> filesystems.
Hey - the proper mailing list for bcachefs is linux-bcachefs@vger.kernel.org now
:)
It's not strictly necessary, we use O_DIRECT because we're emulating the kernel
bio interface. There are other situations where this has been a problem though,
we need to add either a flag to use buffered IO or preferably a way to
automatically fall back to buffered IO.
Could you open a bug for this? I'll try to get to it in the near future
https://github.com/koverstreet/bcachefs/issues/
Cheers,
Kent
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: bcachefs-tools: O_DIRECT necessary?
2022-10-31 1:41 ` Kent Overstreet
@ 2022-10-31 3:10 ` Thomas Weißschuh
0 siblings, 0 replies; 3+ messages in thread
From: Thomas Weißschuh @ 2022-10-31 3:10 UTC (permalink / raw)
To: Kent Overstreet; +Cc: linux-bcache, linux-bcachefs
On 2022-10-30 21:41-0400, Kent Overstreet wrote:
> On Mon, Oct 31, 2022 at 02:18:40AM +0100, Thomas Weißschuh wrote:
> > Hi all,
> >
> > I just tried to run the unittests of bcachefs-tools and they are failing for
> > me.
> > The culprit is that mkfs.bcachefs tries to open the disk image with O_DIRECT
> > which is not allowed on tmpfs.
> >
> > Is O_DIRECT really necessary for mkfs? It does not seem necessary for other
> > filesystems.
>
> Hey - the proper mailing list for bcachefs is linux-bcachefs@vger.kernel.org now
> :)
My bad, I copied the wrong list from one of your mails.
> It's not strictly necessary, we use O_DIRECT because we're emulating the kernel
> bio interface. There are other situations where this has been a problem though,
> we need to add either a flag to use buffered IO or preferably a way to
> automatically fall back to buffered IO.
>
> Could you open a bug for this? I'll try to get to it in the near future
>
> https://github.com/koverstreet/bcachefs/issues/
I opened it in the tools repo because those are broken:
https://github.com/koverstreet/bcachefs-tools/issues/132
Thomas
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2022-10-31 3:10 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-31 1:18 bcachefs-tools: O_DIRECT necessary? Thomas Weißschuh
2022-10-31 1:41 ` Kent Overstreet
2022-10-31 3:10 ` Thomas Weißschuh
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).