[GIT PULL] VFIO fix for v5.8-rc7

[GIT PULL] VFIO fix for v5.8-rc7

[Alex Williamson]

Hi Linus,

The following changes since commit 11ba468877bb23f28956a35e896356252d63c983:

  Linux 5.8-rc5 (2020-07-12 16:34:50 -0700)

are available in the Git repository at:

  git://github.com/awilliam/linux-vfio.git tags/vfio-v5.8-rc7

for you to fetch changes up to b872d0640840018669032b20b6375a478ed1f923:

  vfio/pci: fix racy on error and request eventfd ctx (2020-07-17 08:28:40 -0600)

----------------------------------------------------------------
VFIO fixes for v5.8-rc7

 - Fix race with eventfd ctx cleared outside of mutex (Zeng Tao)

----------------------------------------------------------------
Zeng Tao (1):
      vfio/pci: fix racy on error and request eventfd ctx

 drivers/vfio/pci/vfio_pci.c | 5 +++++
 1 file changed, 5 insertions(+)

Re: [GIT PULL] VFIO fix for v5.8-rc7

[Linus Torvalds]

On Mon, Jul 20, 2020 at 7:34 AM Alex Williamson
<alex.williamson@redhat.com> wrote:
>
> VFIO fixes for v5.8-rc7
>
>  - Fix race with eventfd ctx cleared outside of mutex (Zeng Tao)

Why does this take and then re-take the lock immediately? That just
looks insane.

I realize that this isn't likely to be a performance-critical path,
but this is a basic source cleanliness issue. Doing silly things is
silly, and shouldn't be done, even if they don't matter.

              Linus

Re: [GIT PULL] VFIO fix for v5.8-rc7

[pr-tracker-bot]

The pull request you sent on Mon, 20 Jul 2020 08:34:27 -0600:

> git://github.com/awilliam/linux-vfio.git tags/vfio-v5.8-rc7

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/4fa640dc52302b5e62b01b05c755b055549633ae

Thank you!

-- 
Deet-doot-dot, I am a bot.
https://korg.wiki.kernel.org/userdoc/prtracker

Re: [GIT PULL] VFIO fix for v5.8-rc7

[Alex Williamson]

On Mon, 20 Jul 2020 13:33:29 -0700
Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Mon, Jul 20, 2020 at 7:34 AM Alex Williamson
> <alex.williamson@redhat.com> wrote:
> >
> > VFIO fixes for v5.8-rc7
> >
> >  - Fix race with eventfd ctx cleared outside of mutex (Zeng Tao)  
> 
> Why does this take and then re-take the lock immediately? That just
> looks insane.
> 
> I realize that this isn't likely to be a performance-critical path,
> but this is a basic source cleanliness issue. Doing silly things is
> silly, and shouldn't be done, even if they don't matter.

Yup, it's silly.  Cornelia identified the same during review and I let
it slide after some counter arguments by the submitter.  I see this got
merged, so I'll post a cleanup for v5.9.  Thanks for the input,

Alex