[GIT PULL] xfs: fixes for 5.18-rc5

[GIT PULL] xfs: fixes for 5.18-rc5

[Dave Chinner]

Hi Linus,

Can you please pull the XFS fixes in the request attached below?
It's all minor stuff;  a 5.18 build regession fix, a deadlock fix
and an update to remove redundant fields from the XFS entry in the
MAINTAINERS file.

I just did a test merge against your current master branch and
didn't see any unexpected issues. It's been a while since I sent you
one of these things, so there's every chance I've missed something
I'm supposed to do now. If there's anything you notice that is even
slightly dodgy please let me know and I'll make sure it doesn't
happen next time.

-Dave.

The following changes since commit ce522ba9ef7e2d9fb22a39eb3371c0c64e2a433e:

  Linux 5.18-rc2 (2022-04-10 14:21:36 -1000)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/fs/xfs/xfs-linux.git tags/xfs-5.18-fixes-1

for you to fetch changes up to 9a5280b312e2e7898b6397b2ca3cfd03f67d7be1:

  xfs: reorder iunlink remove operation in xfs_ifree (2022-04-21 08:45:16 +1000)

----------------------------------------------------------------
xfs: fixes for v5.18-rc5

XFS changes for v5.18-rc5:
- define buffer bit flags as unsigned to fix gcc-5 + c11 warnings
- remove redundant XFS fields from MAINTAINERS
- fix inode buffer locking order regression

----------------------------------------------------------------
Dave Chinner (2):
      xfs: convert buffer flags to unsigned.
      xfs: reorder iunlink remove operation in xfs_ifree

Tiezhu Yang (1):
      MAINTAINERS: update IOMAP FILESYSTEM LIBRARY and XFS FILESYSTEM

 MAINTAINERS        |  3 ---
 fs/xfs/xfs_buf.c   |  6 +++---
 fs/xfs/xfs_buf.h   | 42 +++++++++++++++++++++---------------------
 fs/xfs/xfs_inode.c | 24 +++++++++++++-----------
 fs/xfs/xfs_trans.h |  2 +-
 5 files changed, 38 insertions(+), 39 deletions(-)

-- 
Dave Chinner
david@fromorbit.com

Re: [GIT PULL] xfs: fixes for 5.18-rc5

[Linus Torvalds]

On Wed, Apr 27, 2022 at 11:19 PM Dave Chinner <david@fromorbit.com> wrote:
>
> It's all minor stuff;  a 5.18 build regession fix, a deadlock fix
> and an update to remove redundant fields from the XFS entry in the
> MAINTAINERS file.

Done.

I had to look twice at that patch going "why does it complain about
the xfs flags thing" until I realized it was due to XBF_UNMAPPED being
(1 <<31) and the compiler then seeing explicit negative values being
assigned.

We have a lot of "int flags" in various places, very much not just
xfs, and yeah, we should probably try to clean them up.

But 99% of the time it's just not worth the noise.

And xfs only hit it because you guys literally used up the whole 'int'
for it, normally the compiler won't make a peep about it.

Thanks,
                     Linus

Re: [GIT PULL] xfs: fixes for 5.18-rc5

[pr-tracker-bot]

The pull request you sent on Thu, 28 Apr 2022 16:19:21 +1000:

> git://git.kernel.org/pub/scm/fs/xfs/xfs-linux.git tags/xfs-5.18-fixes-1

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/8061e16e203f36e7a5990535760ecb2e60a365f9

Thank you!

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

Re: [GIT PULL] xfs: fixes for 5.18-rc5

[Dave Chinner]

On Thu, Apr 28, 2022 at 09:46:14AM -0700, Linus Torvalds wrote:
> On Wed, Apr 27, 2022 at 11:19 PM Dave Chinner <david@fromorbit.com> wrote:
> >
> > It's all minor stuff;  a 5.18 build regession fix, a deadlock fix
> > and an update to remove redundant fields from the XFS entry in the
> > MAINTAINERS file.
> 
> Done.

Thanks!

> I had to look twice at that patch going "why does it complain about
> the xfs flags thing" until I realized it was due to XBF_UNMAPPED being
> (1 <<31) and the compiler then seeing explicit negative values being
> assigned.
> 
> We have a lot of "int flags" in various places, very much not just
> xfs, and yeah, we should probably try to clean them up.

Yup, I've already got a bunch of similar cleanup patches for
XFS ready for the next merge window.

> But 99% of the time it's just not worth the noise.

Mostly I agree, but in coverting another dozen+ sets of flags the
biggest benefit I've seen comes from the bit flag definitions all
having a consistent format across the code base.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com