Linux 4.9 compat: fix zfs_ctldir xattr handling

[loli10K]

Description

Since torvalds/linux@d0a5b99 IOP_XATTR is used to indicate the inode has xattr support: we should clear it for the ctldir inodes to avoid IO errors.

Motivation and Context

On Linux 4.9+ ls -la on the ".zfs" directory produces IO errors:

root@linux:~# uname -r
4.9.0
root@linux:~# POOLNAME='testpool'
root@linux:~# TMPDIR='/var/tmp'
root@linux:~# mountpoint -q $TMPDIR || mount -t tmpfs tmpfs $TMPDIR
root@linux:~# zpool destroy $POOLNAME
root@linux:~# rm -f $TMPDIR/zpool.dat
root@linux:~# fallocate -l 128m $TMPDIR/zpool.dat
root@linux:~# zpool create -O mountpoint=/tmp/$POOLNAME $POOLNAME $TMPDIR/zpool.dat
root@linux:~# zfs create $POOLNAME/fs1
root@linux:~# zfs snap $POOLNAME@snap1
root@linux:~# ls -lah /tmp/$POOLNAME/.zfs/
ls: /tmp/testpool/.zfs/: Input/output error
ls: /tmp/testpool/.zfs/.: Input/output error
ls: /tmp/testpool/.zfs/snapshot: Input/output error
ls: /tmp/testpool/.zfs/shares: Input/output error
total 512
drwxrwxrwx 1 root root 0 Jun  3 11:56 .
drwxr-xr-x 3 root root 3 Jun  3 11:56 ..
drwxrwxrwx 2 root root 2 Jun  3 11:56 shares
drwxrwxrwx 2 root root 2 Jun  3 11:56 snapshot
root@linux:~# strace -e trace=lgetxattr ls -lah /tmp/$POOLNAME/.zfs/
lgetxattr("/tmp/testpool/.zfs/", "security.selinux", 0x249dd90, 255) = -1 EIO (Input/output error)
ls: /tmp/testpool/.zfs/: Input/output error
lgetxattr("/tmp/testpool/.zfs/.", "security.selinux", 0x24a8aa0, 255) = -1 EIO (Input/output error)
ls: /tmp/testpool/.zfs/.: Input/output error
lgetxattr("/tmp/testpool/.zfs/..", "security.selinux", 0x24a8ae0, 255) = -1 ENODATA (No data available)
lgetxattr("/tmp/testpool/.zfs/snapshot", "security.selinux", 0x24a8b00, 255) = -1 EIO (Input/output error)
ls: /tmp/testpool/.zfs/snapshot: Input/output error
lgetxattr("/tmp/testpool/.zfs/shares", "security.selinux", 0x24a8b20, 255) = -1 EIO (Input/output error)
ls: /tmp/testpool/.zfs/shares: Input/output error
total 512
drwxrwxrwx 1 root root 0 Jun  3 11:56 .
drwxr-xr-x 3 root root 3 Jun  3 11:56 ..
drwxrwxrwx 2 root root 2 Jun  3 11:56 shares
drwxrwxrwx 2 root root 2 Jun  3 11:56 snapshot
+++ exited with 0 +++
root@linux:~# 

How Has This Been Tested?

Manual testing. We could also add a new test to the ZTS in "functional/xattr".

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Performance enhancement (non-breaking change which improves efficiency)
• Code cleanup (non-breaking change which makes code smaller or more readable)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

• My code follows the ZFS on Linux code style requirements.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• I have added tests to cover my changes.
• All new and existing tests passed.
• All commit messages are properly formatted and contain Signed-off-by.
• Change has been approved by a ZFS on Linux member.

[mention-bot]

@loli10K, thanks for your PR! By analyzing the history of the files in this pull request, we identified @behlendorf, @tuxoko and @ahrens to be potential reviewers.

[behlendorf]

Nice find, yes we should definitely do this but I'd suggest adding this same logic a little differently. The following has a few advantages.

• Checking for IOP_XATTR instead of HAVE_GENERIC_SETXATTR ensures we're depending on exactly this interface change. That way in the unlikely case this change gets cherry picked in to an older kernel this will still work correctly.

• Using &= ~IOP_XATTR just clears the one bit we care about.

#if defined(IOP_XATTR)
        ip->i_opflags &= ~IOP_XATTR;
#endif

[behlendorf]

Thanks, looks good. And the test case failures are unrelated.