[Bug 203201] on zfs extattr behaviour broken after unlink

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=203201

            Bug ID: 203201
           Summary: on zfs extattr behaviour broken after unlink
           Product: Base System
           Version: 11.0-CURRENT
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Some People
          Priority: ---
         Component: kern
          Assignee: freebsd-bugs at FreeBSD.org
          Reporter: justin at specialbusservice.com

On ZFS root on
FreeBSD freebsd11 11.0-CURRENT FreeBSD 11.0-CURRENT #0 r285794: Wed Jul 22
16:58:53 UTC 2015     root at releng2.nyi.freebsd.org:/usr/obj/usr/src/sys/GENERIC
 amd64

If I unlink a file, extattr_get_fd does not work correctly as can be seen from
this ktrace:

  7638 luajit-2.0.4 CALL 
openat(AT_FDCWD,0x107550,0x601<O_WRONLY|O_CREAT|O_TRUNC>,0x1c0<S_IRUSR|S_IWUSR|S_IXUSR>)
  7638 luajit-2.0.4 NAMI  "XXXXYYYYZZZ45217638"
  7638 luajit-2.0.4 RET   openat 3
  7638 luajit-2.0.4 CALL  unlink(0x107550)
  7638 luajit-2.0.4 NAMI  "XXXXYYYYZZZ45217638"
  7638 luajit-2.0.4 RET   unlink 0
  7638 luajit-2.0.4 CALL  extattr_get_fd(0x3,0x1,0x1129d8,0,0)
  7638 luajit-2.0.4 RET   extattr_get_fd -1 errno 2 No such file or directory

The correct behaviour is shown on a 10.2 system with ufs root:

 24254 luajit-2.0.3 CALL 
open(0x106d48,0x601<O_WRONLY|O_CREAT|O_TRUNC>,0x1c0<S_IRUSR|S_IWUSR|S_IXUSR>)
 24254 luajit-2.0.3 NAMI  "XXXXYYYYZZZ452124254"
 24254 luajit-2.0.3 RET   open 3
 24254 luajit-2.0.3 CALL  unlink(0x106d48)
 24254 luajit-2.0.3 NAMI  "XXXXYYYYZZZ452124254"
 24254 luajit-2.0.3 RET   unlink 0
 24254 luajit-2.0.3 CALL  extattr_get_fd(0x3,0x1,0x1121d0,0,0)
 24254 luajit-2.0.3 RET   extattr_get_fd -1 errno 87 Attribute not found

The attribute not found error is the correct one as there are no xattrs on this
new file, but somehow zfs is looking up the file and not finding it rather than
using the file descriptor.

(there is a possibility that this is a regression from 10.2 to 11 but it seems
more likely it is due to zfs; I dont have enough machines to test right now).

-- 
You are receiving this mail because:
You are the assignee for the bug.