F_OFD_GETLK implemented wrong with CIFS protocol version 2.0+

[Steve French]

I am glad that there is a fairly simple reproducer, but wondering if
any of the standard Linux file system functional tests (xfstests) use
these - if not would be good to add your test case to xfstests so
these missing features don't slip through.
On Mon, Jun 25, 2018 at 6:58 PM Laura Abbott <labbott at redhat.com> wrote:
>
> Hi,
>
> A while back, someone reported a failure on Fedora when trying to boot
> a QEMU image off of a CIFS share. The issue was reduced down to a
> test case (https://bugzilla.redhat.com/show_bug.cgi?id=1484130#c8)
>
> # cat test-ofd-lock.c
> #define _GNU_SOURCE
> #include <errno.h>
> #include <stdio.h>
> #include <unistd.h>
> #include <fcntl.h>
>
> int main(int argc, char **argv)
> {
>      int ret;
>      int fd;
>      struct flock fl = {
>          .l_whence = SEEK_SET,
>          .l_start  = 0,
>          .l_len    = 0,
>          .l_type   = F_RDLCK,
>      };
>      if (argc < 2) {
>              fprintf(stderr, "Usage: %s <file>\n", argv[0]);
>              return 1;
>      }
>      fd = open(argv[1], O_RDWR);
>      if (fd < 0) {
>              perror("open");
>              return errno;
>      }
>      ret = fcntl(fd, F_OFD_SETLK, &fl);
>      if (ret) {
>              perror("setlk");
>              return errno;
>      }
>      fl.l_type = F_WRLCK;
>      ret = fcntl(fd, F_OFD_GETLK, &fl);
>      if (ret) {
>              perror("getlk");
>              return errno;
>      }
>      if (fl.l_type != F_UNLCK) {
>              fprintf(stderr, "get lock test failed\n");
>              return 1;
>      }
>      return 0;
> }
> [root at localhost ~]# make test-ofd-lock
> cc     test-ofd-lock.c   -o test-ofd-lock
> [root at localhost ~]# touch /tmp/test && ./test-ofd-lock /tmp/test
> [root at localhost ~]# echo $?
> 0
> [root at localhost ~]# touch /mnt/test && ./test-ofd-lock /mnt/test
> get lock test failed
> [root at localhost ~]# mount | grep /mnt
> //192.168.31.1/tddownload on /mnt type cifs (rw,relatime,vers=3.0,
> cache=strict,username=admin,domain=,uid=0,
> noforceuid,gid=0,noforcegid,addr=192.168.31.1,file_mode=0755,
> dir_mode=0755,nounix,serverino,mapposix,rsize=1048576,
> wsize=1048576,echo_interval=60,actimeo=1,user=admin)
>
>
> As explained by one of the QEMU developers
> (https://bugzilla.redhat.com/show_bug.cgi?id=1484130#c37)
>
> '''
> It is a kernel bug. The code snippet in comment 8 shows clearly that the kernel
> is doing the wrong thing, which cannot be fixed/worked around by QEMU.
>
> In man 2 fcntl:
>
>         F_OFD_GETLK (struct flock *)
>                On input to this call, lock describes an open file description lock
> we would like to place on the file.  If the lock could  be  placed,  fcntl()  does  not
>                actually  place  it,  but  returns F_UNLCK in the l_type field of lock
> and leaves the other fields of the structure unchanged.  If one or more incompatible
>                locks would prevent this lock being placed, then details about one of
> these locks are returned via lock, as described above for F_GETLK.
>
> which is not the case with the new CIFS behaviour.
> ''
>
> You can read the full context at https://bugzilla.redhat.com/show_bug.cgi?id=1484130
>
> Any suggestions?
>
> Thanks,
> Laura
> --
> To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
> the body of a message to majordomo at vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
Thanks,

Steve