On Thu, Sep 22, 2005 at 01:09:30AM +0200, Axel Thimm wrote: > On Wed, Sep 21, 2005 at 03:52:42PM -0700, Jeremy Allison wrote: > > On Thu, Sep 22, 2005 at 12:47:08AM +0200, Axel Thimm wrote: > > > I tested it and it runs successfully. > > I'll have to take that back. I tested it on a file on the system disk, > I should had tested this on the filesystem the "lock directory" is on > (see below). > > > > Is there a chance that the fd isn't opened by root in > > > set_delete_on_close? > > > > I can't see a way this can happen. The locking tdb is only > > opened by locking_init() which is called at line 879 in > > smbd/server.c - we are very much root at this time. > > The outcome of the test code you sent segfaults now. > > # mount | grep gfs > /dev/mapper/physik-data on /srv/physik.fu-berlin.de/data type gfs > (rw,acl) > # pwd > /srv/physik.fu-berlin.de/data/samba-test > # ls -l > total 32 > -rwxr-xr-x 1 root root 10080 Sep 22 00:38 a.out > -rw------- 1 root root 1231 Sep 22 00:35 test.c > -rw-r--r-- 1 root root 0 Sep 22 01:07 testfile > # ./a.out testfile thimm > Segmentation fault
What's the gdb backtrace. There's probably a bug in one of the error condition printing in the test code. As I said, I bet GFS isn't POSIX complient. Don't put locking tdb's on anything but local filesystems. Jeremy. -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/listinfo/samba