From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040207 Firefox/0.8 Description of problem: As root, I do this: mount -t cifs '//192.168.117.2/Share' /home/bucky/share -o credentials=/home/bucky/.credentials,uid=bucky,domain=XXX,file_mode=0644 The mount command returns without errors. However, the share is not mounted. A process named: [cifsoplockd] is created, which cannot be killed. When I issue: /sbin/shutdown -r now a shutdown message appears, but then nothing else. When I [ALT][F2], login as root, and do /sbin/telinit 6 the runlevel changes, but after Unmounting file systems: [OK] mount: / is busy the system stops responding, but does not restart. I must power down by hand. Version-Release number of selected component (if applicable): kernel-2.6.2-1.74, samba-client-3.0.2-5 How reproducible: Always Steps to Reproduce: 1.mount -t cifs '//192.168.117.2/Share' /home/bucky/share -o credentials=/home/bucky/.credentials,uid=bucky,domain=XXX,file_mode=0644 Actual Results: No error message, nothing gets mounted and some kind of deadlock with cifsoplockd happens which prevents software-controlled shutdown or restart. Expected Results: Either the share should have been mounted, or else an error message produced saying it had failed. In any event, I should be able to shutdown. Additional info: mount -t smbfs '//SERVER/Share' /home/bucky/share -o credentials=/home/bucky/.credentials,uid=bucky,domain=XXX,file_mode=0644,ip=192.168.117.2 ...works if I compile smbfs into the kernel. I miss smbfs, even if it is an abomination unto the lord.
I get a segfault when I use (as root): > mount.cifs //remote-machine/remote-share /mnt/mountpoint -o username=user,workgroup=group Don't know what the remote windows box runs. (perhaps win2k) Now 'sync' hangs, reboot doesn't work - and I see [cifsoplockd] process. Must be the same problem.
fc1 as server 192.168.101.10 $ rpm -qa samba* samba-common-3.0.0-15 samba-client-3.0.0-15 samba-3.0.0-15 fc2t1 as client # rpm -qa samba* samba-common-3.0.2-1rc1 samba-3.0.2-1rc1 samba-client-3.0.2-1rc1 # smbclient -Uadmin //192.168.101.10/doku Password: Domain=[XP1800] OS=[Unix] Server=[Samba 3.0.0-15] smb: \> mkdir test smb: \> mkdir test NT_STATUS_OBJECT_NAME_COLLISION making remote directory \test smb: \> rmdir test smb: \> rmdir test NT_STATUS_OBJECT_NAME_NOT_FOUND removing remote directory file \test # LANG=C mount -tcifs -ousername=admin //192.168.101.10/doku /mnt/smb mount: block device //192.168.101.10/doku is write-protected, mounting read-only mount: cannot mount block device //192.168.101.10/doku read-only what is happening with cifs ?
no problems here now with fc1 and w2k-as server fc1, w2k-as=192.168.101.120 client fc2t1, updates samba* (glibc*) kernel-2.6.1-1.65 samba-3.0.2-5 samba-common-3.0.2-5 samba-client-3.0.2-5 # smbclient -Uadministrator //192.168.101.120/c$ Password: Domain=[VM-W2KAS] OS=[Windows 5.0] Server=[Windows 2000 LAN Manager] smb: \>quit # mount -t cifs -ousername=administrator //192.168.101.120/c$ /mnt/smb Password: # mkdir; rmdir; vi; ... # ps auxw | grep cifsop root 1621 0.0 0.0 0 0 ? SW 11:26 0:00 [cifsoplockd] # shutdown -r now
I'm getting a segfault too when i do the same on FC2T1 > mount.cifs //remote-machine/remote-share /mnt/mountpoint -o username=user,workgroup=group Can't work out why. I have CIFS compiled into the kernel and smbfs is totally unselected / unavailable, but this shouldn't be a problem.
The segfault has been fixed in more current cifs vfs code (e.g. http://us1.samba.org/samba/ftp/cifs-cvs/cifs-1.0.2-2.6.tar.gz) and mount fails gracefully to NT4 servers. NT4 servers are not a supported target server for the cifs vfs at this time, although support is likely to be added in the future for it. I will push these fixes to Andrew Morton later in the week. Last I checked, Fedora was missing the corresponding mount helper utility (/sbin/mount.cifs) though.
A number of mount errors that dealt with closing down connections on failed mounts (not just to WinNT and Win98 target servers) have been fixed since 0.94 (the version of the cifs vfs in 2.6.2 kernel). I am especially interested in verifying that the more current version of cifs files mentioned in the previous post fixed the problem to the servers reported that were newer than Windows NT servers.
Since cifs is part of the kernel, I'm reassinging this bug to the kernel.
Using linux-2.6.6-rc3 and the CVS version of CIFS (ver 1.13) lets me authenticate a share from an NT4 server. Just using linux-2.6.6-rc3 does not authenticate as of RFC1001/RFC1002, but new CVS does.
Could anyone experiencing this problem please test again with the latest FC2 update kernel (2.6.6-1.435.2.3)? If the problem no longer exists with this kernel, then the bug should be closed.
The windows box I'm connecting to is running (from nmap) Windows NT 4.0 SP5-SP6. The following command works now (earlier it failed) mount.cifs //remote-machine/remote-share /mnt/mountpoint -o username=user,workgroup=group But I get the following wierd behavior touch: setting times of `z': Input/output error [root@localhost balay]# ls -l z -rw-r--r-- 1 root root 0 Jul 14 2004 z [root@localhost balay]# rm z rm: remove regular empty file `z'? y [root@localhost balay]# ls -l z ls: z: No such file or directory [root@localhost balay]# ls -l -rwxrwSrwt 1 root root 0 Jul 14 2004 z If the same share is mounted as 'smbfs' This file is still undeleteable % rm z rm: cannot lstat `z': Permission denied % ls z ls: z: Permission denied % ls -l -rwxr-xr-x 1 balay balay 0 Jul 14 12:57 z* ---------- This is with samba-client-3.0.3-5 and kernel-2.6.6-1.435.2.3 (i686) Perhaps if NT4 shares are mounted as CIFS - the operation should quit with the correct error message (assuming cifs can't automatically delegate it to smbfs?)
cifsoplockd fixed upsteam - new cifs module fixes problem with unkillable process.
should be fixed in 2.6.9 update. reopen if unfixed.