Re: 2.6.24-rc4: mount.ntfs-3g hangs when smack is enabled

Previous thread: none

Next thread: RESEND: [PATCH] libata-sff: Fix oops reported in kerneloops.org for pnp devices with no ctl by Alan Cox on Thursday, May 29, 2008 - 9:25 am. (22 messages)
From: Luiz Fernando N. Capitulino
Date: Thursday, May 29, 2008 - 9:14 am

Hi Casey,

 I'm not sure whether this is a smack problem, but Mandriva users are
reporting that mount.ntfs-3g (ntfs driver, uses FUSE) hangs when
they try to mount their NTFS partition.

 I could reproduce the problem with a vanilla 2.6.24-rc4 kernel, and it
only happens when Smack is enabled.

 Besides that, I've been told that a similar problem happens with unionfs,
but I haven't tried to reproduce it yet.

 What follows is the output of SysRq+D:

"""
Showing all locks held in the system:
1 lock held by mingetty/4193:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
1 lock held by mingetty/4194:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
1 lock held by mingetty/4195:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
1 lock held by mingetty/4196:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
1 lock held by mingetty/4197:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
2 locks held by mount.ntfs-3g/4667:
 #0:  (&type->s_umount_key#19){--..}, at: [<c01726f0>] sget+0x202/0x350
 #1:  (&isp->smk_lock){--..}, at: [<c01c7f87>] smack_d_instantiate+0x39/0x129
1 lock held by bash/4844:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
1 lock held by bash/5389:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
1 lock held by bash/5498:
 #0:  (&tty->atomic_read_lock){--..}, at: [<c0229650>] read_chan+0x1e5/0x605
"""

 Let me know if you need more information.

 Best regards.

-- 
Luiz Fernando N. Capitulino
--

From: Casey Schaufler
Date: Thursday, May 29, 2008 - 9:29 am

From: Miklos Szeredi
Date: Thursday, May 29, 2008 - 9:32 am

Yes, this is a smack/fuse interaction problem.  Also see this thread:

  http://marc.info/?l=fuse-devel&m=120931855107469&w=2

Casey, you were CC-d on the last mail in that thread.  Did you miss
that?

Thanks,
Miklos
--

From: Casey Schaufler
Date: Thursday, May 29, 2008 - 1:29 pm

Yup. Whiffed completely. Thank you for pointing it out.
 

Casey Schaufler
casey@schaufler-ca.com
--

From: Casey Schaufler
Date: Friday, May 30, 2008 - 8:41 pm

Ok. I just set my a machine to try this on and with today's
Linus git tree as the base I'm able to mount an NTFS file system
just fine. Just to be perverse, the machine I'm working with
has Oracle Enterprize 5 installed, with ntfsprogs-2.0.0. 

How about a .config, lsmod, etc?

Thank you.


Casey Schaufler
casey@schaufler-ca.com
--

From: Casey Schaufler
Date: Sunday, June 1, 2008 - 1:06 pm

Looks like a "never mind" with the .config etc as I've been able to 
reproduce the hang
by the simple process of using ntfs-3g, like you did. Thank you.

--

Previous thread: none

Next thread: RESEND: [PATCH] libata-sff: Fix oops reported in kerneloops.org for pnp devices with no ctl by Alan Cox on Thursday, May 29, 2008 - 9:25 am. (22 messages)