[Unionfs] RHEL 2.6.9 hangs with unionfs-1.1.5

[Benoit Guillon]

Hello,

I use unionfs to serve some diskless nodes. Its a 2.6.9-34.EL kernel 
with unionfs-1.1.5 that does the mount and exports the mounted file 
system. The mount is ok, and the diskless nodes can switchroot on this 
NFS mounted file system.

The problem is that often the kernel hangs and freezes completely. I've 
put the debug level to 99, and before the freezing I have the following 
traces showing a kernel oops:

...
inode /tmp/build.mmm/unionfs-tools-1.1.5/super.c:87
Nov 23 19:41:47 node1 
kernel:                                                                               
IN:  unionfs_open /tmp/build.mmm/unionfs-tools-1.1.5/commonfops.c:359
Nov 23 19:41:47 node1 kernel: dentry to open is c5ca963c
Nov 23 19:41:47 node1 kernel: EIP is at _spin_unlock_irqrestore+0x1f/0x2d
Nov 23 19:41:47 node1 kernel: eax: 00000001   ebx: 00000086   ecx: 
c0324b4c   edx: 00000001
Nov 23 19:41:47 node1 kernel: esi: 00000086   edi: 00000025   ebp: 
c02dc62f   esp: da568c38
Nov 23 19:41:47 node1 kernel: ds: 007b   es: 007b   ss: 0068
Nov 23 19:41:47 node1 kernel: Process nfsd (pid: 3351, 
threadinfo=da568000 task=df351430)
Nov 23 19:41:47 node1 kernel: Stack: c04124c5 c012278f e06d5e04 0000004a 
00000000 c0122648 c02dc53a da568c60
Nov 23 19:41:47 node1 kernel:        c0105f2d c02dc53a da568000 da568d58 
c0105fbd e0061680 00000246 c03269bc
Nov 23 19:41:47 node1 kernel:        c01386a7 e069b525 00000000 da568cc0 
c5f5d65c da568d58 00000000 da568cc0
Nov 23 19:41:47 node1 kernel: Call Trace:
Nov 23 19:41:47 node1 kernel:  [<c012278f>] vprintk+0x144/0x14a
Nov 23 19:41:47 node1 kernel:  [<c0122648>] printk+0xe/0x11
Nov 23 19:41:47 node1 kernel:  [<c0105f2d>] handle_BUG+0x46/0x65
Nov 23 19:41:47 node1 kernel:  [<c0105fbd>] die+0x71/0x16b
Nov 23 19:41:47 node1 kernel:  [<c01386a7>] search_module_extables+0x5d/0x64
Nov 23 19:41:47 node1 kernel:  [<e069b525>] 
unionfs_d_revalidate+0x18e5/0x19e0 [unionfs]
Nov 23 19:41:47 node1 kernel:  [<c0106409>] do_invalid_op+0xcf/0xf2
Nov 23 19:41:47 node1 kernel:  [<c02cfd07>] __down_failed_trylock+0x7/0xc
Nov 23 19:41:47 node1 kernel:  [<e069b525>] 
unionfs_d_revalidate+0x18e5/0x19e0 [unionfs]
Nov 23 19:41:47 node1 kernel:  [<c01c18e5>] vsnprintf+0x448/0x488
Nov 23 19:41:47 node1 kernel:  [<c011e7c5>] __wake_up_locked+0x11/0x13
Nov 23 19:41:47 node1 kernel:  [<c010633a>] do_invalid_op+0x0/0xf2
Nov 23 19:41:47 node1 kernel:  [<c02d3133>] error_code+0x2f/0x38
Nov 23 19:41:47 node1 kernel:  [<e069b525>] 
unionfs_d_revalidate+0x18e5/0x19e0 [unionfs]
Nov 23 19:41:47 node1 kernel:  [<c012278f>] vprintk+0x144/0x14a
Nov 23 19:41:47 node1 kernel:  [<e06cd804>] 
unionfs_file_revalidate+0x134/0x14e0 [unionfs]
Nov 23 19:41:47 node1 kernel:  [<c02cfd07>] __down_failed_trylock+0x7/0xc
Nov 23 19:41:47 node1 kernel:  [<c012278f>] vprintk+0x144/0x14a
Nov 23 19:41:47 node1 kernel:  [<e069ccbb>] unionfs_read+0x8b/0x1d0 
[unionfs]
Nov 23 19:41:47 node1 kernel:  [<e069cc30>] unionfs_read+0x0/0x1d0 [unionfs]
Nov 23 19:41:47 node1 kernel:  [<c015a97c>] do_readv_writev+0x1c5/0x21d
Nov 23 19:41:47 node1 kernel:  [<c015997e>] dentry_open+0x49/0x4e
Nov 23 19:41:47 node1 kernel:  [<c015aa0f>] vfs_readv+0x3b/0x40
Nov 23 19:41:47 node1 kernel:  [<e041cd71>] nfsd_read+0x210/0x28d [nfsd]
Nov 23 19:41:47 node1 kernel:  [<e0424457>] nfsd3_proc_read+0xbc/0x114 
[nfsd]
Nov 23 19:41:47 node1 kernel:  [<e042641e>] 
nfs3svc_decode_readargs+0x0/0x1fe [nfsd]
Nov 23 19:41:47 node1 kernel:  [<e0419681>] nfsd_dispatch+0xba/0x16d [nfsd]
Nov 23 19:41:47 node1 kernel:  [<e03c555b>] svc_process+0x432/0x6d7 [sunrpc]
Nov 23 19:41:47 node1 kernel:  [<e041945a>] nfsd+0x1cc/0x339 [nfsd]
Nov 23 19:41:47 node1 kernel:  [<e041928e>] nfsd+0x0/0x339 [nfsd]
Nov 23 19:41:47 node1 kernel:  [<c01041f5>] kernel_thread_helper+0x5/0xb
Nov 23 19:41:47 node1 kernel: Code: 00 00 00 01 74 05 e8 06 e9 ff ff c3 
53 89 c1 89 d3 81 78 04 ad 4e ad de b2 01 74 08 0f 0b 6c 00 60 3d 2e c0 
0f b6 01 84 c0 7e 08 <0f> 0b 6d 00 60 3d 2e c0 86 11 53 9d 5b c3 81 78 
04 ad 4e ad de
Nov 23 19:41:47 node1 kernel:  <0>Fatal exception: panic in 5 seconds
Nov 23 19:41:47 node1 
kernel:                                                                                
IN:  unionfs_decode_fh /tmp/build.mmm/unionfs-tools-1.1.5/super.c:771 2, 
fh{i756, g0, hrcb59d344, hi203070, hpi203061}
Nov 23 19:41:47 node1 
kernel:                                                                                
IN:  unionfs_decode_fh /tmp/build.mmm/unionfs-tools-1.1.5/super.c:771 2, 
fh{i756, g0, hrcb59d344, hi203070, hpi203061}
Nov 23 19:41:47 node1 
kernel:                                                                                
IN:  unionfs_put_inode /tmp/build.mmm/unionfs-tools-1.1.5/super.c:73
...

Can you help me to find out what happens? Do you need more details or 
some testings? Any help would be highly appreciated.

Thanks,

-- 
Benoît Guillon                guillon at thalescomputers.fr