View Issue Details

IDProjectCategoryView StatusLast Update
0007782CentOS-6kernelpublic2015-10-18 20:20
Reporterjfelten Assigned To 
PrioritynormalSeveritycrashReproducibilityalways
Status closedResolutionsuspended 
Platformvmware vmOSCentOSOS Version6.5
Product Version6.5 
Summary0007782: Kernel crash when have both nfs v3 and nfs v4 nfs mounts
DescriptionThe kernel crashes with "PANIC: Oops: 0002 [#1] SMP (check log for details)" and from kdump/crash: "BUG: unable to handle kernel NULL pointer dereference at 0000000000000008" whenever two NFS volumes are mounted, one via nfs v3 and the other nfs v4, then either nfs mount is accessed (even an ls command will do it). It is 100% reproducible. If I mount both with "nfsvers=3" I've only had a single crash which I've not been able to reproduce, but I have the vmcore from that one too.
Kernel: 2.6.32-431.23.3.e16.x86_64
Steps To ReproduceMount one filesystem via nfs v3 and another filesystem via nfs v4 then attempt to access either mount, for example, via the ls command. The crash is immediate. I have reproduced this on multiple identical VM's.
Additional InformationI already had a small nfs v4 mount being served from another CentOS 6.5 VM. Then we added a 1TB nfs v3 mount from a NetApp filer (which doesn't support nfs v4). These systems are on a private network for security reasons so I can't easily copy/paste or attach dump files, but I can retype select pieces if that would help.

Some additional bits from the crash 'log' sub-command output:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
IP: [<ffffffffa016e3f9>] getPatch+0x39/0x80 [lshook]
PGD 23880a067 PUD 235a91067 PMD 0
Oops: 0000 [#1] SMP
last sysfs file: /sys/devices/system/cpu/online
CPU 2
Modules linked in nfs fscache linuxshield(U) lshook(U) nfsd lockd nfs_acl . . . (I can type the rest if it is really helpful) . . . dm_mod [last unloaded: vmmemctl]
TagsNo tags attached.

Activities

toracat

toracat

2014-10-24 15:36

manager   ~0021383

Do you see 'nfs4_match_stateid' in the crash log?
jfelten

jfelten

2014-10-24 16:21

reporter   ~0021385

No. Not in the output from the crash program's 'log' output. I just checked 2 of the crash dumps. I have a couple more to check but just reached the end of the work day here.
toracat

toracat

2014-10-25 03:34

manager   ~0021389

Can you test-install kernel-ml from ELRepo [1]? This will tell you whether or not the latest kernel from kernel.org has the same issue.

[1] http://elrepo.org/tiki/kernel-ml
jfelten

jfelten

2014-10-29 11:43

reporter   ~0021448

Yes, I can try that, but it may be a few days as I'm currently focused on a higher priority project. I also still have the older 2.6.32-358 kernel on those VM's I can try (should have thought of that earlier). Thanks.
toracat

toracat

2015-10-18 20:20

manager   ~0024645

Closing due to inactivity.

Issue History

Date Modified Username Field Change
2014-10-24 14:48 jfelten New Issue
2014-10-24 15:36 toracat Note Added: 0021383
2014-10-24 15:36 toracat Status new => feedback
2014-10-24 16:21 jfelten Note Added: 0021385
2014-10-24 16:21 jfelten Status feedback => assigned
2014-10-25 03:34 toracat Note Added: 0021389
2014-10-29 11:43 jfelten Note Added: 0021448
2015-10-18 20:20 toracat Note Added: 0024645
2015-10-18 20:20 toracat Status assigned => closed
2015-10-18 20:20 toracat Resolution open => suspended