Hi,

after huge upgrade SL6.3 include kernel 2.6.32-358.2.1.el6
and reboot to the newly installed kernel, a few our NFS
server doesn't start but got a panic in a few seconds
after init finished.
  Some digging into the problem show that where is
combination nfs+xfs on servers which leads to
kernel panic. Servers running with nfs and ext3/4 only,
or with xfs and without nfs run with kernel 2.6.32-358.2.1.el6
so far without any problems.

  Anybody seen this?

  Below is a few lines from the panic.
{{{
Kernel 2.6.32-358.2.1.el6.x86_64 on an x86_64

xxxxx.jinr-t1.ru login: BUG: unable to handle kernel NULL pointer 
dereference
at 00000000000000b8
IP: [<ffffffffa029b280>] xfs_dir_fsync+0x20/0xf0 [xfs]
PGD 0
Oops: 0000 [#1] SMP
last sysfs file: /sys/devices/system/cpu/cpu23/cache/index2/shared_cpu_map
CPU 19
Modules linked in: ip6table_filter ip6_tables ebtable_nat ebtables 
openafs(P)(U)
  nfsd nfs lockd fscache auth_rpcgss nfs_acl sunrpc bridge bonding 8021q 
garp stp
  llc ipv6 ipt_REJECT xt_multiport iptable_filter ip_tables xfs exportfs 
tcp_htcp
  vhost_net macvtap macvlan tun kvm_intel kvm e1000e microcode serio_raw 
i2c_i801
  i2c_core sg iTCO_wdt iTCO_vendor_support ioatdma dca i7core_edac 
edac_core shpc
hp ext4 mbcache jbd2 sd_mod crc_t10dif ahci dm_mirror dm_region_hash 
dm_log dm_m
od [last unloaded: scsi_wait_scan]

Pid: 2416, comm: nfsd4 Tainted: P           --------------- 
2.6.32-358.2.1.el
6.x86_64 #1 Supermicro X8DTT-H/X8DTT-H
RIP: 0010:[<ffffffffa029b280>]  [<ffffffffa029b280>] 
xfs_dir_fsync+0x20/0xf0 [xf
s]
RSP: 0018:ffff880c29a19d30  EFLAGS: 00010286
RAX: ffff880c29a18000 RBX: 0000000000000000 RCX: 7fffffffffffffff
RDX: 0000000000000000 RSI: ffff880c2a568b40 RDI: 0000000000000000
RBP: ffff880c29a19d50 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffff880c2a568b40
R13: 0000000000000000 R14: ffffffffa02ab7c0 R15: ffff880c2681eaa8
}}}

-- 
Best regards,
  Valery Mitsyn