Lustre discuss - Aug 2006 - Lustre crash problem

Hi, everyone,

I am a newbie to Lustre and want to portage the 1.6beta4 lustre to x86 and
ppc64 gentoo linux. My gentoo kernel is

"Linux ionode1 2.6.16-gentoo-r12 #1 SMP Sat Aug 5 22:54:56 CST 2006 i686
Intel(R) Xeon(TM) CPU 2.80GHz GNU/Linux"

"Linux compnode1 2.6.16-gentoo-r9 #3 SMP Thu Jun 22 17:24:58 CST 2006
ppc64 PPC970, altivec supported GNU/Linux"

Now I patch the kernel with patches changed from 2.6-fc5(fix all warning
and error when building the kernel), Now on the x86 node, I have
successfully build the Lustre enviroment for regular file. However, for
the executeable file, if I use ./name[TAB] for auto completion, the lustre
system will crash with following error message:


Aug  6 18:42:25 cbiio1 kernel BUG at fs/dcache.c:853!
Aug  6 18:42:25 cbiio1 invalid opcode: 0000 [#1]
Aug  6 18:42:25 cbiio1 SMP
Aug  6 18:42:25 cbiio1 Modules linked in: osc obdfilter ost mds
fsfilt_ldiskfs mgs mgc lustre
lov mdc ksocklnd ptlrpc obdclass lvfs ldiskfs lnet libcfs bonding raid1
Aug  6 18:42:25 cbiio1 CPU:    1
Aug  6 18:42:25 cbiio1 EIP:    0060:[<c0179801>]    Not tainted VLI
Aug  6 18:42:25 cbiio1 EFLAGS: 00010246   (2.6.16-gentoo-r12 #1)
Aug  6 18:42:25 cbiio1 EIP is at d_instantiate_unique+0x12c/0x136
Aug  6 18:42:25 cbiio1 eax: f53d9314   ebx: f53d9314   ecx: f53d9314  
edx: f52ce2bc
Aug  6 18:42:25 cbiio1 esi: f53d9374   edi: f52ce374   ebp: f55563dc  
esp: f7545b78
Aug  6 18:42:25 cbiio1 ds: 007b   es: 007b   ss: 0068
Aug  6 18:42:25 cbiio1 Process bash (pid: 6829, threadinfo=f7544000
task=f7e35580)
Aug  6 18:42:25 cbiio1 Stack: <0>f53d9314 f53d9354 f52ce354 00000080
00010000 01c4ddc5 f52ce37
0 00000004
Aug  6 18:42:25 cbiio1 f7545cd4 f52ce314 f7545c3c 00000000 f90ec5b0
f52ce314 f55563c4 f7545cd4

Aug  6 18:42:25 cbiio1 f5052ec0 f90ec870 f55563c4 f52ce314 f9109ba8
f9101d07 000001cc 00002000

Aug  6 18:42:25 cbiio1 Call Trace:
Aug  6 18:42:25 cbiio1 [<f90ec5b0>] ll_find_alias+0x1f/0x57 [lustre]
Aug  6 18:42:25 cbiio1 [<f90ec870>] lookup_it_finish+0x288/0x55e [lustre]
Aug  6 18:42:25 cbiio1 [<f90ece7f>] ll_lookup_it+0x339/0xa7c [lustre]
Aug  6 18:42:25 cbiio1 [<f90eaa1a>] ll_mdc_blocking_ast+0x0/0x118a
[lustre]
Aug  6 18:42:25 cbiio1 [<f90725a1>] ll_intent_release+0x0/0x22d [lustre]
Aug  6 18:42:25 cbiio1 [<f90ed630>] ll_lookup_nd+0x6e/0x1f7 [lustre]
Aug  6 18:42:25 cbiio1 [<c0179582>] d_alloc+0x101/0x18c
Aug  6 18:42:25 cbiio1 [<c016e249>] real_lookup+0xc3/0xe8
Aug  6 18:42:25 cbiio1 [<c016e53c>] do_lookup+0x9d/0xa8
Aug  6 18:42:25 cbiio1 [<c016edf2>] __link_path_walk+0x802/0xf95
Aug  6 18:42:25 cbiio1 [<f93401f9>] ptlrpc_req_finished+0x17/0x1b [ptlrpc]
Aug  6 18:42:25 cbiio1 [<f90ac724>] ll_inode_revalidate_it+0x2b2/0xcd0
[lustre]
Aug  6 18:42:25 cbiio1 [<f90eaa1a>] ll_mdc_blocking_ast+0x0/0x118a
[lustre]
Aug  6 18:42:25 cbiio1 [<c0141d2d>] read_cache_page+0xcd/0x18b
Aug  6 18:42:25 cbiio1 [<c016f5ea>] link_path_walk+0x65/0x10a
Aug  6 18:42:25 cbiio1 [<c016f9e2>] do_path_lookup+0x164/0x2b0
Aug  6 18:42:25 cbiio1 [<c016fe49>] __user_walk_fd_it+0x3c/0x5c
Aug  6 18:42:25 cbiio1 [<c0169bbf>] vfs_stat_fd+0x5b/0xca
Aug  6 18:42:25 cbiio1 [<c027c608>] copy_to_user+0x42/0x5c
Aug  6 18:42:25 cbiio1 [<c0169c4d>] vfs_stat+0x1f/0x23
Aug  6 18:42:25 cbiio1 [<c016a41c>] sys_stat64+0x18/0x36
Aug  6 18:42:25 cbiio1 [<c0102d01>] syscall_call+0x7/0xb
Aug  6 18:42:25 cbiio1 Code: 4f 18 8b 54 24 38 8b 44 24 34 89 50 0c c6 05
80 7b 53 c0 01 31 c0
 83 c4 20 5b 5e 5f 5d c3 0f 0b 44 03 b0 1a 43 c0 e9 14 ff ff ff <0f> 0b 55
03 b0 1a 43 c0 eb a
1 83 ec 10 89 74 24 0c 8b 74 24 14

On the ppc64 node, I can compile the program, but when mount it with
"mount -v -t lustre ionode1@tcp0:/testfs /mnt/lustre", it failed with
following error:

Aug  6 14:24:46 compnode1 Lustre:   mount data:
Aug  6 14:24:46 compnode1 Lustre: profile: testfs-client
Aug  6 14:24:46 compnode1 Lustre: device:  192.168.0.202@tcp:/testfs
Aug  6 14:24:46 compnode1 Lustre: flags:   2
Aug  6 14:27:43 compnode1 Lustre:   mount data:
Aug  6 14:27:43 compnode1 Lustre: profile: testfs-client
Aug  6 14:27:43 compnode1 Lustre: device:  192.168.0.202@tcp:/testfs
Aug  6 14:27:43 compnode1 Lustre: flags:   2
Aug  6 14:27:44 compnode1 Unable to handle kernel paging request for data
at address 0x00000100
Aug  6 14:27:44 compnode1 Faulting instruction address: 0xd0000000004c4808
Aug  6 14:27:44 compnode1 Oops: Kernel access of bad area, sig: 11 [#1]
Aug  6 14:27:44 compnode1 SMP NR_CPUS=32 NUMA PSERIES LPAR
Aug  6 14:27:44 compnode1 Modules linked in: mgc lustre mdc lov ksocklnd
ptlrpc obdclass lnet lvfs libcfs
Aug  6 14:27:44 compnode1 NIP: D0000000004C4808 LR: D0000000004C4390 CTR:
0000000000000000
Aug  6 14:27:44 compnode1 REGS: c0000000388730f0 TRAP: 0300   Not tainted 
(2.6.16-gentoo-r9)
Aug  6 14:27:44 compnode1 MSR: 8000000000009032 <EE,ME,IR,DR>  CR:
22088442  XER: 20000000
Aug  6 14:27:44 compnode1 DAR: 0000000000000100, DSISR: 0000000040000000
Aug  6 14:27:44 compnode1 TASK = c0000000026967e0[11610]
''mount.lustre''
THREAD: c00000003887000 0 CPU: 0
Aug  6 14:27:44 compnode1 GPR00: 000000000000006B C000000038873370
D0000000004DCCB8 0000000000000000
Aug  6 14:27:44 compnode1 GPR04: D0000000004CEFD4 000000000000000D
D00000000029E0EC C000000038873620
Aug  6 14:27:44 compnode1 GPR08: 0000000000000C90 0000000000000000
0000000000000000 C00000000002FA44
Aug  6 14:27:44 compnode1 GPR12: D0000000004CD630 C00000000046A700
FFFFFFFFFDF9FBFF C00000000044F480
Aug  6 14:27:44 compnode1 GPR16: 00000000000249F0 00000000000253B4
0000000000000000 D0000000002


Thanks for your help.