[BUG] 2.6.24-rc3-git2 softlockup detected

[Andrew Morton]

On Wed, 28 Nov 2007 11:59:00 +0530 Kamalesh Babulal <kamalesh at linux.vnet.ibm.com> wrote:

> Hi,

(cc linux-scsi, for sym53c8xx)

> Soft lockup is detected while bootup with 2.6.24-rc3-git2 on powerbox

I assume this is a post-2.6.23 regression?

> BUG: soft lockup - CPU#1 stuck for 11s! [insmod:375]
> NIP: c00000000002f02c LR: d0000000001414fc CTR: c00000000002f018
> REGS: c00000077cbef0b0 TRAP: 0901   Not tainted  (2.6.24-rc3-git2-autotest)
> MSR: 8000000000009032 <EE,ME,IR,DR>  CR: 24022088  XER: 00000000
> TASK = c00000077cbd8000[375] 'insmod' THREAD: c00000077cbec000 CPU: 1
> GPR00: d0000000001414fc c00000077cbef330 c00000000052b930 d000080080002014 
> GPR04: d00008008000202c 0000000000000000 c00000077ca1cb00 d00000000014ce54 
> GPR08: c00000077ca1c63c 0000000000000000 000000000000002a c00000000002f018 
> GPR12: d000000000143610 c000000000473d00 
> NIP [c00000000002f02c] .ioread8+0x14/0x60
> LR [d0000000001414fc] .sym_hcb_attach+0x1188/0x1378 [sym53c8xx]
> Call Trace:
> [c00000077cbef330] [c00000077cbef3c0] 0xc00000077cbef3c0 (unreliable)
> [c00000077cbef3a0] [d0000000001414fc] .sym_hcb_attach+0x1188/0x1378 [sym53c8xx]
> [c00000077cbef470] [d0000000001395f8] .sym2_probe+0x700/0x99c [sym53c8xx]
> [c00000077cbef710] [c0000000001bc118] .pci_device_probe+0x124/0x1b0
> [c00000077cbef7b0] [c000000000221138] .driver_probe_device+0x144/0x20c
> [c00000077cbef850] [c000000000221450] .__driver_attach+0xcc/0x154
> [c00000077cbef8e0] [c00000000021ff94] .bus_for_each_dev+0x7c/0xd4
> [c00000077cbef9a0] [c000000000220e9c] .driver_attach+0x28/0x40
> [c00000077cbefa20] [c0000000002204d8] .bus_add_driver+0x90/0x228
> [c00000077cbefac0] [c000000000221858] .driver_register+0x94/0xb0
> [c00000077cbefb40] [c0000000001bc430] .__pci_register_driver+0x6c/0xcc
> [c00000077cbefbe0] [d000000000143428] .sym2_init+0x108/0x15b0 [sym53c8xx]
> [c00000077cbefc80] [c00000000008ce80] .sys_init_module+0x17c4/0x1958
> [c00000077cbefe30] [c00000000000872c] syscall_exit+0x0/0x40
> Instruction dump:
> 60000000 786b0420 38210070 7d635b78 e8010010 7c0803a6 4e800020 7c0802a6 
> f8010010 f821ff91 7c0004ac 89230000 <0c090000> 4c00012c 79290620 2f8900ff 

I see no obvious lockup sites near the end of sym_hcb_attach().  Maybe it's
being called lots of times from a higher level..  Do the traces all look
the same?