On Thu, 29 Nov 2007 12:01:08 +0530 Kamalesh Babulal [off-list ref] wrote:

Andrew Morton wrote:

quoted

On Wed, 28 Nov 2007 12:47:19 +0530 Kamalesh Babulal [off-list ref] wrote:

quoted

Andrew Morton wrote:

quoted

On Wed, 28 Nov 2007 11:59:00 +0530 Kamalesh Babulal [off-list ref] wrote:

quoted

Hi,

(cc linux-scsi, for sym53c8xx)

quoted

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

I assume this is a post-2.6.23 regression?

quoted

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?

Hi Andrew,

I see this call trace twice and both looks similar and on another reboot
the following trace is seen twice in different cpu

BUG: soft lockup detected on CPU#3!
Call Trace:
[C00000003FEDEDA0] [C000000000010220] .show_stack+0x68/0x1b0 (unreliable)
[C00000003FEDEE40] [C0000000000A061C] .softlockup_tick+0xf0/0x13c
[C00000003FEDEEF0] [C000000000072E2C] .run_local_timers+0x1c/0x30
[C00000003FEDEF70] [C000000000022FA0] .timer_interrupt+0xa8/0x488
[C00000003FEDF050] [C0000000000034EC] decrementer_common+0xec/0x100

--- Exception: 901 at .ioread8+0x14/0x60
    LR = .sym_hcb_attach+0x1194/0x1384 [sym53c8xx]

[C00000003FEDF340] [D0000000002B3BC0] 0xd0000000002b3bc0 (unreliable)
[C00000003FEDF3B0] [D00000000029A3C0] .sym_hcb_attach+0x1194/0x1384 [sym53c8xx]
[C00000003FEDF480] [D000000000291D30] .sym2_probe+0x75c/0x9f8 [sym53c8xx]
[C00000003FEDF710] [C0000000001B65A4] .pci_device_probe+0x13c/0x1dc
[C00000003FEDF7D0] [C000000000219A0C] .driver_probe_device+0xa0/0x15c
[C00000003FEDF870] [C000000000219C64] .__driver_attach+0xb4/0x138
[C00000003FEDF900] [C00000000021913C] .bus_for_each_dev+0x7c/0xd4
[C00000003FEDF9C0] [C0000000002198B0] .driver_attach+0x28/0x40
[C00000003FEDFA40] [C000000000218BA4] .bus_add_driver+0x98/0x18c
[C00000003FEDFAE0] [C00000000021A064] .driver_register+0xa8/0xc4
[C00000003FEDFB60] [C0000000001B68AC] .__pci_register_driver+0x5c/0xa4
[C00000003FEDFBF0] [D00000000029C204] .sym2_init+0x104/0x1550 [sym53c8xx]
[C00000003FEDFC90] [C00000000008D1F4] .sys_init_module+0x1764/0x1998
[C00000003FEDFE30] [C00000000000869C] syscall_exit+0x0/0x40

hm, odd.

Can you look up sym_hcb_attach+0x1194/0x1384 in gdb?  Something like

Hi Andrew,

I tried with 2.6.24-rc3-git3 and got the following trace

BUG: soft lockup - CPU#2 stuck for 11s! [insmod:375]
NIP: c00000000002f02c LR: d0000000001414fc CTR: c00000000002f018
REGS: c00000077ca3b0b0 TRAP: 0901   Not tainted  (2.6.24-rc3-git3-autokern1)
MSR: 8000000000009032 <EE,ME,IR,DR>  CR: 24022088  XER: 00000000
TASK = c00000077cc58000[375] 'insmod' THREAD: c00000077ca38000 CPU: 2
GPR00: d0000000001414fc c00000077ca3b330 c00000000052b880 d000080080002014 
GPR04: d00008008000202c 0000000000000000 c00000077c82eb00 d00000000014ce54 
GPR08: c00000077c82e63c 0000000000000000 000000000000002a c00000000002f018 
GPR12: d000000000143610 c000000000473f80 
NIP [c00000000002f02c] .ioread8+0x14/0x60
LR [d0000000001414fc] .sym_hcb_attach+0x1188/0x1378 [sym53c8xx]

Call Trace:
[c00000077ca3b330] [c00000077ca3b3c0] 0xc00000077ca3b3c0 (unreliable)
[c00000077ca3b3a0] [d0000000001414fc] .sym_hcb_attach+0x1188/0x1378 [sym53c8xx]
[c00000077ca3b470] [d0000000001395f8] .sym2_probe+0x700/0x99c [sym53c8xx]
[c00000077ca3b710] [c0000000001bc098] .pci_device_probe+0x124/0x1b0
[c00000077ca3b7b0] [c0000000002210c4] .driver_probe_device+0x144/0x20c
[c00000077ca3b850] [c0000000002213dc] .__driver_attach+0xcc/0x154
[c00000077ca3b8e0] [c00000000021ff20] .bus_for_each_dev+0x7c/0xd4
[c00000077ca3b9a0] [c000000000220e28] .driver_attach+0x28/0x40
[c00000077ca3ba20] [c000000000220464] .bus_add_driver+0x90/0x228
[c00000077ca3bac0] [c0000000002217e4] .driver_register+0x94/0xb0
[c00000077ca3bb40] [c0000000001bc3b0] .__pci_register_driver+0x6c/0xcc
[c00000077ca3bbe0] [d000000000143428] .sym2_init+0x108/0x15b0 [sym53c8xx]
[c00000077ca3bc80] [c00000000008ce80] .sys_init_module+0x17c4/0x1958
[c00000077ca3be30] [c00000000000872c] syscall_exit+0x0/0x40

Instruction dump:
60000000 786b0420 38210070 7d635b78 e8010010 7c0803a6 4e800020 7c0802a6
f8010010 f821ff91 7c0004ac 89230000 <0c090000> 4c00012c 79290620 2f8900ff

The gdb list the following for the above trace

0xa4fc is in sym_hcb_attach (drivers/scsi/sym53c8xx_2/sym_hipd.c:1041).
1036            OUTL_DSP(np, pc);
1037            /*
1038             *  Wait 'til done (with timeout)
1039             */
1040            for (i=0; i<SYM_SNOOP_TIMEOUT; i++)
1041                    if (INB(np, nc_istat) & (INTF|SIP|DIP))
1042                            break;
1043            if (i>=SYM_SNOOP_TIMEOUT) {
1044                    printf ("CACHE TEST FAILED: timeout.\n");
1045                    return (0x20);

doh, I missed that.

#define SYM_SNOOP_TIMEOUT (10000000)

ten million is close enough to infinity for me to assume that we broke the
driver and that's never going to terminate.

otoh, if that's true you should have got the "CACHE TEST FAILED: timeout"
message.  Did you?  And does the driver actually work OK after this?

If it is indeed expected that a ~10 second stall in there is correct
behaviour then all we need to do is do make that loop a bit smarter (10,000
msleep(1)'s, for example).