Re: 2.6.21-rc7-mm1 BUG at kernel/sched-clock.c:175 init_sched_clock()

[Tejun Heo <htejun@gmail.com>]

Hello.

Berck E. Nash wrote:
> Okay, while this patch allows my system to boot, it eventually crashed.
>  Unfortunately, I don't know enough to know if this crash is at all
> related, but it seems like it might be?
> 
> [ 1314.960784] irq 316: nobody cared (try booting with the "irqpoll" option)
> [ 1314.967602]
> [ 1314.967602] Call Trace:
> [ 1314.971566]  [<ffffffff8025dae3>] dump_trace+0xb6/0x3d6
> [ 1314.976809]  [<ffffffff8025de3f>] show_trace+0x3c/0x52
> [ 1314.981963]  [<ffffffff8025de6a>] dump_stack+0x15/0x17
> [ 1314.987118]  [<ffffffff802993c8>] __report_bad_irq+0x38/0x7c
> [ 1314.992790]  [<ffffffff802995db>] note_interrupt+0x1cf/0x215
> [ 1314.998463]  [<ffffffff8029a037>] handle_edge_irq+0x104/0x133
> [ 1315.004231]  [<ffffffff8025f4e1>] do_IRQ+0x6d/0xd5
> [ 1315.009039]  [<ffffffff80255071>] ret_from_intr+0x0/0xa
> [ 1315.014281]  [<ffffffff8024e88b>] mwait_idle+0x46/0x4b
> [ 1315.019433]  [<ffffffff802422f4>] cpu_idle+0x87/0xaa
> [ 1315.024414]  [<ffffffff8025c988>] rest_init+0x49/0x4b
> [ 1315.029489]  [<ffffffff805dca5d>] start_kernel+0x291/0x29c
> [ 1315.034989]  [<ffffffff805dc13a>] _sinittext+0x13a/0x141
> [ 1315.040323]
> [ 1315.041813] handlers:
> [ 1315.044090] [<ffffffff803e4951>] (ahci_interrupt+0x0/0x47a)
> [ 1315.049722] Disabling IRQ #316
> [ 1345.037254] ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2
> frozen
> [ 1345.044333] ata1.00: cmd c8/00:08:ff:53:00/00:00:00:00:00/e2 tag 0
> cdb 0x0 data 4096 in
> [ 1345.044334]          res 40/00:00:76:5a:00/00:00:00:00:00/e0 Emask
> 0x4 (timeout)
> [ 1345.059832] ata3.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x2
> frozen
> [ 1345.066919] ata3.00: cmd 60/08:00:bb:26:0a/00:00:09:00:00/40 tag 0
> cdb 0x0 data 4096 in
> [ 1345.066920]          res 40/00:00:00:4f:c2/00:00:00:00:00/00 Emask
> 0x4 (timeout)
> [ 1345.389731] ata1: soft resetting port
> [ 1345.396451] ata3: soft resetting port

Hmmm... Once properly initialized, ahci is highly unlikely to cause
runaway IRQs which results in nobody cared.  It has proper IRQ mask and
pending bits allowing the driver to reliably detect when and why the
controller is raising interrupt and disable it if necessary.  Can you
try without the NVIDIA module?

-- 
tejun