[ATrpms-users] Alsa 1.0.13 crashes my system hard

[Barry Stewart]

Axel Thimm wrote:
> On Mon, Dec 11, 2006 at 12:37:27PM -0500, Barry Stewart wrote:
>> I upgraded alsa to version 1.0.13 on my FC6 dual core AMD x86_64 (Asus
>> A8N-VM CSM) system with nVidia sound (AD1986A using snd-hda-intel).  I did
>> this in an attempt get full surround sound support, which I don't get with
>> stock FC6 rpms.  However, when I attempt to play a sound (i.e. 
>> speaker-test -c 6), my system crashes hard.  No keyboard, mouse, not even
>> a blinking cursor; I have to power cycle the box.  Is any one else having
>> this problem?
>>
>> I'm using the 2.6.18-1.2849 kernel, if that helps.
> 
> I have never heard of alsa being able to break your system that
> hard. What did you update? Both alsa-lib and alsa-driver-(kmdl)?
> 
> What graphics card do you have? Is the kernel otherwise tainted?
> 
> Do you have a second host to log onto the one with the lasa before the
> system hangs? If so can you check that it really hangs, or perhaps
> only the screen freezes?

I updated alsa-lib, alsa-kmdl, alsa-driver and libasound2.  I have three 
nvidia cards in a triple-head setup using the latest stable driver from 
nvidia.

The system definitely hangs.  I cannot ping or ssh into the box from 
another computer, even when booting into runlevel 3 and removing the 
nvidia module.

However, I have noticed some weird messages in the system log:

Dec 12 23:33:48 numbers kernel: ACPI: PCI Interrupt 0000:00:10.1[B] -> 
Link [LAZA] -> GSI 22 (level, low) -> IRQ 217
Dec 12 23:33:48 numbers kernel: irq 217: nobody cared (try booting with 
the "irqpoll" option)
Dec 12 23:33:48 numbers kernel:
Dec 12 23:33:48 numbers kernel: Call Trace:
Dec 12 23:33:48 numbers kernel:  [<ffffffff802691bd>] show_trace+0x34/0x47
Dec 12 23:33:48 numbers kernel:  [<ffffffff802691e2>] dump_stack+0x12/0x17
Dec 12 23:33:48 numbers kernel:  [<ffffffff802b559a>] 
__report_bad_irq+0x30/0x7d
Dec 12 23:33:48 numbers kernel:  [<ffffffff802b57ba>] 
note_interrupt+0x1d3/0x219
Dec 12 23:33:48 numbers kernel:  [<ffffffff802b4d01>] __do_IRQ+0xc8/0x107
Dec 12 23:33:48 numbers kernel:  [<ffffffff8026a1ff>] do_IRQ+0xe7/0xf5
Dec 12 23:33:48 numbers kernel:  [<ffffffff8025c409>] ret_from_intr+0x0/0xa
Dec 12 23:33:48 numbers kernel: DWARF2 unwinder stuck at 
ret_from_intr+0x0/0xa
Dec 12 23:33:48 numbers kernel: Leftover inexact backtrace:
Dec 12 23:33:48 numbers kernel:  <IRQ>  <EOI>  [<ffffffff802689a9>] 
default_idle+0x29/0x50
Dec 12 23:33:48 numbers kernel:  [<ffffffff80247af7>] cpu_idle+0x95/0xb8
Dec 12 23:33:48 numbers kernel:  [<ffffffff8064384b>] 
start_kernel+0x225/0x22a
Dec 12 23:33:48 numbers kernel:  [<ffffffff8064328a>] _sinittext+0x28a/0x28e
Dec 12 23:33:48 numbers kernel:
Dec 12 23:33:48 numbers kernel: handlers:
Dec 12 23:33:48 numbers kernel: [<ffffffff803d4e6f>] (usb_hcd_irq+0x0/0x55)
Dec 12 23:33:48 numbers kernel: Disabling IRQ #217
Dec 12 23:33:48 numbers kernel: ALSA 
/var/tmp/bach-build/BUILD/alsa-driver-1.0.13/pci/hda/../../alsa-kernel/pci/hda/hda_intel.c:540: 
hda_intel: azx_get_response timeout, switching to polling mode...
Dec 12 23:33:48 numbers kernel: ALSA 
/var/tmp/bach-build/BUILD/alsa-driver-1.0.13/pci/hda/../../alsa-kernel/pci/hda/hda_intel.c:546: 
hda_intel: azx_get_response timeout, switching to single_cmd mode...
Dec 12 23:33:48 numbers kernel: hda_codec: num_steps = 0 for NID=0x1b
Dec 12 23:33:48 numbers last message repeated 7 times
Dec 12 23:39:27 numbers kernel: warning: many lost ticks.
Dec 12 23:39:27 numbers kernel: Your time source seems to be instable or 
some driver is hogging interupts
Dec 12 23:39:27 numbers kernel: rip __do_softirq+0x53/0xd5

I'll try the "irqpoll" option as suggested, but what's going on?