Skip to contentSkip to conversation

2.6.23-rc8-mm[12] - the amazing disappearing MSI IRQ

DISCUSSION [email protected] 1 participant 2 messages Oct 7, 2007 – Oct 8, 2007

lore
Share & tools
lore
#0 lore raw

Dell Latitude D820, T7200 processor, x86_64 kernel. The MSI IRQs for HDA-Intel evaporate fairly consistently some time after boot. On a few occasions, the MSI IRQ for the ethernet interface also has dropped.

00:1b.0 Audio device: Intel Corporation 82801G (ICH7 Family) High Definition Audio Controller (rev 01) 09:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5752 Gigabit Ethernet PCI Express (rev 02)

A while after boot I have this in /proc/interrupts:

Sat Oct 6 22:25:19 EDT 2007
CPU0 CPU1

  0:     222717          0   IO-APIC-edge      timer
  1:          9       1783   IO-APIC-edge      i8042
  8:          0          0   IO-APIC-edge      rtc
  9:          2          0   IO-APIC-fasteoi   acpi
 12:         86          0   IO-APIC-edge      i8042
 14:       5490      30700   IO-APIC-edge      libata
 15:         36          0   IO-APIC-edge      libata
 16:       2657          0   IO-APIC-fasteoi   nvidia
 19:          9          0   IO-APIC-fasteoi   ohci1394, yenta
 20:        125       6393   IO-APIC-fasteoi   ehci_hcd:usb1, uhci_hcd:usb2
 21:         90          0   IO-APIC-fasteoi   uhci_hcd:usb3
 22:          0          0   IO-APIC-fasteoi   uhci_hcd:usb4
 23:          0          0   IO-APIC-fasteoi   uhci_hcd:usb5
312:        149      19249   PCI-MSI-edge      eth0
313:         92          0   PCI-MSI-edge      iwl3945
314:       5375          0   PCI-MSI-edge      HDA Intel
NMI:          0          0   Non-maskable interrupts
LOC:      63560     180342   Local interrupts
RES:      60295      49751   Rescheduling interrupts
CAL:        342       2070   function call interrupts
TLB:       2173       1804   TLB shootdowns
TRM:          0          0   Thermal event interrupts
THR:          0          0   Threshold APIC interrupts
SPU:          0          0   Spurious interrupts

ERR: 0

Ethernet, wireless, and sound card all have MSI interrupts attached. Then after a while, this pops in dmesg:

[ 2086.388000] ALSA sound/pci/hda/hda_intel.c:563: hda_intel: No response from codec, disabling MSI: last cmd=0x012f1c00
[ 2087.389000] ALSA sound/pci/hda/hda_intel.c:576: hda_intel: azx_get_response timeout, switching to polling mode: last cmd=0x012f1c00

(The exact time has varied anywhere from 123 seconds after boot to 10,502 secs).

And now /proc/interrupts looks like:

Sat Oct 6 23:26:59 EDT 2007
CPU0 CPU1

  0:    2438973          0   IO-APIC-edge      timer
  1:          9      10061   IO-APIC-edge      i8042
  8:          0          0   IO-APIC-edge      rtc
  9:          2          0   IO-APIC-fasteoi   acpi
 12:         86          0   IO-APIC-edge      i8042
 14:       5490      87784   IO-APIC-edge      libata
 15:         36          0   IO-APIC-edge      libata
 16:      30831          0   IO-APIC-fasteoi   nvidia
 19:          9          0   IO-APIC-fasteoi   ohci1394, yenta
 20:        125      55664   IO-APIC-fasteoi   ehci_hcd:usb1, uhci_hcd:usb2
 21:         91      13292   IO-APIC-fasteoi   uhci_hcd:usb3, HDA Intel
 22:          0          0   IO-APIC-fasteoi   uhci_hcd:usb4
 23:          0          0   IO-APIC-fasteoi   uhci_hcd:usb5
312:        149     101422   PCI-MSI-edge      eth0
313:       2378          0   PCI-MSI-edge      iwl3945
314:      15643          0      none-edge    
NMI:          0          0   Non-maskable interrupts
LOC:     395385    1640119   Local interrupts
RES:     675388     582673   Rescheduling interrupts
CAL:        429       2195   function call interrupts
TLB:       5776       4119   TLB shootdowns
TRM:          0          0   Thermal event interrupts
THR:          0          0   Threshold APIC interrupts
SPU:          0          0   Spurious interrupts
ERR:          0

Sound card has moved to IRQ 21, and 314 is left hanging in the breeze.

I've also seen the eth0 MSI IRQ evaporate, leaving *two* 'none-edge' IRQs, but I can't replicate it at the moment, as it seems to happen pseudo-randomly.

#1 lore raw
Quoted 2 lines · On Sun, 07 Oct 2007 00:51:32 EDT, [email protected] said:

I've also seen the eth0 MSI IRQ evaporate, leaving *two* 'none-edge' IRQs, but I can't replicate it at the moment, as it seems to happen pseudo-randomly.

Ignore this part - the problem is confined to the HDA-Intel driver.

I've verified the eth0 was a red herring - what would happen is if I booted it while undocked, the startup scripts would try to ifconfig it up, find no link on the device, and then ifconfig it down, at which point /proc/interrupts would show it as 'none-edge' because at one time there *had* been interrupts on it when it was 'PCI-MSI-edge', but it was then turned off again.