Bug 7847 - Radeon 9200 DRI driver IRQ, but no-one cared
Summary: Radeon 9200 DRI driver IRQ, but no-one cared
Status: REJECTED INVALID
Alias: None
Product: Drivers
Classification: Unclassified
Component: Video(DRI - non Intel) (show other bugs)
Hardware: i386 Linux
: P2 normal
Assignee: drivers_video-dri
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-01-17 13:50 UTC by Chris Rankin
Modified: 2008-09-24 11:18 UTC (History)
0 users

See Also:
Kernel Version: 2.6.18.6
Subsystem:
Regression: ---
Bisected commit-id:


Attachments

Description Chris Rankin 2007-01-17 13:50:30 UTC
Most recent kernel where this bug did *NOT* occur:
2.6.18.5 (but maybe I was lucky?)

Distribution:
kernel.org kernel, userspace is FC5, Xorg from FC6, Mesa 6.5.2

Hardware Environment:
Dual P4 Xeon, HT-enabled, 2GB RAM

Software Environment:
Playing WoW using Wine 0.9.29-CVS, Mesa 6.5.2

Problem Description:
Was happily playing WoW, when suddenly the LCD monitor went dead. I found this
message on the serial console:

irq 19: nobody cared (try booting with the "irqpoll" option)
 [<c013211f>] __report_bad_irq+0x2b/0x69
 [<c01322fb>] note_interrupt+0x19e/0x1d7
 [<c013192b>] handle_IRQ_event+0x23/0x49
 [<c01319ff>] __do_IRQ+0xae/0xe6
 [<c0104e44>] do_IRQ+0x6e/0x89
 =======================
 [<c010145e>] default_idle+0x0/0x59
 [<c0103372>] common_interrupt+0x1a/0x20
 [<c010145e>] default_idle+0x0/0x59
 [<c010148f>] default_idle+0x31/0x59
 [<c0101526>] cpu_idle+0x6f/0x98
handlers:
[<f88682e5>] (usb_hcd_irq+0x0/0x50 [usbcore])
[<f8e82885>] (radeon_driver_irq_handler+0x0/0x7f [radeon])
Disabling IRQ #19

I am assuming that the IRQ assignments are the same:

           CPU0       CPU1       CPU2       CPU3
  0:     797196     790792     790635     790620    IO-APIC-edge  timer
  1:       1909       1807       1949       2272    IO-APIC-edge  i8042
  4:         11        474          0       1431    IO-APIC-edge  serial
  6:          3          0          0          0    IO-APIC-edge  floppy
  7:          0          0          0          0    IO-APIC-edge  parport0
  8:          1          0          0          0    IO-APIC-edge  rtc
  9:          0          0          0          0   IO-APIC-level  acpi
 12:       9586      26063      11191      23177    IO-APIC-edge  i8042
 14:      18809      18352      17843      10340    IO-APIC-edge  ide0
 15:      17449      92821      17036      93280    IO-APIC-edge  ide1
 16:          0          0          0          0   IO-APIC-level  uhci_hcd:usb3
 17:      25420          0          0          0   IO-APIC-level  eth0
 19:     197311     182715     196433     182636   IO-APIC-level  uhci_hcd:usb1,
radeon@pci:0000:01:00.0
 20:          0          0          0          0   IO-APIC-level  uhci_hcd:usb2
 21:    4730917      29724    4739107      24324   IO-APIC-level  ehci_hcd:usb4
 22:       1839       1680       1549       1451   IO-APIC-level  libata, ohci1394
 23:          3          0          0          0   IO-APIC-level  ohci1394
 24:       1210        644        799        730   IO-APIC-level  EMU10K1
NMI:    3169105    3169076    3169081    3169080
LOC:    3168977    3168978    3168989    3168988
ERR:          0
MIS:          0

Mind you, I think I had an unused Nova-T USB2 DVB Freeview adaptor plugged in at
the same time. But it was my monitor that went dead.

Steps to reproduce:
Play WoW for many, many, many hours, with a Radeon 9200, XOrg 7.1.1 and Mesa 6.5.2
Comment 1 Chris Rankin 2007-01-18 01:07:18 UTC
It's possible that this bug happened because I had a device plugged into the
UHCI root hub that was sharing its IRQ with the Radeon 9200. It's not as if I
haven't played WoW for many hours before. (But I'm sure that it was a USB 2.0
device, so shouldn't that have been triggering the EHCI IRQ instead?)

Could there be a race condition in the IRQ handling?

The DRI modules that I am using with 2.6.18.6 are from the DRI CVS repository,
with "VAP state" and "integer overflow in FB location" patches applied. In other
words, the DRI modules are closer to those in 2.6.19. The native 2.6.18.6
modules don't work at all, but that's a known issue.
Comment 2 Alan 2008-09-24 11:18:57 UTC
Closing out stale bugs, please re-open if still present

Note You need to log in before you can comment on or make changes to this bug.