[rt2x00-users] rt61pci driver fails after variable period of time

[Helmut Schaa]

Am Dienstag, 26. Juli 2011 schrieb Mike Gibson:
> My wireless card stops working after a random period of time.
> Reloading the rt61pci module seems to bring it back.  But sometimes,
> reloading locks up the machine.  The card worked fine when I was using
> Ubuntu 10.10 (2.6.35) but it was immediately after upgrading to Ubuntu
> 11.04 (2.6.38) that things stopped working reliably.  I've also posted
> this bug to ubuntu launchpad:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/803708
> 
> These seem to be the relevant lines in dmesg:
> 
> [266411.791394] irq 18: nobody cared (try booting with the "irqpoll" option)
> [266411.791401] Pid: 0, comm: swapper Tainted: P 2.6.38-8-generic-pae #42-Ubuntu
> [266411.791404] Call Trace:
> [266411.791413] [<c10baccc>] ? __report_bad_irq.clone.2+0x2c/0x90
> [266411.791417] [<c10b998c>] ? handle_IRQ_event+0x4c/0x160
> [266411.791420] [<c10bb01a>] ? note_interrupt+0x15a/0x1a0
> [266411.791425] [<c10bbc3c>] ? handle_fasteoi_irq+0xac/0xd0
> [266411.791428] [<c10bbb90>] ? handle_fasteoi_irq+0x0/0xd0
> [266411.791431] <IRQ> [<c1538952>] ? do_IRQ+0x42/0xc0
> [266411.791440] [<c105fbb0>] ? irq_exit+0x60/0x80
> [266411.791444] [<c1538a2b>] ? smp_apic_timer_interrupt+0x5b/0x8a
> [266411.791449] [<c100b0f0>] ? common_interrupt+0x30/0x38
> [266411.791454] [<c105007b>] ? proc_sched_show_task+0x14b/0x1ac0
> [266411.791460] [<c12cfeb8>] ? intel_idle+0xb8/0x110
> [266411.791466] [<c142fa5d>] ? cpuidle_idle_call+0x7d/0x160
> [266411.791469] [<c100944a>] ? cpu_idle+0x8a/0xc0
> [266411.791474] [<c10420fe>] ? complete+0x4e/0x60
> [266411.791479] [<c151730d>] ? rest_init+0x5d/0x70
> [266411.791484] [<c17c47f2>] ? start_kernel+0x35f/0x366
> [266411.791487] [<c17c43e6>] ? pass_all_bootoptions+0x0/0xa
> [266411.791491] [<c17c40f1>] ? i386_start_kernel+0xe0/0xe8
> [266411.791494] handlers:
> [266411.791495] [<f873fe20>] (rt61pci_interrupt+0x0/0x80 [rt61pci])
> [266411.791502] Disabling IRQ #18
> 
> If the right people don't have the time to fix this immediately, I
> would be willing to help out.  I have done some limited kernel work
> myself, but related to the block device layer, not wifi or networking.
>  I just need some pointers on what I might try or where to start
> looking.  What exactly is happening when the "irq ##: nobody cared"
> message is printed?  Also, I tried booting with irqpoll and the card
> was even less reliable then.

Just a shot in the dark but you might want to try my RFC patch [1] that fixes
a race in interrupt processing? Not sure though if your problem is really
related but might be worth a try ...

Helmut

[1] http://rt2x00.serialmonkey.com/pipermail/users_rt2x00.serialmonkey.com/2011-July/004033.html