Re: 2.6.28-rc2: REGRESSION in early boot

["Yinghai Lu" <yinghai@kernel.org>]

On Wed, Nov 5, 2008 at 10:27 PM, Yinghai Lu <yinghai@kernel.org> wrote:
> Theodore Tso wrote:
>> On Wed, Nov 05, 2008 at 09:22:38AM -0800, Yinghai Lu wrote:
>>>> What is your theory for why this was causing a failure in the timer
>>>> subsystem?  Were we overrunning the end of irq_desc array and stomping
>>>> on some other data structure?
>>> Yes. Please do apply that two patches at first.
>>>
>>
>> Success!  Applying these two patches, as found at:
>>
>
> did you apply
> http://lkml.org/lkml/2008/11/4/431
>
> it seems we aready incread NR_IRQS big enough, but still get...
>
> [    0.000000] ------------[ cut here ]------------
> [    0.000000] WARNING: at arch/x86/kernel/io_apic.c:3614 probe_nr_irqs+0x44/0x4d()
> [    0.000000] Modules linked in:
> [    0.000000] Pid: 0, comm: swapper Not tainted 2.6.28-rc3-00123-g72a0780 #97
> [    0.000000] Call Trace:
> [    0.000000]  [<c01418f1>] warn_on_slowpath+0x46/0x60
> [    0.000000]  [<c0160a3e>] ? __lock_acquire+0x3d9/0xa5b
> [    0.000000]  [<c06dd463>] ? alloc_bootmem_core+0x1bf/0x221
> [    0.000000]  [<c012db8b>] ? native_pmd_val+0x8/0xa
> [    0.000000]  [<c01324a2>] ? pmd_val+0x11/0x21
> [    0.000000]  [<c011bb31>] ? sched_clock+0xc/0x1c
> [    0.000000]  [<c015e5fb>] ? lock_release_holdtime+0x43/0x48
> [    0.000000]  [<c016142c>] ? lock_release+0x14c/0x155
> [    0.000000]  [<c015e46b>] ? trace_hardirqs_off_caller+0x79/0x84
> [    0.000000]  [<c04ab040>] ? _spin_unlock_irqrestore+0x3e/0x55
> [    0.000000]  [<c015e5b6>] ? trace_hardirqs_off+0xb/0xd
> [    0.000000]  [<c04ab040>] ? _spin_unlock_irqrestore+0x3e/0x55
> [    0.000000]  [<c06d5853>] probe_nr_irqs+0x44/0x4d
> [    0.000000]  [<c06cbfe8>] setup_arch+0xa90/0xb00
> [    0.000000]  [<c015e46b>] ? trace_hardirqs_off_caller+0x79/0x84
> [    0.000000]  [<c015e5b6>] ? trace_hardirqs_off+0xb/0xd
> [    0.000000]  [<c015e91a>] ? lockdep_init_map+0x6b/0xc7
> [    0.000000]  [<c06c65e2>] start_kernel+0x85/0x30e
> [    0.000000]  [<c06c6080>] __init_begin+0x80/0x88
> [    0.000000] ---[ end trace 8305b6ae03c24dde ]---
>

ah, you have
CONFIG_PARAVIRT=y

YH