Re: [RFC 0/5] percpu: Optimize percpu accesses

[Ingo Molnar]

* Ingo Molnar <mingo@xxxxxxx> wrote:

> > This patchset provides the following:
> > 
> >   * Cleanup: Fix early references to cpumask_of_cpu(0)
> >   * Generic: Percpu infrastructure to rebase the per cpu area to zero
> >   * x86_64: Fold pda into per cpu area
> >   * x86_64: Rebase per cpu variables to zero
> 
> thanks Mike - i've started testing it in -tip. One small merge fallout 
> fix is below. (it's for the new generic-ipi topic that was not in 
> tip/master yet when you merged percpu-zerobased to it)

ok, -tip testing found an early boot crash caused by your patchset, on 
64-bit x86:

[    0.396000] calling  net_ns_init+0x0/0x143
[    0.400000] net_namespace: 944 bytes
[    0.403578] BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
[    0.404000] IP: [<ffffffff80e05b0e>] net_ns_init+0x9f/0x143
[    0.404000] PGD 0
[    0.404000] Oops: 0000 [1] SMP
[    0.404000] CPU 0
[...]

full bootlog and config can be found at:

  http://redhat.com/~mingo/misc/crashlog-Tue_Jul__1_16_48_45_CEST_2008.bad
  http://redhat.com/~mingo/misc/config-Tue_Jul__1_16_48_45_CEST_2008.bad

(another 64-bit testbox crashed as well, so this should be readily 
reproducible.)

i've pushed this tree out to tip/tmp.core/percpu-zerobased.Jul__1_16_48 
topic branch, that is the 2.6.26-rc8-tip-00250-g90874b0 kernel you can 
see in the crashlog.

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/