Re: almost daily Kernel oops with 2.6.23.9 - and now 2.6.23.11 as well

[Ingo Molnar <mingo@elte.hu>]

* Pekka Enberg <penberg@cs.helsinki.fi> wrote:

> Nah, it's just that vma->anon_vma is probably supposed to be NULL 
> here. And if you look at all the oopses, they do suggest one 
> particular byte lane is dodgy (the corruption is in bits 41-43 and 
> 45).
> 
> The whole thing reminds me of another bug where memtest86 didn't find 
> anything because it's doing cached memory accesses: 
> http://lkml.org/lkml/2007/10/3/259

memtest86+ has an uncached test:

const struct tseq tseq[] = {
        {1,  5,  3,   0, 0, "[Address test, walking ones]          "},
        {1,  6,  3,   2, 0, "[Address test, own address]           "},
        {1,  0,  3,  14, 0, "[Moving inversions, ones & zeros]     "},
        {1,  1,  2,  80, 0, "[Moving inversions, 8 bit pattern]    "},
        {1, 10, 60, 300, 0, "[Moving inversions, random pattern]   "},
        {1,  7, 64,  66, 0, "[Block move, 64 moves]                "},
        {1,  2,  2, 320, 0, "[Moving inversions, 32 bit pattern]   "},
        {1,  9, 40, 120, 0, "[Random number sequence]              "},
        {1,  3,  4, 240, 0, "[Modulo 20, ones & zeros]             "},
        {1,  8,  1,   2, 0, "[Bit fade test, 90 min, 2 patterns]   "},
        {0,  4,  3,   2, 0, "[[Moving inversions, 0 & 1, uncached] "},
        {0,  0,  0,   0, 0, NULL}
};

find that "Moving inversions, 0 & 1" test and run that one alone, 
overnight.

	Ingo