Badness on the Warp

[Sean MacLennan <smaclennan@pikatech.com>]

[    0.000000] ------------[ cut here ]------------
[    0.000000] Badness at c033ebc4 [verbose debug info unavailable]
[    0.000000] NIP: c033ebc4 LR: c033eb94 CTR: 00000000
[    0.000000] REGS: c037fe70 TRAP: 0700   Not tainted  (2.6.30-pika)
[    0.000000] MSR: 00021000 <ME,CE>  CR: 22022024  XER: 00000000
[    0.000000] TASK = c035b440[0] 'swapper' THREAD: c037e000
[    0.000000] <6>GPR00: 00000001 c037ff20 c035b440 00000000 0000000c 00000020 3fffffff 00000000 
[    0.000000] <6>GPR08: c036faa4 c0380000 c07b7586 00000000 22002022 00000000 0ffa7f00 007fff99 
[    0.000000] <6>GPR16: 00400450 00800000 007fff00 0ffa7a90 00000000 ffffffff 00000000 c0351d40 
[    0.000000] <6>GPR24: 00000000 00001229 c0386b1c 0000000c 00000020 00000000 3fffffff 0000000c 
[    0.000000] NIP [c033ebc4] alloc_arch_preferred_bootmem+0x48/0x74
[    0.000000] LR [c033eb94] alloc_arch_preferred_bootmem+0x18/0x74
[    0.000000] Call Trace:
[    0.000000] [c037ff20] [c0350fa8] 0xc0350fa8 (unreliable)
[    0.000000] [c037ff30] [c033ec40] ___alloc_bootmem_nopanic+0x50/0x108
[    0.000000] [c037ff60] [c033ed10] ___alloc_bootmem+0x18/0x50
[    0.000000] [c037ff70] [c033b3bc] uic_init_one+0x40/0x22c
[    0.000000] [c037ff90] [c033b630] uic_init_tree+0x88/0x168
[    0.000000] [c037ffb0] [c0337378] init_IRQ+0x28/0x40
[    0.000000] [c037ffc0] [c0334664] start_kernel+0x15c/0x2a0
[    0.000000] [c037fff0] [c0000200] skpinv+0x190/0x1cc
[    0.000000] Instruction dump:
[    0.000000] 2f830000 38600000 409e0018 80010014 83e1000c 7c0803a6 38210010 4e800020 
[    0.000000] 3d20c038 80095408 21600000 7c0b0114 <0f000000> 2f800000 409e0018 38800000 


I wouldn't trust the backtrace too much since it seems to keep
changing. It used to show __start_notes as the backtrace. This one is
with the head of Linus' tree as of a few minutes ago.

So I did a git bisect, and the problem goes back to June 11. The Warp
boots and seems to run fine. So I didn't notice until looking at the
logs for a different problem. I have since added a test to my
regression tests to look for badness or oops in the logs.

The git bisect returned:

  871fa90791a6f83dd8e2e489feb9534a8c02088d is first bad commit

That is it, no more info.... strange. git show 8714...088d gives:

Merge: 7702667... 79f52b7...
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu Jun 11 11:27:09 2009 -0700

    Merge branch 'for-linus' of
    git://git.kernel.org/pub/scm/linux/kernel/git/shaggy/jfs-2.6 
    * 'for-linus' of
    git://git.kernel.org/pub/scm/linux/kernel/git/shaggy/jfs-2.6: jfs:
    Add missing mutex_unlock call to error path missing unlock in
    jfs_quota_write()

The Warp has jfs disabled, so I don't know why it would affect us,
especially this realy in the boot! Anyway, I tried to revert the patch:

$ git revert 871fa90791a6f83dd8e2e489feb9534a8c02088d
fatal: Commit 871fa90791a6f83dd8e2e489feb9534a8c02088d is a merge but no -m option was given.

So, anybody got an idea what is going on here? Or how I can revert the
patch. I am not a git expert, so if I am doing this wrong, let me know.

Cheers,
   Sean