This was a diskless install from altinstall.xml, I built an SMP kernel with JFS enabled. I think this is why the install failed the first time, so this error should be reproducible (install with JFS on SMP kernel up to emerge system/coreutils. CFLAGS=O2, march=i686, fomit-frame-pointer, pipe >>>sys-apps/coreutils-5.2.1 merged. OOPS: #0002 [#1] PREEMPT SMP Modules linked in: CPU: 0 EIP: 0060[<c0268427>] Not Tainted. EFLAGS 00010282 (2.6.7-gentoo-r9) EIP is at txUnlock+0x67/0x360 eax: 00000aff ebx: ddd18f2c ecx: f880c100 edx: f88019d8 esi: f8824800 edi: 00000aff ebp: f7c38000 esp: f7c39dd0 ds: 0076 es: 0076 ss: 0088 Process jfsCommit (pid: 40, threadinfo=f7c38000 task=c20f2050) Stack: 0000007b f8824800 00000000 f8838fc0 ddd18edc f88014d8 00000001 f7cb02b8 00000aff 000005e0 f88014d8 00000001 00000000 00000010 00000001 c0268db4 f88014d8 f88014d4 f7c39e48 00000000 00000001 f88014d8 f7cb02b8 0000003a Call Trace: [<c0268db4>] txCommit+0x2e4/0x340 [<c0254867>] diFree+0x5d7/0xa60 [<c0249e60>] jfs_delete_inode+0x0/0x40 [<c0249e7d>] jfs_delete_inode+0x1d/0x40 [<c017dc35>] generic_delete_inode+0x75/0x140 [<c017dcf3>] iput+0x63/0x90 [<c0269b3d>] txUpdateMap+0xbd/0x250 [<c026a391>] txLazyCommit+0x21/0x100 [<c026a65d>] jfs_lazycommit+0x1ed/0x230 [<c011be70>] default_wake_function+0x0/0x20 [<c0105162>] ret_from_fork+0x6/0x14 [<c011be70>] default_wake_function+0x0/0x20 [<c026a470>] jfs_lazy_commit+0x0/0x230 [<c01032c5>] kernel_thread_helper+0x5/0x10 Code: 81 58 56 b5 c1 00 0f 85 c4 58 56 b5 c1 53 01 10 0f 58 56 b5 Reproducible: Always Steps to Reproduce: 1. install diskless, etc... emerge system 2. 3.
Can you try a vanilla kernel.org kernel [ development-sources ] and see if you can reproduce the error with that please?
this box has been a real problem.. i can try vanilla tomorrow, but since i reporte3d this bug i have tried it with ext3 instead of jfs and i am getting kernel oops's every time i run some apps. i ran memtest86 and it didn't make it through a complete pass (unexpected interruption i think). so i'm not sure how much vanilla sources would help, but i will surely try it. at this point though, i am starting to think there are hardware problems.
this is undeniably a hardware bug at this point, memtest86 died at the same spot as last time.. should I just close this?
I guess so; not much we can do here even if this was a JFS issue since it would be a bug in the upstream code which we don't maintain - if you find you still happen to have issues like this after solving the hardware problems, then feel free to reopen this bug or open another one at http://bugme.osdl.org for the upstream developers to look at. Thanks!
