2.5.62 oops
Running a stress test of 50 concurrent kernel tree cp's + md5sum the files on 2.5.62 results in the following:
Feb 17 21:53:41 livewire kernel: VFS: brelse: Trying to free free buffer Feb 17 21:53:41 livewire kernel: buffer layer error at fs/buffer.c:1170 Feb 17 21:53:41 livewire kernel: Pass this trace through ksymoops for reporting Feb 17 21:53:42 livewire kernel: Call Trace: [__brelse+53/64]
[bh_lru_install+169/224] [__find_get_block+113/128] [__getblk+43/96] [ext3_getblk+171/752] [ext3_bread+51/176] [dx_probe+158/784] [ext3_bread+51/176] [ext3_dx_find_entry+183/560] [ext3_find_entry+989/1040] [journal_stop+346/496] [ext3_dirty_inode+147/256] [d_alloc+32/496] [ext3_lookup+90/256] [real_lookup+192/240] [do_lookup+159/176] [link_path_walk+833/1584] [__user_walk+73/96] [vfs_lstat+28/96] [sys_lstat64+27/64] [syscall_call+7/11]
Feb 17 21:53:42 livewire kernel: Call Trace: [<c0149e85>] [<c014a049>] [<c014a0f1>] [<c014a12b>] [<c017edab>] [<c017f023>] [<c01824be>] [<c017f023>] [<c0183227>] [<c018313d>] [<c018b81a>] [<c0181b83>] [<c015de20>] [<c01833fa>] [<c0154880>] [<c0154abf>] [<c0154e11>] [<c0155559>] [<c015095c>] [<c0150ffb>] [<c010941f>]
Quoted 8 lines · Bob Johnson
Running a stress test of 50 concurrent kernel tree cp's + md5sum the files on 2.5.62 results in the following:
Feb 17 21:53:41 livewire kernel: VFS: brelse: Trying to free free buffer Feb 17 21:53:41 livewire kernel: buffer layer error at fs/buffer.c:1170 Feb 17 21:53:41 livewire kernel: Pass this trace through ksymoops forreporting
It's not really an oops, although the bug which this is reporting could cause a crash.
It is a known problem in the htree indexed directory code. I can reproduce it, but haven't had time to work on it. Chris Li has time to work on it, but cannot reproduce it.
We'll get there.