Most recent kernel where this bug did not occur: unknown Distribution: Debian sid Hardware Environment: Tyan s3850, Software Environment: Problem Description: BUG: warning at fs/buffer.c:575/__remove_assoc_queue() Call Trace: [<ffffffff80288384>] drop_buffers+0xac/0xf4 [<ffffffff8028842c>] try_to_free_buffers+0x60/0x9f [<ffffffff8035fe4e>] xfs_vm_releasepage+0x98/0xa5 [<ffffffff802518d8>] shrink_inactive_list+0x4af/0x7d5 [<ffffffff802505d0>] __pagevec_release+0x19/0x22 [<ffffffff8025125b>] shrink_active_list+0x448/0x456 [<ffffffff80251ce0>] shrink_zone+0xe2/0x108 [<ffffffff8025296f>] kswapd+0x2f5/0x416 [<ffffffff8023ab74>] autoremove_wake_function+0x0/0x2e [<ffffffff8023a7c5>] keventd_create_kthread+0x0/0x61 [<ffffffff8025267a>] kswapd+0x0/0x416 [<ffffffff8023a7c5>] keventd_create_kthread+0x0/0x61 [<ffffffff8023aa38>] kthread+0xd1/0x101 [<ffffffff8020a518>] child_rip+0xa/0x12 [<ffffffff8023a7c5>] keventd_create_kthread+0x0/0x61 [<ffffffff80216b12>] flat_send_IPI_mask+0x0/0x3d [<ffffffff8023a967>] kthread+0x0/0x101 [<ffffffff8020a50e>] child_rip+0x0/0x12 Steps to reproduce: unknown; happened once on a server that runs a repetitive schedule of tasks; non-fatal.
XFS never uses the bh->b_assoc_map or calls mark_buffer_inode_dirty() so I fail to see how your kernel is getting there from XFS. I suspect that there's a memory corruption problem here - maybe you should turn on memory debugging options and see if that turns anything up. FWIW, are you running any other filesystems on that machine?
Thanks for the comment. I believe the machine has hardware issues -- something as trivial as a poorly seated PCI card. I regret the incorrect report.