Since my kernel have been upgraded from 2.6.33.2 to 2.6.33.3 I can't mount my XFS partition. Here the backtrace: usb 3-1: new high speed USB device using ehci_hcd and address 4 Initializing USB Mass Storage driver... scsi20 : usb-storage 3-1:1.0 usbcore: registered new interface driver usb-storage USB Mass Storage support registered. scsi 20:0:0:0: Direct-Access SAMSUNG HM080HI PQ: 0 ANSI: 2 sd 20:0:0:0: Attached scsi generic sg4 type 0 sd 20:0:0:0: [sdc] 156301488 512-byte logical blocks: (80.0 GB/74.5 GiB) sd 20:0:0:0: [sdc] Write Protect is off sd 20:0:0:0: [sdc] Mode Sense: 38 00 00 00 sd 20:0:0:0: [sdc] Assuming drive cache: write through sd 20:0:0:0: [sdc] Assuming drive cache: write through sdc: sdc1 sdc2 sd 20:0:0:0: [sdc] Assuming drive cache: write through sd 20:0:0:0: [sdc] Attached SCSI disk tun: Universal TUN/TAP device driver, 1.6 tun: (C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com> tap0: no IPv6 routers present SGI XFS with ACLs, security attributes, realtime, large block/inode numbers, no debug enabled SGI XFS Quota Management subsystem BUG: unable to handle kernel NULL pointer dereference at 0000000000000048 IP: [<ffffffff810dd993>] filemap_write_and_wait+0x13/0x50 PGD 119d23067 PUD 12390d067 PMD 0 Oops: 0000 [#1] PREEMPT SMP last sysfs file: /sys/devices/pci0000:00/0000:00:03.0/0000:01:00.0/boot_vga CPU 2 Pid: 2890, comm: mount Tainted: P 2.6.33-ARCH #1 P55A-UD3R/P55A-UD3R RIP: 0010:[<ffffffff810dd993>] [<ffffffff810dd993>] filemap_write_and_wait+0x13/0x50 RSP: 0018:ffff8801237b3b98 EFLAGS: 00010246 RAX: ffff880129070150 RBX: 0000000000000000 RCX: 0000000000000000 RDX: 00000000000001ff RSI: 0000000000000001 RDI: 0000000000000000 RBP: ffff8801237b3ba8 R08: 0000000000000000 R09: ffff880123534f00 R10: 0000000000000080 R11: 2222222222222222 R12: 0000000000000200 R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000800 FS: 00007f9232dde740(0000) GS:ffff880005480000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000048 CR3: 00000001290c3000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process mount (pid: 2890, threadinfo ffff8801237b2000, task ffff8801235cc020) Stack: ffff880129071330 0000000000000200 ffff8801237b3bb8 ffffffff81151a1f <0> ffff8801237b3bc8 ffffffff81151a4e ffff8801237b3bf8 ffffffff8115254f <0> ffff8801237b3c38 0000000000000200 00000000000002d0 ffff880123534f00 Call Trace: [<ffffffff81151a1f>] __sync_blockdev+0x1f/0x40 [<ffffffff81151a4e>] sync_blockdev+0xe/0x10 [<ffffffff8115254f>] set_blocksize+0x6f/0xa0 [<ffffffffa0ebd3c1>] xfs_setsize_buftarg_flags+0x51/0xf0 [xfs] [<ffffffffa0ebabab>] ? kmem_alloc+0x6b/0x120 [xfs] [<ffffffffa0ebd498>] xfs_setsize_buftarg_early+0x38/0x40 [xfs] [<ffffffffa0ec0141>] xfs_alloc_buftarg+0x41/0xa0 [xfs] [<ffffffffa0ec6bc6>] xfs_open_devices+0x86/0x1a0 [xfs] [<ffffffff81152598>] ? sb_set_blocksize+0x18/0x60 [<ffffffffa0ec8385>] xfs_fs_fill_super+0x175/0x3e0 [xfs] [<ffffffff811253d4>] get_sb_bdev+0x194/0x1d0 [<ffffffffa0ec8210>] ? xfs_fs_fill_super+0x0/0x3e0 [xfs] [<ffffffffa0ec5f83>] xfs_fs_get_sb+0x13/0x20 [xfs] [<ffffffff81124e56>] vfs_kern_mount+0x76/0x1a0 [<ffffffff81124fed>] do_kern_mount+0x4d/0x130 [<ffffffff8113f506>] do_mount+0x266/0x880 [<ffffffff810f22df>] ? strndup_user+0x5f/0xb0 [<ffffffff8113fbab>] sys_mount+0x8b/0xe0 [<ffffffff81009fc2>] system_call_fastpath+0x16/0x1b Code: ff ff ff ff ff ff ff 7f 31 f6 48 89 e5 e8 46 f8 ff ff c9 c3 0f 1f 40 00 55 48 89 e5 48 83 ec 10 48 89 1c 24 4c 89 64 24 08 31 db <48> 83 7f 48 00 49 89 fc 75 13 89 d8 4c 8b 64 24 08 48 8b 1c 24 RIP [<ffffffff810dd993>] filemap_write_and_wait+0x13/0x50 RSP <ffff8801237b3b98> CR2: 0000000000000048 ---[ end trace 2283b9cb3afc6eda ]---
I can't say I've seen anything like this, nor can I see a potential cause looking at the changes between .33.2 and .33.3. Can you reproduce it without the module that causes the kernel taint? If you can, can you run a bisect between the two releases to find the commit that causes the problem?
In fact, it seems that my distribution (archlinux) has compiled the kernel .33.3 with gcc 4.5... I guess it wasn't a good idea... A look at http://bugs.archlinux.org/ shows how it wasn't a good idea :p So I think it's not a kernel bug.