Bug 13151
| Summary: | Warning about reiserfs | ||
|---|---|---|---|
| Product: | File System | Reporter: | Dan (dan76) |
| Component: | ReiserFS | Assignee: | ReiseFS developers team (reiserfs-devel) |
| Status: | CLOSED CODE_FIX | ||
| Severity: | low | CC: | 1i5t5.duncan, jeffm, maluta_tiago, mrb74, rjw, sergey.senozhatsky, stefanr |
| Priority: | P1 | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.30-rc3 | Subsystem: | |
| Regression: | Yes | Bisected commit-id: | |
| Bug Depends on: | |||
| Bug Blocks: | 13070 | ||
| Attachments: | [PATCH] reiserfs: Expand i_mutex to enclose lookup_one_len | ||
|
Description
Dan
2009-04-22 19:07:51 UTC
marked as a regression.. This was introduced by commit 2f9092e1020246168b1309b35e085ecd7ff9ff72 expecting the dir inode to be locked when lookup_one_len is called. I can expand the locking around the calls. First-Bad-Commit : 2f9092e1020246168b1309b35e085ecd7ff9ff72 Handled-By : Jeff Mahoney <jeffm@suse.com> Hi. 2.6.30-rc3 Same problem: [ 53.960396] REISERFS (device sda8): using ordered data mode [ 53.960899] REISERFS (device sda8): journal params: device sda8, size 8192, j ournal first block 18, max trans len 1024, max batch 900, max commit age 30, max trans age 30 [ 53.961486] REISERFS (device sda8): checking transaction log (sda8) [ 54.003083] REISERFS (device sda8): Using r5 hash to sort names [ 54.003295] WARNING: at fs/namei.c:1251 lookup_one_len+0xf0/0x110() [ 54.006379] Pid: 2118, comm: mount Not tainted 2.6.30-rc3-test #1 [ 54.006460] Call Trace: [ 54.006537] [<c04b91d1>] ? printk+0x23/0x36 [ 54.006616] [<c015f0a3>] warn_slowpath+0x93/0xe0 [ 54.006696] [<c03d1878>] ? vt_console_print+0x218/0x2e0 [ 54.006777] [<c0149ff7>] ? default_spin_lock_flags+0x17/0x30 [ 54.006860] [<c0282bad>] ? do_journal_end+0x1dd/0xe50 [ 54.006941] [<c02091d0>] lookup_one_len+0xf0/0x110 [ 54.007030] [<c0178960>] ? autoremove_wake_function+0x0/0x60 [ 54.007112] [<c0289569>] reiserfs_xattr_init+0xe9/0x260 [ 54.007192] [<c0178938>] ? wake_up_bit+0x68/0x90 [ 54.007274] [<c0273f29>] reiserfs_fill_super+0xa99/0xe10 [ 54.007367] [<c035bce9>] ? string+0x59/0x100 [ 54.007446] [<c035cda5>] ? snprintf+0x25/0x40 [ 54.007524] [<c0253556>] ? disk_name+0x66/0xd0 [ 54.007603] [<c0200aed>] get_sb_bdev+0x13d/0x190 [ 54.007681] [<c0273490>] ? reiserfs_fill_super+0x0/0xe10 [ 54.007761] [<c0270e9d>] get_super_block+0x2d/0x50 [ 54.007840] [<c0273490>] ? reiserfs_fill_super+0x0/0xe10 [ 54.007919] [<c0200601>] vfs_kern_mount+0x71/0x160 [ 54.007999] [<c020077d>] do_kern_mount+0x4d/0xf0 [ 54.008088] [<c0219cb0>] do_mount+0x450/0x790 [ 54.008166] [<c021a085>] sys_mount+0x95/0xf0 [ 54.008245] [<c01290d3>] sysenter_do_call+0x12/0x28 [ 54.008326] ---[ end trace 7c022b0b46477b0f ]--- Sergey. I have it too. 2.6.30-rc3-341-ge4f062f, incl. cherrypicked e4f062ff9e3c995f3b6f20848a272c8f0fa210e7 , x86: hpet: fix periodic mode programming on AMD 81xx, from tip/x86/urgent to fix #13176 which I had as well). I'm also getting a machine-check taint (tainted G, M), however, that the above are not, and that I don't get on 2.6.29.1. This seems to be consistent as I had the tainted, lockdep disabled warning in the log while I was crashing before the above cherrypick. Since the others don't have it it's likely OT for this bug, but can someone point me at some instructions for tracing it down further, and are there other 2.6.30 bugs known to trigger MCEs or taint M,G anyway? I don't see anything else in the log before the tainted/lockdeps-disabled warning indicating a problem, with it being rather early in the kernel boot process, so whatever it is isn't leaving me much to go on. Still present at 2.6.30-rc4. Created attachment 21183 [details]
[PATCH] reiserfs: Expand i_mutex to enclose lookup_one_len
2.6.30-rc3 introduced some sanity checks in the VFS code to avoid NFS
bugs by ensuring that lookup_one_len is always called under i_mutex.
This patch expands the i_mutex locking to enclose lookup_one_len. This was
always required, but not not enforced in the reiserfs code since it
does locking around the xattr interactions with the xattr_sem.
(In reply to comment #7) > Created an attachment (id=21183) [details] > [PATCH] reiserfs: Expand i_mutex to enclose lookup_one_len Thanks Jeff. Your patch works fine here. > --- Comment #7 from Jeff Mahoney <jeffm@suse.com> 2009-05-01 15:59:51 ---
> Created an attachment (id=21183)
> --> (http://bugzilla.kernel.org/attachment.cgi?id=21183)
> [PATCH] reiserfs: Expand i_mutex to enclose lookup_one_len
>
Hello Jeff.
No crashes during boot at the moment.
(Need to test it more).
Thanks.
Sergey.
(In reply to comment #7) > Created an attachment (id=21183) [details] > [PATCH] reiserfs: Expand i_mutex to enclose lookup_one_len Hi Jeff, this patch wasn't merged in 2.6.30-rc5. Will it be merged before 2.6.30 final release? Thanks. > Hi Jeff, this patch wasn't merged in 2.6.30-rc5. Will it be merged before
> 2.6.30 final release? Thanks.
>
Hi Daniel. Check out 2.6.30-rc5-git1.
Sergey.
Handled-By : Jeff Mahoney <jeffm@suse.com> Patch : http://bugzilla.kernel.org/attachment.cgi?id=21183 *** Bug 13301 has been marked as a duplicate of this bug. *** *** Bug 13299 has been marked as a duplicate of this bug. *** *** Bug 13295 has been marked as a duplicate of this bug. *** *** Bug 13251 has been marked as a duplicate of this bug. *** Fix has been merged by Linux before 2.6.30-rc6. |