Skip to contentSkip to conversation

2.6.0-test7-mm1

DISCUSSION Steven Pratt 5 participants 6 messages Oct 16, 2003 – Oct 19, 2003

lore
Share & tools
lore

Steven Pratt

author
#0 lore raw

On reboot after heavy IO loads (tiobench) I keep getting the following oops. Happens right after the "turning off swap" message. Root FS is ext3, but the oops has happened while testing ext2 and ext3 with tiobench(xfs, jfs and rieser still to come).

kernel BUG at include/linux/list.h:148!
invalid operand: 0000 [#1]
SMP

CPU:    2
EIP:    0060:[<c01756f7>]    Not tainted VLI
EFLAGS: 00010213

EIP is at __iget+0x67/0x80
eax: 00000000 ebx: f7528cb8 ecx: f767359c edx: f7673594
esi: c6668400 edi: 0005dc97 ebp: f7c445f8 esp: f6c6fe40
ds: 007b es: 007b ss: 0068
Process S20reboot (pid: 20537, threadinfo=f6c6e000 task=f752d900)

Stack: f7673594 c017644b f7673594 f7c445f8 0005dc97 0005dc97 f29ee2c0 c6668400
      f29ee2c0 c019d16b c6668400 0005dc97 f76f61dc fffffff4 f76ffa28 
f76ff9b4
      c0169cef f76ff9b4 f29ee2c0 f6c6ff38 00000000 f6c6ff38 f7fde760 
f6c6fee4
Call Trace:
[<c017644b>] iget_locked+0x6b/0xc0
[<c019d16b>] ext3_lookup+0x6b/0xd0
[<c0169cef>] real_lookup+0xcf/0x100
[<c0169fa6>] do_lookup+0xa6/0xc0
[<c016a413>] link_path_walk+0x453/0x8b0
[<c016ad99>] __user_walk+0x49/0x60
[<c0165d6f>] vfs_stat+0x1f/0x60
[<c016650b>] sys_stat64+0x1b/0x40
[<c04103df>] syscall_call+0x7/0xb

Code: 04 a1 10 92 48 c0 89 48 04 89 42 08 c7 41 04 10 92 48 c0 89 0d 10 92 48

Steve

#1 lore raw
Quoted 11 lines · Steven Pratt

On reboot after heavy IO loads (tiobench) I keep getting the following oops. Happens right after the "turning off swap" message. Root FS is ext3, but the oops has happened while testing ext2 and ext3 with tiobench(xfs, jfs and rieser still to come).
kernel BUG at include/linux/list.h:148!
invalid operand: 0000 [#1]
SMP

CPU:    2
EIP:    0060:[<c01756f7>]    Not tainted VLI
EFLAGS: 00010213

EIP is at __iget+0x67/0x80

Could you try with the invalidate_inodes-speedup.patch backed out?

-- wli

#2 lore raw
In reply to Steven Pratt
Quoted 3 lines · Steven Pratt

On reboot after heavy IO loads (tiobench) I keep getting the following
oops.

Yup. The "invalidate_inodes-speedup-fixes" and "invalidate_inodes-speedup" patches were not so great and need to be reverted.

author
#3 lore raw
Quoted 10 lines · Andrew Morton
Quoted 3 lines · Steven Pratt

On reboot after heavy IO loads (tiobench) I keep getting the following oops.

Yup. The "invalidate_inodes-speedup-fixes" and "invalidate_inodes-speedup" patches were not so great and need to be reverted.

Looks like that fixed it!

Steve

#4 lore raw
In reply to Andrew Morton
Quoted 5 lines
Quoted 2 lines

On reboot after heavy IO loads (tiobench) I keep getting the following
oops.

Yup. The "invalidate_inodes-speedup-fixes" and "invalidate_inodes-speedup" patches were not so great and need to be reverted.

I've found that fs/hugetlbfs/inode.c miss list_del(&inode->i_sb_list) when delete/forget inode is called. Probably this can break sb_list and might be the cause.

Kirill

#5 lore raw
In reply to Steven Pratt

I've noticed a wierd thing with 2.6.0-test7-mm1; the disk cache grows wildly. I'm just sitting in X (Gnome 2.4) with maybe a dozen temrinals open, mozilla and a few bits, and when runnign "grep blah *" in a directory with maybe a few hundred megs of data in it, the grep progress dies with memory exhaustion, and using top to view memory consuption, it seems that the disk cache grows uncontrollably.

My machine has 768Mb of RAM and the disk cache greows from ~200Mb to ~600Mb, while X goes very sluggish and then the grep dies.

.config:

dmesg:

(And yes, this happens both with and without the VMware modules)

mvh,
A

Signature · 3 lines
Alexander Hoogerhuis                               | [email protected]
CCNP - CCDP - MCNE - CCSE                          | +47 908 21 485
"You have zero privacy anyway. Get over it."  --Scott McNealy