Bug 210233

Summary:	sys-kernel/gentoo-sources-2.6.23-r8: lockups with forking
Product:	Gentoo Linux	Reporter:	jergendutch
Component:	[OLD] Core system	Assignee:	Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel>
Status:	RESOLVED NEEDINFO
Severity:	normal
Priority:	High
Version:	unspecified
Hardware:	All
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---

Description jergendutch 2008-02-15 11:01:07 UTC

We have a script that forks a bunch of children (up to ten) and runs some transformations on images.

Since upgrading from 2.6.22-r9 to 2.6.23-r8 we have noticed soft lockups, such as:
BUG: soft lockup - CPU#1 stuck for 11s! [watchdirectory:21493]

Pid: 21493, comm:       watchdirectory
EIP: 0060:[<c039f78b>] CPU: 1
EIP is at __write_lock_failed+0x7/0x1c
EFLAGS: 00200287    Tainted: G      D  (2.6.23-gentoo-r8 #1)
EAX: c0570410 EBX: f6c6f6c0 ECX: c057049c EDX: f6c6f6ec
ESI: f74845c0 EDI: dc9e7eb2 EBP: f7b0d740 DS: 007b ES: 007b FS: 00d8
CR0: 8005003b CR2: b7d396cb CR3: 303fd000 CR4: 000006d0
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
[<c039fa5e>] _write_lock+0xe/0x10
[<c01fc55d>] cifs_create+0x32d/0x680
[<c01fc230>] cifs_create+0x0/0x680
[<c0169ae3>] vfs_create+0x93/0xd0
[<c016c9e7>] open_namei+0x577/0x5d0
[<c016064e>] do_filp_open+0x2e/0x60
[<c016031f>] get_unused_fd_flags+0x4f/0xc0
[<c01606cc>] do_sys_open+0x4c/0xf0
[<c01607ac>] sys_open+0x1c/0x20
[<c0103b2e>] sysenter_past_esp+0x5f/0x85
=======================

Reproducible: Always

Comment 1 Daniel Drake (RETIRED) gentoo-dev

2008-02-15 14:45:34 UTC

How easy is it to reproduce?
Can you post the script?
Can you reproduce it on 2.6.24?
Can you reproduce it on a filesystem other than CIFS?
Is this a local or network based CIFS mount?

Comment 2 Daniel Drake (RETIRED) gentoo-dev

2008-02-15 14:45:59 UTC

One more thing, can you reproduce it on an untainted kernel?

Comment 3 jergendutch 2008-02-18 10:48:04 UTC

It happened several times in one night, but has not happened since.
I can't reproduce it at the moment, the script is not able to be posted, and the problem only occurs with CIFS. So we can close for the moment I think..

How is our kernel tainted? VMware perhaps?

Comment 4 Daniel Drake (RETIRED) gentoo-dev

2008-02-18 12:02:27 UTC

Yes, vmware would taint your kernel.
If this is not reproducible then I'll close this for now.