Hi, Andrea.

Thanks for bug reports.
I fix this problem.

This problem causes by miss of trace for children groups.
Please adopt and test this patch.

If OK, this amendment is adopted when I release new patch-set.


Regards,
 Satoshi Uchida.


diff --git a/block/cfq-cgroup.c b/block/cfq-cgroup.c
index f868f4f..64561f5 100644
--- a/block/cfq-cgroup.c
+++ b/block/cfq-cgroup.c
@@ -183,7 +184,7 @@ static void *cfq_cgroup_init_cfq_data(struct cfq_cgroup *cfqc, struct cfq_data *
 
 	/* check and create cfq_data for children */
 	if (cfqc->css.cgroup)
-		list_for_each_entry(child, &cfqc->css.cgroup->children, children){
+		list_for_each_entry(child, &cfqc->css.cgroup->children, sibling){
 			cfq_cgroup_init_cfq_data(cgroup_to_cfq_cgroup(child), cfqd);
 		}


quoted text
> -----Original Message-----
> From: Andrea Righi [mailto:righi.andrea@gmail.com]
> Sent: Friday, May 30, 2008 7:37 PM
> To: Satoshi UCHIDA
> Cc: 'Ryo Tsuruta'; axboe@kernel.dk; vtaras@openvz.org;
> containers@lists.linux-foundation.org; tom-sugawara@ap.jp.nec.com;
> linux-kernel@vger.kernel.org
> Subject: Re: [RFC][v2][patch 0/12][CFQ-cgroup]Yet another I/O bandwidth
> controlling subsystem for CGroups based on CFQ
> 
> Satoshi UCHIDA wrote:
> > Hi, Tsuruta-san.
> >
> >> I'm looking forward to your report.
> 
> Hi Satoshi,
> 
> I'm testing your patch agains latest Linus git and I've got the
> following bug. It can be easily reproduced creating a cgroup, switching
> the i/o scheduler from cfq to any other and switch back to cfq again.
> 
> -Andrea
> 
> BUG: unable to handle kernel paging request at ffffffeb
> IP: [<c0212dc6>] cfq_cgroup_sibling_tree_add+0x36/0x90
> Oops: 0000 [#1] SMP
> Modules linked in: i2c_piix4 ne2k_pci 8390 i2c_core
> 
> Pid: 3543, comm: bash Not tainted (2.6.26-rc4 #1)
> EIP: 0060:[<c0212dc6>] EFLAGS: 00010286 CPU: 0
> EIP is at cfq_cgroup_sibling_tree_add+0x36/0x90
> EAX: 00000003 EBX: c7704c90 ECX: ffffffff EDX: c7102180
> ESI: c7102240 EDI: c7704c80 EBP: c7afbe94 ESP: c7afbe80
>  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
> Process bash (pid: 3543, ti=c7afa000 task=c7aeda00 task.ti=c7afa000)
> Stack: c7704c90 c71022d0 c7ace078 c7704c80 c71020c0 c7afbea8 c021363c
> c7ace078
>        c7803460 c71020c0 c7afbebc c021360a c7102184 c71020c0 c7102180
> c7afbee4
>        c02134e0 00000000 c7102184 c72b8ab0 00000001 c7102140 c72b8ab0
> c04b8ac0
> Call Trace:
>  [<c021363c>] ? cfq_cgroup_init_cfq_data+0x7c/0x80
>  [<c021360a>] ? cfq_cgroup_init_cfq_data+0x4a/0x80
>  [<c02134e0>] ? __cfq_cgroup_init_queue+0x100/0x1e0
>  [<c021097b>] ? cfq_init_queue+0xb/0x10
>  [<c0204ff8>] ? elevator_init_queue+0x8/0x10
>  [<c0205cd0>] ? elv_iosched_store+0x80/0x2b0
>  [<c0209379>] ? queue_attr_store+0x49/0x70
>  [<c01c488b>] ? sysfs_write_file+0xbb/0x110
>  [<c0186276>] ? vfs_write+0x96/0x160
>  [<c01c47d0>] ? sysfs_write_file+0x0/0x110
>  [<c018696d>] ? sys_write+0x3d/0x70
>  [<c0104267>] ? sysenter_past_esp+0x78/0xd1
>  =======================
> Code: ec 08 8b 82 90 00 00 00 83 e0 fc 89 45 f0 8d 82 90 00 00 00 39 45
> f0 75 5f
> 8d 47 10 89 45 ec 89 c3 31 c0 eb 11 8b 56 7c 8d 41 04 <3b> 51 ec 8d 59
> 08 0f 43
> d8 89 c8 8b 0b 85 c9 75 e9 89 86 90 00
> EIP: [<c0212dc6>] cfq_cgroup_sibling_tree_add+0x36/0x90 SS:ESP
> 0068:c7afbe80
> ---[ end trace 9701f4859bb53d27 ]---