Re: incomplete RAID0 array causing bdi_unregister WARN_ON.

[NeilBrown]

On Thu, 21 May 2015 11:47:49 -0400 Dave Jones <davej@xxxxxxxxxxxxxxxxx> wrote:

> One of the disks in my 4 disk RAID0 stripe died.
> It's a hardware problem, but this spew seems unnecessary..
> 
> 	Dave
> 
> WARNING: CPU: 0 PID: 39 at mm/backing-dev.c:372 bdi_unregister+0x3a/0x50()
> CPU: 0 PID: 39 Comm: kworker/0:1 Not tainted 4.1.0-rc4-gelk-debug+ #1
> Workqueue: md_misc mddev_delayed_delete [md_mod]
>  0000000000000009 ffff8800baf9fbf8 ffffffffa6768350 0000000000000000
>  0000000000000000 ffff8800baf9fc38 ffffffffa607eaaa ffff8800baf9fc18
>  ffff8800b57cdcd0 0000000000000000 0000000000000000 00000000ffffffff
> Call Trace:
>  [<ffffffffa6768350>] dump_stack+0x4f/0x7b
>  [<ffffffffa607eaaa>] warn_slowpath_common+0x8a/0xc0
>  [<ffffffffa607eb9a>] warn_slowpath_null+0x1a/0x20
>  [<ffffffffa61c31fa>] bdi_unregister+0x3a/0x50
>  [<ffffffffa636981f>] del_gendisk+0x11f/0x250
>  [<ffffffffc00b8583>] md_free+0x43/0x60 [md_mod]
>  [<ffffffffa638496a>] kobject_release+0x7a/0x1c0
>  [<ffffffffa6384815>] kobject_put+0x35/0x70
>  [<ffffffffc00b5ab4>] mddev_delayed_delete+0x34/0x40 [md_mod]
>  [<ffffffffa609d8e1>] process_one_work+0x1f1/0x820
>  [<ffffffffa609d850>] ? process_one_work+0x160/0x820
>  [<ffffffffa609e00b>] ? worker_thread+0xfb/0x4b0
>  [<ffffffffa609df7b>] worker_thread+0x6b/0x4b0
>  [<ffffffffa609df10>] ? process_one_work+0x820/0x820
>  [<ffffffffa609df10>] ? process_one_work+0x820/0x820
>  [<ffffffffa60a431f>] kthread+0x11f/0x140
>  [<ffffffffa60a4200>] ? kthread_create_on_node+0x250/0x250
>  [<ffffffffa6772292>] ret_from_fork+0x42/0x70
>  [<ffffffffa60a4200>] ? kthread_create_on_node+0x250/0x250
> ---[ end trace 424fb17cf914b7dc ]---
> 
> 
> 370 void bdi_unregister(struct backing_dev_info *bdi)
> 371 {
> 372         if (WARN_ON_ONCE(!bdi->dev))
> 373                 return;
> 374 

 https://lkml.org/lkml/2015/5/19/42

NeilBrown

Attachment: pgpXVzoXq3G9p.pgp
Description: OpenPGP digital signature