On Tue, 2007-10-16 at 02:01 -0700, Andrew Morton wrote: > On Tue, 16 Oct 2007 12:54:29 +0400 Alexey Dobriyan <[email protected]> wrote: > > > Got this during LTP's diotest1. Reproducable. > > > > ======================================================= > > [ INFO: possible circular locking dependency detected ] > > 2.6.23-65a6ec0d72a07f16719e9b7a96e1c4bae044b591 #2 > > ------------------------------------------------------- > > diotest1/13563 is trying to acquire lock: > > (&mm->mmap_sem){----}, at: [<c107a7b7>] dio_get_page+0x4b/0x16b > > > > but task is already holding lock: > > (jbd_handle){--..}, at: [<c1099b05>] journal_start+0xf3/0x120 > > > > which lock already depends on the new lock. > > Someone taught lockdep about journal_start? That's gonna hurt. It was you who suggested doing so,... Are these valid reports, or is the annotation wrong?
Attachment:
signature.asc
Description: This is a digitally signed message part
- Follow-Ups:
- Re: Possible circular locking: ->mmap_sem vs jbd_handle
- From: Andrew Morton <[email protected]>
- Re: Possible circular locking: ->mmap_sem vs jbd_handle
- References:
- Possible circular locking: ->mmap_sem vs jbd_handle
- From: Alexey Dobriyan <[email protected]>
- Re: Possible circular locking: ->mmap_sem vs jbd_handle
- From: Andrew Morton <[email protected]>
- Possible circular locking: ->mmap_sem vs jbd_handle
- Prev by Date: Re: Possible circular locking: ->mmap_sem vs jbd_handle
- Next by Date: Re: Possible circular locking: ->mmap_sem vs jbd_handle
- Previous by thread: Re: Possible circular locking: ->mmap_sem vs jbd_handle
- Next by thread: Re: Possible circular locking: ->mmap_sem vs jbd_handle
- Index(es):