From: Andrew Morton <akpm@linux-foundation.org>
To: Andrea Arcangeli <andrea@qumranet.com>
Cc: torvalds@linux-foundation.org, clameter@sgi.com, steiner@sgi.com,
holt@sgi.com, npiggin@suse.de, a.p.zijlstra@chello.nl,
kvm-devel@lists.sourceforge.net, kanojsarcar@yahoo.com,
rdreier@cisco.com, swise@opengridcomputing.com,
linux-kernel@vger.kernel.org, avi@qumranet.com,
linux-mm@kvack.org, general@lists.openfabrics.org,
hugh@veritas.com, rusty@rustcorp.com.au, aliguori@us.ibm.com,
chrisw@redhat.com, marcelo@kvack.org, dada1@cosmosbay.com,
paulmck@us.ibm.com
Subject: Re: [PATCH 08 of 11] anon-vma-rwsem
Date: Wed, 7 May 2008 15:59:14 -0700 [thread overview]
Message-ID: <20080507155914.d7790069.akpm@linux-foundation.org> (raw)
In-Reply-To: <20080507224406.GI8276@duo.random>
On Thu, 8 May 2008 00:44:06 +0200
Andrea Arcangeli <andrea@qumranet.com> wrote:
> On Wed, May 07, 2008 at 03:31:03PM -0700, Andrew Morton wrote:
> > Nope. We only need to take the global lock before taking *two or more* of
> > the per-vma locks.
> >
> > I really wish I'd thought of that.
>
> I don't see how you can avoid taking the system-wide-global lock
> before every single anon_vma->lock/i_mmap_lock out there without
> mm_lock.
>
> Please note, we can't allow a thread to be in the middle of
> zap_page_range while mmu_notifier_register runs.
>
> vmtruncate takes 1 single lock, the i_mmap_lock of the inode. Not more
> than one lock and we've to still take the global-system-wide lock
> _before_ this single i_mmap_lock and no other lock at all.
>
> Please elaborate, thanks!
umm...
CPU0: CPU1:
spin_lock(a->lock); spin_lock(b->lock);
spin_lock(b->lock); spin_lock(a->lock);
bad.
CPU0: CPU1:
spin_lock(global_lock) spin_lock(global_lock);
spin_lock(a->lock); spin_lock(b->lock);
spin_lock(b->lock); spin_lock(a->lock);
Is OK.
CPU0: CPU1:
spin_lock(global_lock)
spin_lock(a->lock); spin_lock(b->lock);
spin_lock(b->lock); spin_unlock(b->lock);
spin_lock(a->lock);
spin_unlock(a->lock);
also OK.
As long as all code paths which can take two-or-more locks are all covered
by the global lock there is no deadlock scenario. If a thread takes just a
single instance of one of these locks without taking the global_lock then
there is also no deadlock.
Now, if we need to take both anon_vma->lock AND i_mmap_lock in the newly
added mm_lock() thing and we also take both those locks at the same time in
regular code, we're probably screwed.
next prev parent reply other threads:[~2008-05-07 23:01 UTC|newest]
Thread overview: 106+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-05-07 14:35 [PATCH 00 of 11] mmu notifier #v16 Andrea Arcangeli
2008-05-07 14:35 ` [PATCH 01 of 11] mmu-notifier-core Andrea Arcangeli
2008-05-07 17:35 ` Rik van Riel
2008-05-07 20:02 ` Andrew Morton
2008-05-07 20:05 ` Andrew Morton
2008-05-07 20:30 ` Linus Torvalds
2008-05-07 21:58 ` Andrea Arcangeli
2008-05-07 22:11 ` Linus Torvalds
2008-05-07 22:27 ` Andrea Arcangeli
2008-05-07 22:31 ` [ofa-general] " Roland Dreier
2008-05-07 22:39 ` Andrea Arcangeli
2008-05-07 23:03 ` Linus Torvalds
2008-05-07 22:37 ` Andrea Arcangeli
2008-05-07 23:38 ` Linus Torvalds
2008-05-07 23:00 ` Linus Torvalds
2008-05-07 14:35 ` [PATCH 02 of 11] get_task_mm Andrea Arcangeli
2008-05-07 15:59 ` Robin Holt
2008-05-07 16:20 ` Andrea Arcangeli
2008-05-07 14:35 ` [PATCH 03 of 11] invalidate_page outside PT lock Andrea Arcangeli
2008-05-07 17:39 ` Rik van Riel
2008-05-07 17:57 ` Andrea Arcangeli
2008-05-07 14:35 ` [PATCH 04 of 11] free-pgtables Andrea Arcangeli
2008-05-07 17:41 ` Rik van Riel
2008-05-07 14:35 ` [PATCH 05 of 11] unmap vmas tlb flushing Andrea Arcangeli
2008-05-07 17:46 ` Rik van Riel
2008-05-07 14:35 ` [PATCH 06 of 11] rwsem contended Andrea Arcangeli
2008-05-07 14:35 ` [PATCH 07 of 11] i_mmap_rwsem Andrea Arcangeli
2008-05-07 14:35 ` [PATCH 08 of 11] anon-vma-rwsem Andrea Arcangeli
2008-05-07 20:56 ` Linus Torvalds
2008-05-07 21:26 ` Andrea Arcangeli
2008-05-07 21:36 ` Linus Torvalds
2008-05-07 22:22 ` Andrea Arcangeli
2008-05-07 22:31 ` Andrew Morton
2008-05-07 22:44 ` Andrea Arcangeli
2008-05-07 22:59 ` Andrew Morton [this message]
2008-05-07 23:19 ` Linus Torvalds
2008-05-07 23:39 ` Christoph Lameter
2008-05-08 0:03 ` Linus Torvalds
2008-05-08 0:52 ` Robin Holt
2008-05-08 0:56 ` Christoph Lameter
2008-05-08 1:07 ` Linus Torvalds
2008-05-08 1:39 ` Linus Torvalds
2008-05-08 1:52 ` Andrea Arcangeli
2008-05-08 1:57 ` Linus Torvalds
2008-05-08 2:24 ` Andrea Arcangeli
2008-05-08 2:32 ` Linus Torvalds
2008-05-07 23:39 ` Andrea Arcangeli
2008-05-08 1:02 ` Linus Torvalds
2008-05-08 1:12 ` Christoph Lameter
2008-05-08 1:32 ` Linus Torvalds
2008-05-08 2:56 ` Andrea Arcangeli
2008-05-08 3:10 ` Christoph Lameter
2008-05-08 3:41 ` Andrea Arcangeli
2008-05-08 4:14 ` Linus Torvalds
2008-05-08 5:20 ` Andrea Arcangeli
2008-05-08 5:27 ` Pekka Enberg
2008-05-08 5:30 ` Pekka Enberg
2008-05-08 5:49 ` Andrea Arcangeli
2008-05-08 15:03 ` Linus Torvalds
2008-05-08 16:11 ` Linus Torvalds
2008-05-08 22:01 ` Andrea Arcangeli
2008-05-09 18:37 ` Peter Zijlstra
2008-05-09 18:55 ` Andrea Arcangeli
2008-05-09 19:04 ` Peter Zijlstra
2008-05-08 1:26 ` Andrea Arcangeli
2008-05-07 23:28 ` Benjamin Herrenschmidt
2008-05-07 23:45 ` Andrea Arcangeli
2008-05-08 1:34 ` Andrea Arcangeli
2008-05-13 12:14 ` Nick Piggin
2008-05-14 5:43 ` Benjamin Herrenschmidt
2008-05-14 6:06 ` Nick Piggin
2008-05-14 13:15 ` Jack Steiner
2008-05-07 22:44 ` Linus Torvalds
2008-05-07 22:58 ` Andrea Arcangeli
2008-05-07 23:02 ` Andrea Arcangeli
2008-05-07 23:09 ` Linus Torvalds
2008-05-08 0:38 ` Robin Holt
2008-05-08 0:55 ` Linus Torvalds
2008-05-13 12:06 ` Nick Piggin
2008-05-13 15:32 ` Robin Holt
2008-05-14 4:11 ` Nick Piggin
2008-05-14 11:26 ` Robin Holt
2008-05-14 15:18 ` Linus Torvalds
2008-05-14 16:22 ` Robin Holt
2008-05-14 16:56 ` Linus Torvalds
2008-05-14 17:57 ` Christoph Lameter
2008-05-14 18:27 ` Linus Torvalds
2008-05-17 1:38 ` mm notifier: Notifications when pages are unmapped Christoph Lameter
2008-05-15 7:57 ` [PATCH 08 of 11] anon-vma-rwsem Nick Piggin
2008-05-15 11:01 ` Robin Holt
2008-05-15 11:12 ` Avi Kivity
2008-05-15 17:33 ` Christoph Lameter
2008-05-15 23:52 ` Nick Piggin
2008-05-16 11:23 ` Robin Holt
2008-05-16 11:50 ` Robin Holt
2008-05-20 5:31 ` Nick Piggin
2008-05-20 10:01 ` Robin Holt
2008-05-20 10:50 ` Nick Piggin
2008-05-20 11:05 ` Robin Holt
2008-05-20 11:14 ` Nick Piggin
2008-05-20 11:26 ` Robin Holt
2008-05-07 22:42 ` Jack Steiner
2008-05-07 14:35 ` [PATCH 09 of 11] mm_lock-rwsem Andrea Arcangeli
2008-05-07 14:36 ` [PATCH 10 of 11] export zap_page_range for XPMEM Andrea Arcangeli
2008-05-07 14:36 ` [PATCH 11 of 11] mmap sems Andrea Arcangeli
-- strict thread matches above, loose matches on Subject: below --
2008-05-02 15:05 [PATCH 00 of 11] mmu notifier #v15 Andrea Arcangeli
2008-05-02 15:05 ` [PATCH 08 of 11] anon-vma-rwsem Andrea Arcangeli
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080507155914.d7790069.akpm@linux-foundation.org \
--to=akpm@linux-foundation.org \
--cc=a.p.zijlstra@chello.nl \
--cc=aliguori@us.ibm.com \
--cc=andrea@qumranet.com \
--cc=avi@qumranet.com \
--cc=chrisw@redhat.com \
--cc=clameter@sgi.com \
--cc=dada1@cosmosbay.com \
--cc=general@lists.openfabrics.org \
--cc=holt@sgi.com \
--cc=hugh@veritas.com \
--cc=kanojsarcar@yahoo.com \
--cc=kvm-devel@lists.sourceforge.net \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=marcelo@kvack.org \
--cc=npiggin@suse.de \
--cc=paulmck@us.ibm.com \
--cc=rdreier@cisco.com \
--cc=rusty@rustcorp.com.au \
--cc=steiner@sgi.com \
--cc=swise@opengridcomputing.com \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).