From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Nick Piggin <npiggin@suse.de>
Cc: Linux Memory Management <linux-mm@kvack.org>,
Neil Brown <neilb@suse.de>, Anton Altaparmakov <aia21@cam.ac.uk>,
Chris Mason <chris.mason@oracle.com>,
Linux Kernel <linux-kernel@vger.kernel.org>,
Andrew Morton <akpm@osdl.org>,
ralf@linux-mips.org, David Howells <dhowells@redhat.com>
Subject: Re: RRe: [patch 6/6] mm: fix pagecache write deadlocks
Date: Sun, 15 Oct 2006 18:13:55 +0200 [thread overview]
Message-ID: <1160928835.5230.41.camel@lappy> (raw)
In-Reply-To: <20061015155727.GA539@wotan.suse.de>
On Sun, 2006-10-15 at 17:57 +0200, Nick Piggin wrote:
> On Sun, Oct 15, 2006 at 05:47:03PM +0200, Peter Zijlstra wrote:
> > >
> > > And we should really decouple it from preempt entirely, in case we
> > > ever want to check for it some other way in the pagefault handler.
> >
> > How about we make sure all kmap_atomic implementations behave properly
> > and make in_atomic true.
>
> Hmm, but you may not be doing a copy*user within the kmap. And you may
> want an atomic copy*user not within a kmap (maybe).
>
> I think it really would be more logical to do it in a wrapper function
> pagefault_disable() pagefault_enable()? ;)
I did that one first, but then noticed that most non trivial kmap_atomic
implementations already did the inc_preempt_count()/dec_preempt_count()
thing (except frv which did preempt_disable()/preempt_enable() ?)
Anyway, here goes:
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
mm/filemap.c | 4 +---
mm/filemap.h | 20 ++++++++++++++++++++
2 files changed, 21 insertions(+), 3 deletions(-)
Index: linux-2.6/mm/filemap.c
===================================================================
--- linux-2.6.orig/mm/filemap.c 2006-10-14 20:20:20.000000000 +0200
+++ linux-2.6/mm/filemap.c 2006-10-15 17:16:59.000000000 +0200
@@ -2140,9 +2140,8 @@ retry_noprogress:
* the page lock, so we might recursively deadlock on the same
* lock, or get an ABBA deadlock against a different lock, or
* against the mmap_sem (which nests outside the page lock).
- * So increment preempt count, and use _atomic usercopies.
+ * So use _atomic usercopies.
*/
- inc_preempt_count();
if (likely(nr_segs == 1))
copied = filemap_copy_from_user_atomic(page, offset,
buf, bytes);
@@ -2150,7 +2149,6 @@ retry_noprogress:
copied = filemap_copy_from_user_iovec_atomic(page,
offset, cur_iov, iov_offset,
bytes);
- dec_preempt_count();
if (!PageUptodate(page)) {
/*
Index: linux-2.6/mm/filemap.h
===================================================================
--- linux-2.6.orig/mm/filemap.h 2006-10-14 20:20:20.000000000 +0200
+++ linux-2.6/mm/filemap.h 2006-10-15 17:17:45.000000000 +0200
@@ -21,6 +21,22 @@ __filemap_copy_from_user_iovec_inatomic(
size_t bytes);
/*
+ * By increasing the preempt_count we make sure the arch preempt
+ * handler bails out early, before taking any locks, so that the copy
+ * operation gets terminated early.
+ */
+pagefault_static inline void disable(void)
+{
+ inc_preempt_count();
+}
+
+pagefault_static inline void enable(void)
+{
+ dec_preempt_count();
+ preempt_check_resched();
+}
+
+/*
* Copy as much as we can into the page and return the number of bytes which
* were sucessfully copied. If a fault is encountered then return the number of
* bytes which were copied.
@@ -40,9 +56,11 @@ filemap_copy_from_user_atomic(struct pag
char *kaddr;
int left;
+ pagefault_disable();
kaddr = kmap_atomic(page, KM_USER0);
left = __copy_from_user_inatomic_nocache(kaddr + offset, buf, bytes);
kunmap_atomic(kaddr, KM_USER0);
+ pagefault_enable();
return bytes - left;
}
@@ -75,10 +93,12 @@ filemap_copy_from_user_iovec_atomic(stru
char *kaddr;
size_t copied;
+ pagefault_disable();
kaddr = kmap_atomic(page, KM_USER0);
copied = __filemap_copy_from_user_iovec_inatomic(kaddr + offset, iov,
base, bytes);
kunmap_atomic(kaddr, KM_USER0);
+ pagefault_enable();
return copied;
}
WARNING: multiple messages have this Message-ID (diff)
From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: Nick Piggin <npiggin@suse.de>
Cc: Linux Memory Management <linux-mm@kvack.org>,
Neil Brown <neilb@suse.de>, Anton Altaparmakov <aia21@cam.ac.uk>,
Chris Mason <chris.mason@oracle.com>,
Linux Kernel <linux-kernel@vger.kernel.org>,
Andrew Morton <akpm@osdl.org>,
ralf@linux-mips.org, David Howells <dhowells@redhat.com>
Subject: Re: RRe: [patch 6/6] mm: fix pagecache write deadlocks
Date: Sun, 15 Oct 2006 18:13:55 +0200 [thread overview]
Message-ID: <1160928835.5230.41.camel@lappy> (raw)
In-Reply-To: <20061015155727.GA539@wotan.suse.de>
On Sun, 2006-10-15 at 17:57 +0200, Nick Piggin wrote:
> On Sun, Oct 15, 2006 at 05:47:03PM +0200, Peter Zijlstra wrote:
> > >
> > > And we should really decouple it from preempt entirely, in case we
> > > ever want to check for it some other way in the pagefault handler.
> >
> > How about we make sure all kmap_atomic implementations behave properly
> > and make in_atomic true.
>
> Hmm, but you may not be doing a copy*user within the kmap. And you may
> want an atomic copy*user not within a kmap (maybe).
>
> I think it really would be more logical to do it in a wrapper function
> pagefault_disable() pagefault_enable()? ;)
I did that one first, but then noticed that most non trivial kmap_atomic
implementations already did the inc_preempt_count()/dec_preempt_count()
thing (except frv which did preempt_disable()/preempt_enable() ?)
Anyway, here goes:
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
mm/filemap.c | 4 +---
mm/filemap.h | 20 ++++++++++++++++++++
2 files changed, 21 insertions(+), 3 deletions(-)
Index: linux-2.6/mm/filemap.c
===================================================================
--- linux-2.6.orig/mm/filemap.c 2006-10-14 20:20:20.000000000 +0200
+++ linux-2.6/mm/filemap.c 2006-10-15 17:16:59.000000000 +0200
@@ -2140,9 +2140,8 @@ retry_noprogress:
* the page lock, so we might recursively deadlock on the same
* lock, or get an ABBA deadlock against a different lock, or
* against the mmap_sem (which nests outside the page lock).
- * So increment preempt count, and use _atomic usercopies.
+ * So use _atomic usercopies.
*/
- inc_preempt_count();
if (likely(nr_segs == 1))
copied = filemap_copy_from_user_atomic(page, offset,
buf, bytes);
@@ -2150,7 +2149,6 @@ retry_noprogress:
copied = filemap_copy_from_user_iovec_atomic(page,
offset, cur_iov, iov_offset,
bytes);
- dec_preempt_count();
if (!PageUptodate(page)) {
/*
Index: linux-2.6/mm/filemap.h
===================================================================
--- linux-2.6.orig/mm/filemap.h 2006-10-14 20:20:20.000000000 +0200
+++ linux-2.6/mm/filemap.h 2006-10-15 17:17:45.000000000 +0200
@@ -21,6 +21,22 @@ __filemap_copy_from_user_iovec_inatomic(
size_t bytes);
/*
+ * By increasing the preempt_count we make sure the arch preempt
+ * handler bails out early, before taking any locks, so that the copy
+ * operation gets terminated early.
+ */
+pagefault_static inline void disable(void)
+{
+ inc_preempt_count();
+}
+
+pagefault_static inline void enable(void)
+{
+ dec_preempt_count();
+ preempt_check_resched();
+}
+
+/*
* Copy as much as we can into the page and return the number of bytes which
* were sucessfully copied. If a fault is encountered then return the number of
* bytes which were copied.
@@ -40,9 +56,11 @@ filemap_copy_from_user_atomic(struct pag
char *kaddr;
int left;
+ pagefault_disable();
kaddr = kmap_atomic(page, KM_USER0);
left = __copy_from_user_inatomic_nocache(kaddr + offset, buf, bytes);
kunmap_atomic(kaddr, KM_USER0);
+ pagefault_enable();
return bytes - left;
}
@@ -75,10 +93,12 @@ filemap_copy_from_user_iovec_atomic(stru
char *kaddr;
size_t copied;
+ pagefault_disable();
kaddr = kmap_atomic(page, KM_USER0);
copied = __filemap_copy_from_user_iovec_inatomic(kaddr + offset, iov,
base, bytes);
kunmap_atomic(kaddr, KM_USER0);
+ pagefault_enable();
return copied;
}
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2006-10-15 16:14 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-10-13 16:43 [rfc] buffered write deadlock fix Nick Piggin
2006-10-13 16:43 ` Nick Piggin
2006-10-13 16:44 ` [patch 1/6] mm: revert "generic_file_buffered_write(): handle zero length iovec segments" Nick Piggin
2006-10-13 16:44 ` Nick Piggin, Andrew Morton
2006-10-13 16:44 ` [patch 2/6] mm: revert "generic_file_buffered_write(): deadlock on vectored write" Nick Piggin
2006-10-13 16:44 ` Nick Piggin, Andrew Morton
2006-10-13 16:44 ` [patch 3/6] mm: generic_file_buffered_write cleanup Nick Piggin
2006-10-13 16:44 ` Nick Piggin, Andrew Morton
2006-10-13 16:44 ` [patch 4/6] mm: comment mmap_sem / lock_page lockorder Nick Piggin
2006-10-13 16:44 ` Nick Piggin
2006-10-13 16:44 ` [patch 5/6] mm: debug write deadlocks Nick Piggin
2006-10-13 16:44 ` Nick Piggin
2006-10-13 16:44 ` [patch 6/6] mm: fix pagecache " Nick Piggin
2006-10-13 16:44 ` Nick Piggin, Andrew Morton
2006-10-13 22:14 ` Andrew Morton
2006-10-13 22:14 ` Andrew Morton
2006-10-14 4:19 ` Nick Piggin
2006-10-14 4:19 ` Nick Piggin
2006-10-14 4:30 ` Nick Piggin
2006-10-14 4:30 ` Nick Piggin
2006-10-15 11:35 ` Peter Zijlstra
2006-10-15 11:35 ` Peter Zijlstra
2006-10-14 5:04 ` Nick Piggin
2006-10-14 5:04 ` Nick Piggin
2006-10-15 11:37 ` Peter Zijlstra
2006-10-15 11:37 ` Peter Zijlstra
2006-10-15 11:56 ` Nick Piggin
2006-10-15 11:56 ` Nick Piggin
2006-10-15 13:51 ` Peter Zijlstra
2006-10-15 13:51 ` Peter Zijlstra
2006-10-15 14:19 ` SPAM: " Nick Piggin
2006-10-15 14:19 ` Nick Piggin
2006-10-15 15:47 ` Peter Zijlstra
2006-10-15 15:47 ` Peter Zijlstra
2006-10-15 15:57 ` RRe: " Nick Piggin
2006-10-15 15:57 ` Nick Piggin
2006-10-15 16:13 ` Peter Zijlstra [this message]
2006-10-15 16:13 ` Peter Zijlstra
2006-10-16 15:24 ` pagefault_disable (was Re: [patch 6/6] mm: fix pagecache write deadlocks) Nick Piggin
2006-10-16 15:24 ` Nick Piggin
2006-10-16 16:05 ` Peter Zijlstra
2006-10-16 16:05 ` Peter Zijlstra
2006-10-16 16:12 ` Nick Piggin
2006-10-16 16:12 ` Nick Piggin
2006-10-18 14:25 ` [patch 6/6] mm: fix pagecache write deadlocks Chris Mason
2006-10-18 14:25 ` Chris Mason
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=1160928835.5230.41.camel@lappy \
--to=a.p.zijlstra@chello.nl \
--cc=aia21@cam.ac.uk \
--cc=akpm@osdl.org \
--cc=chris.mason@oracle.com \
--cc=dhowells@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=neilb@suse.de \
--cc=npiggin@suse.de \
--cc=ralf@linux-mips.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.