From: Andrea Arcangeli <aarcange@redhat.com>
To: linux-mm@kvack.org
Cc: Marcelo Tosatti <mtosatti@redhat.com>,
Adam Litke <agl@us.ibm.com>, Avi Kivity <avi@redhat.com>,
Izik Eidus <ieidus@redhat.com>,
Hugh Dickins <hugh.dickins@tiscali.co.uk>,
Nick Piggin <npiggin@suse.de>, Rik van Riel <riel@redhat.com>,
Mel Gorman <mel@csn.ul.ie>, Andi Kleen <andi@firstfloor.org>,
Dave Hansen <dave@linux.vnet.ibm.com>,
Benjamin Herrenschmidt <benh@kernel.crashing.org>,
Ingo Molnar <mingo@elte.hu>, Mike Travis <travis@sgi.com>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
Christoph Lameter <cl@linux-foundation.org>,
Chris Wright <chrisw@sous-sol.org>,
Andrew Morton <akpm@linux-foundation.org>,
bpicco@redhat.com, Christoph Hellwig <chellwig@redhat.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Subject: [PATCH 04 of 31] update futex compound knowledge
Date: Mon, 25 Jan 2010 18:18:55 +0100 [thread overview]
Message-ID: <3fbd2869d7b6664243ae.1264439935@v2.random> (raw)
In-Reply-To: <patchbomb.1264439931@v2.random>
From: Andrea Arcangeli <aarcange@redhat.com>
Futex code is smarter than most other gup_fast O_DIRECT code and knows about
the compound internals. However now doing a put_page(head_page) will not
release the pin on the tail page taken by gup-fast, leading to all sort of
refcounting bugchecks. Getting a stable head_page is a little tricky.
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
---
diff --git a/kernel/futex.c b/kernel/futex.c
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -218,7 +218,7 @@ get_futex_key(u32 __user *uaddr, int fsh
{
unsigned long address = (unsigned long)uaddr;
struct mm_struct *mm = current->mm;
- struct page *page;
+ struct page *page, *page_head;
int err;
/*
@@ -250,10 +250,32 @@ again:
if (err < 0)
return err;
- page = compound_head(page);
- lock_page(page);
- if (!page->mapping) {
- unlock_page(page);
+ page_head = page;
+ if (unlikely(PageTail(page))) {
+ put_page(page);
+ /* serialize against __split_huge_page_splitting() */
+ local_irq_disable();
+ if (likely(__get_user_pages_fast(address, 1, 1, &page) == 1)) {
+ page_head = compound_head(page);
+ local_irq_enable();
+ } else {
+ local_irq_enable();
+ goto again;
+ }
+ }
+
+ lock_page(page_head);
+ if (unlikely(page_head != page)) {
+ compound_lock(page_head);
+ if (unlikely(!PageTail(page))) {
+ compound_unlock(page_head);
+ unlock_page(page_head);
+ put_page(page);
+ goto again;
+ }
+ }
+ if (!page_head->mapping) {
+ unlock_page(page_head);
put_page(page);
goto again;
}
@@ -265,19 +287,21 @@ again:
* it's a read-only handle, it's expected that futexes attach to
* the object not the particular process.
*/
- if (PageAnon(page)) {
+ if (PageAnon(page_head)) {
key->both.offset |= FUT_OFF_MMSHARED; /* ref taken on mm */
key->private.mm = mm;
key->private.address = address;
} else {
key->both.offset |= FUT_OFF_INODE; /* inode-based key */
- key->shared.inode = page->mapping->host;
- key->shared.pgoff = page->index;
+ key->shared.inode = page_head->mapping->host;
+ key->shared.pgoff = page_head->index;
}
get_futex_key_refs(key);
- unlock_page(page);
+ if (unlikely(PageTail(page)))
+ compound_unlock(page_head);
+ unlock_page(page_head);
put_page(page);
return 0;
}
--
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:[~2010-01-25 17:30 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-01-25 17:18 [PATCH 00 of 31] Transparent Hugepage support #6 Andrea Arcangeli
2010-01-25 17:18 ` [PATCH 01 of 31] define MADV_HUGEPAGE Andrea Arcangeli
2010-01-26 11:41 ` Mel Gorman
2010-01-26 12:30 ` Andrea Arcangeli
2010-01-26 13:46 ` Andrea Arcangeli
2010-01-25 17:18 ` [PATCH 02 of 31] compound_lock Andrea Arcangeli
2010-01-25 17:18 ` [PATCH 03 of 31] alter compound get_page/put_page Andrea Arcangeli
2010-01-25 17:18 ` Andrea Arcangeli [this message]
2010-01-25 17:18 ` [PATCH 05 of 31] fix bad_page to show the real reason the page is bad Andrea Arcangeli
2010-01-25 17:18 ` [PATCH 06 of 31] clear compound mapping Andrea Arcangeli
2010-01-25 17:18 ` [PATCH 07 of 31] add native_set_pmd_at Andrea Arcangeli
2010-01-25 17:18 ` [PATCH 08 of 31] add pmd paravirt ops Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 09 of 31] no paravirt version of pmd ops Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 10 of 31] export maybe_mkwrite Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 11 of 31] comment reminder in destroy_compound_page Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 12 of 31] config_transparent_hugepage Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 13 of 31] add pmd mangling functions to x86 Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 14 of 31] add pmd mangling generic functions Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 15 of 31] special pmd_trans_* functions Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 16 of 31] bail out gup_fast on splitting pmd Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 17 of 31] pte alloc trans splitting Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 18 of 31] add pmd mmu_notifier helpers Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 19 of 31] clear page compound Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 20 of 31] add pmd_huge_pte to mm_struct Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 21 of 31] split_huge_page_mm/vma Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 22 of 31] split_huge_page paging Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 23 of 31] clear_copy_huge_page Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 24 of 31] kvm mmu transparent hugepage support Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 25 of 31] transparent hugepage core Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 26 of 31] madvise(MADV_HUGEPAGE) Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 27 of 31] pmd_trans_huge migrate bugcheck Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 28 of 31] memcg compound Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 29 of 31] memcg huge memory Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 30 of 31] transparent hugepage vmstat Andrea Arcangeli
2010-01-25 17:19 ` [PATCH 31 of 31] khugepaged Andrea Arcangeli
2010-01-26 13:51 [PATCH 00 of 31] Transparent Hugepage support #7 Andrea Arcangeli
2010-01-26 13:51 ` [PATCH 04 of 31] update futex compound knowledge Andrea Arcangeli
2010-01-26 16:12 ` Rik van Riel
2010-01-26 17:10 ` Andrea Arcangeli
2010-01-26 19:48 ` Rik van Riel
2010-01-26 18:37 ` Mel Gorman
2010-01-27 19:45 ` Andrea Arcangeli
2010-01-28 15:33 ` Mel Gorman
2010-01-28 15:57 ` Andrea Arcangeli
2010-01-28 14:33 [PATCH 00 of 31] Transparent Hugepage support #8 Andrea Arcangeli
2010-01-28 14:33 ` [PATCH 04 of 31] update futex compound knowledge Andrea Arcangeli
2010-01-28 16:11 ` Mel Gorman
2010-02-01 7:45 ` 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=3fbd2869d7b6664243ae.1264439935@v2.random \
--to=aarcange@redhat.com \
--cc=agl@us.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=andi@firstfloor.org \
--cc=avi@redhat.com \
--cc=benh@kernel.crashing.org \
--cc=bpicco@redhat.com \
--cc=chellwig@redhat.com \
--cc=chrisw@sous-sol.org \
--cc=cl@linux-foundation.org \
--cc=dave@linux.vnet.ibm.com \
--cc=hugh.dickins@tiscali.co.uk \
--cc=ieidus@redhat.com \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-mm@kvack.org \
--cc=mel@csn.ul.ie \
--cc=mingo@elte.hu \
--cc=mtosatti@redhat.com \
--cc=npiggin@suse.de \
--cc=riel@redhat.com \
--cc=travis@sgi.com \
/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