From: NIIBE Yutaka <gniibe@m17n.org>
To: Marcelo Tosatti <marcelo@conectiva.com.br>
Cc: linux-kernel@vger.kernel.org
Subject: Re: [PATCH] swapin flush cache bug
Date: Wed, 27 Jun 2001 09:51:46 +0900 (JST) [thread overview]
Message-ID: <200106270051.f5R0pkl19282@mule.m17n.org> (raw)
In-Reply-To: <Pine.LNX.4.21.0102140510240.30964-100000@freak.distro.conectiva>
In-Reply-To: <200102140208.LAA18226@mule.m17n.org> <Pine.LNX.4.21.0102140510240.30964-100000@freak.distro.conectiva>
Hello Marcelo,
This is follow-up to the mail in February. You may perhaps forget the
context, it's the bug of MM about cache flushing for swapped-in-pages.
I see this bug on SuperH port (SH-4).
I think that we have this issue on the machine whose flush_dcache_page()
is defined. In current code, the cache aren't flushed for
the asynchronously-swapped-in pages which is cached in swap cache.
This is the problem.
Marcelo Tosatti writes:
> Yet another thing (1) on end_buffer_io_async() to handle a case which is
> only true for a specific user of it. Since the other special case handling
> is for swap IO too, I think a separate IO end operation for swap would be
> interesting.
>
> (1) The current one is SetPageDecrAfter handling.
How about this? I've updated MM bugzilla already.
2001-06-26 NIIBE Yutaka <gniibe@m17n.org>
* include/linux/mm.h (PG_flush_after, PageFlushAfter,
SetPageFlushAfter, PageTestandClearFlushAfter): New bit.
* mm/page_io.c (rw_swap_page_base): Set flush-after bit.
* fs/buffer.c (end_buffer_io_async): Implement flush-ing
with PG_flush_after.
* mm/memory.c (do_swap_page): Remove flush-ing the page.
diff -ruNp --exclude=CVS --exclude=.cvsignore v2.4.6-pre5/fs/buffer.c kernel/fs/buffer.c
--- v2.4.6-pre5/fs/buffer.c Mon Jun 25 18:48:07 2001
+++ kernel/fs/buffer.c Tue Jun 26 15:11:17 2001
@@ -831,6 +831,9 @@ static void end_buffer_io_async(struct b
if (PageTestandClearDecrAfter(page))
atomic_dec(&nr_async_pages);
+ if (PageTestandClearFlushAfter(page))
+ flush_dcache_page(page);
+
UnlockPage(page);
return;
diff -ruNp --exclude=CVS --exclude=.cvsignore v2.4.6-pre5/include/linux/mm.h kernel/include/linux/mm.h
--- v2.4.6-pre5/include/linux/mm.h Mon Jun 25 18:48:09 2001
+++ kernel/include/linux/mm.h Tue Jun 26 14:58:56 2001
@@ -282,6 +282,7 @@ typedef struct page {
#define PG_inactive_clean 11
#define PG_highmem 12
#define PG_checked 13 /* kill me in 2.5.<early>. */
+#define PG_flush_after 14
/* bits 21-29 unused */
#define PG_arch_1 30
#define PG_reserved 31
@@ -364,6 +365,10 @@ static inline void set_page_dirty(struct
#define SetPageReserved(page) set_bit(PG_reserved, &(page)->flags)
#define ClearPageReserved(page) clear_bit(PG_reserved, &(page)->flags)
+
+#define PageFlushAfter(page) test_bit(PG_flush_after, &(page)->flags)
+#define SetPageFlushAfter(page) set_bit(PG_flush_after, &(page)->flags)
+#define PageTestandClearFlushAfter(page) test_and_clear_bit(PG_flush_after, &(page)->flags)
/*
* Error return values for the *_nopage functions
diff -ruNp --exclude=CVS --exclude=.cvsignore v2.4.6-pre5/mm/memory.c kernel/mm/memory.c
--- v2.4.6-pre5/mm/memory.c Mon Jun 25 18:48:10 2001
+++ kernel/mm/memory.c Tue Jun 26 14:48:15 2001
@@ -1109,8 +1109,6 @@ static int do_swap_page(struct mm_struct
return -1;
}
wait_on_page(page);
- flush_page_to_ram(page);
- flush_icache_page(vma, page);
}
/*
diff -ruNp --exclude=CVS --exclude=.cvsignore v2.4.6-pre5/mm/page_io.c kernel/mm/page_io.c
--- v2.4.6-pre5/mm/page_io.c Mon Apr 30 16:15:32 2001
+++ kernel/mm/page_io.c Tue Jun 26 15:01:00 2001
@@ -50,6 +50,7 @@ static int rw_swap_page_base(int rw, swp
if (rw == READ) {
ClearPageUptodate(page);
+ SetPageFlushAfter(page);
kstat.pswpin++;
} else
kstat.pswpout++;
--
next prev parent reply other threads:[~2001-06-27 0:52 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2001-02-12 23:21 Marcelo Tosatti
2001-02-13 9:50 ` Russell King
2001-02-13 10:53 ` NIIBE Yutaka
2001-02-13 11:16 ` Russell King
2001-02-13 11:26 ` Alan Cox
2001-02-13 23:50 ` NIIBE Yutaka
2001-02-14 2:08 ` NIIBE Yutaka
2001-02-14 10:12 ` Marcelo Tosatti
2001-06-27 0:51 ` NIIBE Yutaka [this message]
2001-06-27 10:11 ` Marcelo Tosatti
2001-06-28 0:42 ` David S. Miller
2001-06-28 0:07 ` NIIBE Yutaka
2001-06-27 22:41 ` Marcelo Tosatti
2001-06-28 0:23 ` Stephen C. Tweedie
2001-06-28 0:47 ` David S. Miller
2001-06-28 1:10 ` David S. Miller
2001-06-29 14:18 ` NIIBE Yutaka
2001-07-02 22:47 ` Cache issues David S. Miller
2001-06-28 0:41 ` [PATCH] swapin flush cache bug NIIBE Yutaka
2001-06-28 1:04 ` NIIBE Yutaka
2001-07-02 11:23 ` Cache issues NIIBE Yutaka
2001-07-03 0:04 ` NIIBE Yutaka
2001-06-28 0:46 ` [PATCH] swapin flush cache bug David S. Miller
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=200106270051.f5R0pkl19282@mule.m17n.org \
--to=gniibe@m17n.org \
--cc=linux-kernel@vger.kernel.org \
--cc=marcelo@conectiva.com.br \
/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
all inboxes | Powered by JetHome®