The patch titled
mm: share PG_readahead and PG_reclaim
has been added to the -mm tree. Its filename is
mm-share-pg_readahead-and-pg_reclaim.patch
*** Remember to use Documentation/SubmitChecklist when testing your code ***
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
------------------------------------------------------
Subject: mm: share PG_readahead and PG_reclaim
From: Fengguang Wu <wfg@xxxxxxxxxxxxxxxx>
Share the same page flag bit for PG_readahead and PG_reclaim.
One is used only on file reads, another is only for emergency writes. One
is used mostly for fresh/young pages, another is for old pages.
Combinations of possible interactions are:
a) clear PG_reclaim => implicit clear of PG_readahead
it will delay an asynchronous readahead into a synchronous one
it actually does _good_ for readahead:
the pages will be reclaimed soon, it's readahead thrashing!
in this case, synchronous readahead makes more sense.
b) clear PG_readahead => implicit clear of PG_reclaim
one(and only one) page will not be reclaimed in time
it can be avoided by checking PageWriteback(page) in readahead first
c) set PG_reclaim => implicit set of PG_readahead
will confuse readahead and make it restart the size rampup process
it's a trivial problem, and can mostly be avoided by checking
PageWriteback(page) first in readahead
d) set PG_readahead => implicit set of PG_reclaim
PG_readahead will never be set on already cached pages.
PG_reclaim will always be cleared on dirtying a page.
so not a problem.
In summary,
a) we get better behavior
b,d) possible interactions can be avoided
c) racy condition exists that might affect readahead, but the chance
is _really_ low, and the hurt on readahead is trivial.
Compound pages also use PG_reclaim, but for now they do not interact with
reclaim/readahead code.
Signed-off-by: Fengguang Wu <wfg@xxxxxxxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/page-flags.h | 3 ++-
mm/page-writeback.c | 1 +
mm/readahead.c | 4 ++++
3 files changed, 7 insertions(+), 1 deletion(-)
diff -puN include/linux/page-flags.h~mm-share-pg_readahead-and-pg_reclaim include/linux/page-flags.h
--- a/include/linux/page-flags.h~mm-share-pg_readahead-and-pg_reclaim
+++ a/include/linux/page-flags.h
@@ -92,7 +92,8 @@
#define PG_lazyfree 19 /* MADV_FREE potential throwaway */
#define PG_booked 20 /* Has blocks reserved on-disk */
-#define PG_readahead 21 /* Reminder to do async read-ahead */
+/* PG_readahead is only used for file reads; PG_reclaim is only for writes */
+#define PG_readahead PG_reclaim /* Reminder to do async read-ahead */
/* PG_owner_priv_1 users should have descriptive aliases */
#define PG_checked PG_owner_priv_1 /* Used by some filesystems */
diff -puN mm/page-writeback.c~mm-share-pg_readahead-and-pg_reclaim mm/page-writeback.c
--- a/mm/page-writeback.c~mm-share-pg_readahead-and-pg_reclaim
+++ a/mm/page-writeback.c
@@ -922,6 +922,7 @@ int clear_page_dirty_for_io(struct page
BUG_ON(!PageLocked(page));
+ ClearPageReclaim(page);
if (mapping && mapping_cap_account_dirty(mapping)) {
/*
* Yes, Virginia, this is indeed insane.
diff -puN mm/readahead.c~mm-share-pg_readahead-and-pg_reclaim mm/readahead.c
--- a/mm/readahead.c~mm-share-pg_readahead-and-pg_reclaim
+++ a/mm/readahead.c
@@ -447,6 +447,10 @@ page_cache_readahead_ondemand(struct add
if (!ra->ra_pages)
return 0;
+ /* It's PG_reclaim! */
+ if (PageWriteback(page))
+ return 0;
+
if (page) {
ClearPageReadahead(page);
_
Patches currently in -mm which might be from wfg@xxxxxxxxxxxxxxxx are
readahead-introduce-pg_readahead.patch
readahead-add-look-ahead-support-to-__do_page_cache_readahead.patch
readahead-min_ra_pages-max_ra_pages-macros.patch
readahead-data-structure-and-routines.patch
readahead-on-demand-readahead-logic.patch
readahead-convert-filemap-invocations.patch
readahead-convert-splice-invocations.patch
readahead-convert-ext3-ext4-invocations.patch
readahead-remove-the-old-algorithm.patch
mm-share-pg_readahead-and-pg_reclaim.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
