linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Peter Zijlstra <a.p.zijlstra@chello.nl>
To: linux-mm@kvack.org
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Subject: [PATCH 24/39] mm: pgrep: generic shrinker logic
Date: Wed, 12 Jul 2006 16:41:42 +0200	[thread overview]
Message-ID: <20060712144142.16998.48555.sendpatchset@lappy> (raw)
In-Reply-To: <20060712143659.16998.6444.sendpatchset@lappy>

From: Peter Zijlstra <a.p.zijlstra@chello.nl>

Add a general shrinker that policies can make use of.
The policy defines MM_POLICY_HAS_SHRINKER when it does _NOT_ want
to make use of this framework.

API:

Return the number of pages in the scanlist for this zone.

	unsigned long __pgrep_nr_scan(struct zone *);

Fill the @list with at most @nr pages from @zone.

	void pgrep_get_candidates(struct zone *, int, unsigned long, 
                                  struct list_head *, unsigned long *);

Reinsert @list into @zone where @nr pages were freed - reinsert those 
pages that could not be freed.

	void pgrep_put_candidates(struct zone *, struct list_head *,
                                  unsigned long, int);

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Marcelo Tosatti <marcelo.tosatti@cyclades.com>

 include/linux/mm_page_replace.h    |    7 ++++
 include/linux/mm_use_once_policy.h |    2 +
 mm/vmscan.c                        |   60 +++++++++++++++++++++++++++++++++++++
 3 files changed, 69 insertions(+)

Index: linux-2.6/include/linux/mm_page_replace.h
===================================================================
--- linux-2.6.orig/include/linux/mm_page_replace.h	2006-07-12 16:09:19.000000000 +0200
+++ linux-2.6/include/linux/mm_page_replace.h	2006-07-12 16:11:29.000000000 +0200
@@ -114,5 +114,12 @@ static inline void pgrep_add_drain(void)
 	put_cpu();
 }
 
+#if ! defined MM_POLICY_HAS_SHRINKER
+/* unsigned long __pgrep_nr_scan(struct zone *); */
+void __pgrep_get_candidates(struct zone *, int, unsigned long, struct list_head *,
+		unsigned long *);
+void pgrep_put_candidates(struct zone *, struct list_head *, unsigned long, int);
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_PAGE_REPLACE_H */
Index: linux-2.6/mm/vmscan.c
===================================================================
--- linux-2.6.orig/mm/vmscan.c	2006-07-12 16:09:19.000000000 +0200
+++ linux-2.6/mm/vmscan.c	2006-07-12 16:09:19.000000000 +0200
@@ -592,6 +592,66 @@ int should_reclaim_mapped(struct zone *z
 	return 0;
 }
 
+#if ! defined MM_POLICY_HAS_SHRINKER
+unsigned long pgrep_shrink_zone(int priority, struct zone *zone,
+		struct scan_control *sc)
+{
+	unsigned long nr_reclaimed = 0;
+	unsigned long nr_scan = 0;
+
+	atomic_inc(&zone->reclaim_in_progress);
+
+	if (unlikely(sc->swap_cluster_max > SWAP_CLUSTER_MAX)) {
+		nr_scan = zone->policy.nr_scan;
+		zone->policy.nr_scan =
+			sc->swap_cluster_max + SWAP_CLUSTER_MAX - 1;
+	} else
+		zone->policy.nr_scan +=
+			(__pgrep_nr_scan(zone) >> priority) + 1;
+
+	while (zone->policy.nr_scan >= SWAP_CLUSTER_MAX) {
+		LIST_HEAD(page_list);
+		unsigned long nr_scan, nr_freed;
+
+		zone->policy.nr_scan -= SWAP_CLUSTER_MAX;
+
+		pgrep_add_drain();
+		spin_lock_irq(&zone->lru_lock);
+
+		__pgrep_get_candidates(zone, priority, SWAP_CLUSTER_MAX,
+				&page_list, &nr_scan);
+
+		spin_unlock(&zone->lru_lock);
+		if (current_is_kswapd())
+			__mod_page_state_zone(zone, pgscan_kswapd, nr_scan);
+		else
+			__mod_page_state_zone(zone, pgscan_direct, nr_scan);
+		local_irq_enable();
+
+		if (list_empty(&page_list))
+			continue;
+
+		nr_freed = shrink_page_list(&page_list, sc);
+		nr_reclaimed += nr_freed;
+
+		local_irq_disable();
+		if (current_is_kswapd())
+			__mod_page_state(kswapd_steal, nr_freed);
+		__mod_page_state_zone(zone, pgsteal, nr_freed);
+		local_irq_enable();
+
+		pgrep_put_candidates(zone, &page_list, nr_freed, sc->may_swap);
+	}
+	if (nr_scan)
+		zone->policy.nr_scan = nr_scan;
+
+	atomic_dec(&zone->reclaim_in_progress);
+
+	throttle_vm_writeout();
+	return nr_reclaimed;
+}
+#endif
+
 /*
  * This is the direct reclaim path, for page-allocating processes.  We only
  * try to reclaim pages from zones which will satisfy the caller's allocation
Index: linux-2.6/include/linux/mm_use_once_policy.h
===================================================================
--- linux-2.6.orig/include/linux/mm_use_once_policy.h	2006-07-12 16:09:19.000000000 +0200
+++ linux-2.6/include/linux/mm_use_once_policy.h	2006-07-12 16:11:31.000000000 +0200
@@ -173,5 +173,7 @@ static inline unsigned long __pgrep_nr_p
 	return zone->policy.nr_active + zone->policy.nr_inactive;
 }
 
+#define MM_POLICY_HAS_SHRINKER
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_USEONCE_POLICY_H */

--
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>

  parent reply	other threads:[~2006-07-12 14:41 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-07-12 14:36 [PATCH 0/39] mm: 2.6.17-pr1 - generic page-replacement framework and 4 new policies Peter Zijlstra
2006-07-12 14:37 ` [PATCH 1/39] mm: disuse activate_page() Peter Zijlstra
2006-07-12 14:37 ` [PATCH 2/39] mm: adjust blk_congestion_wait() logic Peter Zijlstra
2006-07-12 14:37 ` [PATCH 3/39] mm: pgrep: prepare for page replace framework Peter Zijlstra
2006-07-12 14:37 ` [PATCH 4/39] mm: pgrep: convert insertion Peter Zijlstra
2006-07-12 14:37 ` [PATCH 5/39] mm: pgrep: add a use-once insertion hint Peter Zijlstra
2006-07-12 14:38 ` [PATCH 6/39] mm: pgrep: generice __pagevec_*_add Peter Zijlstra
2006-07-12 14:38 ` [PATCH 7/39] mm: pgrep: abstract the activation logic Peter Zijlstra
2006-07-12 14:38 ` [PATCH 8/39] mm: pgrep: move useful macros around Peter Zijlstra
2006-07-12 14:38 ` [PATCH 9/39] mm: pgrep: move struct scan_control around Peter Zijlstra
2006-07-12 14:38 ` [PATCH 10/39] mm: pgrep: isolate the reclaim_mapped logic Peter Zijlstra
2006-07-12 14:39 ` [PATCH 11/39] mm: pgrep: replace mark_page_accessed Peter Zijlstra
2006-07-12 14:39 ` [PATCH 12/39] mm: pgrep: move the shrink logic Peter Zijlstra
2006-07-12 14:39 ` [PATCH 13/39] mm: pgrep: abstract rotate_reclaimable_page() Peter Zijlstra
2006-07-12 14:39 ` [PATCH 14/39] mm: pgrep: manage page-state Peter Zijlstra
2006-07-12 14:39 ` [PATCH 15/39] mm: pgrep: abstract page removal Peter Zijlstra
2006-07-12 14:40 ` [PATCH 16/39] mm: pgrep: remove mm_inline.h Peter Zijlstra
2006-07-12 14:40 ` [PATCH 17/39] mm: pgrep: re-insertion logic Peter Zijlstra
2006-07-12 14:40 ` [PATCH 18/39] mm: pgrep: initialisation hooks Peter Zijlstra
2006-07-12 14:40 ` [PATCH 19/39] mm: pgrep: info functions Peter Zijlstra
2006-07-12 14:40 ` [PATCH 20/39] mm: pgrep: page count functions Peter Zijlstra
2006-07-12 14:41 ` [PATCH 21/39] mm: pgrep: per policy data Peter Zijlstra
2006-07-12 14:41 ` [PATCH 22/39] mm: pgrep: per policy PG_flags Peter Zijlstra
2006-07-12 14:41 ` [PATCH 23/39] mm: pgrep: nonresident page tracking hooks Peter Zijlstra
2006-07-12 14:41 ` Peter Zijlstra [this message]
2006-07-12 14:41 ` [PATCH 25/39] mm: pgrep: documentation Peter Zijlstra
2006-07-12 14:42 ` [PATCH 26/39] sum_cpu_var Peter Zijlstra
2006-07-12 14:42 ` [PATCH 27/39] mm: clockpro: nonresident page tracking for CLOCK-Pro Peter Zijlstra
2006-07-12 14:42 ` [PATCH 28/39] mm: clockpro: re-introduce page_referenced() ignore_token Peter Zijlstra
2006-07-12 14:42 ` [PATCH 29/39] mm: clockpro: second per policy PG_flag Peter Zijlstra
2006-07-12 14:42 ` [PATCH 30/39] mm: clockpro: CLOCK-Pro policy implementation Peter Zijlstra
2006-07-12 14:43 ` [PATCH 31/39] mm: cart: nonresident page tracking for CART Peter Zijlstra
2006-07-12 14:43 ` [PATCH 32/39] mm: cart: third per policy PG_flag Peter Zijlstra
2006-07-12 14:43 ` [PATCH 33/39] mm: cart: CART policy implementation Peter Zijlstra
2006-07-12 14:43 ` [PATCH 34/39] mm: cart: CART-r " Peter Zijlstra
2006-07-12 14:43 ` [PATCH 35/39] mm: random: random page replacement policy Peter Zijlstra
2006-07-12 14:44 ` [PATCH 36/39] mm: refault histogram for non-resident policies Peter Zijlstra
2006-07-12 14:44 ` [PATCH 37/39] mm: use-once: cleanup of the use-once logic Peter Zijlstra
2006-07-12 14:44 ` [PATCH 38/39] mm: use-once: use the generic shrinker logic Peter Zijlstra
2006-07-12 14:44 ` [PATCH 39/39] mm: use-once: cleanup of the insertion logic Peter Zijlstra
2006-07-13 15:38 ` [PATCH 0/39] mm: 2.6.17-pr1 - generic page-replacement framework and 4 new policies Christoph Lameter
2006-07-15 17:03   ` Peter Zijlstra
2006-07-16  3:50     ` Christoph Lameter
2006-07-26 10:03       ` Marcelo Tosatti

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=20060712144142.16998.48555.sendpatchset@lappy \
    --to=a.p.zijlstra@chello.nl \
    --cc=linux-mm@kvack.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