[PATCH] vmscan: scan_control cleanup
authorAndrew Morton <akpm@osdl.org>
Wed, 22 Mar 2006 08:08:18 +0000 (00:08 -0800)
committerLinus Torvalds <torvalds@g5.osdl.org>
Wed, 22 Mar 2006 15:54:00 +0000 (07:54 -0800)
Initialise as much of scan_control as possible at the declaration site.  This
tidies things up a bit and assures us that all unmentioned fields are zeroed
out.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
mm/vmscan.c

index f7c4f37..5feef4d 100644 (file)
@@ -1431,13 +1431,14 @@ int try_to_free_pages(struct zone **zones, gfp_t gfp_mask)
        int ret = 0;
        int total_scanned = 0, total_reclaimed = 0;
        struct reclaim_state *reclaim_state = current->reclaim_state;
-       struct scan_control sc;
        unsigned long lru_pages = 0;
        int i;
-
-       sc.gfp_mask = gfp_mask;
-       sc.may_writepage = !laptop_mode;
-       sc.may_swap = 1;
+       struct scan_control sc = {
+               .gfp_mask = gfp_mask,
+               .may_writepage = !laptop_mode,
+               .swap_cluster_max = SWAP_CLUSTER_MAX,
+               .may_swap = 1,
+       };
 
        inc_page_state(allocstall);
 
@@ -1455,7 +1456,6 @@ int try_to_free_pages(struct zone **zones, gfp_t gfp_mask)
                sc.nr_mapped = read_page_state(nr_mapped);
                sc.nr_scanned = 0;
                sc.nr_reclaimed = 0;
-               sc.swap_cluster_max = SWAP_CLUSTER_MAX;
                if (!priority)
                        disable_swap_token();
                shrink_caches(priority, zones, &sc);
@@ -1478,7 +1478,8 @@ int try_to_free_pages(struct zone **zones, gfp_t gfp_mask)
                 * that's undesirable in laptop mode, where we *want* lumpy
                 * writeout.  So in laptop mode, write out the whole world.
                 */
-               if (total_scanned > sc.swap_cluster_max + sc.swap_cluster_max/2) {
+               if (total_scanned > sc.swap_cluster_max +
+                                       sc.swap_cluster_max / 2) {
                        wakeup_pdflush(laptop_mode ? 0 : total_scanned);
                        sc.may_writepage = 1;
                }
@@ -1532,14 +1533,16 @@ static int balance_pgdat(pg_data_t *pgdat, int nr_pages, int order)
        int i;
        int total_scanned, total_reclaimed;
        struct reclaim_state *reclaim_state = current->reclaim_state;
-       struct scan_control sc;
+       struct scan_control sc = {
+               .gfp_mask = GFP_KERNEL,
+               .may_swap = 1,
+               .swap_cluster_max = nr_pages ? nr_pages : SWAP_CLUSTER_MAX,
+       };
 
 loop_again:
        total_scanned = 0;
        total_reclaimed = 0;
-       sc.gfp_mask = GFP_KERNEL;
-       sc.may_writepage = !laptop_mode;
-       sc.may_swap = 1;
+       sc.may_writepage = !laptop_mode,
        sc.nr_mapped = read_page_state(nr_mapped);
 
        inc_page_state(pageoutrun);
@@ -1621,7 +1624,6 @@ scan:
                                zone->prev_priority = priority;
                        sc.nr_scanned = 0;
                        sc.nr_reclaimed = 0;
-                       sc.swap_cluster_max = nr_pages? nr_pages : SWAP_CLUSTER_MAX;
                        shrink_zone(priority, zone, &sc);
                        reclaim_state->reclaimed_slab = 0;
                        nr_slab = shrink_slab(sc.nr_scanned, GFP_KERNEL,
@@ -1869,46 +1871,21 @@ int zone_reclaim_interval __read_mostly = 30*HZ;
 /*
  * Try to free up some pages from this zone through reclaim.
  */
-int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
+static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
 {
-       int nr_pages;
+       const int nr_pages = 1 << order;
        struct task_struct *p = current;
        struct reclaim_state reclaim_state;
-       struct scan_control sc;
-       cpumask_t mask;
-       int node_id;
        int priority;
-
-       if (time_before(jiffies,
-               zone->last_unsuccessful_zone_reclaim + zone_reclaim_interval))
-                       return 0;
-
-       if (!(gfp_mask & __GFP_WAIT) ||
-               zone->all_unreclaimable ||
-               atomic_read(&zone->reclaim_in_progress) > 0 ||
-               (p->flags & PF_MEMALLOC))
-                       return 0;
-
-       node_id = zone->zone_pgdat->node_id;
-       mask = node_to_cpumask(node_id);
-       if (!cpus_empty(mask) && node_id != numa_node_id())
-               return 0;
-
-       sc.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE);
-       sc.may_swap = !!(zone_reclaim_mode & RECLAIM_SWAP);
-       sc.nr_scanned = 0;
-       sc.nr_reclaimed = 0;
-       sc.nr_mapped = read_page_state(nr_mapped);
-       sc.gfp_mask = gfp_mask;
+       struct scan_control sc = {
+               .may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
+               .may_swap = !!(zone_reclaim_mode & RECLAIM_SWAP),
+               .nr_mapped = read_page_state(nr_mapped),
+               .swap_cluster_max = max(nr_pages, SWAP_CLUSTER_MAX),
+               .gfp_mask = gfp_mask,
+       };
 
        disable_swap_token();
-
-       nr_pages = 1 << order;
-       if (nr_pages > SWAP_CLUSTER_MAX)
-               sc.swap_cluster_max = nr_pages;
-       else
-               sc.swap_cluster_max = SWAP_CLUSTER_MAX;
-
        cond_resched();
        /*
         * We need to be able to allocate from the reserves for RECLAIM_SWAP
@@ -1949,5 +1926,44 @@ int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
 
        return sc.nr_reclaimed >= nr_pages;
 }
+
+int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
+{
+       cpumask_t mask;
+       int node_id;
+
+       /*
+        * Do not reclaim if there was a recent unsuccessful attempt at zone
+        * reclaim.  In that case we let allocations go off node for the
+        * zone_reclaim_interval.  Otherwise we would scan for each off-node
+        * page allocation.
+        */
+       if (time_before(jiffies,
+               zone->last_unsuccessful_zone_reclaim + zone_reclaim_interval))
+                       return 0;
+
+       /*
+        * Avoid concurrent zone reclaims, do not reclaim in a zone that does
+        * not have reclaimable pages and if we should not delay the allocation
+        * then do not scan.
+        */
+       if (!(gfp_mask & __GFP_WAIT) ||
+               zone->all_unreclaimable ||
+               atomic_read(&zone->reclaim_in_progress) > 0 ||
+               (current->flags & PF_MEMALLOC))
+                       return 0;
+
+       /*
+        * Only run zone reclaim on the local zone or on zones that do not
+        * have associated processors. This will favor the local processor
+        * over remote processors and spread off node memory allocations
+        * as wide as possible.
+        */
+       node_id = zone->zone_pgdat->node_id;
+       mask = node_to_cpumask(node_id);
+       if (!cpus_empty(mask) && node_id != numa_node_id())
+               return 0;
+       return __zone_reclaim(zone, gfp_mask, order);
+}
 #endif