[RFC/T/D,2/2] Linux/Guest cooperative unmapped page cache control

Commit Message

Balbir Singh June 8, 2010, 3:51 p.m. UTC

None

Patch

diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloon.c
index 0f1da45..609a9c2 100644
--- a/drivers/virtio/virtio_balloon.c
+++ b/drivers/virtio/virtio_balloon.c
@@ -104,7 +104,8 @@  static void fill_balloon(struct virtio_balloon *vb, size_t num)
 
 	for (vb->num_pfns = 0; vb->num_pfns < num; vb->num_pfns++) {
 		struct page *page = alloc_page(GFP_HIGHUSER | __GFP_NORETRY |
-					__GFP_NOMEMALLOC | __GFP_NOWARN);
+					__GFP_NOMEMALLOC | __GFP_NOWARN |
+					__GFP_FREE_CACHE);
 		if (!page) {
 			if (printk_ratelimit())
 				dev_printk(KERN_INFO, &vb->vdev->dev,
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 975609c..9048259 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -61,12 +61,18 @@  struct vm_area_struct;
 #endif
 
 /*
+ * While allocating pages, try to free cache pages first. Note the
+ * heavy dependency on zone_reclaim_mode logic
+ */
+#define __GFP_FREE_CACHE ((__force gfp_t)0x400000u) /* Free cache first */
+
+/*
  * This may seem redundant, but it's a way of annotating false positives vs.
  * allocations that simply cannot be supported (e.g. page tables).
  */
 #define __GFP_NOTRACK_FALSE_POSITIVE (__GFP_NOTRACK)
 
-#define __GFP_BITS_SHIFT 22	/* Room for 22 __GFP_FOO bits */
+#define __GFP_BITS_SHIFT 23	/* Room for 22 __GFP_FOO bits */
 #define __GFP_BITS_MASK ((__force gfp_t)((1 << __GFP_BITS_SHIFT) - 1))
 
 /* This equals 0, but use constants in case they ever change */
diff --git a/include/linux/swap.h b/include/linux/swap.h
index f92f1ee..f77c603 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -254,16 +254,13 @@  extern long vm_total_pages;
 extern bool should_balance_unmapped_pages(struct zone *zone);
 
 extern int sysctl_min_unmapped_ratio;
-#ifdef CONFIG_NUMA
-extern int zone_reclaim_mode;
 extern int sysctl_min_slab_ratio;
 extern int zone_reclaim(struct zone *, gfp_t, unsigned int);
+
+#ifdef CONFIG_NUMA
+extern int zone_reclaim_mode;
 #else
 #define zone_reclaim_mode 0
-static inline int zone_reclaim(struct zone *z, gfp_t mask, unsigned int order)
-{
-	return 0;
-}
 #endif
 
 extern int page_evictable(struct page *page, struct vm_area_struct *vma);
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index fee9420..d977b36 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1649,7 +1649,8 @@  zonelist_scan:
 				    classzone_idx, alloc_flags))
 				goto try_this_zone;
 
-			if (zone_reclaim_mode == 0)
+			if (zone_reclaim_mode == 0 &&
+				!(gfp_mask & __GFP_FREE_CACHE))
 				goto this_zone_full;
 
 			ret = zone_reclaim(zone, gfp_mask, order);
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 27bc536..393bee5 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2624,6 +2624,7 @@  module_init(kswapd_init)
  * the watermarks.
  */
 int zone_reclaim_mode __read_mostly;
+#endif
 
 /*
  * If the number of slab pages in a zone grows beyond this percentage then
@@ -2780,7 +2781,6 @@  int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
 
 	return ret;
 }
-#endif
 
 /*
  * page_evictable - test whether a page is evictable