memcg: condense page_cgroup-to-page lookup points
Johannes Weiner [Wed, 23 Mar 2011 23:42:29 +0000 (16:42 -0700)]
The per-cgroup LRU lists string up 'struct page_cgroup's.  To get from
those structures to the page they represent, a lookup is required.
Currently, the lookup is done through a direct pointer in struct
page_cgroup, so a lot of functions down the callchain do this lookup by
themselves instead of receiving the page pointer from their callers.

The next patch removes this pointer, however, and the lookup is no longer
that straight-forward.  In preparation for that, this patch only leaves
the non-optional lookups when coming directly from the LRU list and passes
the page down the stack.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/memcontrol.c

index e9d33dc..e286e16 100644 (file)
@@ -1077,9 +1077,11 @@ unsigned long mem_cgroup_isolate_pages(unsigned long nr_to_scan,
                if (scan >= nr_to_scan)
                        break;
 
-               page = pc->page;
                if (unlikely(!PageCgroupUsed(pc)))
                        continue;
+
+               page = pc->page;
+
                if (unlikely(!PageLRU(page)))
                        continue;
 
@@ -2108,6 +2110,7 @@ struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page)
 }
 
 static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
+                                      struct page *page,
                                       struct page_cgroup *pc,
                                       enum charge_type ctype,
                                       int page_size)
@@ -2154,7 +2157,7 @@ static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
         * Insert ancestor (and ancestor's ancestors), to softlimit RB-tree.
         * if they exceeds softlimit.
         */
-       memcg_check_events(mem, pc->page);
+       memcg_check_events(mem, page);
 }
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
@@ -2201,6 +2204,7 @@ void mem_cgroup_split_huge_fixup(struct page *head, struct page *tail)
 
 /**
  * mem_cgroup_move_account - move account of the page
+ * @page: the page
  * @pc:        page_cgroup of the page.
  * @from: mem_cgroup which the page is moved from.
  * @to:        mem_cgroup which the page is moved to. @from != @to.
@@ -2216,7 +2220,7 @@ void mem_cgroup_split_huge_fixup(struct page *head, struct page *tail)
  * true, this function does "uncharge" from old cgroup, but it doesn't if
  * @uncharge is false, so a caller should do "uncharge".
  */
-static int mem_cgroup_move_account(struct page_cgroup *pc,
+static int mem_cgroup_move_account(struct page *page, struct page_cgroup *pc,
                                   struct mem_cgroup *from, struct mem_cgroup *to,
                                   bool uncharge, int charge_size)
 {
@@ -2225,7 +2229,7 @@ static int mem_cgroup_move_account(struct page_cgroup *pc,
        int ret;
 
        VM_BUG_ON(from == to);
-       VM_BUG_ON(PageLRU(pc->page));
+       VM_BUG_ON(PageLRU(page));
        /*
         * The page is isolated from LRU. So, collapse function
         * will not handle this page. But page splitting can happen.
@@ -2233,7 +2237,7 @@ static int mem_cgroup_move_account(struct page_cgroup *pc,
         * hold it.
         */
        ret = -EBUSY;
-       if (charge_size > PAGE_SIZE && !PageTransHuge(pc->page))
+       if (charge_size > PAGE_SIZE && !PageTransHuge(page))
                goto out;
 
        lock_page_cgroup(pc);
@@ -2273,8 +2277,8 @@ unlock:
        /*
         * check events
         */
-       memcg_check_events(to, pc->page);
-       memcg_check_events(from, pc->page);
+       memcg_check_events(to, page);
+       memcg_check_events(from, page);
 out:
        return ret;
 }
@@ -2283,11 +2287,11 @@ out:
  * move charges to its parent.
  */
 
-static int mem_cgroup_move_parent(struct page_cgroup *pc,
+static int mem_cgroup_move_parent(struct page *page,
+                                 struct page_cgroup *pc,
                                  struct mem_cgroup *child,
                                  gfp_t gfp_mask)
 {
-       struct page *page = pc->page;
        struct cgroup *cg = child->css.cgroup;
        struct cgroup *pcg = cg->parent;
        struct mem_cgroup *parent;
@@ -2317,7 +2321,7 @@ static int mem_cgroup_move_parent(struct page_cgroup *pc,
        if (page_size > PAGE_SIZE)
                flags = compound_lock_irqsave(page);
 
-       ret = mem_cgroup_move_account(pc, child, parent, true, page_size);
+       ret = mem_cgroup_move_account(page, pc, child, parent, true, page_size);
        if (ret)
                mem_cgroup_cancel_charge(parent, page_size);
 
@@ -2363,7 +2367,7 @@ static int mem_cgroup_charge_common(struct page *page, struct mm_struct *mm,
        if (ret || !mem)
                return ret;
 
-       __mem_cgroup_commit_charge(mem, pc, ctype, page_size);
+       __mem_cgroup_commit_charge(mem, page, pc, ctype, page_size);
        return 0;
 }
 
@@ -2501,7 +2505,7 @@ __mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
        cgroup_exclude_rmdir(&ptr->css);
        pc = lookup_page_cgroup(page);
        mem_cgroup_lru_del_before_commit_swapcache(page);
-       __mem_cgroup_commit_charge(ptr, pc, ctype, PAGE_SIZE);
+       __mem_cgroup_commit_charge(ptr, page, pc, ctype, PAGE_SIZE);
        mem_cgroup_lru_add_after_commit_swapcache(page);
        /*
         * Now swap is on-memory. This means this page may be
@@ -2956,7 +2960,7 @@ int mem_cgroup_prepare_migration(struct page *page,
                ctype = MEM_CGROUP_CHARGE_TYPE_CACHE;
        else
                ctype = MEM_CGROUP_CHARGE_TYPE_SHMEM;
-       __mem_cgroup_commit_charge(mem, pc, ctype, PAGE_SIZE);
+       __mem_cgroup_commit_charge(mem, page, pc, ctype, PAGE_SIZE);
        return ret;
 }
 
@@ -3323,6 +3327,8 @@ static int mem_cgroup_force_empty_list(struct mem_cgroup *mem,
        loop += 256;
        busy = NULL;
        while (loop--) {
+               struct page *page;
+
                ret = 0;
                spin_lock_irqsave(&zone->lru_lock, flags);
                if (list_empty(list)) {
@@ -3338,7 +3344,9 @@ static int mem_cgroup_force_empty_list(struct mem_cgroup *mem,
                }
                spin_unlock_irqrestore(&zone->lru_lock, flags);
 
-               ret = mem_cgroup_move_parent(pc, mem, GFP_KERNEL);
+               page = pc->page;
+
+               ret = mem_cgroup_move_parent(page, pc, mem, GFP_KERNEL);
                if (ret == -ENOMEM)
                        break;
 
@@ -4956,7 +4964,7 @@ retry:
                        if (isolate_lru_page(page))
                                goto put;
                        pc = lookup_page_cgroup(page);
-                       if (!mem_cgroup_move_account(pc,
+                       if (!mem_cgroup_move_account(page, pc,
                                        mc.from, mc.to, false, PAGE_SIZE)) {
                                mc.precharge--;
                                /* we uncharge from mc.from later. */