mm: fix a vmscan warning
Shaohua Li [Thu, 25 Aug 2011 22:59:10 +0000 (15:59 -0700)]
I get the below warning:

  BUG: using smp_processor_id() in preemptible [00000000] code: bash/746
  caller is native_sched_clock+0x37/0x6e
  Pid: 746, comm: bash Tainted: G        W   3.0.0+ #254
  Call Trace:
   [<ffffffff813435c6>] debug_smp_processor_id+0xc2/0xdc
   [<ffffffff8104158d>] native_sched_clock+0x37/0x6e
   [<ffffffff81116219>] try_to_free_mem_cgroup_pages+0x7d/0x270
   [<ffffffff8114f1f8>] mem_cgroup_force_empty+0x24b/0x27a
   [<ffffffff8114ff21>] ? sys_close+0x38/0x138
   [<ffffffff8114ff21>] ? sys_close+0x38/0x138
   [<ffffffff8114f257>] mem_cgroup_force_empty_write+0x17/0x19
   [<ffffffff810c72fb>] cgroup_file_write+0xa8/0xba
   [<ffffffff811522d2>] vfs_write+0xb3/0x138
   [<ffffffff8115241a>] sys_write+0x4a/0x71
   [<ffffffff8114ffd9>] ? sys_close+0xf0/0x138
   [<ffffffff8176deab>] system_call_fastpath+0x16/0x1b

sched_clock() can't be used with preempt enabled.  And we don't need
fast approach to get clock here, so let's use ktime API.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Tested-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmscan.c

index 7ef6912..22631e0 100644 (file)
@@ -2283,7 +2283,7 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
                .mem_cgroup = mem,
                .memcg_record = rec,
        };
-       unsigned long start, end;
+       ktime_t start, end;
 
        sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
                        (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
@@ -2292,7 +2292,7 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
                                                      sc.may_writepage,
                                                      sc.gfp_mask);
 
-       start = sched_clock();
+       start = ktime_get();
        /*
         * NOTE: Although we can get the priority field, using it
         * here is not a good idea, since it limits the pages we can scan.
@@ -2301,10 +2301,10 @@ unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
         * the priority and make it zero.
         */
        shrink_zone(0, zone, &sc);
-       end = sched_clock();
+       end = ktime_get();
 
        if (rec)
-               rec->elapsed += end - start;
+               rec->elapsed += ktime_to_ns(ktime_sub(end, start));
        *scanned = sc.nr_scanned;
 
        trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed);
@@ -2319,7 +2319,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
 {
        struct zonelist *zonelist;
        unsigned long nr_reclaimed;
-       unsigned long start, end;
+       ktime_t start, end;
        int nid;
        struct scan_control sc = {
                .may_writepage = !laptop_mode,
@@ -2337,7 +2337,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
                .gfp_mask = sc.gfp_mask,
        };
 
-       start = sched_clock();
+       start = ktime_get();
        /*
         * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't
         * take care of from where we get pages. So the node where we start the
@@ -2352,9 +2352,9 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
                                            sc.gfp_mask);
 
        nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink);
-       end = sched_clock();
+       end = ktime_get();
        if (rec)
-               rec->elapsed += end - start;
+               rec->elapsed += ktime_to_ns(ktime_sub(end, start));
 
        trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed);