sched/numa: Add debugging
authorIngo Molnar <mingo@kernel.org>
Mon, 7 Oct 2013 10:29:30 +0000 (11:29 +0100)
committerIngo Molnar <mingo@kernel.org>
Wed, 9 Oct 2013 12:48:04 +0000 (14:48 +0200)
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: http://lkml.kernel.org/r/1381141781-10992-53-git-send-email-mgorman@suse.de
include/linux/sched.h
kernel/sched/debug.c
kernel/sched/fair.c

index f6385107c35243d9eecbce5861572b7fd4c1e103..1127a46ac3d2bd0faba717cebe56e53e15d7aaff 100644 (file)
@@ -1366,6 +1366,7 @@ struct task_struct {
        unsigned long *numa_faults_buffer;
 
        int numa_preferred_nid;
+       unsigned long numa_pages_migrated;
 #endif /* CONFIG_NUMA_BALANCING */
 
        struct rcu_head rcu;
@@ -2661,6 +2662,11 @@ static inline unsigned int task_cpu(const struct task_struct *p)
        return task_thread_info(p)->cpu;
 }
 
+static inline int task_node(const struct task_struct *p)
+{
+       return cpu_to_node(task_cpu(p));
+}
+
 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
 
 #else
index 196559994f7ce33e210f3a290043ed5862950ec6..e6ba5e31c7ca24e67bb3f56fbe8f422371c8fb74 100644 (file)
@@ -15,6 +15,7 @@
 #include <linux/seq_file.h>
 #include <linux/kallsyms.h>
 #include <linux/utsname.h>
+#include <linux/mempolicy.h>
 
 #include "sched.h"
 
@@ -137,6 +138,9 @@ print_task(struct seq_file *m, struct rq *rq, struct task_struct *p)
        SEQ_printf(m, "%15Ld %15Ld %15Ld.%06ld %15Ld.%06ld %15Ld.%06ld",
                0LL, 0LL, 0LL, 0L, 0LL, 0L, 0LL, 0L);
 #endif
+#ifdef CONFIG_NUMA_BALANCING
+       SEQ_printf(m, " %d", cpu_to_node(task_cpu(p)));
+#endif
 #ifdef CONFIG_CGROUP_SCHED
        SEQ_printf(m, " %s", task_group_path(task_group(p)));
 #endif
@@ -159,7 +163,7 @@ static void print_rq(struct seq_file *m, struct rq *rq, int rq_cpu)
        read_lock_irqsave(&tasklist_lock, flags);
 
        do_each_thread(g, p) {
-               if (!p->on_rq || task_cpu(p) != rq_cpu)
+               if (task_cpu(p) != rq_cpu)
                        continue;
 
                print_task(m, rq, p);
@@ -345,7 +349,7 @@ static void sched_debug_header(struct seq_file *m)
        cpu_clk = local_clock();
        local_irq_restore(flags);
 
-       SEQ_printf(m, "Sched Debug Version: v0.10, %s %.*s\n",
+       SEQ_printf(m, "Sched Debug Version: v0.11, %s %.*s\n",
                init_utsname()->release,
                (int)strcspn(init_utsname()->version, " "),
                init_utsname()->version);
@@ -488,6 +492,56 @@ static int __init init_sched_debug_procfs(void)
 
 __initcall(init_sched_debug_procfs);
 
+#define __P(F) \
+       SEQ_printf(m, "%-45s:%21Ld\n", #F, (long long)F)
+#define P(F) \
+       SEQ_printf(m, "%-45s:%21Ld\n", #F, (long long)p->F)
+#define __PN(F) \
+       SEQ_printf(m, "%-45s:%14Ld.%06ld\n", #F, SPLIT_NS((long long)F))
+#define PN(F) \
+       SEQ_printf(m, "%-45s:%14Ld.%06ld\n", #F, SPLIT_NS((long long)p->F))
+
+
+static void sched_show_numa(struct task_struct *p, struct seq_file *m)
+{
+#ifdef CONFIG_NUMA_BALANCING
+       struct mempolicy *pol;
+       int node, i;
+
+       if (p->mm)
+               P(mm->numa_scan_seq);
+
+       task_lock(p);
+       pol = p->mempolicy;
+       if (pol && !(pol->flags & MPOL_F_MORON))
+               pol = NULL;
+       mpol_get(pol);
+       task_unlock(p);
+
+       SEQ_printf(m, "numa_migrations, %ld\n", xchg(&p->numa_pages_migrated, 0));
+
+       for_each_online_node(node) {
+               for (i = 0; i < 2; i++) {
+                       unsigned long nr_faults = -1;
+                       int cpu_current, home_node;
+
+                       if (p->numa_faults)
+                               nr_faults = p->numa_faults[2*node + i];
+
+                       cpu_current = !i ? (task_node(p) == node) :
+                               (pol && node_isset(node, pol->v.nodes));
+
+                       home_node = (p->numa_preferred_nid == node);
+
+                       SEQ_printf(m, "numa_faults, %d, %d, %d, %d, %ld\n",
+                               i, node, cpu_current, home_node, nr_faults);
+               }
+       }
+
+       mpol_put(pol);
+#endif
+}
+
 void proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 {
        unsigned long nr_switches;
@@ -591,6 +645,8 @@ void proc_sched_show_task(struct task_struct *p, struct seq_file *m)
                SEQ_printf(m, "%-45s:%21Ld\n",
                           "clock-delta", (long long)(t1-t0));
        }
+
+       sched_show_numa(p, m);
 }
 
 void proc_sched_set_task(struct task_struct *p)
index 147349987bfeda1564c5f7920ceec20789431da3..2876a37cdfc4c2dcf1eccfedbe40f475dd363c72 100644 (file)
@@ -1137,7 +1137,7 @@ static int task_numa_migrate(struct task_struct *p)
                .p = p,
 
                .src_cpu = task_cpu(p),
-               .src_nid = cpu_to_node(task_cpu(p)),
+               .src_nid = task_node(p),
 
                .imbalance_pct = 112,
 
@@ -1515,6 +1515,9 @@ void task_numa_fault(int last_cpupid, int node, int pages, int flags)
        if (p->numa_migrate_retry && time_after(jiffies, p->numa_migrate_retry))
                numa_migrate_preferred(p);
 
+       if (migrated)
+               p->numa_pages_migrated += pages;
+
        p->numa_faults_buffer[task_faults_idx(node, priv)] += pages;
 }