Commit 873b47717732c2f33a4b14de02571a4295a02f0c

Authored by Keika Kobayashi
Committed by Linus Torvalds
1 parent 3e85ba034d

per-task-delay-accounting: add memory reclaim delay

Sometimes, application responses become bad under heavy memory load.
Applications take a bit time to reclaim memory.  The statistics, how long
memory reclaim takes, will be useful to measure memory usage.

This patch adds accounting memory reclaim to per-task-delay-accounting for
accounting the time of do_try_to_free_pages().

<i.e>

- When System is under low memory load,
  memory reclaim may not occur.

$ free
             total       used       free     shared    buffers     cached
Mem:       8197800    1577300    6620500          0       4808    1516724
-/+ buffers/cache:      55768    8142032
Swap:     16386292          0   16386292

$ vmstat 1
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa
 0  0      0 5069748  10612 3014060    0    0     0     0    3   26  0  0 100  0
 0  0      0 5069748  10612 3014060    0    0     0     0    4   22  0  0 100  0
 0  0      0 5069748  10612 3014060    0    0     0     0    3   18  0  0 100  0

Measure the time of tar command.

$ ls -s test.dat
1501472 test.dat

$ time tar cvf test.tar test.dat
real    0m13.388s
user    0m0.116s
sys     0m5.304s

$ ./delayget -d -p <pid>
CPU             count     real total  virtual total    delay total
                  428     5528345500     5477116080       62749891
IO              count    delay total
                  338     8078977189
SWAP            count    delay total
                    0              0
RECLAIM         count    delay total
                    0              0

- When system is under heavy memory load
  memory reclaim may occur.

$ vmstat 1
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa
 0  0 7159032  49724   1812   3012    0    0     0     0    3   24  0  0 100  0
 0  0 7159032  49724   1812   3012    0    0     0     0    4   24  0  0 100  0
 0  0 7159032  49848   1812   3012    0    0     0     0    3   22  0  0 100  0

In this case, one process uses more 8G memory
by execution of malloc() and memset().

$ time tar cvf test.tar test.dat
real    1m38.563s        <-  increased by 85 sec
user    0m0.140s
sys     0m7.060s

$ ./delayget -d -p <pid>
CPU             count     real total  virtual total    delay total
                 9021     7140446250     7315277975      923201824
IO              count    delay total
                 8965    90466349669
SWAP            count    delay total
                    3       21036367
RECLAIM         count    delay total
                  740    61011951153

In the later case, the value of RECLAIM is increasing.
So, taskstats can show how much memory reclaim influences TAT.

Signed-off-by: Keika Kobayashi <kobayashi.kk@ncos.nec.co.jp>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujistu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 4 changed files with 41 additions and 0 deletions Side-by-side Diff

include/linux/delayacct.h
... ... @@ -39,6 +39,8 @@
39 39 extern void __delayacct_blkio_end(void);
40 40 extern int __delayacct_add_tsk(struct taskstats *, struct task_struct *);
41 41 extern __u64 __delayacct_blkio_ticks(struct task_struct *);
  42 +extern void __delayacct_freepages_start(void);
  43 +extern void __delayacct_freepages_end(void);
42 44  
43 45 static inline int delayacct_is_task_waiting_on_io(struct task_struct *p)
44 46 {
... ... @@ -107,6 +109,18 @@
107 109 return 0;
108 110 }
109 111  
  112 +static inline void delayacct_freepages_start(void)
  113 +{
  114 + if (current->delays)
  115 + __delayacct_freepages_start();
  116 +}
  117 +
  118 +static inline void delayacct_freepages_end(void)
  119 +{
  120 + if (current->delays)
  121 + __delayacct_freepages_end();
  122 +}
  123 +
110 124 #else
111 125 static inline void delayacct_set_flag(int flag)
112 126 {}
... ... @@ -129,6 +143,11 @@
129 143 { return 0; }
130 144 static inline int delayacct_is_task_waiting_on_io(struct task_struct *p)
131 145 { return 0; }
  146 +static inline void delayacct_freepages_start(void)
  147 +{}
  148 +static inline void delayacct_freepages_end(void)
  149 +{}
  150 +
132 151 #endif /* CONFIG_TASK_DELAY_ACCT */
133 152  
134 153 #endif
include/linux/sched.h
... ... @@ -672,6 +672,10 @@
672 672 /* io operations performed */
673 673 u32 swapin_count; /* total count of the number of swapin block */
674 674 /* io operations performed */
  675 +
  676 + struct timespec freepages_start, freepages_end;
  677 + u64 freepages_delay; /* wait for memory reclaim */
  678 + u32 freepages_count; /* total count of memory reclaim */
675 679 };
676 680 #endif /* CONFIG_TASK_DELAY_ACCT */
677 681  
... ... @@ -164,4 +164,17 @@
164 164 spin_unlock_irqrestore(&tsk->delays->lock, flags);
165 165 return ret;
166 166 }
  167 +
  168 +void __delayacct_freepages_start(void)
  169 +{
  170 + delayacct_start(&current->delays->freepages_start);
  171 +}
  172 +
  173 +void __delayacct_freepages_end(void)
  174 +{
  175 + delayacct_end(&current->delays->freepages_start,
  176 + &current->delays->freepages_end,
  177 + &current->delays->freepages_delay,
  178 + &current->delays->freepages_count);
  179 +}
... ... @@ -38,6 +38,7 @@
38 38 #include <linux/kthread.h>
39 39 #include <linux/freezer.h>
40 40 #include <linux/memcontrol.h>
  41 +#include <linux/delayacct.h>
41 42  
42 43 #include <asm/tlbflush.h>
43 44 #include <asm/div64.h>
... ... @@ -1316,6 +1317,8 @@
1316 1317 struct zone *zone;
1317 1318 enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
1318 1319  
  1320 + delayacct_freepages_start();
  1321 +
1319 1322 if (scan_global_lru(sc))
1320 1323 count_vm_event(ALLOCSTALL);
1321 1324 /*
... ... @@ -1395,6 +1398,8 @@
1395 1398 }
1396 1399 } else
1397 1400 mem_cgroup_record_reclaim_priority(sc->mem_cgroup, priority);
  1401 +
  1402 + delayacct_freepages_end();
1398 1403  
1399 1404 return ret;
1400 1405 }