diff options
author | Ravikiran G Thirumalai <kiran@scalex86.org> | 2007-07-19 04:47:53 -0400 |
---|---|---|
committer | Linus Torvalds <torvalds@woody.linux-foundation.org> | 2007-07-19 13:04:43 -0400 |
commit | 4004c69ad68dd03733179277280ea2946990ba36 (patch) | |
tree | 521f206229832f757c0f7652101d187a7a6ff5c8 /fs | |
parent | a0a9983509f45b2225ca87fdcf7b40ea916834ed (diff) |
Avoid too many remote cpu references due to /proc/stat
Optimize show_stat to collect per-irq information just once.
On x86_64, with newer kernel versions, kstat_irqs is a bit of a problem.
On every call to kstat_irqs, the process brings in per-cpu data from all
online cpus. Doing this for NR_IRQS, which is now 256 + 32 * NR_CPUS
results in (256+32*63) * 63 remote cpu references on a 64 cpu config.
Considering the fact that we already compute this value per-cpu, we can
save on the remote references as below.
Signed-off-by: Alok N Kataria <alok.kataria@calsoftinc.com>
Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'fs')
-rw-r--r-- | fs/proc/proc_misc.c | 15 |
1 files changed, 12 insertions, 3 deletions
diff --git a/fs/proc/proc_misc.c b/fs/proc/proc_misc.c index d24b8d46059a..f133afebed7a 100644 --- a/fs/proc/proc_misc.c +++ b/fs/proc/proc_misc.c | |||
@@ -445,6 +445,11 @@ static int show_stat(struct seq_file *p, void *v) | |||
445 | cputime64_t user, nice, system, idle, iowait, irq, softirq, steal; | 445 | cputime64_t user, nice, system, idle, iowait, irq, softirq, steal; |
446 | u64 sum = 0; | 446 | u64 sum = 0; |
447 | struct timespec boottime; | 447 | struct timespec boottime; |
448 | unsigned int *per_irq_sum; | ||
449 | |||
450 | per_irq_sum = kzalloc(sizeof(unsigned int)*NR_IRQS, GFP_KERNEL); | ||
451 | if (!per_irq_sum) | ||
452 | return -ENOMEM; | ||
448 | 453 | ||
449 | user = nice = system = idle = iowait = | 454 | user = nice = system = idle = iowait = |
450 | irq = softirq = steal = cputime64_zero; | 455 | irq = softirq = steal = cputime64_zero; |
@@ -462,8 +467,11 @@ static int show_stat(struct seq_file *p, void *v) | |||
462 | irq = cputime64_add(irq, kstat_cpu(i).cpustat.irq); | 467 | irq = cputime64_add(irq, kstat_cpu(i).cpustat.irq); |
463 | softirq = cputime64_add(softirq, kstat_cpu(i).cpustat.softirq); | 468 | softirq = cputime64_add(softirq, kstat_cpu(i).cpustat.softirq); |
464 | steal = cputime64_add(steal, kstat_cpu(i).cpustat.steal); | 469 | steal = cputime64_add(steal, kstat_cpu(i).cpustat.steal); |
465 | for (j = 0 ; j < NR_IRQS ; j++) | 470 | for (j = 0; j < NR_IRQS; j++) { |
466 | sum += kstat_cpu(i).irqs[j]; | 471 | unsigned int temp = kstat_cpu(i).irqs[j]; |
472 | sum += temp; | ||
473 | per_irq_sum[j] += temp; | ||
474 | } | ||
467 | } | 475 | } |
468 | 476 | ||
469 | seq_printf(p, "cpu %llu %llu %llu %llu %llu %llu %llu %llu\n", | 477 | seq_printf(p, "cpu %llu %llu %llu %llu %llu %llu %llu %llu\n", |
@@ -501,7 +509,7 @@ static int show_stat(struct seq_file *p, void *v) | |||
501 | 509 | ||
502 | #if !defined(CONFIG_PPC64) && !defined(CONFIG_ALPHA) && !defined(CONFIG_IA64) | 510 | #if !defined(CONFIG_PPC64) && !defined(CONFIG_ALPHA) && !defined(CONFIG_IA64) |
503 | for (i = 0; i < NR_IRQS; i++) | 511 | for (i = 0; i < NR_IRQS; i++) |
504 | seq_printf(p, " %u", kstat_irqs(i)); | 512 | seq_printf(p, " %u", per_irq_sum[i]); |
505 | #endif | 513 | #endif |
506 | 514 | ||
507 | seq_printf(p, | 515 | seq_printf(p, |
@@ -516,6 +524,7 @@ static int show_stat(struct seq_file *p, void *v) | |||
516 | nr_running(), | 524 | nr_running(), |
517 | nr_iowait()); | 525 | nr_iowait()); |
518 | 526 | ||
527 | kfree(per_irq_sum); | ||
519 | return 0; | 528 | return 0; |
520 | } | 529 | } |
521 | 530 | ||