lkml.org 
[lkml]   [2006]   [Jul]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [patch] reduce IPI noise due to /dev/cdrom open/close
Date
Jes Sorensen (on 03 Jul 2006 11:33:54 -0400) wrote:
>Anyway, this patch reduces the IPI noise by keeping a cpumask of CPUs
>which have items in the bh lru and only flushing on the relevant
>CPUs. On systems with larger CPU counts it's quite normal that only a
>few CPUs are actively doing block IO, so spewing IPIs everywhere to
>flush this is unnecessary.
>
>Index: linux-2.6/fs/buffer.c
>===================================================================
>--- linux-2.6.orig/fs/buffer.c
>+++ linux-2.6/fs/buffer.c
>@@ -1323,6 +1323,7 @@ struct bh_lru {
> };
>
> static DEFINE_PER_CPU(struct bh_lru, bh_lrus) = {{ NULL }};
>+static cpumask_t lru_in_use;
>
> #ifdef CONFIG_SMP
> #define bh_lru_lock() local_irq_disable()
>@@ -1352,9 +1353,14 @@ static void bh_lru_install(struct buffer
> lru = &__get_cpu_var(bh_lrus);
> if (lru->bhs[0] != bh) {
> struct buffer_head *bhs[BH_LRU_SIZE];
>- int in;
>- int out = 0;
>+ int in, out, cpu;
>
>+ cpu = raw_smp_processor_id();

Why raw_smp_processor_id? That normally indicates code that wants a
lazy cpu number, but this code requires the exact cpu number, IMHO
using raw_smp_processor_id is confusing. smp_processor_id can safely
be used here, bh_lru_lock has disabled irq or preempt.

>+ /* Test first to avoid cache lines bouncing around */
>+ if (!cpu_isset(cpu, lru_in_use))
>+ cpu_set(cpu, lru_in_use);
>+
>+ out = 0;
> get_bh(bh);
> bhs[out++] = bh;
> for (in = 0; in < BH_LRU_SIZE; in++) {
>@@ -1500,19 +1506,28 @@ EXPORT_SYMBOL(__bread);
> */
> static void invalidate_bh_lru(void *arg)
> {
>- struct bh_lru *b = &get_cpu_var(bh_lrus);
>+ struct bh_lru *b;
> int i;
>
>+ local_irq_disable();
>+ b = &get_cpu_var(bh_lrus);
> for (i = 0; i < BH_LRU_SIZE; i++) {
> brelse(b->bhs[i]);
> b->bhs[i] = NULL;
> }
> put_cpu_var(bh_lrus);
>+ local_irq_enable();
> }
>
> static void invalidate_bh_lrus(void)
> {
>- on_each_cpu(invalidate_bh_lru, NULL, 1, 1);
>+ /*
>+ * Need to hand down a copy of the mask or we wouldn't be run
>+ * anywhere due to the original mask being cleared
>+ */
>+ cpumask_t mask = lru_in_use;
>+ cpus_clear(lru_in_use);
>+ schedule_on_each_cpu_mask(invalidate_bh_lru, NULL, mask);
> }

Racy? Start with an empty lru_in_use.

Cpu A Cpu B
invalidate_bh_lrus()
mask = lru_in_use;
preempted
block I/O
bh_lru_install()
cpu_set(cpu, lru_in_use);
resume
cpus_clear(lru_in_use);
schedule_on_each_cpu_mask() - does not send IPI to cpu B

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2006-07-04 07:35    [from the cache]
©2003-2011 Jasper Spaans