From: Andrew Morton <akpm@osdl.org>
Date: Wed, 22 Mar 2006 08:07:39 +0000 (-0800)
Subject: [PATCH] __get_page_state() cpumask cleanup and fix
X-Git-Url: http://git.lede-project.org./?a=commitdiff_plain;h=b40607fc02f8248828d52d88f91b7d68df1933b0;p=openwrt%2Fstaging%2Fblogic.git

[PATCH] __get_page_state() cpumask cleanup and fix

__get_page_state() has an open-coded for_each_cpu_mask() loop in it.

Tidy that up, then notice that the code was buggy:

	while (cpu < NR_CPUS) {
		unsigned long *in, *out, off;

		if (!cpu_isset(cpu, *cpumask))
			continue;

an obvious infinite loop.  I guess we just never call it with a holey cpu
mask.

Even after my cpumask size-reduction work, this patch increases code size :(

Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
---

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 234bd4895d14..61775866ea18 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1214,24 +1214,22 @@ DEFINE_PER_CPU(long, nr_pagecache_local) = 0;
 
 static void __get_page_state(struct page_state *ret, int nr, cpumask_t *cpumask)
 {
-	int cpu = 0;
+	unsigned cpu;
 
 	memset(ret, 0, nr * sizeof(unsigned long));
 	cpus_and(*cpumask, *cpumask, cpu_online_map);
 
-	cpu = first_cpu(*cpumask);
-	while (cpu < NR_CPUS) {
-		unsigned long *in, *out, off;
-
-		if (!cpu_isset(cpu, *cpumask))
-			continue;
+	for_each_cpu_mask(cpu, *cpumask) {
+		unsigned long *in;
+		unsigned long *out;
+		unsigned off;
+		unsigned next_cpu;
 
 		in = (unsigned long *)&per_cpu(page_states, cpu);
 
-		cpu = next_cpu(cpu, *cpumask);
-
-		if (likely(cpu < NR_CPUS))
-			prefetch(&per_cpu(page_states, cpu));
+		next_cpu = next_cpu(cpu, *cpumask);
+		if (likely(next_cpu < NR_CPUS))
+			prefetch(&per_cpu(page_states, next_cpu));
 
 		out = (unsigned long *)ret;
 		for (off = 0; off < nr; off++)