Module Name: src
Committed By: riastradh
Date: Sat Feb 1 12:51:19 UTC 2020
Modified Files:
src/sys/kern: subr_percpu.c
Log Message:
Use __read_mostly and gather related __cacheline_aligned together.
Should save a few bytes of kernel.
To generate a diff of this commit:
cvs rdiff -u -r1.21 -r1.22 src/sys/kern/subr_percpu.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
Modified files:
Index: src/sys/kern/subr_percpu.c
diff -u src/sys/kern/subr_percpu.c:1.21 src/sys/kern/subr_percpu.c:1.22
--- src/sys/kern/subr_percpu.c:1.21 Sat Feb 1 12:49:02 2020
+++ src/sys/kern/subr_percpu.c Sat Feb 1 12:51:18 2020
@@ -1,4 +1,4 @@
-/* $NetBSD: subr_percpu.c,v 1.21 2020/02/01 12:49:02 riastradh Exp $ */
+/* $NetBSD: subr_percpu.c,v 1.22 2020/02/01 12:51:18 riastradh Exp $ */
/*-
* Copyright (c)2007,2008 YAMAMOTO Takashi,
@@ -31,7 +31,7 @@
*/
#include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: subr_percpu.c,v 1.21 2020/02/01 12:49:02 riastradh Exp $");
+__KERNEL_RCSID(0, "$NetBSD: subr_percpu.c,v 1.22 2020/02/01 12:51:18 riastradh Exp $");
#include <sys/param.h>
#include <sys/cpu.h>
@@ -55,9 +55,11 @@ struct percpu {
};
static krwlock_t percpu_swap_lock __cacheline_aligned;
-static kmutex_t percpu_allocation_lock __cacheline_aligned;
-static vmem_t * percpu_offset_arena __cacheline_aligned;
-static unsigned int percpu_nextoff __cacheline_aligned;
+static vmem_t * percpu_offset_arena __read_mostly;
+static struct {
+ kmutex_t lock;
+ unsigned int nextoff;
+} percpu_allocation __cacheline_aligned;
static percpu_cpu_t *
cpu_percpu(struct cpu_info *ci)
@@ -71,7 +73,7 @@ percpu_offset(percpu_t *pc)
{
const unsigned int off = pc->pc_offset;
- KASSERT(off < percpu_nextoff);
+ KASSERT(off < percpu_allocation.nextoff);
return off;
}
@@ -172,10 +174,10 @@ percpu_backend_alloc(vmem_t *dummy, vmem
return ENOMEM;
size = roundup(size, PERCPU_IMPORT_SIZE);
- mutex_enter(&percpu_allocation_lock);
- offset = percpu_nextoff;
- percpu_nextoff = nextoff = percpu_nextoff + size;
- mutex_exit(&percpu_allocation_lock);
+ mutex_enter(&percpu_allocation.lock);
+ offset = percpu_allocation.nextoff;
+ percpu_allocation.nextoff = nextoff = percpu_allocation.nextoff + size;
+ mutex_exit(&percpu_allocation.lock);
percpu_cpu_enlarge(nextoff);
@@ -213,8 +215,8 @@ percpu_init(void)
ASSERT_SLEEPABLE();
rw_init(&percpu_swap_lock);
- mutex_init(&percpu_allocation_lock, MUTEX_DEFAULT, IPL_NONE);
- percpu_nextoff = PERCPU_QUANTUM_SIZE;
+ mutex_init(&percpu_allocation.lock, MUTEX_DEFAULT, IPL_NONE);
+ percpu_allocation.nextoff = PERCPU_QUANTUM_SIZE;
percpu_offset_arena = vmem_xcreate("percpu", 0, 0, PERCPU_QUANTUM_SIZE,
percpu_backend_alloc, NULL, NULL, PERCPU_QCACHE_MAX, VM_SLEEP,
@@ -231,7 +233,7 @@ void
percpu_init_cpu(struct cpu_info *ci)
{
percpu_cpu_t * const pcc = cpu_percpu(ci);
- size_t size = percpu_nextoff; /* XXX racy */
+ size_t size = percpu_allocation.nextoff; /* XXX racy */
ASSERT_SLEEPABLE();
pcc->pcc_size = size;
