linux-kernel - [RFC MM] mmap_sem scaling: only scan cpus used by an mm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 6 Nov 2009 13:53:35 -0500 (EST)
From:	Christoph Lameter <cl@...ux-foundation.org>
To:	Andi Kleen <andi@...stfloor.org>
cc:	npiggin@...e.de, linux-kernel@...r.kernel.org, linux-mm@...ck.org,
	Tejun Heo <tj@...nel.org>, Ingo Molnar <mingo@...e.hu>,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	"hugh.dickins@...cali.co.uk" <hugh.dickins@...cali.co.uk>
Subject: [RFC MM] mmap_sem scaling: only scan cpus used by an mm

One way to reduce the cost of the writer lock is to track the cpus used
and loop over the processors in that bitmap.

---
 arch/x86/include/asm/mmu_context.h |    1 +
 include/linux/mm_types.h           |    3 ++-
 kernel/fork.c                      |    2 ++
 mm/init-mm.c                       |    1 +
 4 files changed, 6 insertions(+), 1 deletion(-)

Index: linux-2.6/arch/x86/include/asm/mmu_context.h
===================================================================
--- linux-2.6.orig/arch/x86/include/asm/mmu_context.h	2009-11-06 12:26:24.000000000 -0600
+++ linux-2.6/arch/x86/include/asm/mmu_context.h	2009-11-06 12:26:36.000000000 -0600
@@ -43,6 +43,7 @@ static inline void switch_mm(struct mm_s
 		percpu_write(cpu_tlbstate.active_mm, next);
 #endif
 		cpumask_set_cpu(cpu, mm_cpumask(next));
+		cpumask_set_cpu(cpu, &next->cpus_used);

 		/* Re-load page tables */
 		load_cr3(next->pgd);
Index: linux-2.6/include/linux/mm_types.h
===================================================================
--- linux-2.6.orig/include/linux/mm_types.h	2009-11-06 12:26:35.000000000 -0600
+++ linux-2.6/include/linux/mm_types.h	2009-11-06 12:26:36.000000000 -0600
@@ -241,6 +241,7 @@ struct mm_struct {
 	struct linux_binfmt *binfmt;

 	cpumask_t cpu_vm_mask;
+	cpumask_t cpus_used;

 	/* Architecture-specific MM context */
 	mm_context_t context;
@@ -291,7 +292,7 @@ static inline int mm_has_reader(struct m
 {
 	int cpu;

-	for_each_possible_cpu(cpu)
+	for_each_cpu(cpu, &mm->cpus_used)
 		if (per_cpu(mm->rss->readers, cpu))
 			return 1;

Index: linux-2.6/mm/init-mm.c
===================================================================
--- linux-2.6.orig/mm/init-mm.c	2009-11-06 12:26:35.000000000 -0600
+++ linux-2.6/mm/init-mm.c	2009-11-06 12:26:36.000000000 -0600
@@ -19,5 +19,6 @@ struct mm_struct init_mm = {
 	.page_table_lock =  __SPIN_LOCK_UNLOCKED(init_mm.page_table_lock),
 	.mmlist		= LIST_HEAD_INIT(init_mm.mmlist),
 	.cpu_vm_mask	= CPU_MASK_ALL,
+	.cpus_used	= CPU_MASK_ALL,
 	.rss		= &init_mm_counters,
 };
Index: linux-2.6/kernel/fork.c
===================================================================
--- linux-2.6.orig/kernel/fork.c	2009-11-06 12:26:35.000000000 -0600
+++ linux-2.6/kernel/fork.c	2009-11-06 12:26:40.000000000 -0600
@@ -297,6 +297,8 @@ static int dup_mmap(struct mm_struct *mm
 	mm->cached_hole_size = ~0UL;
 	mm->map_count = 0;
 	cpumask_clear(mm_cpumask(mm));
+	cpumask_clear(&mm->cpus_used);
+	cpumask_set_cpu(smp_processor_id(), &mm->cpus_used);
 	mm->mm_rb = RB_ROOT;
 	rb_link = &mm->mm_rb.rb_node;
 	rb_parent = NULL;

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