lkml.org 
[lkml]   [2012]   [Mar]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH v4 07/12] ima: replace iint spinlock with rwlock/read_lock
Date
For performance, replace the iint spinlock with rwlock/read_lock.

Eric Paris questioned this change, from spinlocks to rwlocks, saying
"rwlocks have been shown to actually be slower on multi processor
systems in a number of cases due to the cache line bouncing required."

Based on performance measurements compiling the kernel on a cold
boot with multiple jobs with/without this patch, Dmitry Kasatkin
and I found that rwlocks performed better than spinlocks, but very
insignificantly. For example with total compilation time around 6
minutes, with rwlocks time was 1 - 3 seconds shorter... but always
like that.

Changelog v2:
- new patch taken from the 'allocating iint improvements' patch

Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@intel.com>
Signed-off-by: Mimi Zohar <zohar@us.ibm.com>
---
security/integrity/iint.c | 16 +++++++---------
1 files changed, 7 insertions(+), 9 deletions(-)
diff --git a/security/integrity/iint.c b/security/integrity/iint.c
index c91a436..d82a5a1 100644
--- a/security/integrity/iint.c
+++ b/security/integrity/iint.c
@@ -22,7 +22,7 @@
#include "integrity.h"

static struct rb_root integrity_iint_tree = RB_ROOT;
-static DEFINE_SPINLOCK(integrity_iint_lock);
+static DEFINE_RWLOCK(integrity_iint_lock);
static struct kmem_cache *iint_cache __read_mostly;

int iint_initialized;
@@ -35,8 +35,6 @@ static struct integrity_iint_cache *__integrity_iint_find(struct inode *inode)
struct integrity_iint_cache *iint;
struct rb_node *n = integrity_iint_tree.rb_node;

- assert_spin_locked(&integrity_iint_lock);
-
while (n) {
iint = rb_entry(n, struct integrity_iint_cache, rb_node);

@@ -63,9 +61,9 @@ struct integrity_iint_cache *integrity_iint_find(struct inode *inode)
if (!IS_IMA(inode))
return NULL;

- spin_lock(&integrity_iint_lock);
+ read_lock(&integrity_iint_lock);
iint = __integrity_iint_find(inode);
- spin_unlock(&integrity_iint_lock);
+ read_unlock(&integrity_iint_lock);

return iint;
}
@@ -100,7 +98,7 @@ struct integrity_iint_cache *integrity_inode_get(struct inode *inode)
if (!iint)
return NULL;

- spin_lock(&integrity_iint_lock);
+ write_lock(&integrity_iint_lock);

p = &integrity_iint_tree.rb_node;
while (*p) {
@@ -119,7 +117,7 @@ struct integrity_iint_cache *integrity_inode_get(struct inode *inode)
rb_link_node(node, parent, p);
rb_insert_color(node, &integrity_iint_tree);

- spin_unlock(&integrity_iint_lock);
+ write_unlock(&integrity_iint_lock);
return iint;
}

@@ -136,10 +134,10 @@ void integrity_inode_free(struct inode *inode)
if (!IS_IMA(inode))
return;

- spin_lock(&integrity_iint_lock);
+ write_lock(&integrity_iint_lock);
iint = __integrity_iint_find(inode);
rb_erase(&iint->rb_node, &integrity_iint_tree);
- spin_unlock(&integrity_iint_lock);
+ write_unlock(&integrity_iint_lock);

iint_free(iint);
}
--
1.7.6.5


\
 
 \ /
  Last update: 2012-03-29 16:45    [from the cache]
©2003-2011 Jasper Spaans