From: Eric Dumazet <eric.dumazet@gmail.com>
To: David Miller <davem@davemloft.net>
Cc: netdev <netdev@vger.kernel.org>
Subject: [PATCH net-next] fib: introduce fib_alias_accessed() helper
Date: Thu, 21 Oct 2010 10:03:38 +0200 [thread overview]
Message-ID: <1287648218.6871.18.camel@edumazet-laptop> (raw)
Perf tools session at NFWS 2010 pointed out a false sharing on struct
fib_alias that can be avoided pretty easily, if we set FA_S_ACCESSED bit
only if needed (ie : not already set)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
net/ipv4/fib_hash.c | 3 ++-
net/ipv4/fib_lookup.h | 7 +++++++
net/ipv4/fib_semantics.c | 2 +-
net/ipv4/fib_trie.c | 3 ++-
4 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/net/ipv4/fib_hash.c b/net/ipv4/fib_hash.c
index 4f1aafd..43e1c59 100644
--- a/net/ipv4/fib_hash.c
+++ b/net/ipv4/fib_hash.c
@@ -335,7 +335,8 @@ void fib_table_select_default(struct fib_table *tb,
if (!next_fi->fib_nh[0].nh_gw ||
next_fi->fib_nh[0].nh_scope != RT_SCOPE_LINK)
continue;
- fa->fa_state |= FA_S_ACCESSED;
+
+ fib_alias_accessed(fa);
if (fi == NULL) {
if (next_fi != res->fi)
diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
index 5072d8e..a29edf2 100644
--- a/net/ipv4/fib_lookup.h
+++ b/net/ipv4/fib_lookup.h
@@ -17,6 +17,13 @@ struct fib_alias {
#define FA_S_ACCESSED 0x01
+/* Dont write on fa_state unless needed, to keep it shared on all cpus */
+static inline void fib_alias_accessed(struct fib_alias *fa)
+{
+ if (!(fa->fa_state & FA_S_ACCESSED))
+ fa->fa_state |= FA_S_ACCESSED;
+}
+
/* Exported by fib_semantics.c */
extern int fib_semantic_match(struct list_head *head,
const struct flowi *flp,
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 6734c9c..3e0da3e 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -901,7 +901,7 @@ int fib_semantic_match(struct list_head *head, const struct flowi *flp,
if (fa->fa_scope < flp->fl4_scope)
continue;
- fa->fa_state |= FA_S_ACCESSED;
+ fib_alias_accessed(fa);
err = fib_props[fa->fa_type].error;
if (err == 0) {
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 31494f3..cd5e13a 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1838,7 +1838,8 @@ void fib_table_select_default(struct fib_table *tb,
if (!next_fi->fib_nh[0].nh_gw ||
next_fi->fib_nh[0].nh_scope != RT_SCOPE_LINK)
continue;
- fa->fa_state |= FA_S_ACCESSED;
+
+ fib_alias_accessed(fa);
if (fi == NULL) {
if (next_fi != res->fi)
next reply other threads:[~2010-10-21 8:03 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-10-21 8:03 Eric Dumazet [this message]
2010-10-21 10:11 ` [PATCH net-next] fib: introduce fib_alias_accessed() helper David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1287648218.6871.18.camel@edumazet-laptop \
--to=eric.dumazet@gmail.com \
--cc=davem@davemloft.net \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox