[PATCH] FIB_TRIE: RCU refine

[PATCH] FIB_TRIE: RCU refine

[Wang Chen]

David, sorry to send this patch again.
I sent it yesterday, but seems that it was refused by
netdev maillist. I subscribed the mailist and send it
again.

On reader side, RCU list-traversal functions, such as
*list_for_each_entry_rcu(), should be within an RCU
read-side critical section.

Surround them by rcu_read_*lock().

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
---
 net/ipv4/fib_trie.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 1ff446d..092e02b 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -2416,6 +2416,7 @@ static int fib_trie_seq_show(struct seq_file *seq, void *v)
 		seq_indent(seq, iter->depth);
 		seq_printf(seq, "  |-- %d.%d.%d.%d\n", NIPQUAD(val));
 
+		rcu_read_lock();
 		hlist_for_each_entry_rcu(li, node, &l->list, hlist) {
 			struct fib_alias *fa;
 
@@ -2433,6 +2434,7 @@ static int fib_trie_seq_show(struct seq_file *seq, void *v)
 				seq_putc(seq, '\n');
 			}
 		}
+		rcu_read_unlock();
 	}
 
 	return 0;
-- 
WCN

Re: [PATCH] FIB_TRIE: RCU refine

[David Miller]

From: Wang Chen <wangchen@cn.fujitsu.com>
Date: Thu, 20 Mar 2008 13:15:07 +0800

> I sent it yesterday, but seems that it was refused by
> netdev maillist. I subscribed the mailist and send it
> again.

The server was down starting yesterday afternoon.

It did not reject your posting, and you do not
need to subscribe in order to post.

Re: [PATCH] FIB_TRIE: RCU refine

[Stephen Hemminger]

> David, sorry to send this patch again.
> I sent it yesterday, but seems that it was refused by
> netdev maillist. I subscribed the mailist and send it
> again.
> 
> On reader side, RCU list-traversal functions, such as
> *list_for_each_entry_rcu(), should be within an RCU
> read-side critical section.
> 
> Surround them by rcu_read_*lock()

This patch is unnecessary since rcu_read_lock is already fib_trie_seq_start
and fib_trie_seq_stop.

Re: [PATCH] FIB_TRIE: RCU refine

[David Miller]

From: Stephen Hemminger <shemminger@vyatta.com>
Date: Thu, 20 Mar 2008 08:35:44 -0700

> 
> > David, sorry to send this patch again.
> > I sent it yesterday, but seems that it was refused by
> > netdev maillist. I subscribed the mailist and send it
> > again.
> > 
> > On reader side, RCU list-traversal functions, such as
> > *list_for_each_entry_rcu(), should be within an RCU
> > read-side critical section.
> > 
> > Surround them by rcu_read_*lock()
> 
> This patch is unnecessary since rcu_read_lock is already fib_trie_seq_start
> and fib_trie_seq_stop.

COrrect.

Re: [PATCH] FIB_TRIE: RCU refine

[Wang Chen]

Stephen Hemminger said the following on 2008-3-20 23:35:
>> David, sorry to send this patch again.
>> I sent it yesterday, but seems that it was refused by
>> netdev maillist. I subscribed the mailist and send it
>> again.
>>
>> On reader side, RCU list-traversal functions, such as
>> *list_for_each_entry_rcu(), should be within an RCU
>> read-side critical section.
>>
>> Surround them by rcu_read_*lock()
> 
> This patch is unnecessary since rcu_read_lock is already fib_trie_seq_start
> and fib_trie_seq_stop.
> 

I see.
Thank you, Stephen.