netdev - Re: [BISECTED] Linux 3.2: Networking out of LAN does not work

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 for Android: free password hash cracker in your pocket

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

Date:	Wed, 25 Jan 2012 12:05:31 +0100
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Rune Magnussen <rum@...kdata.dk>,
	David Miller <davem@...emloft.net>
Cc:	netdev@...r.kernel.org, Mihai Maruseac <mihai.maruseac@...il.com>
Subject: Re: [BISECTED] Linux 3.2: Networking out of LAN does not work

Le mercredi 25 janvier 2012 à 10:40 +0100, Eric Dumazet a écrit :
> Le mercredi 25 janvier 2012 à 10:35 +0100, Rune Magnussen a écrit :
> > Hi
> > 
> > Networking to and from the outside of my LAN does not work with Linux 
> > 3.2.x.
> > Linux 3.1.x works. Using ssh from a PC on the LAN also works.
> > The machine has a VIA C7 CPU and the NIC uses the 8139too-driver.
> > 
> > Another machine with a r8169-NIC and an Atom-CPU (32bit) did not 
> > show any problems.
> > It was used very ligtly though.
> > 
> > Bisection led to this:
> > ---
> > f04565ddf52e401880f8ba51de0dff8ba51c99fd is the first bad commit
> > commit f04565ddf52e401880f8ba51de0dff8ba51c99fd
> > Author: Mihai Maruseac <mihai.maruseac@...il.com>
> > Date:   Thu Oct 20 20:45:10 2011 +0000
> > 
> >     dev: use name hash for dev_seq_ops
> >  
> >     Instead of using the dev->next chain and trying to resync at each call 
> > to
> >     dev_seq_start, use the name hash, keeping the bucket and the offset in
> >     seq->private field.
> >  
> >     Tests revealed the following results for ifconfig > /dev/null
> >        * 1000 interfaces:
> >           * 0.114s without patch
> >           * 0.089s with patch
> >        * 3000 interfaces:
> >           * 0.489s without patch
> >           * 0.110s with patch
> >        * 5000 interfaces:
> >           * 1.363s without patch
> >           * 0.250s with patch
> >        * 128000 interfaces (other setup):
> >           * ~100s without patch
> >           * ~30s with patch
> >  
> >     Signed-off-by: Mihai Maruseac <mmaruseac@...acom.com>
> >     Signed-off-by: Eric Dumazet <eric.dumazet@...il.com>
> >     Signed-off-by: David S. Miller <davem@...emloft.net>
> > 
> > :040000 040000 2e86e1c9ecaf719d312a8b5863ec9e89d92bc4b3 
> > 8b0d09ca105502dd5389bb6281ef059c8bb38667 M   net
> > ---
> > 
> > Simply reverting the commit on top of Linux 3.2.1 resultet in a build 
> > error.
> > 
> > Used kernel configuration: http://knus.info/kernel/config-3.1.txt
> > 
> > Greetings Rune

OK please test the following patch.

[PATCH] net: use index hash for /proc/net/dev

Commit f04565ddf52e (dev: use name hash for dev_seq_ops) added a
regression for legacy apps expecting to find devices in a particular
order.

Instead of using name hash, we can use dev_index_head hash, based on
device index, to not change device ordering.

Name hash order is not particularly useful, while index ordering is
meaningful on machines where no more than 256 devices are ever created.

"cat /proc/net/dev" and "ip link" will output devices in same order.

Reported-by: Rune Magnussen <rum@...kdata.dk>
Bisected-by: Rune Magnussen <rum@...kdata.dk>
Cc: Mihai Maruseac <mihai.maruseac@...il.com>
Signed-off-by: Eric Dumazet <eric.dumazet@...il.com>
---
 net/core/dev.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 115dee1..c58dba3 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4052,9 +4052,9 @@ static inline struct net_device *dev_from_same_bucket(struct seq_file *seq)
 
 	bucket = get_bucket(state->pos);
 	offset = get_offset(state->pos);
-	h = &net->dev_name_head[bucket];
+	h = dev_index_hash(net, bucket);
 	count = 0;
-	hlist_for_each_entry_rcu(dev, p, h, name_hlist) {
+	hlist_for_each_entry_rcu(dev, p, h, index_hlist) {
 		if (count++ == offset) {
 			state->pos = set_bucket_offset(bucket, count);
 			return dev;


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html