netdev - Re: [PATCH v1 net 1/3] gtp: Use for_each_netdev_rcu() in gtp_genl_dump

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]

Message-ID: <20250108122536.51089-1-kuniyu@amazon.com>
Date: Wed, 8 Jan 2025 21:25:36 +0900
From: Kuniyuki Iwashima <kuniyu@...zon.com>
To: <horms@...nel.org>
CC: <andrew+netdev@...n.ch>, <davem@...emloft.net>, <edumazet@...gle.com>,
	<kuba@...nel.org>, <kuni1840@...il.com>, <kuniyu@...zon.com>,
	<laforge@...monks.org>, <netdev@...r.kernel.org>, <pabeni@...hat.com>,
	<pablo@...filter.org>, <shaw.leon@...il.com>
Subject: Re: [PATCH v1 net 1/3] gtp: Use for_each_netdev_rcu() in gtp_genl_dump_pdp().

From: Simon Horman <horms@...nel.org>
Date: Wed, 8 Jan 2025 10:40:40 +0000
> On Wed, Jan 08, 2025 at 03:28:32PM +0900, Kuniyuki Iwashima wrote:
> > gtp_newlink() links the gtp device to a list in dev_net(dev).
> > 
> > However, even after the gtp device is moved to another netns,
> > it stays on the list but should be invisible.
> > 
> > Let's use for_each_netdev_rcu() for netdev traversal in
> > gtp_genl_dump_pdp().
> > 
> > Note that gtp_dev_list is no longer used under RCU, so list
> > helpers are converted to the non-RCU variant.
> > 
> > Fixes: 459aa660eb1d ("gtp: add initial driver for datapath of GPRS Tunneling Protocol (GTP-U)")
> > Reported-by: Xiao Liang <shaw.leon@...il.com>
> > Closes: https://lore.kernel.org/netdev/CABAhCOQdBL6h9M2C+kd+bGivRJ9Q72JUxW+-gur0nub_=PmFPA@mail.gmail.com/
> > Signed-off-by: Kuniyuki Iwashima <kuniyu@...zon.com>
> > ---
> > Cc: Pablo Neira Ayuso <pablo@...filter.org>
> > Cc: Harald Welte <laforge@...monks.org>
> 
> ...
> 
> > @@ -2280,7 +2281,10 @@ static int gtp_genl_dump_pdp(struct sk_buff *skb,
> >  		return 0;
> >  
> >  	rcu_read_lock();
> > -	list_for_each_entry_rcu(gtp, &gn->gtp_dev_list, list) {
> > +	for_each_netdev_rcu(net, dev) {
> > +		if (dev->rtnl_link_ops != &gtp_link_ops)
> > +			continue;
> > +
> >  		if (last_gtp && last_gtp != gtp)
> >  			continue;
> >  		else
> 
> Hi Iwashima-san,
> 
> With this change gtp seems to be uninitialised here.
> And, also, it looks like gn is now set but otherwise unused in this function.
> 
> Flagged by W=1 builds with clang-19.

Ah, thank you for catching these !

I'll squash this to v2.

---8<---
diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c
index b905b7bc46cb..0d03e150efc6 100644
--- a/drivers/net/gtp.c
+++ b/drivers/net/gtp.c
@@ -2298,9 +2298,6 @@ static int gtp_genl_dump_pdp(struct sk_buff *skb,
 	struct net *net = sock_net(skb->sk);
 	struct net_device *dev;
 	struct pdp_ctx *pctx;
-	struct gtp_net *gn;
-
-	gn = net_generic(net, gtp_net_id);
 
 	if (cb->args[4])
 		return 0;
@@ -2310,6 +2307,8 @@ static int gtp_genl_dump_pdp(struct sk_buff *skb,
 		if (dev->rtnl_link_ops != &gtp_link_ops)
 			continue;
 
+		gtp = netdev_priv(dev);
+
 		if (last_gtp && last_gtp != gtp)
 			continue;
 		else
---8<---