linux-kernel - [PATCH net v2] net: ethtool: Don't call .cleanup_data when prepare

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-next>] [day] [month] [year] [list]

Message-ID: <20250407130511.75621-1-maxime.chevallier@bootlin.com>
Date: Mon,  7 Apr 2025 15:05:10 +0200
From: Maxime Chevallier <maxime.chevallier@...tlin.com>
To: davem@...emloft.net,
	Andrew Lunn <andrew@...n.ch>,
	Jakub Kicinski <kuba@...nel.org>,
	Eric Dumazet <edumazet@...gle.com>,
	Paolo Abeni <pabeni@...hat.com>
Cc: Maxime Chevallier <maxime.chevallier@...tlin.com>,
	netdev@...r.kernel.org,
	linux-kernel@...r.kernel.org,
	thomas.petazzoni@...tlin.com,
	Simon Horman <horms@...nel.org>,
	Michal Kubecek <mkubecek@...e.cz>,
	Florian Fainelli <f.fainelli@...il.com>,
	Kory Maincent <kory.maincent@...tlin.com>
Subject: [PATCH net v2] net: ethtool: Don't call .cleanup_data when prepare_data fails

There's a consistent pattern where the .cleanup_data() callback is
called when .prepare_data() fails, when it should really be called to
clean after a successful .prepare_data() as per the documentation.

Rewrite the error-handling paths to make sure we don't cleanup
un-prepared data.

Fixes: c781ff12a2f3 ("ethtool: Allow network drivers to dump arbitrary EEPROM data")
Reviewed-by: Kory Maincent <kory.maincent@...tlin.com>
Reviewed-by: Simon Horman <horms@...nel.org>
Reviewed-by: Michal Kubecek <mkubecek@...e.cz>
Signed-off-by: Maxime Chevallier <maxime.chevallier@...tlin.com>
---
V2: - Fixed typo in commit log (Simon)
    - Changed the Fixes tag, as per Jakub's comment. Eeprom appears to
      be the first potentially problematic case, as we risk freeing
      twice the eeprom data. I couldn't test that with what I have
      locally though.
    - Aggregated Simon and Michal's reviews


 net/ethtool/netlink.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/net/ethtool/netlink.c b/net/ethtool/netlink.c
index a163d40c6431..977beeaaa2f9 100644
--- a/net/ethtool/netlink.c
+++ b/net/ethtool/netlink.c
@@ -500,7 +500,7 @@ static int ethnl_default_doit(struct sk_buff *skb, struct genl_info *info)
 		netdev_unlock_ops(req_info->dev);
 	rtnl_unlock();
 	if (ret < 0)
-		goto err_cleanup;
+		goto err_dev;
 	ret = ops->reply_size(req_info, reply_data);
 	if (ret < 0)
 		goto err_cleanup;
@@ -560,7 +560,7 @@ static int ethnl_default_dump_one(struct sk_buff *skb, struct net_device *dev,
 	netdev_unlock_ops(dev);
 	rtnl_unlock();
 	if (ret < 0)
-		goto out;
+		goto out_cancel;
 	ret = ethnl_fill_reply_header(skb, dev, ctx->ops->hdr_attr);
 	if (ret < 0)
 		goto out;
@@ -569,6 +569,7 @@ static int ethnl_default_dump_one(struct sk_buff *skb, struct net_device *dev,
 out:
 	if (ctx->ops->cleanup_data)
 		ctx->ops->cleanup_data(ctx->reply_data);
+out_cancel:
 	ctx->reply_data->dev = NULL;
 	if (ret < 0)
 		genlmsg_cancel(skb, ehdr);
@@ -793,7 +794,7 @@ static void ethnl_default_notify(struct net_device *dev, unsigned int cmd,
 	ethnl_init_reply_data(reply_data, ops, dev);
 	ret = ops->prepare_data(req_info, reply_data, &info);
 	if (ret < 0)
-		goto err_cleanup;
+		goto err_rep;
 	ret = ops->reply_size(req_info, reply_data);
 	if (ret < 0)
 		goto err_cleanup;
@@ -828,6 +829,7 @@ static void ethnl_default_notify(struct net_device *dev, unsigned int cmd,
 err_cleanup:
 	if (ops->cleanup_data)
 		ops->cleanup_data(reply_data);
+err_rep:
 	kfree(reply_data);
 	kfree(req_info);
 	return;
-- 
2.49.0