Received-SPF: pass (google.com: domain of linux-kernel+bounces-182094-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Fri, 17 May 2024 13:22:38 +0100
From: Simon Horman <horms@kernel.org>
To: Hagar Hemdan <hagarhem@amazon.com>
Cc: Norbert Manthey <nmanthey@amazon.de>,
	Steffen Klassert <steffen.klassert@secunet.com>,
	Herbert Xu <herbert@gondor.apana.org.au>,
	"David S. Miller" <davem@davemloft.net>,
	David Ahern <dsahern@kernel.org>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Sabrina Dubroca <sd@queasysnail.net>, netdev@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] net: esp: cleanup esp_output_tail_tcp() in case of
 unsupported ESPINTCP
Message-ID: <20240517122238.GE443576@kernel.org>
References: <20240516080309.1872-1-hagarhem@amazon.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240516080309.1872-1-hagarhem@amazon.com>

On Thu, May 16, 2024 at 08:03:09AM +0000, Hagar Hemdan wrote:
> xmit() functions should consume skb or return error codes in error
> paths.
> When the configuration "CONFIG_INET_ESPINTCP" is not used, the
> implementation of the function "esp_output_tail_tcp" violates this rule.
> The function frees the skb and returns the error code.
> This change removes the kfree_skb from both functions, for both
> esp4 and esp6.
> 
> This should not be reachable in the current code, so this change is just
> a cleanup.
> 
> This bug was discovered and resolved using Coverity Static Analysis
> Security Testing (SAST) by Synopsys, Inc.
> 
> Fixes: e27cca96cd68 ("xfrm: add espintcp (RFC 8229)")
> Signed-off-by: Hagar Hemdan <hagarhem@amazon.com>

Hi Hagar,

If esp_output() may be the x->type->output callback called from esp_output()
then I agree that this seems to be a problem as it looks like a double free
may occur.

However, I believe that your proposed fix introduces will result in skb
being leaked leak in the case of esp_output_done() calling
esp_output_tail_tcp(). Perhaps a solution is for esp_output_done()
to free the skb if esp_output_tail_tcp() fails.

I did not analyse other call-chains, but I think such analysis is needed.

..