Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] net/rds: Avoid potential use after free in
 rds_send_remove_from_sock
To:     Aditya Pakki <pakki001@umn.edu>
Cc:     Santosh Shilimkar <santosh.shilimkar@oracle.com>,
        "David S. Miller" <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>, netdev@vger.kernel.org,
        linux-rdma@vger.kernel.org, rds-devel@oss.oracle.com,
        linux-kernel@vger.kernel.org
References: <20210407000913.2207831-1-pakki001@umn.edu>
From:   Eric Dumazet <eric.dumazet@gmail.com>
Message-ID: <bd3c84bc-6ae0-63e9-61f2-5cf64a976531@gmail.com>
Date:   Fri, 9 Apr 2021 12:26:15 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.8.0
MIME-Version: 1.0
In-Reply-To: <20210407000913.2207831-1-pakki001@umn.edu>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 4/7/21 2:09 AM, Aditya Pakki wrote:
> In case of rs failure in rds_send_remove_from_sock(), the 'rm' resource
> is freed and later under spinlock, causing potential use-after-free.
> Set the free pointer to NULL to avoid undefined behavior.
> 
> Signed-off-by: Aditya Pakki <pakki001@umn.edu>
> ---
>  net/rds/message.c | 1 +
>  net/rds/send.c    | 2 +-
>  2 files changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/net/rds/message.c b/net/rds/message.c
> index 071a261fdaab..90ebcfe5fe3b 100644
> --- a/net/rds/message.c
> +++ b/net/rds/message.c
> @@ -180,6 +180,7 @@ void rds_message_put(struct rds_message *rm)
>  		rds_message_purge(rm);
>  
>  		kfree(rm);
> +		rm = NULL;

This is a nop really.

This does not clear @rm variable in the caller.


>  	}
>  }
>  EXPORT_SYMBOL_GPL(rds_message_put);
> diff --git a/net/rds/send.c b/net/rds/send.c
> index 985d0b7713ac..fe5264b9d4b3 100644
> --- a/net/rds/send.c
> +++ b/net/rds/send.c
> @@ -665,7 +665,7 @@ static void rds_send_remove_from_sock(struct list_head *messages, int status)
>  unlock_and_drop:
>  		spin_unlock_irqrestore(&rm->m_rs_lock, flags);
>  		rds_message_put(rm);
> -		if (was_on_sock)
> +		if (was_on_sock && rm)
>  			rds_message_put(rm);

Maybe the bug is that the refcount has not be elevated when was_on_sock
has been set.

>  	}
>  
>