Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
DMARC-Filter: OpenDMARC Filter v1.3.2 smtp.codeaurora.org A8C5960128
Date:   Wed, 14 Aug 2019 14:32:03 -0700
From:   Jack Pham <jackp@codeaurora.org>
To:     Wenwen Wang <wenwen@cs.uga.edu>
Cc:     Oliver Neukum <oneukum@suse.com>,
        "David S. Miller" <davem@davemloft.net>,
        "open list:USB \"USBNET\" DRIVER FRAMEWORK" <netdev@vger.kernel.org>,
        "open list:USB NETWORKING DRIVERS" <linux-usb@vger.kernel.org>,
        open list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] net: usbnet: fix a memory leak bug
Message-ID: <20190814213203.GA9754@jackp-linux.qualcomm.com>
References: <1565804493-7758-1-git-send-email-wenwen@cs.uga.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1565804493-7758-1-git-send-email-wenwen@cs.uga.edu>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Aug 14, 2019 at 12:41:33PM -0500, Wenwen Wang wrote:
> In usbnet_start_xmit(), 'urb->sg' is allocated through kmalloc_array() by
> invoking build_dma_sg(). Later on, if 'CONFIG_PM' is defined and the if
> branch is taken, the execution will go to the label 'deferred'. However,
> 'urb->sg' is not deallocated on this execution path, leading to a memory
> leak bug.
> 
> Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
> ---
>  drivers/net/usb/usbnet.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/net/usb/usbnet.c b/drivers/net/usb/usbnet.c
> index 72514c4..f17fafa 100644
> --- a/drivers/net/usb/usbnet.c
> +++ b/drivers/net/usb/usbnet.c
> @@ -1433,6 +1433,7 @@ netdev_tx_t usbnet_start_xmit (struct sk_buff *skb,
>  		usb_anchor_urb(urb, &dev->deferred);
>  		/* no use to process more packets */
>  		netif_stop_queue(net);
> +		kfree(urb->sg);
>  		usb_put_urb(urb);

The URB itself is not getting freed here; it is merely added to the
anchor list and will be submitted later upon usbnet_resume(). Therefore
freeing the SG list is premature and incorrect, as it will get freed in
either the tx_complete/tx_done path or upon URB submission failure.

>  		spin_unlock_irqrestore(&dev->txq.lock, flags);
>  		netdev_dbg(dev->net, "Delaying transmission for resumption\n");

Jack
-- 
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project