Re: [PATCH v3] usb: rtl8150: Initialize buffers to fix KMSAN uninit-value in rtl8150_open

[Greg KH <gregkh@linuxfoundation.org>]

On Wed, Nov 05, 2025 at 07:56:26PM +0000, Dharanitharan R wrote:
> KMSAN reported an uninitialized value use in rtl8150_open().
> Initialize rx_skb->data and intr_buff before submitting URBs to
> ensure memory is in a defined state.
> 
> Changes in v3:
>  - Fixed whitespace and indentation (checkpatch clean)
>  - Corrected syzbot tag
>  - Corrected syzbot hash and tag

Please follow the documentation for where to put this version
information.

> Reported-by: syzbot+b4d5d8faea6996fd55e3@syzkaller.appspotmail.com
> Signed-off-by: Dharanitharan R <dharanitharan725@gmail.com>
> ---
>  drivers/net/usb/rtl8150.c | 34 +++++++++++++++-------------------
>  1 file changed, 15 insertions(+), 19 deletions(-)
> 
> diff --git a/drivers/net/usb/rtl8150.c b/drivers/net/usb/rtl8150.c
> index f1a868f0032e..a7116d03c3d3 100644
> --- a/drivers/net/usb/rtl8150.c
> +++ b/drivers/net/usb/rtl8150.c
> @@ -735,33 +735,30 @@ static int rtl8150_open(struct net_device *netdev)
>  	rtl8150_t *dev = netdev_priv(netdev);
>  	int res;
>  
> -	if (dev->rx_skb == NULL)
> -		dev->rx_skb = pull_skb(dev);
> -	if (!dev->rx_skb)
> -		return -ENOMEM;

It looks like you are attempting to break existing situations?


> -
>  	set_registers(dev, IDR, 6, netdev->dev_addr);
>  
>  	/* Fix: initialize memory before using it (KMSAN uninit-value) */
>  	memset(dev->rx_skb->data, 0, RTL8150_MTU);
>  	memset(dev->intr_buff, 0, INTBUFSIZE);
>  
> -	usb_fill_bulk_urb(dev->rx_urb, dev->udev, usb_rcvbulkpipe(dev->udev, 1),
> -		      dev->rx_skb->data, RTL8150_MTU, read_bulk_callback, dev);
> -	if ((res = usb_submit_urb(dev->rx_urb, GFP_KERNEL))) {
> -		if (res == -ENODEV)
> -			netif_device_detach(dev->netdev);
> +	usb_fill_bulk_urb(dev->rx_urb, dev->udev,
> +			  usb_rcvbulkpipe(dev->udev, 1),
> +			  dev->rx_skb->data, RTL8150_MTU,
> +			  read_bulk_callback, dev);
> +
> +	res = usb_submit_urb(dev->rx_urb, GFP_KERNEL);
> +	if (res) {

Why did you break the -ENODEV situation?

How was this tested?

greg k-h