Re: [Linux-kernel-mentees] [PATCH 05/15] usb: serial: cypress_m8: use usb_control_msg_recv() and usb_control_msg_send()

[Johan Hovold <johan@kernel.org>]

On Wed, Nov 04, 2020 at 12:16:53PM +0530, Himadri Pandya wrote:
> The new usb_control_msg_recv() and usb_control_msg_send() nicely wraps
> usb_control_msg() with proper error check. Hence use the wrappers
> instead of calling usb_control_msg() directly.
> 
> Signed-off-by: Himadri Pandya <himadrispandya@gmail.com>
> ---
>  drivers/usb/serial/cypress_m8.c | 38 +++++++++++++++++----------------
>  1 file changed, 20 insertions(+), 18 deletions(-)
> 
> diff --git a/drivers/usb/serial/cypress_m8.c b/drivers/usb/serial/cypress_m8.c
> index cc028601c388..4d66cab8eece 100644
> --- a/drivers/usb/serial/cypress_m8.c
> +++ b/drivers/usb/serial/cypress_m8.c
> @@ -341,20 +341,21 @@ static int cypress_serial_control(struct tty_struct *tty,
>  			feature_buffer[4]);
>  
>  		do {
> -			retval = usb_control_msg(port->serial->dev,
> -					usb_sndctrlpipe(port->serial->dev, 0),
> -					HID_REQ_SET_REPORT,
> -					USB_DIR_OUT | USB_RECIP_INTERFACE | USB_TYPE_CLASS,
> -					0x0300, 0, feature_buffer,
> -					feature_len, 500);
> +			retval = usb_control_msg_send(port->serial->dev, 0,
> +						      HID_REQ_SET_REPORT,
> +						      USB_DIR_OUT |
> +						      USB_RECIP_INTERFACE |
> +						      USB_TYPE_CLASS, 0x0300,
> +						      0, feature_buffer,
> +						      feature_len, 500,
> +						      GFP_KERNEL);
>  
>  			if (tries++ >= 3)
>  				break;
>  
> -		} while (retval != feature_len &&
> -			 retval != -ENODEV);
> +		} while (retval != -ENODEV);

Here you actually break the logic; after this change the driver will
always retry as the test for success is removed.

>  
> -		if (retval != feature_len) {
> +		if (retval) {
>  			dev_err(dev, "%s - failed sending serial line settings - %d\n",
>  				__func__, retval);
>  			cypress_set_dead(port);
> @@ -379,19 +380,20 @@ static int cypress_serial_control(struct tty_struct *tty,
>  		}
>  		dev_dbg(dev, "%s - retrieving serial line settings\n", __func__);
>  		do {
> -			retval = usb_control_msg(port->serial->dev,
> -					usb_rcvctrlpipe(port->serial->dev, 0),
> -					HID_REQ_GET_REPORT,
> -					USB_DIR_IN | USB_RECIP_INTERFACE | USB_TYPE_CLASS,
> -					0x0300, 0, feature_buffer,
> -					feature_len, 500);
> +			retval = usb_control_msg_recv(port->serial->dev, 0,
> +						      HID_REQ_GET_REPORT,
> +						      USB_DIR_IN |
> +						      USB_RECIP_INTERFACE |
> +						      USB_TYPE_CLASS, 0x0300,
> +						      0, feature_buffer,
> +						      feature_len, 500,
> +						      GFP_KERNEL);
>  
>  			if (tries++ >= 3)
>  				break;
> -		} while (retval != feature_len
> -						&& retval != -ENODEV);
> +		} while (retval != -ENODEV);

Same here.

>  
> -		if (retval != feature_len) {
> +		if (retval) {
>  			dev_err(dev, "%s - failed to retrieve serial line settings - %d\n",
>  				__func__, retval);
>  			cypress_set_dead(port);

As the driver is already using a DMA-able buffer and is reusing it for
every retry as well, this change is now introducing redundant memdup +
memcpy for each call.

I suggest just dropping this one.

Johan
_______________________________________________
Linux-kernel-mentees mailing list
Linux-kernel-mentees@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees