Re: [PATCH] can: usb: f81604: add Fintek F81604 support

[Michal Swiatkowski <michal.swiatkowski@linux.intel.com>]

On Mon, Mar 20, 2023 at 10:59:33AM +0800, Peter Hong wrote:
> Hi,
> 
> Michal Swiatkowski 於 2023/3/17 下午 09:15 寫道:
> > On Fri, Mar 17, 2023 at 05:33:52PM +0800, Ji-Ze Hong (Peter Hong) wrote:
> > 
> > --- a/drivers/net/can/usb/Kconfig
> > +++ b/drivers/net/can/usb/Kconfig
> > @@ -147,4 +147,13 @@ config CAN_UCAN
> >   	          from Theobroma Systems like the A31-ÂľQ7 and the RK3399-Q7
> >   	          (https://www.theobroma-systems.com/rk3399-q7)
> > Hi,
> > 
> > I am not familiar with CAN, so only style review :)
> 
> Thanks for your reviews :D
> > +
> > +	if (status) {
> > +		dev_err(&dev->dev, "%s: reg: %x data: %x failed: %d\n",
> > +			__func__, reg, data, status);
> > +	}
> > The { and } aren't needed as inside if is only one line.
> 
> Could I remove the { and } when the logical line to split multi-line ?
> 

Yes You can, and You should :)

> > > +static int f81604_set_normal_mode(struct net_device *netdev)
> > > +{
> > > +	struct f81604_port_priv *priv = netdev_priv(netdev);
> > > +	int status, i;
> > > +	u8 mod_reg_val = 0x00;
> > RCT, mod_reg should be one line above
> 
> What mean about "RCT"?
> 
> Is this section should change to above like ??
> 
>     u8 mod_reg_val;
>     ...
> 
>     mod_reg_val = 0;

reverse christmas tree, it is about how variable definition should look
like. In Your case:

struct f81604_port_priv *priv = netdev_priv(netdev);
u8 mod_reg_val = 0x00;
int status, i;

instead of


struct f81604_port_priv *priv = netdev_priv(netdev);
int status, i;
u8 mod_reg_val = 0x00;

Cosmetic Linux style rule

> > > +static int f81604_register_urbs(struct net_device *netdev)
> > > +{
> > > +	struct f81604_port_priv *priv = netdev_priv(netdev);
> > > +	int status, i;
> > > +
> > > +	for (i = 0; i < F81604_MAX_RX_URBS; ++i) {
> > > +		status = usb_submit_urb(priv->read_urb[i], GFP_KERNEL);
> > > +		if (status) {
> > > +			netdev_warn(netdev, "%s: submit rx urb failed: %d\n",
> > > +				    __func__, status);
> > > +			return status;
> > Don't know usb subsytem, but shouldn't previously submitted urb be
> > killed?
> 
> Yes, I had made kill operations in
>     f81604_start()
>         -> f81604_unregister_urbs()
>

Ok, thanks

> > > +static void f81604_process_rx_packet(struct urb *urb)
> > > +{
> > > +	struct net_device_stats *stats;
> > > +	struct net_device *netdev;
> > > +	struct can_frame *cf;
> > > +	struct sk_buff *skb;
> > > +	u8 *data;
> > > +	u8 *ptr;
> > > +	int i;
> > > +	int count;
> > RCT
> > 
> > > +
> > > +	netdev = urb->context;
> > > +	stats = &netdev->stats;
> > > +	data = urb->transfer_buffer;
> > netdev and data can be set in declaration
> 
> why only netdev & data ?? Could I set netdev, stats & data in declaration ?
> 

You can, but it will be hard to still have declaration as RCT (netdev
declaration have to be before stats declaration).

> 
> > > +/* Called by the usb core when driver is unloaded or device is removed */
> > > +static void f81604_disconnect(struct usb_interface *intf)
> > > +{
> > > +	struct f81604_priv *priv = usb_get_intfdata(intf);
> > > +	int i;
> > > +
> > > +	for (i = 0; i < F81604_MAX_DEV; ++i) {
> > > +		if (!priv->netdev[i])
> > > +			continue;
> > > +
> > > +		unregister_netdev(priv->netdev[i]);
> > > +		free_candev(priv->netdev[i]);
> > > +	}
> > What about closing USB device? It is called brefore disconnect or it
> > should be done here?
> 
> When candev close in f81604_close(), It will call f81604_set_reset_mode() to
> make candev to reset mode.
> 

Understand, thanks

> Thanks