Re: [PATCH] usb: gadget: lpc32xx_udc: Port to new start/stop interface

[Sebastian Andrzej Siewior <bigeasy@linutronix.de>]

On 08/17/2012 01:32 PM, Roland Stigge wrote:
> How about the following: Below, I show how the initialization of the
> current controller is done (statically). Removing this struct
> initialization completely would make the code much uglier, introducing
> many individual assignments.

I said what I had to say, I leave it up to Felipe.

> static struct lpc32xx_udc controller = {
>          .gadget = {
>                  .ops    =&lpc32xx_udc_ops,
>                  .ep0    =&controller.ep[0].ep,
>                  .name   = driver_name,
>                  .dev    = {
>                          .init_name = "gadget",
>                          .release = nop_release,
>                  }
>          },
>          .ep[0] = {
>                  .ep = {
>                          .name   = "ep0",
>                          .ops    =&lpc32xx_ep_ops,
>                  },
>                  .udc            =&controller,
>                  .maxpacket      = 64,
>                  .hwep_num_base  = 0,
>                  .hwep_num       = 0, /* Can be 0 or 1, has special
> handling */

Can it be 0 or 1 or has it to be 0? Who would change and why?

>                  .lep            = 0,
>                  .eptype         = EP_CTL_TYPE,
>          },
>          .ep[1] = {
>                  .ep = {
>                          .name   = "ep1-int",
>                          .ops    =&lpc32xx_ep_ops,
>                  },
>                  .udc            =&controller,
>                  .maxpacket      = 64,
>                  .hwep_num_base  = 2,
>                  .hwep_num       = 0, /* 2 or 3, will be set later */

Why not now?

>                  .lep            = 1,
>                  .eptype         = EP_INT_TYPE,

Do you have any restrictions on these endpoints? I mean can you pick
any endpoints for BULK/INT/ISO or have they been pre-defined by HW?

Sebastian