[U-Boot] twister: usb host support

[Stefano Babic]

Hi Jeroen,


On 09/07/2013 21:46, Jeroen Hofstee wrote:
> Hello Stefano,
> 
> The twister always had a fragile usb support in u-boot
> in the past. Since some floating patches mentioned
> alternating success after usb hub resets, I had a look
> if these patches fixed the twister as well, they don't.
> 
> Some fiddling around [1] makes the usb work however,
> by not attempting to reset the USB hub with gpio_25.
> I cannot find any document mentioning gpio_25 though.
> Any reason for this usage?

That is right. There is no written documentation by Technexion about
this pin. I have added in CC Tapani, maybe he can tell us something more.
The GPIO is used internally on the TAM3517 SOM, and we have no
schematics about it. We have to guess about its usage.

The reason to have it is that the sources for the kernel provided by
Technexion (an ancient 2.6.32 Version) set this pin to reset the EHCI.
As I said, there is no further documentation and rather we can only make
some supposition. The usage in U-Boot then reflects the usage made by
Technexion.


> p.s.
> I found by sheer accidence that the usb is behaving
> completely normal (after the patch) on rev B1. It is
> broken on rev B without the patch and working buggy
> with it. On rev B1 + patch the USB to SATA converter
> is discovered as well, which I have never seen before.
> 

I can confirm issues with version B of the TAM3517-SOM. I am testing a
B1 SOM with a B twister. I do not know the details about changelog
between B and B1 version of the SOM, but I got both USB and Ethernet
problems with TAM3517-B. The same image runs without the same problems
on a B1.

There is also this pending patch by Michael:

http://patchwork.ozlabs.org/patch/250290/

It is applied to u-boot-ti, not yet to mainline.

I have tested current mainline with Mihael's patch applied, and the
recognition of the USB storage seens reliable to me. No case with "0
device found".


> 
> === [1] patch to disable the usb reset
> 
> The change to port_mode[1] is not needed, but since it is unused,
> MODE_UNUSED seems appropriate.

Agree - as we do not use it, we can simply disable.

> diff --git a/board/technexion/twister/twister.c
> b/board/technexion/twister/twister.c
> index a28c704..31cf6c0 100644
> --- a/board/technexion/twister/twister.c
> +++ b/board/technexion/twister/twister.c
> @@ -63,7 +63,7 @@ static const u32 gpmc_XR16L2751[] = {
>  #ifdef CONFIG_USB_EHCI
>  static struct omap_usbhs_board_data usbhs_bdata = {
>      .port_mode[0] = OMAP_EHCI_PORT_MODE_PHY,
> -    .port_mode[1] = OMAP_EHCI_PORT_MODE_PHY,
> +    .port_mode[1] = OMAP_USBHS_PORT_MODE_UNUSED,
>      .port_mode[2] = OMAP_USBHS_PORT_MODE_UNUSED,
>  };
> 
> @@ -92,9 +92,6 @@ int board_init(void)
>      enable_gpmc_cs_config(gpmc_XR16L2751, &gpmc_cfg->cs[3],
>          XR16L2751_UART2_BASE, GPMC_SIZE_16M);
> 
> -    gpio_request(CONFIG_OMAP_EHCI_PHY1_RESET_GPIO, "USB_PHY1_RESET");
> -    gpio_direction_output(CONFIG_OMAP_EHCI_PHY1_RESET_GPIO, 1);
> -

See my concerns above. Do not reset the hub in the kernel ?

Best regards,
Stefano

-- 
=====================================================================
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-53 Fax: +49-8142-66989-80 Email: sbabic at denx.de
=====================================================================