Mainline U-Boot, EHCI, usbkbd not working (workaround)

[Lars Doelle]

Hi everyone,

while testing with mainline u-boot, I came over the
problem, that the USB keyboard is not recognized.

The device is an A20-OLinuXIno-LIME2. I used the
current A20-OLinuXino-Lime2_defconfig for building.

In my understanding, the issue should be reproducible
with all devices having an EHCI root hub.

---
sun7i# usb reset
(Re)start USB...
USB0: USB EHCI 1.00
scanning bus 0 for devices... cannot reset port 1!?
1 USB Device(s) found
USB1: USB EHCI 1.00
scanning bus 1 for devices... 1 USB Device(s) found
scanning usb for storage devices... 0 Storage Device(s) found
sun7i# usb tree
USB device tree:
1 Hub (480 Mb/s, 0mA)
u-boot EHCI Host Controller

2 Hub (480 Mb/s, 0mA)
u-boot EHCI Host Controller
---

As a workaround, i plugged an USB hub in between:

---
sun7i# usb reset
(Re)start USB...
USB0: USB EHCI 1.00
scanning bus 0 for devices... 4 USB Device(s) found
USB1: USB EHCI 1.00
scanning bus 1 for devices... 1 USB Device(s) found
scanning usb for storage devices... 0 Storage Device(s) found
sun7i# usb tree
USB device tree:
1 Hub (480 Mb/s, 0mA)
| u-boot EHCI Host Controller
|
+-2 Hub (480 Mb/s, 100mA)
| USB 2.0 Hub [MTT]
|
+-3 Hub (480 Mb/s, 100mA)
| USB 2.0 Hub [MTT]
|
+-4 Human Interface (1.5 Mb/s, 100mA)

5 Hub (480 Mb/s, 0mA)
u-boot EHCI Host Controller

sun7i#
USB device tree:
1 Hub (480 Mb/s, 0mA)
| u-boot EHCI Host Controller
|
+-2 Hub (480 Mb/s, 100mA)
| USB 2.0 Hub [MTT]
|
+-3 Hub (480 Mb/s, 100mA)
| USB 2.0 Hub [MTT]
|
+-4 Human Interface (1.5 Mb/s, 100mA)

5 Hub (480 Mb/s, 0mA)
u-boot EHCI Host Controller
---

Kind regards,

Lars

[Toroshin Dmitry]

The same on Wexler tab 7200: usb keyboard works only with exterenal hub.

[B.R. Oake]

(This started on the linux-sunxi mailing list but will hopefully be
of interest on the U-Boot list)

On 04/01/15 13:45, Lars Doelle wrote:
> while testing with mainline u-boot, I came over the
> problem, that the USB keyboard is not recognized.
>
> The device is an A20-OLinuXIno-LIME2. I used the
> current A20-OLinuXino-Lime2_defconfig for building.
>
> In my understanding, the issue should be reproducible
> with all devices having an EHCI root hub.
>
> ---
> sun7i# usb reset
> (Re)start USB...
> USB0: USB EHCI 1.00
> scanning bus 0 for devices... cannot reset port 1!?
> 1 USB Device(s) found
> USB1: USB EHCI 1.00
> scanning bus 1 for devices... 1 USB Device(s) found
> scanning usb for storage devices... 0 Storage Device(s) found
> sun7i# usb tree
> USB device tree:
> 1 Hub (480 Mb/s, 0mA)
> u-boot EHCI Host Controller
>
> 2 Hub (480 Mb/s, 0mA)
> u-boot EHCI Host Controller
> ---
>
> As a workaround, i plugged an USB hub in between:

> [...]


I also have this problem. I've tried three different USB keyboards on
an A20-Olinuxino-Micro and a Banana Pi, and I always get that error
"cannot reset port N!?" where N is whichever USB socket I've plugged
it into, and U-Boot cannot see the keyboard. Once Linux has loaded,
the keyboard works without any trouble.

Can anyone suggest what is causing this?

Thank you,
B.R. Oake.

[Hans de Goede]

Hi,

The problem is that u-boot does not allow building both ohci and
ehci drivers into the same u-boot binary, so we cannot enable both
usb-1 and usb-2 support at the same time.

So we're stuck with having only usb-2 support until someone reworks
u-boot's usb code, and keyboards and mice are typically usb-1 devices,
the workaround for this is to plug in a usb-2 hub so that the board
sees a usb-2 device, and then plug the mouse / keyboard into that
hub.

Regards,

Hans





>
> Thank you,
> B.R. Oake.
>

[B.R. Oake]

On 04/01/15 19:05, Hans de Goede wrote:
> The problem is that u-boot does not allow building both ohci and
> ehci drivers into the same u-boot binary, so we cannot enable both
> usb-1 and usb-2 support at the same time.
>
> So we're stuck with having only usb-2 support until someone reworks
> u-boot's usb code, and keyboards and mice are typically usb-1 devices,
> the workaround for this is to plug in a usb-2 hub so that the board
> sees a usb-2 device, and then plug the mouse / keyboard into that
> hub.

Thanks for the explanation Hans.

At U-Boot time, the only USB device I would need is the keyboard, and
I would rather avoid adding a hub if I don't otherwise need one, so
perhaps I can build U-Boot with the OHCI driver enabled instead of EHCI?
I tried removing CONFIG_USB_EHCI and adding:

CONFIG_USB_OHCI_NEW
CONFIG_SYS_USB_OHCI_BOARD_INIT
CONFIG_SYS_USB_OHCI_CPU_INIT
CONFIG_SYS_USB_OHCI_REGS_BASE=0x01c14400
CONFIG_SYS_USB_OHCI_SLOT_NAME="sunxi_slot_0"
CONFIG_SYS_USB_OHCI_MAX_ROOT_PORTS=1

which built and ran, but gave the error "unknown command usb". Can
this approach be made to work?

Cheers,
B.R.

[Simon Glass]

Hi Hans,

This could be solved by moving USB to driver model.

Marek do you know if anyone is looking at this?

Regards,
Simon

[Marek Vasut]

On Monday, January 05, 2015 at 03:49:32 AM, Simon Glass wrote:
> Hi Hans,

Hi all,

[...]

> > The problem is that u-boot does not allow building both ohci and
> > ehci drivers into the same u-boot binary, so we cannot enable both
> > usb-1 and usb-2 support at the same time.
> >
> > So we're stuck with having only usb-2 support until someone reworks
> > u-boot's usb code, and keyboards and mice are typically usb-1 devices,
> > the workaround for this is to plug in a usb-2 hub so that the board
> > sees a usb-2 device, and then plug the mouse / keyboard into that
> > hub.
>
> This could be solved by moving USB to driver model.
>
> Marek do you know if anyone is looking at this?

To my knowledge, noone at this point. I cannot tell when I will be able
to dive into this due to the university stuff, sorry about that :(

Best regards,
Marek Vasut

[Hans de Goede]

Hi,

Yes, that is what I was thinking too.

> Marek do you know if anyone is looking at this?

Actually I was about to ask you (Simon) if you are looking into converting
the usb stuff to dm, AFAIK no one else is working on this.

Regards,

Hans

[Simon Glass]

Hi Hans,

I think Marek is busy for a while.

I'd like to get PCI over the line first, and I have a few other things
on my plate.

But I think it would not be too difficult to move the
usb_lowlevel_init() stuff into driver model. That would solve the
primary problem I think. I'll see how things look later in the month.

Regards,
Simon