Re: [SeaBIOS] questions about seabios ohci code

[Kevin O'Connor]

I'm CC'ing the mailing list.

On Wed, Apr 14, 2010 at 02:04:48AM +0800, Qing Pei Wang wrote:
> hi kevin,
>    i am learning the code of seabios/ohci stack followed with OHCI spec. I
> have some question.

Note that the OHCI support has only been tested on a modified qemu - I
haven't had any reports on real hardware.
> in the function alloc_intr_pipe (usb.c line118)what does the frameexp
>
> mean?

It's log2(epdesc->bInterval) - the log of the sampling time (in ms)
requested by the device.
> how does ohci get the intervals. the code just use epdesc->bInterval, how
>
> does it init? can pass the parameter from the usb-hid.c?

The epdesc structure is populated by get_device_config() - the config
read from the device includes the device config, interface
descriptions, and end point descriptions.

-Kevin

questions about seabios ohci code

[Qing Pei Wang]

Kevin,
   i had tested the OHCI keyboard, the works but not pretty good as enough.
the reaction time is too long.
So i am trying to learn more about this things.
BTW, seabios/OHCI use count=DIV_ROUND_UP(PIT_TICK_INTERVAL * 1000 * 2,
PIT_TICK_RATE * ms)+1
as the td count. i can not find any document  about that. Where can i find
or some place i missing with OHCI spec?
On Wed, Apr 14, 2010 at 9:22 AM, Kevin O'Connor  wrote:
> I'm CC'ing the mailing list.
>
> On Wed, Apr 14, 2010 at 02:04:48AM +0800, Qing Pei Wang wrote:
> > hi kevin,
> >    i am learning the code of seabios/ohci stack followed with OHCI spec.
>
> I
> > have some question.
>
> Note that the OHCI support has only been tested on a modified qemu - I
> haven't had any reports on real hardware.
> > in the function alloc_intr_pipe (usb.c line118)what does the frameexp
> >
> > mean?
>
> It's log2(epdesc->bInterval) - the log of the sampling time (in ms)
> requested by the device.
> > how does ohci get the intervals. the code just use epdesc->bInterval,
>
> how
> > does it init? can pass the parameter from the usb-hid.c?
>
> The epdesc structure is populated by get_device_config() - the config
> read from the device includes the device config, interface
> descriptions, and end point descriptions.
>
> -Kevin

questions about seabios ohci code

[Kevin O'Connor]

On Thu, Apr 15, 2010 at 10:40:21AM +0800, Qing Pei Wang wrote:
> Kevin,
>    i had tested the OHCI keyboard, the works

Great!  Please send the full SeaBIOS log.
> but not pretty good as enough.
> the reaction time is too long.

Is the time it takes a key to be shown too long?  Or are keys being
lost when typing fast?
> So i am trying to learn more about this things.
> BTW, seabios/OHCI use count=DIV_ROUND_UP(PIT_TICK_INTERVAL * 1000 * 2,
> PIT_TICK_RATE * ms)+1
> as the td count. i can not find any document  about that. Where can i find
> or some place i missing with OHCI spec?

That's internal to SeaBIOS.  SeaBIOS doesn't register a hardware irq
for usb, instead it polls for USB activity on every timer interrupt
(~55ms).  The above calculation determines how many key events to
queue so that no keyboard drops occur.  Usually a keyboard requests
service every 10ms, causing the above calculation will return 11 queue
items.

-Kevin