--- On Fri, 5/28/10, Sergei Shtylyov <[email protected]> wrote:
>
> > Would anyone like to comment on the status of MUSB OTG
> (for DM355) in the current git version?
>
> Haven't tried...
As of something like 18 months ago, two key mechanisms
mostly worked when I tested against (a) OMAP3, and
(b) the fullspeed OTG tester. The two mechanisms were
(i) cable based role switch: ID pin sensing, etc,
during link setup. (This was working before, but
I found a few bugs to fix.)
(ii) basic HNP behavior, although some of the timings
were out of spec, and some other minor non-compliance
situations existed too.
ISTR that SRP worked in some cases too, but that was rather
awkward to test with the current PM framework.
Haven't tried recently. My box that had the fullspeed
OTG tester died (sigh), so I'll not be able to retest that
any time soon.
Those code paths were kind of fragile, so it would not
surprise me to learn that someone patched something and
(accidentally) broke some of them.
> > And/or on the TI CPPI DMA option?
>
> Should basically work...
.. Within limits of its design flaws, ISTR it was more or
less behaving. Read the comments; "RX woes" explains
some of how the original design was USB-antagonistic,
TX was less so. Later CPPI (4.1) addressed at least
some of the issues. I never trusted the code to abort
transactions (TI never documented how to do that stuff
so that code was always iffy).
Since I last touched that code, I believe various folk
have not just worked on it, but have successfully made
it work better (yay!). Fixed corner cases and patched
over undocumented bhaviors, and so on.
> > I note that OTG depends on EXPERIMENTAL and the CPPI
> driver has a lot of discouraging comments and REVISITs
> throughout it.
>
> Well, the whole MUSB driver has a lot of
> REVISITs. :-)
Coming from odd hardware design artifacts, both from
Mentor (the "M" in MUSB) and from how various teams
at TI integrated that IP with DMA.
- Dave
> > Thanks,
>
> WBR, Sergei
>
> ______________________________________________
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
