Blatant simple bug in Cavium SDK USB code?

[Andrew Duane]

I have USB hardware, and some familiarity with at least the bottom half of the code. I know nothing of the scheduling, perdiodic stuff, the USB protocol itself. But I can probably look at the hardware and driver aspects of it.  My hardware is very simple, a direct connect between the 5650 and a single USB plug on the chassis. There's a similar platform I have someone else working on that has a 4-port hub too, to look at "some day".

We got a driver from a 3rd party company called Synopsis I think, I doubt there's any way we can use it, but I should be able to get the bottom half going to at least recognize a stick.

--
Andrew Duane             Juniper Networks
978-589-0551              10 Technology Park Dr
aduane at juniper.net      Westford, MA  01886-3418

________________________________________
From: juli at clockworksquid.com [juli at clockworksquid.com] On Behalf Of Juli Mallett [jmallett at FreeBSD.org]
Sent: Friday, April 01, 2011 6:18 PM
To: bhanuprakash bodireddy
Cc: Andrew Duane; mips at freebsd.org
Subject: Re: Blatant simple bug in Cavium SDK USB code?

Following up to Andrew,

On Fri, Apr 1, 2011 at 14:23, bhanuprakash bodireddy
<bhanuprakashbodireddy at gmail.com> wrote:
> On Fri, Apr 1, 2011 at 1:03 PM, Andrew Duane <aduane at juniper.net> wrote:
>>
>> This isn't even close to the only problem I'm having. I don't make it out
>> of cvmx_usb_initialize at all. One of two things happens:
>
> We are yet to add the USB support to mainline. I have sent USB support
> patches to Juli and she had reviewed them earlier.
> But some of the routines are yet to be rewritten(due to other issues) and
> hence the support is still missing.

The outstanding patches are for Octeon 2 only, so if you're using
CN5XXX you'll have to wait for someone to fix up the SDK-based USB
driver.  HPS did a great job on it initially, even working without
hardware, and we managed to get it mostly working, but there were/are
remaining issues that somebody with interest, time and hardware will
need to solve.  My USB need has disappeared and I've been short on
time, so it hasn't been a priority for me.  (I can't speak to CN3XXX
as I don't have any usable CN3XXX hardware presently and it doesn't
seem worth investing as much effort in.)

Oleksandr has Octeon hardware with USB, and there's a system with USB
in the Netperf cluster (although obviously that isn't really
convenient for USB testing, but can be used for basic controller work)
and Alexander Kabaev has expressed an interest in USB in the past.
There's a D-Link Ukraine developer who has done some work with the
DWC_OTG driver from Linux for Octeon and some other boards with
similar hardware.  Adrian Chadd should be getting Octeon hardware in
the near future, including USB.  So there's enough people with the
right skills and hardware out there, it's just a matter of someone
having enough time and interest to work on it.

Juli.