On 2008-03-07 20:10 +0000, Chris Jones wrote:
> This afternoon I've employed the pxaregs tool provided by Wookey to 
> investigate what's going on with USB port 2. It's a little mystifying, 
> but we're getting further.
> 
> Firstly, I had to rummage around for the source to pxaregs so I could 
> add support for the host controller registers. The one I found was at
> http://www.mn-logistik.de/unsupported/pxa250/
> but didn't include support for the device controller registers, which 
> Wookey's one did, so I had to use that as well. Not to worry.

The source (which indeed came from the above spot plus patches from
OE) is in balloon svn:
http://balloonboard.org/cgi-bin/viewcvs.cgi/balloon/trunk/bbl/pxaregs/

If you've added some more regs then put them in there.

All the registers are now in:
http://balloonboard.org/cgi-bin/viewcvs.cgi/balloon/branches/kernel/2.6.22.2/pxa-registers.patch?rev=444&view=log
but that module doesn't actually work beyond listing all the registers
due to the inode it looks up for procfs accesses not matching the one
saved when the entry was created. Fixing that got interrupted by
minipug/samosa support, so pxaregs is the best tool right now (and a
more versatile tool anyway).

> The host controller configuration looks sane and appears to be 
> symmetrical between all three ports on the root hub. I found a problem 
> in that the root hub believed that ports 2 and 3 were in an overcurrent 
> condition because their overcurrent detect pins aren't connected to 
> anything (we've wired the port 2 overcurrent detect pin to a different 
> GPIO). I fixed that by setting a bit in UHCRHDA which tells the hub that 
> overcurrent detection isn't supported. My hopes were high that 
> everything would spring into life, but it didn't.

I can put this in the kernel for now, but presumably this isn';t the
correct fix. Anyone know what is?

> I'm still attempting to use the second differential transceiver 
> (normally used for USB client) and have followed the instructions for 
> remapping it to be USB host 2. At some point I'll have another go at 
> using the ISP1301 transceiver now that I've found this overcurrent 
> detection problem.
> 
> That's all for now - I'm away next week but may do a little more hacking 
> on this this evening.

OK. Cheers for that. I'll finish off the regsiter-prodding support for
next week.

Wookey
-- 
Principal hats:  Balloonz - Toby Churchill - Aleph One - Debian
http://wookware.org/