omap3 otg host not working since 2.6.28?

omap3 otg host not working since 2.6.28?

[Brad Parker]

I can't get OTG host mode to work with any recent kernels.  I'm curious 
if anyone else has tried this.
It may well be something most people don't do.

I've tried this on our custom hardware (omap3530/tps65930) and on a 
beagleboard with the same
results.  Our OTG hardware is identical to the beagleboard.

I am using a special mini-B cable which has the ID pin grounded.  I know 
about forcing the
mode with echo "host" >/sys/devices/platform/musb_hdrc/mode. 

When I use a 2.6.28 kernel built using the same patches as are in OE, 
OTG host works fine.  I can get the controller
into host mode.  When I use any kernels from the omap git tree past 
2.6.28 it doesn't work.  The controller
stays in "b-mode" and won't switch.

I can dig into it and figure out why, but I thought I'd ask in case 
anyone else has already solved it.  I've written
some diagnostics for the OTG controller and found it's finicky about 
getting into host mode, sometimes
wanting the phy to be reset before it will switch.

-brad

 
-----------
Brad Parker
Heeltoe Consulting
http://www.heeltoe.com
781-483-3101

Re: omap3 otg host not working since 2.6.28? (sys_boot affects this)

[Brad Parker]

Brad Parker wrote:
> I can't get OTG host mode to work with any recent kernels.  I'm 
> curious if anyone else has tried this.
> It may well be something most people don't do.
I know it's poor form to respond to you own posts, but I thought I'd 
add/ask the following:

I've found that the sys_boot settings can affect whether OTG host mode 
works or does not work (when it works :-)

Has anyone else seen this? 

OMAP3530; the beagleboard uses 01111 for sys_boot[4:0];  If you set them 
to say,  00110,
USB OTG host won't work, even in older kernels which normally work.

01111 is  nand,usb,uart3,mmc
00110 is mmc,usb

So in the second case it looks like the ROM has not touched the USB 
hardware when it
boots from from mmc.  In the first case it has set up the USB OTG hardware.

It seems that the OMAP ROM is doing something in the way of hardware 
initialization which
the linux driver is depending on...

-brad