USB on 2.6.29 kernel

[Peter van Loben Sels]

I'm pretty new at this, so thanks for any help.

I have a Rev6 BB using a wall wart for power, a powered USB hub and USB
keyboard, mouse, and network dongle.

I've used the guides on the wiki to get started. My first step was to
setup and partition an SD card and install the Angstrom Demo image on
it. I powered up the BB, and everything including USB worked just fine.

After tinkering with that for awhile, I then installed openembedded (on
Ubuntu Hardy) and built the OE console-image. I installed that on the SD
card and everything again worked just fine, including USB.

Feeling pretty confident now, I checked out the most recent kernel I
could find:

git://git2.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6

and used the OE tool chain to build this kernel, starting with "make
CROSS_COMPILE=arm-angstrom-linux-gnueabi- ARCH=arm
omap3_beagle_defconfig". I installed the result as uImage on the SD
card, copied the modules to /lib/modules, booted, and everything worked
just fine, except I had no USB. None.

Ok, too big of a step. I then built what appears to me to be latest
kernel included in OE: "bitbake linux-omap_2.6.29.bb". This downloaded
the same git tree, added some patches, and compiled just fine; I
installed uImage and the modules as before on the SD card and booted it
successfully. But, still no USB.

Dmesg for two kernels that would not bring up USB both had this error:

"
musb_hdrc: version 6.0, musb-dma, otg (peripheral+host), debug=0
musb_hdrc: ConfigData=0x33 (UTMI-16, HB-ISO Rx (X), SoftConn)
musb_hdrc: MHDRC RTL version 1.400
musb_core_init 1406: reconfigure software for static FIFOs
musb_hdrc musb_hdrc: musb_init_controller failed with status -19
"

But I do not know if it relates to my problem; my build of console-image
and the Angstrom demo did not have this error.

I'm clearly missing or mis-using something. I've tried to compare the
.config files of the various kernels to see if it is a configuration
issue, but there doesn't seem to be anything obvious (to me). It feels
like I'm perhaps missing some patches, but the linux-omap_2.6.29.bb
install included patches, and I sort of thought that the latest git
master would include all necessary patches as of the latest pull.
Googleing the error above hasn't brought up anything that seems useful.

Any hints as to what I am doing wrong?

Thanks,

Peter

[Koen Kooi]

Op 3 apr 2009, om 00:50 heeft Peter van Loben Sels het volgende
geschreven:

You're doing nothing wrong, you're just in a bit of uncharted
territory :) The 2.6.29 kernel is quite new and we don't have
everything working with it yet, that's why it isn't the default yet.

regards,

Koen

[Frans Meulenbroeks]

Please do not open two threads for the same issue.
http://groups.google.com/group/beagleboard/browse_thread/thread/f54ed4e92f762292/fab89c8b96c039ac#fab89c8b96c039ac
Looking at the text I am quite confident that although they are from
different accounts, they are from the same sender.

FM

[ptm5851]

FM

You are correct; I sent both. Somehow the first post I sent on 4/2 did
not show up on the list, so I tried again on 4/4. My second attempt
posted almost immediately; the first apparently just arrived today
(4/6). Go figure. Guess I was impatient ... please consider my hands
slapped, and my apologies. Won't happen again.

Peter

On Apr 6, 12:57 am, Frans Meulenbroeks <fransmeulenbro...@gmail.com>
wrote:
> Please do not open two threads for the same issue.http://groups.google.com/group/beagleboard/browse_thread/thread/f54ed...

[Frans Meulenbroeks]

On Apr 7, 1:30 am, ptm5851 <ptm5...@blackfoot.net> wrote:
> FM
>
> You are correct; I sent both. Somehow the first post I sent on 4/2 did
> not show up on the list, so I tried again on 4/4. My second attempt
> posted almost immediately; the first apparently just arrived today
> (4/6). Go figure. Guess I was impatient ... please consider my hands
> slapped, and my apologies. Won't happen again.

I didn't want to slap, just to educate. Seeing that the first post got
delayed it is quite understandable.
Now wrt usb: yesterday I tried 2.6.29 r21 (the latest version on the
feed) and that one has OTG host working (assuming you have a proper
cable).
Didn't check the logs, but I guess kudo's go toe Koen.

Frans.

[ptm5851]

Thanks for the heads-up. I did a quick bitbake -b recipes/linux/linux-
omap_2.6.29.bb after a fresh OE git pull; result was still no usb (otg
host), although there seem to be many new patches. Same musb_hdrc
error as before. When you say "latest version on the feed" ... what
feed do you mean?

Peter

On Apr 7, 2:52 am, Frans Meulenbroeks <fransmeulenbro...@gmail.com>
wrote:

[Frans Meulenbroeks]

Peter,

with latest version I indeed mean the latest version on the oe tree.
If you did a pull followed with the bitbake command you should have
build kernel r24 or later, which has all those patches.
That version works for me.

Note: initially I also thought it was not working. Then I used a
different otg cable and things worked.
I haven't really exercised the problem but I guess this was because
maybe the first cable did not have pins 4 and 5 shortened.
The fact that it worked before is maybe because at some point someone
gave me a kernel which used OTG in host only mode.
Not fully sure on how to achieve that (guess it is in .config
somewhere).

You might give that a try too.

Frans.