Hi,
This is the OMAPL138 eXperimenter Kit. I have changed the OS running on the kit from Linux to VxWorks.
And I am developing an USB driver (peripheral only) on it. Of course I made reference to the Linux implementation and the L138 USB 2.0 user guide during development.
I have done the development, and passed the Chapter-9 Tests with the Command Verifier (from usb.org), and the device is also able to enumerate successfully in high speed mode.
8863.Chapter 9 Tests 2011-01-28.pdf
But, until recently I have found that although enumeration (high speed mode) is successful, the device actually never receive standard request to get the string descriptor.
Using software analyzer (USBTrace), I can see the host sending out request to get the string descriptor, but never got it. Other standard requests are working fine. This is the strange part.
If I disable the high speed, and let the device enumerates in full speed, everything is fine, which means the device is able to receive the request to get the string descriptor. This is the distinct behaviour I see between high speed and full speed.
There is another problem I am facing using OMAPL138 USB 2.0. I just cannot use the hardware anayzer from Ellisys to look at the USB bus traffic to find out what is actually going on (I have reported this in another post). It is kind of painful without this capability.
Hopefully someone out there can help me out or shine some light on this.
rgds,
kc Wong
Wong
Do you say that the phy configuration done in linux is correct? If so why in linux you have observed the issue.
I did not understand the fix. Do you say phy configuration is not ok in linux as well. Then how the uImage built by you worked on EVM.
I did not understand completely. could you explain, what is the root cause where it got fixed.
Regards
Ravi B
If my reply answers your question then please click on the green button "Verify Answer"
small correction
If phy configuration is wrong in linux, then how the uImage sent by you worked on my EVM.
Hi Ravi,
I took TI UBL and added PHY configuration. Only today, when I tried to experiment something else, I put back the TI UBL and noticed this difference. This is the very first code for the USB driver development, never suspect that this is the cause. Partly also because the code is from SPRUFM9H.
TI UBL + Linux (OK)
Modified UBL + Linux (not OK)
You can take a look at this PHY configuration code which is in the 1st page of this post (05-06-2011).
Worst thing is full speed works fine with this configuration, and high speed also except for GetDescriptor(string) requests. And all the testing and experiment seem to tell me the HW has problem.
Not sure you have detailed explaination on this. If yes, hopefully you can share.
I understood now, you have taken TI-UBL and modified the phy configuration. This modification has leads to this problem and it worked for FS and for HS the string descriptor has caused the problem. This explains the root cause of the problem. Thanks.
Ravi,
Just to clarify, I did not modify the phy configuration, there was no usb phy configuration in the original UBL.
I actaully took the sample code for phy configuration from SPRUFM9H, and added into the UBL.
It turns out to be I should not have followed the configuration shown in SPRUFM9H. The sample configuration shown in SPRUFM9H has caused the GetDescriptor(string) request to fail on eXperimenter Kit.
You can compare the code in SPRUFM9H and the code in the 1st page of this post.
I don't know which one is actually causing the problem, either CFGCHIP2.USB0OTGMODE or CFGCHIP2.USB0PHYCLKMUX .
My mistake is I should not have just copied the sample code without doubt. And I did post the code and ask for consultation, and nobody pointed out the error.
Anyway, I still very much appreciate your support.