gumstix

HI David,

On 7/16/07, David Kilp <david.kilp@...> wrote:
>
> Apologize in advance for the length of this but I think there is some good
> information here for those people trying to get USB client mode working on
> Verdex boards.

The real problem isn't so much the GPIOs as it is that the USB client
hardware internal to the PXA270 is different from the PXA255 and the
PXA270 version hasn't been maintained (i.e. doesn't work).

-- 
Dave Hylands
Vancouver, BC, Canada
http://www.DaveHylands.com/

On Jul 17, 2007, at 12:35 AM, Dave Hylands wrote:

> HI David,
>
> On 7/16/07, David Kilp <david.kilp@...> wrote:
>>
>> Apologize in advance for the length of this but I think there is  
>> some good
>> information here for those people trying to get USB client mode  
>> working on
>> Verdex boards.
>
> The real problem isn't so much the GPIOs as it is that the USB client
> hardware internal to the PXA270 is different from the PXA255 and the
> PXA270 version hasn't been maintained (i.e. doesn't work).

There's been a fairly extensive thread on this topic on the linux-arm- 
kernel mailing list very recently, which made it sound like work will  
be done to get the 270 client driver working; timeline for that was  
not super clear though:

<http://gumstix.com/~craig/270udc/>

C

Dave Hylands wrote:
> 
> On 7/16/07, David Kilp <david.kilp@...> wrote:
>>
>> Apologize in advance for the length of this but I think there is some
>> good
>> information here for those people trying to get USB client mode working
>> on
>> Verdex boards.
> 
> The real problem isn't so much the GPIOs as it is that the USB client
> hardware internal to the PXA270 is different from the PXA255 and the
> PXA270 version hasn't been maintained (i.e. doesn't work).
> 
> -- 
> Dave Hylands
> Vancouver, BC, Canada
> http://www.DaveHylands.com/
> 
> 

Actually this might be useful information in debugging. Since I don't have
an older gumstix board to test, can I assume that the actual USB gadget code
(ether.c, file_storage.c, etc.) works for the older gumstix boards
(PXA255-based ones)?  Therefore the problem is related to the USB device
controller for PXA270 (pxa27x_udc.c) ? That would narrow things down a lot
in trying to get it working.

--David
-- 
View this message in context: http://www.nabble.com/Verdex-%2B-USB-client-tf4094690.html#a11653849
Sent from the Gumstix mailing list archive at Nabble.com.

Hi David,

> Actually this might be useful information in debugging. Since I don't have
> an older gumstix board to test, can I assume that the actual USB gadget code
> (ether.c, file_storage.c, etc.) works for the older gumstix boards
> (PXA255-based ones)?  Therefore the problem is related to the USB device
> controller for PXA270 (pxa27x_udc.c) ? That would narrow things down a lot
> in trying to get it working.

That would be correct (that pxa27x_udc.c is the culprit.

Apparently there are also some silicon bugs which require workarounds,
etc. I highly recommend reading the thread on the arm linux list.

-- 
Dave Hylands
Vancouver, BC, Canada
http://www.DaveHylands.com/

Hi David,

Thanks for the information below on usb client working on Verdex.

I am also trying to make the USB gadget mass storage driver working on a
Verdex XLP6.

Made the kernel changes+removing FB2 from breakout-vx as described below and
loaded the g_file_storage driver, but unfortunately no Voila for me.

I have enabled some of the debug prints in the driver and this is how it
looks like:

After loading the driver:
+---------------------+
g_file_storage gadget-lun0: open backing file: /dev/mmcblk0p1
udc: pxa27x_config_ep is called
udc:  usb endpoint descriptor is:
	bLength:7
	bDescriptorType:5
	bEndpointAddress:80
	bmAttributes:2
	wMaxPacketSize:0
udc: UDCCR: 0xf2600404 is 0x200d103
udc: pxa27x_config_ep is called
udc:  usb endpoint descriptor is:
	bLength:7
	bDescriptorType:5
	bEndpointAddress:0
	bmAttributes:2
	wMaxPacketSize:0
udc: UDCCR: 0xf2600408 is 0x2014103
g_file_storage gadget: File-backed Storage Gadget, version: 28 November 2005
g_file_storage gadget: Number of LUNs=1
g_file_storage gadget-lun0: ro=0, file: /dev/mmcblk0p1
g_file_storage gadget: transport=Bulk-only (x50)
g_file_storage gadget: protocol=Transparent SCSI (x06)
g_file_storage gadget: VendorID=x0525, ProductID=xa4a5, Release=x0311
g_file_storage gadget: removable=1, stall=1, buflen=16384
g_file_storage gadget: I/O thread pid: 432
udc: registered gadget driver 'g_file_storage'
udc: EP0_IDLE, udcicr B8000000.03, udcsir 00.00, udcfnr 00
udc: udccr 0x00000003 = uda ude, con=0,inter=0,altinter=0
udc: ep0 driver 'g_file_storage'
udc: udccsr0 EP0_IDLE 0x00000000 =
udc: ep0 IN 0/0, OUT 0/0
udc: udccs1 = 00
udc: udccs2 = 0

After Connecting to the Host PC:
+------------------------------+
udc: Interrupt, UDCISR0:0x00000000, UDCISR1:0x08000000, UDCCR:0x00000001
udc: USB reset start
udc: Trace path 1
g_file_storage gadget: disconnect or port reset
udc: USB reset
g_file_storage gadget-lun0: fdatasync -> 

Please advise if I am missing something here.

On another thread on Nabble I am happen to read that USB client will not
work on Verdex due to PXA270 related bugs.

Really appreciate any hint/pointers in getting the USB mass storage working
with Verdex.

Thanks,
Shameer





David Kilp wrote:
> 
> Apologize in advance for the length of this but I think there is some good
> information here for those people trying to get USB client mode working on
> Verdex boards.
> 
> I have searched through the mailing list and the wiki regarding using the
> USB client on the Verdex (PXA270) and noticed that no one seems to have it
> working yet and even claims that it won't work. However, there was a post
> (http://article.gmane.org/gmane.linux.distributions.gumstix.general/23659/match=voila)
> in which Craig said:
> 
>> By the way, the Product Comparison Chart for the breakout-vx says  
>> it has
>> USB Client.  Someone might want to fix that.
> 
> It does have USB client.  Plug a breakout-vx into a verdex and load  
> the usb gadget driver on the verdex.  Voila, USB client.
> 
> ----
> 
> Well, I tried this but no voila for me. 
> 
> I need this to work (at least I would like to use the g_ether and
> g_file_storage gadgets) so I did a bit of investigation. I am using the
> combination of a Verdex XM4-BT motherboard + breakout-vx. Normally this
> combination will provide USB host which seems to work pretty well. Looking
> through the PXA270 documentation to get a USB client, it seems like the
> D+/D- USB lines (USBH_P1/USBH_N1) on the 60-pin connector) are used for
> BOTH USB host and USB client depending on how the software controls it.
> Therefore, I figured that USB client SHOULD work as Craig said. However,
> further investigation showed that the gumstix.h defines the following: 
> 
> #define GPIO_GUMSTIX_USB_GPIOn          35
> #define GPIO_GUMSTIX_USB_GPIOx          41
> 
> the first (GPIOn) is used to detect cable presence, the 2nd (GPIOx) is use
> to connect a pull-up on the D+ line to signal to the host that client is
> connected. These are correct for a PXA255 board (the older gumstix boards)
> 
> However, the definitions are wrong for the Verdex boards. According to the
> 60-pin connector these should be:
> 
> #define GPIO_GUMSTIX_USB_GPIOn          100
> #define GPIO_GUMSTIX_USB_GPIOx          27
> 
> (GPIO41 is also used for the OTG_ID which I'm not considering here)
> 
> I changed the lines above, rebuilt my kernel and also removed FB2 from my
> breakout-vx board so that cable detect circuit would work (rather than be
> a host supplying USB +5V)
> 
> Next, I loaded up the g_file_storage gadget driver to see if the mass
> storage device would work and low and behold it showed up as a removable
> storage device on my desktop just like a regular flash drive! Now voila!
> 
> I next tried the g_ether gadget driver (CDC Ethernet) and usbnet on host
> but there seems to be some problem that configuration. While the host
> would do the detection and the usb0 devices were created on both ends they
> wouldn't talk to each other. A usb monitor showed a constant stream of USB
> NAK messages which I haven't investigated further. It seems like it's
> pretty close though. Perhaps someone with expertise in this area could get
> it working (hint, hint...).
> 
> Therefore, it seems like the definitions for the cable detect and pull-up
> resistor are wrong. I was using the  verdex build that came with my board
> (r1321 in my case). However, I just checked the latest mainline build
> (r1457) and the definitions are still wrong.
> 
> Perhaps someone can put some #ifdef's for the Verdex board in the
> gumstix.h file to correct the problem
> 
> One other thought regarding these GPIO's. The PXA270 device has an
> internal pull-up resistor (UP2OCR_DPPUE) that can be used to signal client
> mode. I understand you are trying to create a compatible board between the
> PXA270 and PXA255. However, for the breakout-vx board you could have left
> GPIO27 available and used the internal D+ pull up to signal client status
> in udc_enable() in pxa27x_udc.c by adding a line like this:
>     UP2OCR = UP2OCR_HXOE | UP2OCR_DPPUE;   // set to non-otg client mode
> and set D+ pull-up 
> 
> 
> Regards,
> 
> --David Kilp
> 

-- 
View this message in context: http://www.nabble.com/Verdex-%2B-USB-client-tp11643240p24252464.html
Sent from the Gumstix mailing list archive at Nabble.com.