Lpc2000

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Carsten Gr\ufffdn wrote:
> Hello,
> 
> I have been playing around with some of the USB samples from Keil
> etc, and been trying to get just one of them to run on a LPC2148
> board from www.embeddedartists.com (very nice boards btw) but I'm
> having some problems. Well, actually it's the sam eproblem no matter
> which sample I try, I always get a message from Win2K "Unknown
> device....."....
> Now, before I bury myself deeeeeep into the "unknown of USB", is
> there something I need to be aware of when running some of these
> samples ? The LPC2148 board has P0.31 switching the USB+ to 3V3, and
> P0.23 senses Vcc from the USB host, which should be the same as the
> Keil MC2140 board......
> The same thing happens with the Keil usbmem, usbhid and the philips
> vcom sample (the keil audio reports an internal linker error when I
> try to build that....)
> 
> So, something I must be doing wrong, any leads ???
> 
> (PS: I have ported uIP to a LPC2138 Ethernet board from
> embeddedartists running with the Keil RTX RTOS, working very nicely)

I too have an LP2148 from embedded artists and I just got the usbmem
example working with gcc to the point that windows recognises my
board as a mass-storage device with the proper USB device descriptors
(product/vendor id, strings descriptor, etc.)

USB mass storage itself is not quite working yet, although windows
properly recognises the size of the disk (at 16 kB).

I think the relevant steps to make it work were:
* the P0.31 pin should be configured for the USB 'CONNECT' function
* aggressive packing of all USB related structs by means of
  '__attribute ((packed))' and of every 'sub-struct' within those
  structs.
* aligning several structures to 32-bit boundaries, using
  '__attribute ((aligned(4)))'
* replacing __irq by '__attribute__ ((interrupt))'
* (I just removed some of the '__at' directives)
* Modifying the for-loop in USB_ReadEP into
  for (n = 0; n < (cnt + 3) / 4; n++) {
	dw = RX_DATA;
	*pData++ = dw >> 0;
	*pData++ = dw >> 8;
	*pData++ = dw >> 16;
	*pData++ = dw >> 24;
  }
* Modifying the for-loop in USB_WriteEP into
  for (n = 0; n < (cnt + 3) / 4; n++) {
    TX_DATA = (pData[3] << 24) | (pData[2] << 16) | (pData[1] << 8) |
pData[0];
    pData += 4;
  }
(there used to be some __packed directive in these parts in the
Keil code)

I may have changed some other stuff, but I think the list above
are the most significant changes.

Some very useful tools in making it work so far, were:
* OpenOCD in combination with a wiggler clone and arm-elf-insight.
  OpenOCD is similar to what OCDRemote does, except this is open-source
  http://developer.berlios.de/projects/openocd
* "usb in a nutshell", a really good document describing the essence
  of USB, at
  http://www.beyondlogic.org/usbnutshell/usb-in-a-nutshell.pdf
* usbsnoop, a windows USB packet sniffer, see:
  http://benoit.papillault.free.fr/usbsnoop

Kind regards,
Bertrik
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDx/4iETD6mlrWxPURAnaXAJ46r76Yzi/5+YR16aypISUGCIC3fQCgl1HE
Dss2pAvZbJVHFWL2KAZQevc=
=XTft
-----END PGP SIGNATURE-----