Re: [RE: [iPAQ] USB for SA11x0]

Hi Brad,

Regarding the half duplex test I have put a first draft of detailed procedure at-
http://www.handhelds.org/projects/usbhdx.html
I ran the test on the current drivers (using the above proceedure) and I successfully transmitted
6900 16k packets back and forth between the iPAQ and the PC. The only thing I have done
differently is that I built and flashed a new test4 kernel from the source since I thought I
needed to export some dma variables but ended up not needing them. I will repeat the test tomorrow
with the zImage from handhelds.org.

>>I would recommend that
>>charlie flynn take a look at it and compare it to what he's done/taken over to
>>see if there's overlap and how to proceed so as to not duplicate effort.

OK I've downloaded the kit and applied RMK5 and Nico's patches. I have found the Doco and the net
host driver but not had time to look at anything else. The usb-net-host driver is derived from the
original itsy driver and should be relatively easy to understand.
Regarding the lost pings etc at this stage I can only suggest the problem I saw with the PC host
driver dropping the 1st 64 byte packet ( I kludged the driver by getting the iPAQ to transmit the
1st packet twice). This problem is a result of either
(1) The D0 and D1 bit being out of sync as sent from the iPAQ
(2) The PC host missing the 1st 64 byte packet in the sequence.
(3) Something else I haven't found yet.
A CATC analyser showed me that the 1st 64 packet _was_ being sent by the iPAQ but I havn't
checked the validity of the 'D' bit on that first packet.
I'll try and ping a kludged driver (at least I have confidence the data link is working such that
it can tx 16k packets from the iPAQ and 64 byte packets to the iPAQ) .

Tomorrow I can look at pinging the iPAQ unless somebody fixes it first.

That's all I know for now.

Regards
Charlie

Brad Morrey wrote:

> This works great. I got the usb client and host working. I used both a test8
> kernel and test5 kernel on the host side to test it out. Seems to be quite
> buggy. pings worked, sometimes. rlogin to the ipaq failed 2 of 3 times.
> debugging errors were very verbose. I can forward on the errors to interested
> parties. But definitely not ready for prime time. I would recommend that
> charlie flynn take a look at it and compare it to what he's done/taken over to
> see if there's overlap and how to proceed so as to not duplicate effort.
> Great work brad parker and Nicolas! Almost there!
>
> -brad
>
> Nicolas Pitre <nico@cam.org> wrote:
>
> On 5 Oct 2000, Brad Morrey wrote:
>
> > Second I went out to kernel.org and got the 2.4.0-test8 kernel source, and
> > then applied the rmk5 patch from the ARM site, then applied nic's diff.
> This
> > works perfectly. I can then make menuconfig to my heart's content. (No
> MTD
> > problems here.) I made a kernel with a fair amount of stuff compiled in to
> > try and get this working. Unfortunately, my 550K kernel wouldn't flash
> > because the space allocated in flash for kernels is smaller than that.
> >
> > I paired down the kernel significantly, keeping cramfs, ext2, and ramfs
> > support compiled into the kernel. (I've included my .config for reference.)
>
> > When I flashed this kernel, it boots, but then kernel panics trying to
> mount
> > the root filesystem. The precise error is:
> >
> > wrong magic
> > Kernel panic: VFS: Unable to mount root fs on 3c:04
>
> Try copying drivers/block/flash-mem.h from the handhelds.org tree over the
> one in my patch.
>
> > Now the server portion is another matter. Again, taking the test8 kernel,
> and
> > patching the rmk5 patch, I get a menuconfig that doesn't have the
> usb-net-host
> > code. So, I apply the np1 patch, and get the net-host code, but a pile of
> ARM
> > additions that make it impossible to compile the kernel for x86.
>
> Take a fresh 2.4.0-test9 tree and copy drivers/usb/usb-net-host.c from
> the patched tree to the unpatched one. Next edit drivers/usb/Config.in
> and drivers/usb/Makefile to add the single line dealing with
> CONFIG_USB_NET_HOST to reflect the patched tree. Then you should be able
> to build yourself a 2.4.0-test9 tree with usb-net-host module for your PC.
>
> Nicolas
>
> > ---------------------------------------------
> > Attachment: .config
> > MIME Type: APPLICATION/OCTET-STREAM
> > ---------------------------------------------
>
> Computer Science PhD Student
> University of Colorado, Boulder
> (Anti-spam account)
>
> ____________________________________________________________________
> Get your own FREE, personal Netscape WebMail account today at http://home.netscape.com/webmail
> _______________________________________________
> iPAQ mailing list
> iPAQ@handhelds.org
> http://handhelds.org/mailman/listinfo/ipaq
Received on Mon Oct 9 07:32:29 2000