Re: Step by step instructions

On Tuesday 06 September 2005 18:11, JeDi wrote:
> On 9/7/05, Richard Bateman <taxilian_at_zzt.net> wrote:
> > well, the screen resolution of an x50v is 480x640 and the base memory
> > address is 0x0c00_0000. these are all set correctly in the
> > kernel26.tar.bz2 file on my site, which is why I sent you that link.
>
> Are the zImage and initrd files on your tmp site from the same kernel?
> I mounted the initrd file and the vsfb module didn't seem to be there.
> I haven't done a cross compile ever, so if I have to build the kernel
> and/or modules from your archive, will "make menuconfig", "make" and
> "make modules_install" just work? Or do I have to do something special
> because it is for another device than my pc?

Honestly, I don't know what the initrd and zImage files on my site are from.
they probably don't work. I haven't had time to find my working kernel
build. I'm sorry. I might have time this weekend.

You need the cross-compile arm toolchain, which you can get from handhelds.org

I didn't have to type anything special, once I had those in the path.

> > it's easier to put haret, kernel, and initrd on an sd card, and then if
> > you create a ext2 image called (I think) rootfs in the root of the sd
> > card you can mount it from there rather than needing a full partition on
> > the sd card.
>
> Yep, I read that in the linuxrc file of the initrd. I'll try that as
> soon as I'm back at work (in a couple of hours :-)). I see that the SD
> card gets mounted at /mnt/mmc. Is that writable? Because then I could
> dump the kernel messages to a file on the SD card (dmesg >
> /mnt/mmc/kernel_log), so I could at least verify that it tries to boot
> the kernel.

You may have to mount it yourself. it's something like /dev/mmc/??0 or
someodd. it may mount automatically if you have it set to do so in your
initrd

> > Actually, the framebuffer is the easy part, and that's just because the
> > 2700g controller is already initialized by wince. we're just directly
> > accessing the video ram for our framebuffer.
>
> So I can get the kernel messages on screen too? That would be sweet.
> As for the controller not being initialized by the driver: let's
> tackle that later. First get linux decently running from wince ;-)

Yes, the kernel messages go to the screen.. and yes, it's sweet =] biggest
problem right now with making linux semi-usable is the touchscreen driver.
of course, it still also has no networking.

> > It'll take more than a bootprocess modification to get usb networking,
> > bluetooth, and wifi working, automatically or otherwise.
>
> What about this guy?
> http://handhelds.org/hypermail/aximx30-port/0/0067.html He says he's got a
> wifi tcp/ip control connection running with the x50v. If that wasn't a
> joke, maybe wifi is just working out of the box? I'll try that at work too
> (plenty of access points).

That was just using haret, with one of my early test builds. Haret runs in
wince, so linux was not involved when his network was working.

> > USB we seem to be missing some GPIOs for, so I suspect it is accessed
> > through the Toshiba ASIC (which we know nothing about). bluetooth may
> > be very easy to make work, but I don't know enough about it to try. as
> > for wifi, we don't even know how it's interfaced. My guess would be
> > that it's at one of the external memory locations. (see static memory
> > addresses in cpu docs)
>
> I'm afraid I'm a software boy mainly. I don't even know what GPIOs and
> ASIC is :-) But if I can help with anything, I'll do my best. Helping
> with the kernel and modules will be more my thing I guess...

Kernel and module work will require some basic understanding of the hardware,
since it's the kernel and modules that control the hardware... i.e. drivers

> > when the framebuffer is working, you still don't yet have the pxa keys
> > working; this shouldn't be hard to make work, but I haven't tried it yet
> > so I could be wrong. Also, the touchscreen doesn't work. we know what
> > chip it is, and how it's interfaced, but for some reason we are unable
> > to make a driver work for it; someone with more knowledge about
> > touchscreens might be able to spot an ovious difference, or it may be
> > something to do with that toshiba asic chip again.
> >
> > These are our challenges; this is what we need to figure out to make
> > this usable.
>
> I have a good feeling about this. I'm still convinced that the first
> thing we'll have to do is get a control connection other than a serial
> one up and running (usb preferable, but as that seems to be hard right
> now WiFi or bluethooth), cause then more people can join the effort.
>
> I'll give it another full-time working day tomorrow, and who knows
> what will happen :-)

There are definitely possibilities. bluetooth may even work (if you turn it
on in wince before starting linux) with existing drivers. biggest problems
now are:

1. figuring out usb host
2. figuring out touchscreen

audio probably won't be hard; wireless network, I dont' even know where to
start. :-/

Richard
Received on Wed Sep 07 2005 - 02:13:33 EDT