Summary

[G Almasi]

Wow guys, there is so much activity on the mailing lists now,
I have trouble just reading through it.

Let me try to answer the few points that stuck to my mind while
reading.

1) USB - support is not really hard to do; the code already
exists in some form (written for the SA1111, of course) and
it is relatively straightforward to back-port it to the
SA-1101, especially since the latter is "less broken".

But I don't have any USB devices to talk to, not even a pen
drive, and I am not ready to shell out another $50 for one.
I just spent my allowance on this machine - I bought a 256MB
CF card.

2) Userland distribution: I have managed to install Debian
base on the CF card. Debian base + emacs + openssh + X
takes about 130 Megabytes. I have 60 MBytes free on the disk
and the rest is taken up with the DOS partition (to boot),
an 8MB partition for the debootstrap partition and a swap
partition.

I will publish instructions just as soon as I can make this
happen reproducibly. I used Debian Woody's netwinder port,
directly from ftp.debian.org.

Of course, X doesn't work - I don't know how to make it use
the framebuffer device. Anyone care to shoot ideas?

Another problem I'm having is the way I'm chroot-ing from
the initrd to the CF card's root. I have to have cardmgr
running all the time, and yet I'd like it to read the config
files from the new root file system. Ideas?

3) I have clocked two days of uptime so far. After Oleg's
suggestion to change memory to 16MBytes instead of 32, the
machine became rock stable - before the kernel used to crash
whenever I pressed it too hard. Guess it went into RAM that
didn't exist ;-)

Does anyone know what the physical address of the RAM
in the expansion slot is? Oleg, this seems like right up
your alley ;-) Of course I would need to test the
memory before using the second bank.

4) Somebody mentioned the mouse device. I'm fairly
convinced that the mouse is in the "standard" place on the
SA-1101, and building a driver is not very hard (same as with
USB, I guess). In time, I will do it - let me get X hacked up
first, so there is a reason to do the mouse ;-)

5) The frame buffer display and PCMCIA don't like each other
too much. Every time the machine accesses the disk (and, with
16 MB of RAM, that is fairly often) the display goes away
momentarily. I guess this must be a DMA problem. Related problem:
Pulsing reset on the SA-1101 kills the frame buffer completely.
This means that we are "inheriting" some state in the SA-1101 from
WinCE. Ugly, that.

George