2006/8/16, Jay Coles <jaycoles_at_yahoo.com>:
>
>
> Actually there would be some reasons to make writing
> to flash low priority
>
> the erasoning behind this is that you caould have a
> loop mounted image on your SD card so thatyou can dual
> boot when you use a backup program on your Z
>
> i persanally cant see a good reason to do that :)
> hawever once we ID the write protect flag of the flash
> and have a kernel running then it shouldnt take long
> at all to get writing to flash working
>
> in the end i see that there will be two ways to
> install, the special X30 flasher or using HARET to
> launch a kernel with an initrd that does all the flash
> writing itself, the second would be a better way IMHO
> but a bit less safe (for the developers)
>
> a quick breakdown of terms:
> HARET=kexec for windows CE
> flash bootloader=BIOS bootloader, it does a minimal
> setup (like a BIOS) and then passes execution to the
> next stage (normally another bootloader on a PC, on
> PDA's this is normally the kernel)
>
> now unlike a PC we dont have a BIOS that allows us to
> tinker with setteings and sets up everything for the
> OS (most moderen day OS's setup everything again
> anyway and ignore thi BIOS settings) however you can
> probelly see now that if the linux kernel can set
> everythin up then there is no need for a bios (hence
> why there are projects to flash normal PC motherboards
> BIOS chips with a linux kernel, linux righ from power
> on :))
>
> now a status update, it turns out that LFS (linux from
> scratch) have a cross compilation guide which i am
> following, at the moment i have bin-utils and the GLIB
> headers. i expect GCC by tonight, for those that care
> i am compiling it with EABI so that we can use the
> iwmmxt instructions that make this thing fast, i am
> doing that naw rather than latter so i dant have to
> recompile the toolchain down the track
>
> cool, back to work
>
I see, so we boot windows CE, we ask Haret to write the Kernel that will be
built with your glibc / gcc toolchain compiled for the Pocket PC
architecture (ARM). This is really birght you know and quite simple.
>From there, we will have to develop module for the kernel to get control of
BlueTooth and Wireless, sound and touchscreen based on what is working and
what not.
Do you plan to use the latest kernel (2.6.17/18?) ? I think that for LFS
it's using older Kernel just to be sure that it will be running well. I've
checked the manual to build the toolchain... do you really need all those
softwares ?
If I may suggest something, I would really like to use the pacman packet
manager from ArchLinux <http://www.archlinux.org>. When should we decide
this? This way we would be able to update the package really easily and we
just need to create a mirror site where we could put base
packages created by us. It is able to manage a source directory and a
package directory easily. But it might be easier to use a complete source
distro and compile using our pocket pc since I'm pretty sure, we will not
have a place fast enough to place all those packages.
What do you think? Am I lost?
Michel
Received on Fri Aug 18 2006 - 15:31:07 EDT
This archive was generated by hypermail 2.2.0 : Fri Aug 18 2006 - 15:31:24 EDT