my j820 toolchain breaks

David Jones
Fri Mar 26 04:18:21 CET 2004

Boy that is strange.  I used the toolchain and cross
compiler that is on the project site.  The kernel is
the one from with the patches that are
on the project site.  It may be late Saturday before I
can get a diff or compressed tar of the source I'm
using up on the project site.  I'll probably go for a
diff since it's smaller.  In the mean time, I'll check
the toolchain install instructions and upload the
zImage I'm using.  It's in the dave directory and is
called ramdiskimage.  I'll also download your binary
and see what it does. 


--- Francois-Rene Rideau <> wrote:
> Can you tell me why the kernels I compile won't get
> booted by hpcboot???
> The very same hpcboot will start without a problem
> those kernels made by Dave.
> So do you have a working setup? Which is it?
> Can you cp -al or rsync -a your tree and make a diff
> wrt to latest CVS?
> If you use cp -al, it might be a good idea to make
> the original read-only
> (e.g. owner by another user or chattr -i).
> What toolchain(s) do you use, as installed how?
> Wondering about my toolchain, I used
> arm-linux-objdump -x to compare
> my non-bootable kernel to dave's bootable one,
> and saw no noticeable difference.
> All: Can you test that my non-working binary doesn't
> work for you?
>  (the latter might not be present yet, till my new
> build is done.)
> Dave: Can you upload to a working zImage and
> so I may test
>  just the link step?
> Oleg: if you're familiar with the hpcboot sources,
> can you tell me what
>  hpcboot does between the moment it prints
> 	load link 991 zero clear link 0.
>  and the moment it prints
> 	kernel entry address: 0xc0800000
> 	framebuffer: 640x480 type=3 linebytes=640
> addr=0xc0002e00
> 	console = 1
>  near the end of the boot sequence?
> I've installed the 2.95.3 toolchain and will be
> trying with it
> (already fixed a function in j820_keyb.c that relied
> on a recent gcc).
> PS: I'd like to hack 32MB memory detection, and
> discovered that the method
> used by blob to autodetect memory clobbers one word
> of memory every 1MB
> (among possible other things) -- which might pretty
> well explain
> why I had trouble with blob when I last tried, last
> year.
> Blob really expects to sit in ROM or Flash at be
> able to waste the RAM.
> Is there a better canonical way to discover the
> amount of RAM?
> PPS: the replacement 2GB tinydrive I got seems
> fux0red at about the same
> place as the first one. Could it be an artefact of
> the kernel driver?
> Are there settings to use on the linux side so as to
> be "safe" with
> a microdrive?
> [ François-René ÐVB Rideau | Reflection&Cybernethics
> | ]
> [  TUNES project for a Free Reflective Computing
> System  |  ]
> That man who dares to waste one hour of time, has
> not discovered
> the value of life. -- Charles Darwin
> -- 
> Mailing list info:

Do you Yahoo!?
Yahoo! Finance Tax Center - File online. File on time.

More information about the Jornada820 mailing list