On Tue, Dec 04, 2007 at 09:25:42AM +1100, Andrew Dennison wrote:
> To the list too...
> 
> ---------- Forwarded message ----------
> From: Andrew Dennison <[EMAIL PROTECTED]>
> Date: Dec 4, 2007 9:24 AM
> Subject: Re: [Prex-devel] problems booting with large task
> To: Alexander Botero-Lowry <[EMAIL PROTECTED]>
> 
> 
> On Dec 3, 2007 8:05 PM, Alexander Botero-Lowry <[EMAIL PROTECTED]> wrote:
> > Howdy all,
> >
> > For the last few days I've been porting over SEE (Simple ECMAScript Engine)
> > [ http://www.adaptive-enterprises.com.au/~d/software/see/ ] to prex. This
> > has mainly involved porting over some of msun (the math lib freebsd uses),
> > and filling in a few holes in stdio implementation as well as stealing a
> > few necessary include files and the likes here and there. I also extended
> > fslib to have fprintf, vfprintf, fputc and fwrite implementations, which
> > are used in a few places by libsee. Anyway, I finally got it all compiling
> > and linking, but when I attemped to boot with my new task I ran into a
> > fun new set of problems. If any calls are actually made to libsee, prex
> > either gets a triple fault (or so qemu says), or qemu simply goes for a
> > blank screen with no indication of how far into the boot process it got.
> 
> Have you turned on boot debugging?
> In boot/include/boot.h:
> #define DEBUG_BOOT             1
> #define DEBUG_BOOT_IMAGE    1
> #define DEBUG_BOOT_IMAGE_RELOCATE 1
> 
This doesn't seem to give me anything new when I try to boot, either in the case
of a triple fault or in the case of a hang on a blank screen. Which is leading 
me
to believe that it may be a problem in the boot sector. Since the triple fault
occurs right after "Loading PREXOS..." 

> How much memory does prex have available on i386? It looks like
> CONFIG_MIN_MEMORY is set in the default config so you may be
> exhausting the 512KB that appears to be available (I've never run prex
> seriously on i386).
> 
On boot it always says one meg available, I haven't quite figured out how
to convince it to see all the memory..

> 
> > My guess is that either bootldr or the boot sector can not handle the
> > large binary in the disk image. A simple test of the interpreter is about
> > a 300kb binary.
> 
> This would most likely show up as an unsupported type of elf
> relocation. Are you using an executable task or a relocatable task?
>
executable task, but i'm thinking that the increased size of the FAT
table is tripping up the boot sectors ability to actually load the
image at all?
 
> Does your task fit within any arbitrary limits imposed by the link script?
> 
Not that I can tell, but that asm stuff that constitutes the linker script
doesn't really mean much to me.

> >
> > The information dumped by qemu when a triple fault occurs are as follows:
> 
> Wikipedia suggests a triple fault can be caused when the IDT is
> corrupted: sounds like a memory sizing / linking issue.
> 
> The prex i386 port has a gdb stub - are you using gdb in your debugging?
> 
I haven't quite figured out how to attach gdb via qemu, though if I'm right
about it being the boot sector that won't help much.

> Andrew
> 
> -------------------------------------------------------------------------
> SF.Net email is sponsored by: The Future of Linux Business White Paper
> from Novell.  From the desktop to the data center, Linux is going
> mainstream.  Let it simplify your IT future.
> http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
> _______________________________________________
> Prex-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/prex-devel
> 

-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell.  From the desktop to the data center, Linux is going
mainstream.  Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
_______________________________________________
Prex-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/prex-devel

Reply via email to