Eric,

I am still having problems with mkelfImage, and I am now getting very
confused.

I have created a LinuxBIOS for a SiS55x platform, and have created the
appropriate Etherboot module, put them into an Eprom Emulator.

Using a 2.4.5 Kernel I boot perfectly.

Using a 2.4.17 or 2.4.18 Kernel, I run in to problems.

I have done most of the debug in the 2.4.17 kernel.

Basically it just hangs up in the setup_arch function. The worst issue
though is that the hangup point moves when I add debug code! The mode debug
code I add, the further it executes.

I have added some simple serial print routines (at this point the console
has not yet been initialised, so printk's are being buffered). I always get
into the setup_arch function, but then if I only add a print just before the
init of the boot-time allocator, I do not see that print. If I scatter
enough prints in the code preceding this, it does the init of the boot-time
allocator correctly. These symptoms look a little ominous to me!

I have tested this both with gcc 2.91.66/ld 2.9.5 and 2.95.3/ld 2.11.92 with
identical results. I am using mkelfImage from the current LinuxBIOS tree.

I have examined the e820 tables generated by mkelfImage (both when they are
created by mkelfImage and also once I get into the kernel code) and they
seem to be good.

I hope it is not something silly like a watchdog killing things - well
actually it cannot be that because it just hangs, it doesn't reboot.

Have you tested 2.4.17 with similar tools?

I know Ollie has created stuff based on 2.4.17 on the 550 but I am not sure
if he is using Etherboot/mkelfImage.

Ant ideas?

Hamish

Hamish Guthrie
Guthrie Consulting
Buglgrond 86
7550 Scuol
Switzerland
TEL: +41 81 862 2613
FAX: +41 81 862 2640
MOBILE: +41 78 809 0910
E-MAIL: [EMAIL PROTECTED]

Reply via email to