> : so you'll have to wait a bit until I upload the stuff.
> Ralf, you are such a tease :-)
Oh, I just thought I should keep you informed :->
> I've manage to compile the latest kernel (1.3.91 is the latest that I
> grabbed) for the Deskstation. I did it w/o procfs, and found and
> fixed a couple of bugs in that support. One of the things I noticed
> was the load address for my machine is now 0x80100000. Cool! It also
> looks like the interrupt handlers are copied down. Isn't it dangerous
> to do that inplace copy? Shouldn't the code have a if (begin_except
> != KSEG0) protecting the copy?
Why should an inplace copy be dangerous? Ok, for Jazz machines the kernel
needs several *microseconds* longer to startup. A long time if you're
in hurry :-)
> Anyway, I'm not having much luck. For reasons unknown I find that I'm
> unable to get past paging_init() in the boot sequence (at least I get
> a 'b' in the upper left hand corner of the screen and I've hacked
> start_kernel to look like:
> *(unsigned char *)0xa00a0000 = 'b';
> memory_start = paging_init(memory_start,memory_end);
> *(unsigned char *)0xa00a0000 = 'c';
> which is a pretty strong smoking gun).
I definately bugged mem_init() for your machine which frees the complete
low 640k including the exception handlers ... As paging_init() gets
called before mem_init() this cannot be your problem. Hmmm...
I recall 1.3.64 hanging on
> boot for reasons unknown, or was that 1.3.82. It didn't happen on
> 1.3.62. I'm off to find out why this is happening to me. I wish
> there was some easy way to get early panics to always print
> something... Ah, a problem to chase when I have time...
>From my Tyne hackery I've got still a tiny console in arch/mips/lib/tinycon.c.
Adjust the videoparams and you have a debugging console. Usage should
be clear after a short look into the code. With print_string() and
sprintf() in combination you can the same things as with printk().
tinycon.c is also easily useable for other machines with a PC style
character based display.