On Mon, Feb 20, 2006 at 14:41:02 +0300, Valeriy E. Ushakov wrote:

> I think this is a bug in hpcboot ELF loader.  I've just encountered
> a a problem with one of my kernels, where hpcboot would miscalculate
> a number of pages to allocate and the kernel ends up booted with a
> corrupt stack.  I'll try to fix it today.

kloader(4) had a similar bug that i've fixed some time ago.  On
hpcboot it's harder to trigger the bug conditions, b/c hpcboot rounds
memory request up to 64K, so sizes of kernel segments and symbol table
must have "bad" rounding properties *and* the size must be just under
64K boundary so that there are no spare pages.

I'll commit the fix shortly and ask for hpcboot200 to be recompiled (I
cannot do that myself).

SY, Uwe
-- 
uwe@ptc.spbu.ru                         |       Zu Grunde kommen
http://snark.ptc.spbu.ru/~uwe/          |       Ist zu Grunde gehen