But for me, your patch is making the situation worse, not better.
for me too, however, if you increase #1024 stack size fo #64*1024 size, it's ok.
So, there must be a "correct" fix.
i agree, this is just quick hack.
Can you produce a sample application that demonstrates a problem with
the original crt0.S ??
I think it'll be easier to debug if we have an application that does
this first.
well... if I remember correctly, there were crashes in gcc compilation... but maybe it would be nice idea to try that compilation with all patches but this one.
Ah, yes, I've got it -- I know what was the problem. My patch consists of two things -- one is that "explicit" stack size in ctr0.s and second was some friendly message about insufficient memory for TPA. When I used the original crt0.s, sometimes, instead of displaying this message, I've seen crash/memory violation. I think it was something to do with stack align, I even reported some speculation what could be wrong here in mintlist. So I've chosen to define stack in the easier way, i.e. using hard constant to be sure it was that problem and -- for that time -- bug disappeared.