[PATCH] New -fstack-check implementation

[Duncan Sands]

Hi Eric,

> Due to popular demand :-), I'm again submitting the new stack checking 
> implementation that we have been using at AdaCore for a few years now.

in the past valgrind has noticed invalid memory accesses when using
-fstack-check in a multi-tasking program for example when a task exits
(the invalid accesses coming from the stack checking code).  Is this a
known (fixed) problem?  If not, I will try to cook up a testcase.

Ciao,

Duncan.