kaffe-0.8.x garbage collection problem (at least on m68k)
tim at tjwassoc.demon.co.uk
Mon Mar 3 15:07:58 PST 1997
On Mon, 3 Mar 1997, Richard Henderson wrote:
> Likely dependant on the compiler. The processor only needs 2 byte alignment.
> gcc 2.7.2 at least rounds to 4 bytes.
> But there is another problem with the gc, at least on A/UX -- the main
> program stack is grown dynamically, but only if the fault is through a
> write through %sp. Which means that you cannot at startup time say
> stackbase = ...
> stackend = stackbase - X
> because X is likely below the current %sp, and so you'll crash on the
> first gc pass. It would be nice if the gc would look at the current
> stack rather than the entire stack. This would also have the effect
> of supressing false positives from stack garbage.
There are already GC related fixes a foot and I'll try to address this in
them. I still worry about integrating the GC with native threads (e.g. how
does one obtain the current thread register set) when I come to abstract this.
Tim Wilkinson Tel/Fax: +44 181 440 0658
T. J. Wilkinson & Associates, Mobile: +44 370 621006
London, UK. Email: tim at tjwassoc.demon.co.uk
More information about the kaffe