This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RFC]: New zone based garbage collector and thoughts onincremental collection
Daniel Berlin <dberlin@dberlin.org> writes:
> On Fri, 5 Sep 2003, Zack Weinberg wrote:
>
>> Daniel Berlin <dberlin@dberlin.org> writes:
>>
>> >> > When I was thinking about this in the spring, I convinced myself
>> >> > that knowing the type at ggc_alloc time was never necessary.
>> >>
>> >> For generations, sure, for incremental, you do need to know it, because
>> >> you need to restart marking at any point.
>> >
>> > We also need to know it to be able to mark a single zone at a time,
>> > because we don't know which roots point to things in which zones (we only
>> > know which roots *start* in which zones, which isn't helpful at all).
>>
>> And furthermore, always having a type code around allows *drastic*
>> simplification of the heap walker routines.
>
> I've got type codes going now, and a marker array.
> The problem is the uses in htabs and whatnot may require a wrapper
> function for their allocator, rather than passing in regular ggc_alloc
> (since we need to pass in the right type code).
I think we can live with that. Or we can add a closure-argument
variant to the htab allocation logic.
zw