[Gc] bdwgc for RTEMS gc collect fail

Jie Liu lj8175 at gmail.com
Thu May 19 04:24:32 PDT 2011


Hi Ivan,

I have tested it, It's worked well without setting GC_stackbottom in test.c.

Best Regards,
Jie


2011/5/19 Ivan Maidanski <ivmai at mail.ru>

> Hi.
>
> I've committed the patch except for one line in test.c which sets
> GC_stackbottom - it should be redundant since the bottom value is set from
> STACKBOTTOM.
> Please test it.
>
> Regards
>
> >Hi Ivan,
> >Thank you very much for your advice. I have run tests/test.c on RTEMS/i386
> successfully under your help.And I have updated some RTEMS related code,
> which is attached in this email.
> >Best Regards,Jie
> >2011/5/17 Ivan Maidanski <ivmai at mail.ru>
> >Hi Jie,
> >
> >Check DATASTART/END values - the range should not intersect with the heap
> addresses.
> >
> >Regards.
> >
> >Sat, 14 May 2011 00:23:49  0800 Jie Liu <lj8175 at gmail.com>:
> >> Hi all,
> >>
> >>
> >> I am using bdwgc on RTEMS[1] i386 platform for previous work of
>  GSoC2011[2] project "RTEMS Porting of GCJ".
> >>
> >> But even the simplest test like :
> >>
> >> for(i=0;i<20000000;i  ){
> >>   int *q=(int *)GC_MALLOC(sizeof(int));
> >> }
> >> printf("GC_get_heap_size=%d\n",GC_get_heap_size());
> >>
> >> will get the result "GC_get_heap_size=161669120", this said the object
> allocated by GC_MALLOC could not be collected automatically, even use
> GC_MALLOC_ATOMIC the result is the same.
> >>
> >> I try to read the source code of bdwgc and find GC_finalize_now is 0.
> And bdwgc's mark phase is not easy to read, can someone give me a help hand
> ?
> >>
> >> Thank you very much for your time.
> >>
> >> [1]http://rtems.com/
> >> [2]http://google-melange.com/
> >>
> >> Best Regards,
> >> Jie
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://napali.hpl.hp.com/pipermail/gc/attachments/20110519/d01b7ba4/attachment.htm


More information about the Gc mailing list