From:	jd marrow <pachydrm AT ix DOT netcom DOT com>
Newsgroups:	comp.os.msdos.djgpp
Subject:	bug fix (i think) for boehm's GC4.11 under djgpp
Date:	Fri, 20 Jun 1997 01:12:24 -0400
Organization:	pachyderm software
Lines:	30
Message-ID:	<33AA1137.41DA@ix.netcom.com>
Reply-To:	pachydrm AT ix DOT netcom DOT com
NNTP-Posting-Host:	nyc-ny20-03.ix.netcom.com
Mime-Version:	1.0
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

i downloaded boehm's GC4.11 tonight, and decided to test it with djgpp
(under win95) before trying win32 & linux... 

i found that the hardcoded STACKBOTTOM was causing problems (erratic
gctest performance when it wasn't dumping or failing on stack direction
tests), and i didn't like the idea of having to set GC_stackbottom in my
client code, so i fiddled around a little and the following seems to
work:
	
	in config.h, replace the djgpp STACKBOTTOM #define with:

	#define STACKBOTTOM __dpmi_get_segment_limit(_go32_my_ss())

i make no guarantees :), but it seems to work pretty well; gctest.exe
results seem to consistently look like:

	Completed 1 tests
	Finalized 2206/2206 objects - finalization is probably ok
	Total number of bytes allocated is 55227956
	Final heap size is 5500928 bytes
	Collector appears to work

i hope this is useful... 

regards,

jdm


jd marrow 			jd AT riverbed DOT com

- Raw text -