Message-ID:	<393636D8.72B4BF13@softhome.net>
Date:	Thu, 01 Jun 2000 13:11:36 +0300
From:	Laurynas Biveinis <lauras AT softhome DOT net>
X-Mailer:	Mozilla 4.72 [en] (Win98; U)
X-Accept-Language:	en
MIME-Version:	1.0
To:	djgpp-workers AT delorie DOT com, Alexandre Oliva <oliva AT lsd DOT ic DOT unicamp DOT br>
Subject:	Re: Using ggc-page with plain malloc
References:	<orhfgef6lo DOT fsf AT garnize DOT lsd DOT ic DOT unicamp DOT br>
	<3914FA82 DOT DFFBCF52 AT softhome DOT net> <or4s7iao1f DOT fsf AT saci DOT lsd DOT ic DOT unicamp DOT br>
Reply-To:	djgpp-workers AT delorie DOT com

Alexandre Oliva wrote:
> 
> On May  7, 2000, Laurynas Biveinis <lauras AT softhome DOT net> wrote:
> 
> > Alexandre Oliva wrote:
> 
> > While looking into gc-simple performance issues, I've tried your
> > program on DJGPP (i686-pc-msdosdjgpp) and it fails:
> 
> > Assertion failed at align-malloc.c line 60: p == r
> > Exiting due to signal SIGABRT
> 
> Sorry about the late reply.  I see you were not using the latest
> version of my patch (I'm not even sure I posted it :-).  Would you
> like to give this one a try?

I've tried it; and there are several issues: a loop in xvmalloc_align()
is semi-infinite - it fails to allocate aligned memory, but does not
count that as a failure, so it tries over and over again. As a result,
my memory debugger gives me a 2.3MB log file for simple program:
   unsigned i = xvalloc_overhead();
   xvfree(xvalloc(4096), 4096);
I can see from a log that there were 7600 calls to xmalloc, memory usage
peak level over 7 MB.

So it is still not very useful to DJGPP. Sorry, I can't provide much help
now because idea behind your code is not very clear to me. DJGPP's malloc()
sources are confusing too. But if you have specific questions about DJGPP
malloc() behaviour, just ask.

Laurynas

- Raw text -