Date: Thu, 11 Apr 1996 08:41:03 +0200 (IST) From: Eli Zaretskii To: j DOT aldrich6 AT genie DOT com Cc: djgpp AT delorie DOT com Subject: Re: malloc crash In-Reply-To: <199604110429.AA236316991@relay1.geis.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Thu, 11 Apr 1996 j DOT aldrich6 AT genie DOT com wrote: > still crashed, but at i=997, not 1027. What's more, it crashes not only itself > but every program associated with it! When I ran just the image using > go32-v2, it nuked go32 at the same place. When I tried running it through > redir to save a copy of the core dump, it crashed redir, too. In DJGPP v2, parent programs generally get the signals of their children, unless they take special precautions, so that is not surprising. Of course, when the DPMI host is the cause of the crash (see my other message), you should expect all its clients to crash right there and then. > haven't tried the code under Windoze yet. What's odd is that I have a > large program that allocates thousands of memory blocks, totalling > many megabytes, and it has never had any problem. What matters is not the total size of the allocated memory, but the number of independent allocations which live at any given moment. I doubt that any reasonable program has more than a hundred malloced regions simultaneously. The GNU C++ compiler that is the first DJGPP program to be known to hit that CWSDPMI limit (see the FAQ), reaches slightly more than 200 chunks when compiling some *really* large C++ programs.