X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-workers-bounces using -f Date: Tue, 11 Dec 2001 10:12:30 +0200 (IST) From: Eli Zaretskii X-Sender: eliz AT is To: Charles Sandmann cc: djgpp-workers AT delorie DOT com, acottrel AT ihug DOT com DOT au Subject: Re: go32-v2 memory chompage [was: Re: v2.03 refresh ...] In-Reply-To: <10112102224.AA25539@clio.rice.edu> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Reply-To: djgpp-workers AT delorie DOT com Errors-To: nobody AT delorie DOT com X-Mailing-List: djgpp-workers AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk On Mon, 10 Dec 2001, Charles Sandmann wrote: > > So, are you saying that the W2K DPMI host returns a bogus selector > > when we allocate DOS memory? Another W2K bug? ;-) > > No, it's a valid selector, but it's not a valid "internal NTVDM identified > PSP for talking with DPMI/DOSX". Ah, but invoking a program via go32-v2 is not really nesting, as far as the DPMI host and NTVDM are concerned, right? To them, this looks like a single crazy DOS program which just likes to longjmp and lcall too much. So perhaps the cleanup code which tries to fix NTVDM should detect this situation and do nothing. > I wrote the original versions of all that stuff so I remember :-) I think the original version didn't allocate all those additional DOS buffers. Perhaps I'm mistaken. > And it hasn't changed radically ... For go32-v2 we don't ever return > to the go32-v2 image! So, when we set the bogus PSP nothing gets > cleaned up properly. Right, that's the reason. > Now, has dosexec or make been enhanced to run the .exe instead of > calling go32-v2 in cvs Yes, I think so. If so, this is why it doesn't happen with the CVS version: we simply don't invoke go32-v2 (which is a Good Thing by itself).