Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT sources DOT redhat DOT com Delivered-To: mailing list cygwin AT sources DOT redhat DOT com Date: Wed, 19 Sep 2001 07:33:53 -0400 From: Christopher Faylor To: cygwin AT cygwin DOT com Subject: Re: [BUG] cygwin-1.3.3-2 -- making auto-import dlls Message-ID: <20010919073353.E12960@redhat.com> Reply-To: cygwin AT cygwin DOT com Mail-Followup-To: cygwin AT cygwin DOT com References: <20010918225336 DOT G8924 AT redhat DOT com> <010a01c140fd$13e3e310$0200a8c0 AT lifelesswks> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <010a01c140fd$13e3e310$0200a8c0@lifelesswks> User-Agent: Mutt/1.3.21i On Wed, Sep 19, 2001 at 09:20:23PM +1000, Robert Collins wrote: >----- Original Message ----- >From: "Christopher Faylor" >> >> Just in case anyone knows how to do this: >> >> If I could *reserve* but not allocate a fixed amount of memory for the >> cygwin heap on DLL load, this would fix the problem. I could >> conceivably add another 512K of allocated memory to the end of every >> cygwin DLL load but that just didn't seem right. >> >>I played around with a lot of ld script file settings trying to achieve >>this effect but I could never do it. I wouldn't be surprised that >>Windows doesn't allow actually it because the concept requires a >>sophisticated run time loader. > >wrong list .... isn't this a binutils question? :]. Seriously >though, someone there may well be able to answer this. On a related >note, I didn't realise the fix would be so simple. My understanding >was that the variables in question cannot cope with the address's being >different between parent and child, and tacking a 1/2Mb on the end of >the cygwin1.dll address space wouldn't affect that - that area would >get relocated too. perhaps I'm missing something though. The issue here is that a "foreign" DLL is getting loaded into area that cygwin wants to use for its cyheap. It is immaterial where or if the cygwin DLL is relocated. As I said, making the cygwin DLL occupy a fixed location would not solve this particular problem. The ld script file issue that I raised would fix the issue of collisions from a DLL that loads after the cygwin DLL, occupying space that cygwin desperately wants to use. It would not affect the fork relocation issue. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/