Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT sourceware DOT cygnus DOT com Delivered-To: mailing list cygwin AT sourceware DOT cygnus DOT com From: Chris Faylor Date: Thu, 21 Oct 1999 10:54:33 -0400 To: Earnie Boyd Cc: Steve Jorgensen , cygwin AT sourceware DOT cygnus DOT com Subject: Re: Mmap problem Message-ID: <19991021105433.A4912@cygnus.com> Reply-To: cygwin AT sourceware DOT cygnus DOT com Mail-Followup-To: Earnie Boyd , Steve Jorgensen , cygwin AT sourceware DOT cygnus DOT com References: <19991021120802 DOT 18332 DOT rocketmail AT web117 DOT yahoomail DOT com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.95.6i In-Reply-To: <19991021120802.18332.rocketmail@web117.yahoomail.com>; from Earnie Boyd on Thu, Oct 21, 1999 at 05:08:02AM -0700 On Thu, Oct 21, 1999 at 05:08:02AM -0700, Earnie Boyd wrote: >--- Steve Jorgensen wrote: >> >>I'm trying to port Khoros Pro 2001 to NT using cygwin, and am having a >>problem with the mmap call. >> >>Khoros has a transport abstraction that makes using mmap, tcpip, file, >>pipes, or memory all look like a standard unix file, allowing users to >>use our k* file i/o calls to read and write to each different kind of >>transport. Anyway, the mmap transport, in order to avoid excessive >>resizing, forces the underlying mmap'ed file to grow more than the >>write call needs. At close time, we ftruncate the file back to the >>size it's supposed to be. The problem is that ftruncate is returning >>with an access error. This works on all the unix systems we've ported >>to. >> >>I've tried the ftruncate both before and after the munmap call with the >>same results. Any ideas what else I can try, or am I just out of luck? > >This is an issue with the underlying file system. Assuming that it can >be done, you must have all accesses to the file closed. Win32 will not >allow such changes to an open file, even if you are the one that has it >open. I think I remember that there is coding in cygwin to queue >operations on a file that fail due to an open fd to the file, so this >could be a bug in that related coding. There is a queue but it is applicable only to unlinking files. -chris -- Want to unsubscribe from this list? Send a message to cygwin-unsubscribe AT sourceware DOT cygnus DOT com