From: cgf AT bbc DOT com (Christopher Faylor) Subject: Re: Fork Speedup? 2 Feb 1998 06:14:30 -0800 Message-ID: References: <01BD2FCE DOT E270E5A0 AT gater DOT krystalbank DOT msk DOT ru> Reply-To: gnu-win32-developers AT cygnus DOT com To: cygwin32-developers AT cygnus DOT com In article <01BD2FCE DOT E270E5A0 AT gater DOT krystalbank DOT msk DOT ru>, Sergey Okhapkin wrote: >Christopher Faylor wrote: >> It just occurred to me that fork might be a little faster if the >> sequence of events was something like: >> > >I've tried this some time ago and had a lot of troubles with cygwin.dll >initialization sequence (mainly with u->self pointer). Probably, with your >new u->self initialization method it will be much easier now if this >initialization will be moved from pinfo_init() to some earlier place. But >we shouldn't change fork mechanism now till dllinit patch will be applied. That's what got me thinking, actually. I know that your new mechanism changes things in the fork code. The last time I looked, it actually had a few more "sychronization events" between parent and child than previously. -- http://www.bbc.com/ cgf AT bbc DOT com "Strange how unreal VMS=>UNIX Solutions Boston Business Computing the real can be."