Mailing-List: contact cygwin-developers-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-developers-owner AT sourceware DOT cygnus DOT com Delivered-To: mailing list cygwin-developers AT sourceware DOT cygnus DOT com Date: Mon, 28 Feb 2000 00:13:27 -0500 From: Chris Faylor To: Kazuhiro Fujieda Cc: cygwin-developers AT sourceware DOT cygnus DOT com Subject: Re: bash often consume a cpu. Message-ID: <20000228001327.A13901@cygnus.com> Reply-To: cygwin-developers AT sourceware DOT cygnus DOT com References: <20000224151353 DOT A5264 AT cygnus DOT com> <20000225201506 DOT A3484 AT cygnus DOT com> <20000227161541 DOT A7844 AT cygnus DOT com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.1.4i In-Reply-To: <20000227161541.A7844@cygnus.com>; from cgf@cygnus.com on Sun, Feb 27, 2000 at 04:15:41PM -0500 On Sun, Feb 27, 2000 at 04:15:41PM -0500, Chris Faylor wrote: >On Sat, Feb 26, 2000 at 04:58:24PM +0900, Kazuhiro Fujieda wrote: >>>>> On Fri, 25 Feb 2000 20:15:06 -0500 >>>>> Chris Faylor said: >> >>> I've made a few more changes. Please try the next snapshot. >> >>I tried your fixes through `cvs update'. Unfortunately, the >>fixes make nothing about the problem, too. I replaced the strace >>log with one which may be more helpful for your debugging. >> >>In the strace log, I executed the bash command without any >>environment variables nor /etc/{termcap,passwd,group}, and >>simply typed `a\n' three times. The bash didn't consume a cpu >>until two times. After I typed three times, it began to do so. >> >>Anyway, my NT box has two CPU. Multi-processor systems hit >>problems about race conditions more often than uni-processor >>ones. > >I do all of my debugging on an NT with two CPUs and I'm not seeing >this. > >Regardless, I think I see the problem. When things get into this >state, could you try hitting a "CTRL-C" at the bash prompt and see >if the CPU utilization goes back down? FWIW, I've made a few more changes and attempted to eliminate a couple more races in the signal code. Please let me know if this changes any behavior since I can't duplicate what you're seeing. cgf