Mailing-List: contact cygwin-developers-help AT cygwin DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-developers-owner AT cygwin DOT com Delivered-To: mailing list cygwin-developers AT cygwin DOT com Message-ID: <3D8AFE12.2040403@netscape.net> Date: Fri, 20 Sep 2002 06:53:06 -0400 From: Nicholas Wourms User-Agent: Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US; rv:1.0.1) Gecko/20020823 MultiZilla/v1.1.22 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Pierre A. Humblet" CC: cygwin-developers AT cygwin DOT com Subject: Re: Random freezes on WinME References: <3 DOT 0 DOT 5 DOT 32 DOT 20020919221522 DOT 0081ec70 AT mail DOT attbi DOT com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Pierre A. Humblet wrote: > I would be interested to have Nicholas experience about this. > For several weeks (don't recall when it started), I have gotten > random freezes while making from cvs. > > ps reveals: > > 33984103 1 33984103 4260983193 con 510 21:40:51 /c/PROGRAM > FILES/CYGWIN/BIN/RXVT > 84280863 33984103 84280863 4210684593 0 510 21:40:52 /c/PROGRAM > FILES/CYGWIN/BIN/BASH > 36584127 84280863 36584127 4210563805 0 510 21:57:34 /c/PROGRAM > FILES/CYGWIN/BIN/MAKE > 84403975 36584127 36584127 4210563321 0 510 21:58:05 /c/PROGRAM > FILES/CYGWIN/BIN/SH > 85036611 84403975 36584127 4209930685 0 510 21:58:05 /c/PROGRAM > FILES/CYGWIN/BIN/SH > 84943659 85036611 36584127 4210023637 0 510 21:58:05 > > Is it just me, or is it ME? AFAICT its ME, but the same might apply to Win9x (dunno tho). > Any advice on debugging this? Well I'm not certain what's to blame but here's what I do know: 1)This problem occurs during any complicated build with multiple subdirectories and configure scripts (i.e. binutils, gcc, cygwin, etc.). 2)It seems that the last process in the list is racing when it tries to terminate, thus it prevents the other processes from continuing. Using taskman to kill it will allow the build to continue. 3)After killing the offending process, the process is not "really" killed. Doing a ps in Cygwin will list the process as and utilities like process explorer will list it as an unknown thread. It will not show up in taskman and it cannot be killed by any utility I know of. It won't free up the resources it used, so the only solution is to reboot. 15% of the times I reboot results in an unsucessful shutdown [blinking cursor effect]. 4)Allowing too many of these "defunct" process to build up will result in WinME segfaulting kernel32 and crashing. 5)The condition has exist for ~1.5 months now, but happened rarely up until about the time Chris changed Cygwin's internal threading mechanism. 6)Conrad suggests attaching gdb to the final process to see what it is doing. I've been unable to do that b/c WinME bsod's on any attempt to start gdb. Sorry I couldn't be more helpful and I wasn't reporting it b/c I was still investigating. I also was worried that it had something to do with my configuration. Cheers, Nicholas