site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com -- Terry Thanks Terry, On 5 Sep 2009, at 10:57, Terry Lambert wrote: However, to answer your specific points: ( I also confirm that I can launch new non-gui processes at will ) but all gui-based application launches fail (with that error code). thanks again, Iain _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... Oh. And top will not report zombies, which is why I specified the ps with the bsd argument list. On Sep 5, 2009, at 4:59 AM, IainS <developer@sandoe-acoustics.co.uk> wrote: building and checking gcc is quite stressful of system resources, it seems. Your process table is full. I am used to the temporary over-run of processes (manifesting as "can't fork"). This current situation is new to me. Any particular reason that the error numbers < -9000 are not apparent on the apple support site? should I file an enhancement request? The most likely cause is the dejagnu test suite not reaping its zombie child processes. Other possibilities are you are running the older buggy dejagnu test suite which closes the master side of the pry before the child process has drained its output, so the individual tests are stuck waiting for someone to reopen it and read their output so their closes can complete, or you have some other runaway software, or you have software that calls out to userspace from a kernel next and deadlocks itself when it makes a system call because it did not make itself immune to its own outcalls (usually firewall or antivirus software). I'm working with gcc-svn-trunk and there are some code-generation hassles right now so it's possible that the nasty is a result of those. I have the latest XCode tools, latest released dejagnu/expect and the correct gnu toolset pre-requisites for the build. There is scope for other influences on the i686 (8core Xeon) - but the powerpc (Quad G5) is a vanilla install of 10.5. ** I've repeated the effect but I can't see the bogey using either ps or activity monitor... top shows 85-ish processes, between 3 & 5 running 2 stuck and 80-ish sleeping - 320 ish threads. there is one suspicious beast (a gdb related check that hangs around) .. but killing that does not resolve the problem. one other observation is that when I log out and back in again - it is evident that the logout hasn't properly killed all subordinate processes (e.g. an ssh session left hanging.. and an iTunes instance left hanging w/out it's gui). This email sent to site_archiver@lists.apple.com