From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 26 Mar 2002 21:21:12 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 26 Mar 2002 21:21:01 -0500 Received: from p-nya.swiftel.com.au ([202.154.76.83]:9485 "EHLO mail.robres.com.au") by vger.kernel.org with ESMTP id ; Tue, 26 Mar 2002 21:20:42 -0500 Message-Id: <200203270220.g2R2K7t32470@mail.robres.com.au> Content-Type: text/plain; charset=US-ASCII From: Franco Broi To: linux-kernel@vger.kernel.org Subject: Process hangs Date: Wed, 27 Mar 2002 10:22:23 +0800 X-Mailer: KMail [version 1.3.1] MIME-Version: 1.0 Content-Transfer-Encoding: 7BIT Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi I'm running 2.4.18-rc4 on Alpha SMP. I'm experiencing problems with processes which don't exit properly leaving an entry in the /proc filesystem that hangs any process that tries to read it, ie top, ps etc. I've seen this exact problem mentioned on the mailing list for Intel machines so it doesn't look like an Alpha specific problem. Strace of ps stat("/proc/4019", {st_mode=S_IFDIR|0555, st_size=0, ...}) = 0 open("/proc/4019/stat", O_RDONLY) = 7 read(7, "4019 (tcsh) S 4017 4019 4019 348"..., 511) = 215 close(7) = 0 open("/proc/4019/statm", O_RDONLY) = 7 read(7, "253 107 106 0 106 1 0\n", 511) = 22 close(7) = 0 open("/proc/4019/status", O_RDONLY) = 7 read(7, "Name:\ttcsh\nState:\tS (sleeping)\nT"..., 511) = 456 close(7) = 0 open("/proc/4019/cmdline", O_RDONLY) = 7 read(7, "-csh\0", 2047) = 5 close(7) = 0 open("/proc/4019/environ", O_RDONLY) = -1 EACCES (Permission denied) stat("/proc/10577", {st_mode=S_IFDIR|0555, st_size=0, ...}) = 0 open("/proc/10577/stat", O_RDONLY) = 7 read(7, hangs here. I traced the offending process to a user program which had done a divide by zero. When I ran the program with gdb I got a floating exception, when I ran it without gdb it hung. I then did a few more tests using different shells and different user accounts, the results were inconclusive. It soemtimes worked and sometimes failed. The kernel is patched to run GFS but I don't think this is a factor. Regards Franco