From owner-freebsd-stable Mon Aug 14 5:32:26 2000 Delivered-To: freebsd-stable@freebsd.org Received: from kirk.dsl.visi.com (kirk.dsl.visi.com [209.98.248.172]) by hub.freebsd.org (Postfix) with ESMTP id DA2F237B787 for ; Mon, 14 Aug 2000 05:32:22 -0700 (PDT) (envelope-from dgl@visi.com) Received: from localhost (dgl@localhost) by kirk.dsl.visi.com (8.9.3/8.9.3) with ESMTP id HAA11032 for ; Mon, 14 Aug 2000 07:32:21 -0500 (CDT) (envelope-from dgl@visi.com) X-Authentication-Warning: kirk.dsl.visi.com: dgl owned process doing -bs Date: Mon, 14 Aug 2000 07:32:21 -0500 (CDT) From: Doug Lee To: freebsd-stable@freebsd.org Subject: pty race condition in 4.1? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG I was compiling a kernel under 4.1-20000806-STABLE inside of `script' inside of `screen' when the `make depend' process blocked unexpectedly: status I+, wchan "ttywri" (from `ps -axl'). The parent `script' process stopped similarly: status I+, wchan "ptcin." I traced "ptcin" to tty_pty.c in /usr/src/sys/kern--a tsleep call with that label--unless I missed another reference ("traced" here means `grep', not gdb/backtrace, which I didn't think of in time). I couldn't find a way to unwedge the process without killing something, so I finally typed ^C in the screen window running script/make, and script exited and left the make process running with no place to send output. The wedge happened in the middle of a HUGE (14,400 or so-character) line from `make depend' listing C files; the cursor parked itself just after "../../miscfs/procfs/procfs_regs.c," which is about 6,144 characters into the line. I subsequently killed off the whole kernel build, did a `make clean' in the compile subdirectory for my particular kernel, and started over at `config.' The process ran fine this time. I realize this is a complicated issue, what with script and screen and make and its descendants all running at once; but I'm wondering if anyone has seen anything like this. I can sure provide further system information if requested, but I'm not sure what more would be helpful, nor do I know of a way to predictably reproduce the problem. I was running the generic kernel that came with the snapshot when this happened. `script' is as it came with the snapshot; `screen' is out of ports with no local tweaks unless my /etc/make.conf having "CFLAGS=-pipe" in it affects that particular port's build process. -- Doug Lee dgl@visi.com http://www.visi.com/~dgl To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message