Message-Id: <200503182308.j2IN82kd006191@delorie.com> Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT cygwin DOT com Mail-Followup-To: cygwin AT cygwin DOT com Delivered-To: mailing list cygwin AT cygwin DOT com Resent-date: Fri, 18 Mar 2005 15:07:34 -0800 Date: Fri, 18 Mar 2005 15:05:28 -0800 Resent-from: David Rothenberger From: David Rothenberger Subject: Re: ssh-add -l hanging with 20050316 snapshot In-reply-to: <200503171842.j2HIg01J010269@mail.daveroth.dyndns.org> Resent-to: cygwin To: cygwin AT cygwin DOT com Resent-message-id: <423B5F36 DOT 6050408 AT acm DOT org> MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7bit Resent-User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) References: <200503171842 DOT j2HIg01J010269 AT mail DOT daveroth DOT dyndns DOT org> User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-IsSubscribed: yes Reply-To: cygwin AT cygwin DOT com On 3/17/2005 10:38 AM, David Rothenberger wrote: > I've had keychain hang twice now while running "ssh-add -l". It happened > with an older snapshot and again this morning with a DLL I built from > CVS yesterday that corresponds to the 20050316 snapshot. I've been able to reproduce this hang at will with the 20050316 snapshot. I start an ssh-agent process, add a key to it, and then run "ssh-add -l" in a loop in two windows. This generally triggers the hang in a few minutes. I've captured strace output from ssh-agent and the two ssh-add processes. The output is available here: http://mysite.verizon.net/res00a7j/ssh-agent.zip http://mysite.verizon.net/res00a7j/ssh-add.1.strace.txt http://mysite.verizon.net/res00a7j/ssh-add.1.strace.old.txt http://mysite.verizon.net/res00a7j/ssh-add.2.strace.txt http://mysite.verizon.net/res00a7j/ssh-add.2.strace.old.txt The ssh-agent strace out is compressed because it's 2.4MB uncompressed. The uncompressed version is here if you're brave: http://mysite.verizon.net/res00a7j/ssh-agent.strace.txt The strace's from the two ssh-add processes that hung are ssh-add.1.strace.txt and ssh-add.2.strace.txt. The "*.old.txt" traces are from the previous successful invocation of ssh-add in the two loops. I used this simple script to capture the ssh-add traces: #!/bin/bash num=$1 while true; do [ -f /c/temp/ssh-add.$num.strace ] && \ mv /c/temp/ssh-add.$num.strace /c/temp/ssh-add.$num.strace.2 date strace -o /c/temp/ssh-add.$num.strace ssh-add -l done I ran it in two windows as "./ssh-add-loop.sh 1" and "./ssh-add-loop.sh 2". Two things caught my eye when looking at the traces. In ssh-add.2.strace.txt, I saw this: 252 122112 [main] ssh-add 4120 fhandler_socket::signal_secret_event: signaled secret_event 121 122233 [main] ssh-add 4120 fhandler_socket::eid_pipe_name: \\.\pipe\cygwin-unix-/tmp/ssh-wxIlND2024/agent.2024 2556 124789 [main] ssh-add 4120 fhandler_socket::connect: Receiving eid credentials failed: Win32 error 231 although the eid credentials were received successfully by ssh-add.1. Also, at the end of the traces for the two hung ssh-add processes and ssh-agent, I saw this. 7963651 8097324 [sig] ssh-add 4120 pinfo::init: execed process windows pid 4884, cygwin pid 5728 182 8097506 [sig] ssh-add 4120 pinfo::init: execed process windows pid 5892, cygwin pid 3552 115 8097621 [sig] ssh-add 4120 pinfo::init: execed process windows pid 3104, cygwin pid 4996 This repeated many times (180 times for ssh-add.2). This occurred when I interrupted the processes with CTRL-C. -- David Rothenberger spammer? -> spam AT daveroth DOT dyndns DOT org GPG/PGP: 0x7F67E734, C233 365A 25EF 2C5F C8E1 43DF B44F BA26 7F67 E734 kernel, n.: A part of an operating system that preserves the medieval traditions of sorcery and black art. -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/