From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1Myeko-0007Xh-KL for qemu-devel@nongnu.org; Fri, 16 Oct 2009 00:49:26 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1Myekj-0007XF-Ip for qemu-devel@nongnu.org; Fri, 16 Oct 2009 00:49:26 -0400 Received: from [199.232.76.173] (port=55884 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1Myekj-0007XC-EH for qemu-devel@nongnu.org; Fri, 16 Oct 2009 00:49:21 -0400 Received: from 217-162-29-130.static.cablecom.ch ([217.162.29.130]:45971 helo=x41.ch) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1Myeki-0000wa-S3 for qemu-devel@nongnu.org; Fri, 16 Oct 2009 00:49:21 -0400 Received: from [193.201.33.73] (account bolle@geodb.org) by x41.ch (CommuniGate Pro WebUser 5.0.10) with HTTP id 1433162 for qemu-devel@nongnu.org; Fri, 16 Oct 2009 06:49:13 +0200 From: "Bolle" Date: Fri, 16 Oct 2009 06:49:13 +0200 Message-ID: In-Reply-To: References: MIME-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1"; format="flowed" Content-Transfer-Encoding: 8bit Subject: [Qemu-devel] Re: [REASON][PATCH-REQUEST] Starting QEMU by PHP/Apache List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org After some research, I think that Apache with mod_php is blocking the SIGALRM. This is not the fault of QEMU. But maybe, QEMU could check on the start-up, if SIGALRM is enabled and usable. If not, then it can bail out and inform the user about this. I believe, there are other environments where SIGALRM is blocked too. Andreas PS: Sorry for not being able to provide a patch by my self. On Wed, 14 Oct 2009 07:12:56 +0200 "Bolle" wrote: > Hello all > >First, I forgot to say, that I use the QEMU stable >version 11.0 on a Gentoo box. I did some more testings. >The QEMU instance is starting and sometimes the OS starts >to load, but then QEMU stucks and consumes 100% CPU. > > I checked the startup it with strace. The start of QEMU >itself produces the same output in strace, regardless if >started by the PHP CLI or by mod_php from Apache. > > I then attached strace to the QEMU process, and there is >a big difference. > > When starting QEMU with the PHP script from the CLI (as >the same user as the web server), it works and I get >something like that: > ... > select(12, [6 11], [], [], {4, 993369}) = ? >ERESTARTNOHAND (To be restarted) > --- SIGALRM (Alarm clock) @ 0 (0) --- > write(7, "\0"..., 1) = 1 > sigreturn() = ? (mask now >[]) > clock_gettime(CLOCK_MONOTONIC, {2818001, 146485529}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2818001, 146548886}) = 0 > timer_gettime(0, {it_interval={0, 0}, it_value={0, 0}}) >= 0 > timer_settime(0, 0, {it_interval={0, 0}, it_value={0, >250000}}, NULL) = 0 > clock_gettime(CLOCK_MONOTONIC, {2818001, 146758750}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2818001, 146819115}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2818001, 146879549}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2818001, 146938839}) = 0 > --- SIGALRM (Alarm clock) @ 0 (0) --- > write(7, "\0"..., 1) = 1 > sigreturn() = ? (mask now >[]) > ... > > If QEMU has been started by the web server with the same >script and user, I get: > ... > select(11, [5 10], [], [], {3, 767239}) = 0 (Timeout) > clock_gettime(CLOCK_MONOTONIC, {2817617, 818297714}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2817617, 818382772}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2817617, 818444894}) = 0 > timer_gettime(0, {it_interval={0, 0}, it_value={0, 0}}) >= 0 > timer_settime(0, 0, {it_interval={0, 0}, it_value={0, >250000}}, NULL) = 0 > clock_gettime(CLOCK_MONOTONIC, {2817617, 818698197}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2817617, 818759263}) = 0 > timer_gettime(0, {it_interval={0, 0}, it_value={0, >67453}}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2817617, 818894002}) = 0 > clock_gettime(CLOCK_MONOTONIC, {2817617, 818954673}) = 0 > timer_gettime(0, {it_interval={0, 0}, it_value={0, 0}}) >= 0 > timer_settime(0, 0, {it_interval={0, 0}, it_value={0, >250000}}, NULL) = 0 > ... > > There are no SIGALRM executed. What can be the reason >for that ? > > Andreas