From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:60685) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fBns0-0006GW-5L for qemu-devel@nongnu.org; Thu, 26 Apr 2018 16:52:01 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1fBnrz-00089Q-F5 for qemu-devel@nongnu.org; Thu, 26 Apr 2018 16:52:00 -0400 Received: from mail-pf0-x22b.google.com ([2607:f8b0:400e:c00::22b]:35184) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1fBnrz-00088T-92 for qemu-devel@nongnu.org; Thu, 26 Apr 2018 16:51:59 -0400 Received: by mail-pf0-x22b.google.com with SMTP id j5so19296267pfh.2 for ; Thu, 26 Apr 2018 13:51:58 -0700 (PDT) References: <20180424152405.10304-1-alex.bennee@linaro.org> <20180424152405.10304-44-alex.bennee@linaro.org> <84e43559-54ee-315b-24c9-c7e62e5a6d64@linaro.org> <87604el3g1.fsf@linaro.org> From: Richard Henderson Message-ID: Date: Thu, 26 Apr 2018 10:51:52 -1000 MIME-Version: 1.0 In-Reply-To: <87604el3g1.fsf@linaro.org> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit Subject: Re: [Qemu-devel] [PATCH v3 43/46] tests/tcg/Makefile: update to be called from Makefile.target List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?UTF-8?Q?Alex_Benn=c3=a9e?= Cc: peter.maydell@linaro.org, cota@braap.org, famz@redhat.com, berrange@redhat.com, f4bug@amsat.org, balrogg@gmail.com, aurelien@aurel32.net, agraf@suse.de, qemu-devel@nongnu.org On 04/25/2018 11:30 PM, Alex Bennée wrote: > > Richard Henderson writes: > >> On 04/24/2018 05:24 AM, Alex Bennée wrote: >>> +run-%: % >>> + $(call quiet-command, $(QEMU) $< > $<.out, "TEST", "$< on $(TARGET_NAME)") >> >> I've just had an x86_64 guest test run for 70 minutes. We need to limit the >> amount of time spent here in some way, with excessive time reported as test >> failure. >> >> One potential way would be to add "ulimit -t $TIMEOUT" to the subshell before >> running qemu. A default TIMEOUT might be e.g. 15 seconds. I'm not sure what >> we should expect for the largest test on the smallest hosts... > > Sounds reasonable - test-i386-fprem is probably the longest - I'm not > sure it needs to be so exhaustive. Better, not via ulimit but via something that measures wall time. That way we don't stall on e.g. a locked mutex when the child isn't spinning in a loop as it was here. I'd guess a small python script would be most maintainable... r~