From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:51417) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1QbCTb-0000w9-2K for qemu-devel@nongnu.org; Mon, 27 Jun 2011 10:07:48 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1QbCTX-0001LO-4p for qemu-devel@nongnu.org; Mon, 27 Jun 2011 10:07:46 -0400 Received: from mail-ww0-f53.google.com ([74.125.82.53]:47402) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1QbCTW-0001LE-RA for qemu-devel@nongnu.org; Mon, 27 Jun 2011 10:07:42 -0400 Received: by wwf26 with SMTP id 26so4168441wwf.10 for ; Mon, 27 Jun 2011 07:07:41 -0700 (PDT) Sender: Paolo Bonzini Message-ID: <4E088EA8.50000@redhat.com> Date: Mon, 27 Jun 2011 16:07:36 +0200 From: Paolo Bonzini MIME-Version: 1.0 References: <1307617825-17149-1-git-send-email-pbonzini@redhat.com> In-Reply-To: <1307617825-17149-1-git-send-email-pbonzini@redhat.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 0/2] iothread improvements for Mac OS X List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Alexander Graf , =?ISO-8859-1?Q?Andreas_F=E4rber?= , Alexandre Raymond Cc: qemu-devel@nongnu.org On 06/09/2011 01:10 PM, Paolo Bonzini wrote: > These are two old patches that I never submitted because I didn't really > think they were useful except as cleanups. Recently, however, Alex Graf > mentioned some problems that Mac OS X has with iothread, and they sounded > to me like they were related to these patches. And quite surprisingly, > both of them were fixing bugs! > > Mac OS X still has problems with iothread according to Alex's testing > (Linux times out in libata, and reactos likewise hangs early on I/O), > but at least the patches fix deadlocks and keep a responsive UI. > > Paolo Bonzini (2): > iothread: replace fair_mutex with a condition variable > qemu-timer: change unix timer to dynticks > > cpus.c | 24 +++++++++--------------- > qemu-timer.c | 40 ++++++++++++++++++++++++++++------------ > 2 files changed, 37 insertions(+), 27 deletions(-) > Ping? 1/2 is probably somehow working around the sigmask problem fixed by Alexandre (Mac people, can you check?), but it is way more readable than the fair_mutex IMNSHO. I would be surprised if 2/2 also turned out to be a workaround, but even if this were the case, it makes CPU usage lower. Paolo