From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:60547) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YINJ3-0002Wb-SV for qemu-devel@nongnu.org; Mon, 02 Feb 2015 15:09:14 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YINIw-00079J-Fq for qemu-devel@nongnu.org; Mon, 02 Feb 2015 15:09:13 -0500 Received: from mx1.redhat.com ([209.132.183.28]:40622) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YINIw-00079A-83 for qemu-devel@nongnu.org; Mon, 02 Feb 2015 15:09:06 -0500 Date: Mon, 2 Feb 2015 18:08:45 -0200 From: Marcelo Tosatti Message-ID: <20150202200845.GA13011@amt.cnet> References: <20150129203728.6cb9271e@crunchbang> <54CABD0A.1090207@redhat.com> <54CB34CB.20805@siemens.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <54CB34CB.20805@siemens.com> Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] QEMU and Real Time OS List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jan Kiszka Cc: Marc =?iso-8859-1?Q?Mar=ED?= , Paolo Bonzini , KONRAD =?iso-8859-1?Q?Fr=E9d=E9ric?= , qemu-devel , Luiz Capitulino On Fri, Jan 30, 2015 at 08:37:47AM +0100, Jan Kiszka wrote: > On 2015-01-30 00:06, Paolo Bonzini wrote: > >=20 > >=20 > > On 29/01/2015 20:37, Marc Mar=ED wrote: > >> Is this an expected behaviour? I can't see why. > >> > >> I'd like to know if there is a certain reason why it doesn't work. O= r > >> if it should work and the problem is too much I/O overhead. Or any > >> other hint to understand it. > >=20 > > It is due to latencies in the host. You need at least to use preempt= -rt > > kernels in the host as well. >=20 > That alone won't help much. You also need to fine-tune the guest to > avoid running into QEMU locks that continuously synchronizes the guest > on things like VGA or disk I/O emulation. >=20 > When using KVM, thus being able to run VCPUs widely independent of each > other and the device models, you need to push cyclictest on an isolated > second virtual CPU of the guest. Luiz and Marcelo can probably confirm > this based on their ongoing experiments. Yes, we have achieved low latencies by using a dedicated pCPU for a guest vCPU.=20 This also avoids iothread <-> guest vCPU -RT priority issues. > With TCG, we would first of all have to make it true SMP and independen= t > of the I/O device lock. That's what Frederic is working on [1]. >=20 > Jan >=20 > [1] http://permalink.gmane.org/gmane.comp.emulators.qemu/314406 >=20 > --=20 > Siemens AG, Corporate Technology, CT RTC ITP SES-DE > Corporate Competence Center Embedded Linux