From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:53094) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XcKDy-0002EG-M3 for qemu-devel@nongnu.org; Thu, 09 Oct 2014 16:22:15 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1XcKDt-0002U0-JJ for qemu-devel@nongnu.org; Thu, 09 Oct 2014 16:22:10 -0400 Received: from mx1.redhat.com ([209.132.183.28]:3057) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XcKDt-0002Tp-Co for qemu-devel@nongnu.org; Thu, 09 Oct 2014 16:22:05 -0400 Message-ID: <5436EE6A.6030705@redhat.com> Date: Thu, 09 Oct 2014 15:22:02 -0500 From: Wei Huang MIME-Version: 1.0 References: <1412711059-12524-1-git-send-email-wei@redhat.com> <543453D9.8020706@redhat.com> <54345843.8070703@redhat.com> <54345CC8.8050806@redhat.com> <54348850.9080808@redhat.com> <5434EC14.5060806@redhat.com> In-Reply-To: <5434EC14.5060806@redhat.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 1/1] target-i386: prevent users from setting threads>1 for AMD CPUs List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini , afaerber@suse.de, ehabkost@redhat.com Cc: qemu-devel@nongnu.org On 10/08/2014 02:47 AM, Paolo Bonzini wrote: > Il 08/10/2014 02:41, Wei Huang ha scritto: >> I am OK with either way. The key question is: should QEMU presents >> CPUIDs strictly as specified by the command line or QEMU can tweak a >> little bit on behalf of end-users? For instance, if end-users say "-smp >> 8,cores=2,threads=2,sockets=2", they meant "two socket, each has two >> 2-hyperthread cores". Current QEMU will convert CPUID as "two socket, >> each has 4 cores". My patch will forbid the tweaking... > > Understood---it actually looks like it was intentional: > > commit 400281af34e5ee6aa9f5496b53d8f82c6fef9319 > Author: Andre Przywara > Date: Wed Aug 19 15:42:42 2009 +0200 > > set CPUID bits to present cores and threads topology > > Controlled by the enhanced -smp option set the CPUID bits to present the > guest the desired topology. This is vendor specific, but (with the exception > of the CMP_LEGACY bit) not conflicting, so we set all bits everytime. > There is no real multithreading support for AMD CPUs, so report cores > instead. > > Signed-off-by: Andre Przywara > Signed-off-by: Anthony Liguori Given that back-ward compatibility is a concern, will the following work? 1. Instead of bailing out, print a warning message (e.g. to log file via error_report) in QEMU. 2. [optional] Eduardo Habkost suggested that we can create a new machine model which more strictly checks threads=n option for AMD. For any existing machine config, we don't force it; but warning message still applies. This is optional because it is a bit over-killed IMO. 3. Gives out a warning in virt-manager as well. This is similar to "Overcomming CPUs will slow down performance" in current virt-manager screen. The message will read "Chosen CPU model doesn't support hyperthreading" or something similar. -Wei > > Paolo >