From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:43195) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gtg2v-0001cu-S1 for qemu-devel@nongnu.org; Tue, 12 Feb 2019 16:56:54 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gtg2v-0006AJ-14 for qemu-devel@nongnu.org; Tue, 12 Feb 2019 16:56:53 -0500 From: Laurent Vivier Date: Tue, 12 Feb 2019 22:48:23 +0100 Message-Id: <20190212214827.30543-1-lvivier@redhat.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Subject: [Qemu-devel] [RFC 0/4] numa, spapr: add thread-id in the possible_cpus list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: qemu-ppc@nongnu.org, Marcel Apfelbaum , Laurent Vivier , Paolo Bonzini , Igor Mammedov , Thomas Huth , David Gibson , Eduardo Habkost There are inconsistencies between the command line using "-numa node,cpus=3DXX" and what is checked internally: the XX is supposed to be a CPU number, but for SPAPR it's taken as a core number, ignoring the threads. (See the description message of PATCH 1 for more details) This series fixes this problem by introducing the threads in the possible_cpus list instead of only the cores. To avoid inconsistent topology, it doesn't allow anymore to have an incomplete CPU NUMA config on the command line (there was already a message announcing it will be absoleted for 2 years). Laurent Vivier (4): numa,spapr: add thread-id in the possible_cpus list numa: exit on incomplete CPU mapping numa: move cpu_slot_to_string() upper in the function numa: check threads of the same core are on the same node hw/core/machine.c | 115 ++++++++++++++++++++++++++-------------------- hw/ppc/spapr.c | 33 ++++++------- tests/numa-test.c | 24 +--------- 3 files changed, 81 insertions(+), 91 deletions(-) --=20 2.20.1