From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:44693) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gt8yP-0001fM-3W for qemu-devel@nongnu.org; Mon, 11 Feb 2019 05:38:02 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gt8yL-0002bn-CM for qemu-devel@nongnu.org; Mon, 11 Feb 2019 05:37:59 -0500 Date: Mon, 11 Feb 2019 11:37:26 +0100 From: Cornelia Huck Message-ID: <20190211113726.62d2e1d5.cohuck@redhat.com> In-Reply-To: References: <20190208221111.25237-1-walling@linux.ibm.com> <20190211092339.054706bb.cohuck@redhat.com> <4441a5a6-6c26-0e2c-8b95-40af98b2962c@de.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 1/2] s390x/cpumodel: default enable mepoch for z14 and later List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: David Hildenbrand Cc: Christian Borntraeger , Collin Walling , qemu-devel@nongnu.org, qemu-s390x@nongnu.org, pasic@linux.ibm.com, rth@twiddle.net On Mon, 11 Feb 2019 09:55:25 +0100 David Hildenbrand wrote: > On 11.02.19 09:28, Christian Borntraeger wrote: > > > > > > On 11.02.2019 09:23, Cornelia Huck wrote: > >> On Fri, 8 Feb 2019 17:11:10 -0500 > >> Collin Walling wrote: > >> > >>> Latest systems and host kernels support mepoch, which is a > >>> feature that was meant to be supported for z14 GA1 from the > >>> get-go. Let's copy it to the z14 GA1 default CPU model. > >> > >> Do older kernels simply not support it, or was there something actively > >> missing? IOW, is it fine to turn this on with older kernels? > > > > Older kernels will not provide the necessary implementation, so mepoch will not > > work with such older kernel. (So worst case would be that a new qemu+old kernel > > requires to add mepoch=off to -cpu z14) > > Or use a compat machine. > > > The idea of this patch is to update QEMU in accordance with newer kernels to use > > more features by default if people use -cpu z14. > > Agreed, this was the original purpose of the default models. Can "break" > on old kernels, but the compat machines can help for these users, so it > does not really "break". That seems fine then. > > > > > Now that I look a 2nd time: shall we also add > > S390_FEAT_GROUP_MULTIPLE_EPOCH_PTFF ? > > I think so. > > > > > > >> > >> [It's probably all good, although I like to wait for David's ack for > >> any cpu model patches :)] > > I'm fine with this :) > > With S390_FEAT_GROUP_MULTIPLE_EPOCH_PTFF also added > > Reviewed-by: David Hildenbrand Ok, can I please get a v2 with that added as well? > > >> > >>> > >>> Machines s390-ccw-virtio-3.1 and older will retain the old CPU > >>> models and will not provide this bit in the default model. > >>> > >>> Signed-off-by: Collin Walling > >>> --- > >>> hw/s390x/s390-virtio-ccw.c | 1 + > >>> target/s390x/gen-features.c | 1 + > >>> 2 files changed, 2 insertions(+) > >>> > >>> diff --git a/hw/s390x/s390-virtio-ccw.c b/hw/s390x/s390-virtio-ccw.c > >>> index fd9d0b0542..32c5027345 100644 > >>> --- a/hw/s390x/s390-virtio-ccw.c > >>> +++ b/hw/s390x/s390-virtio-ccw.c > >>> @@ -757,6 +757,7 @@ DEFINE_CCW_MACHINE(4_0, "4.0", true); > >>> static void ccw_machine_3_1_instance_options(MachineState *machine) > >>> { > >>> ccw_machine_4_0_instance_options(machine); > >>> + s390_cpudef_featoff_greater(14, 1, S390_FEAT_MULTIPLE_EPOCH); > >>> } > >>> > >>> static void ccw_machine_3_1_class_options(MachineClass *mc) > >>> diff --git a/target/s390x/gen-features.c b/target/s390x/gen-features.c > >>> index 70015eaaf5..d258d67dca 100644 > >>> --- a/target/s390x/gen-features.c > >>> +++ b/target/s390x/gen-features.c > >>> @@ -550,6 +550,7 @@ static uint16_t default_GEN14_GA1[] = { > >>> S390_FEAT_GROUP_MSA_EXT_6, > >>> S390_FEAT_GROUP_MSA_EXT_7, > >>> S390_FEAT_GROUP_MSA_EXT_8, > >>> + S390_FEAT_MULTIPLE_EPOCH, > >>> }; > >>> > >>> /* QEMU (CPU model) features */ > >> > > > >