From mboxrd@z Thu Jan 1 00:00:00 1970 From: sudeep.holla@arm.com (Sudeep Holla) Date: Fri, 3 Feb 2017 10:44:38 +0000 Subject: [PATCH v2 3/3] arm64: kvm: add support for the extended 64bit ccsidr In-Reply-To: <85bd7758-e0a3-f274-ba7a-db080e66d97c@arm.com> References: <1485793527-18028-1-git-send-email-sudeep.holla@arm.com> <1485793527-18028-3-git-send-email-sudeep.holla@arm.com> <85bd7758-e0a3-f274-ba7a-db080e66d97c@arm.com> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi Marc, On 01/02/17 16:02, Marc Zyngier wrote: [...] > > I'm a bit worried about this patch. If we snapshot a VM on a 32bit > CCSIDR system, and restore it on a 64bit CSSIDR system (or the reverse), > what happens? My hunch is that we cannot restore the VM properly. > I agree. I had a look at QEMU as you suggested offline. Looks like QEMU emulate these registers with predefined values for each core type. Also it looks like it's not using the existing DEMUX_ID_CCSIDR > Now, I'm questioning the need for having those altogether, as we do a > lot of work to prevent the guest from actually using that geometry (and > on a big-little system, this hardly works). > If we can conclude there are no users for DEMUX_ID_CCSIDR, we can remove it all together instead of introducing new one for 64bit. Are there any other users of these interface provided by KVM after from kvmtool and qemu ? -- Regards, Sudeep