From mboxrd@z Thu Jan 1 00:00:00 1970 From: marc.zyngier@arm.com (Marc Zyngier) Date: Mon, 22 Jun 2015 07:47:59 +0100 Subject: [PATCH] arm64/kvm: Add generic v8 KVM target In-Reply-To: <55847C1F.1090606@codeaurora.org> References: <1434531646-4873-1-git-send-email-suzuki.poulose@arm.com> <55847C1F.1090606@codeaurora.org> Message-ID: <20150622074759.01d677a3@arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Fri, 19 Jun 2015 21:31:27 +0100 Timur Tabi wrote: > On 06/17/2015 04:00 AM, Suzuki K. Poulose wrote: > > &genericv8_target_table); > > kvm_register_target_sys_reg_table(KVM_ARM_TARGET_XGENE_POTENZA, > > &genericv8_target_table); > > + > > kvm_register_target_sys_reg_table(KVM_ARM_TARGET_GENERIC_V8, > > + &genericv8_target_table); > > > > Shouldn't you also remove all of the previous lines that return > &genericv8_target_table? No. KVM_ARM_TARGET_* are part of the uapi, and existing userspace knows about them. You can't just drop them. What you *could* do would be to map the existing targets to the generic one in a way that leaves userspace blissfully unaware of the underlying change (for example, KVM_ARM_PREFERRED_TARGET should still return KVM_ARM_TARGET_XGENE_POTENZA on an XGene platform, and KVM_ARM_VCPU_INIT should still be accept it). What would be the gain of such a mapping? Added complexity, hardly anything else. We're much better off considering the generic target as a new one, both inside the kernel and in the view we export to userspace. Thanks, M. -- Jazz is not dead. It just smells funny.