Patch "KVM: PPC: Book3S HV: Context-switch EBB registers properly" has been added to the 4.9-stable tree

This is a note to let you know that I've just added the patch titled

KVM: PPC: Book3S HV: Context-switch EBB registers properly

to the 4.9-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
kvm-ppc-book3s-hv-context-switch-ebb-registers-properly.patch
and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

From: Paul Mackerras <[email protected]>
Date: Tue, 6 Jun 2017 16:47:22 +1000
Subject: KVM: PPC: Book3S HV: Context-switch EBB registers properly

From: Paul Mackerras <[email protected]>

commit ca8efa1df1d15a1795a2da57f9f6aada6ed6b946 upstream.

This adds code to save the values of three SPRs (special-purpose registers) used by userspace to control event-based branches (EBBs), which are essentially interrupts that get delivered directly to userspace. These registers are loaded up with guest values when entering the guest, and their values are saved when exiting the guest, but we were not saving the host values and restoring them before going back to userspace.

On POWER8 this would only affect userspace programs which explicitly request the use of EBBs and also use the KVM_RUN ioctl, since the only source of EBBs on POWER8 is the PMU, and there is an explicit enable bit in the PMU registers (and those PMU registers do get properly context-switched between host and guest). On POWER9 there is provision for externally-generated EBBs, and these are not subject to the control in the PMU registers.

Since these registers only affect userspace, we can save them when we first come in from userspace and restore them before returning to userspace, rather than saving/restoring the host values on every guest entry/exit. Similarly, we don't need to worry about their values on offline secondary threads since they execute in the context of the idle task, which never executes in userspace.

---
 arch/powerpc/kvm/book3s_hv.c |   15 +++++++++++++++
 1 file changed, 15 insertions(+)

@@ -2807,6 +2807,7 @@ static int kvmppc_vcpu_run_hv(struct kvm
 {
 	int r;
 	int srcu_idx;
+	unsigned long ebb_regs[3] = {};	/* shut up GCC */
 
 	if (!vcpu->arch.sane) {
 		run->exit_reason = KVM_EXIT_INTERNAL_ERROR;

@@ -2855,6 +2856,13 @@ static int kvmppc_vcpu_run_hv(struct kvm
 
 	flush_all_to_thread(current);
 
+	/* Save userspace EBB register values */
+	if (cpu_has_feature(CPU_FTR_ARCH_207S)) {
+		ebb_regs[0] = mfspr(SPRN_EBBHR);
+		ebb_regs[1] = mfspr(SPRN_EBBRR);
+		ebb_regs[2] = mfspr(SPRN_BESCR);
+	}
+
 	vcpu->arch.wqp = &vcpu->arch.vcore->wq;
 	vcpu->arch.pgdir = current->mm->pgd;
 	vcpu->arch.state = KVMPPC_VCPU_BUSY_IN_HOST;

@@ -2877,6 +2885,13 @@ static int kvmppc_vcpu_run_hv(struct kvm
 			r = kvmppc_xics_rm_complete(vcpu, 0);
 	} while (is_kvmppc_resume_guest(r));
 
+	/* Restore userspace EBB register values */
+	if (cpu_has_feature(CPU_FTR_ARCH_207S)) {
+		mtspr(SPRN_EBBHR, ebb_regs[0]);
+		mtspr(SPRN_EBBRR, ebb_regs[1]);
+		mtspr(SPRN_BESCR, ebb_regs[2]);
+	}
+
  out:
 	vcpu->arch.state = KVMPPC_VCPU_NOTREADY;
 	atomic_dec(&vcpu->kvm->arch.vcpus_running);

Patches currently in stable-queue which might be from [email protected] are

queue-4.9/kvm-ppc-book3s-hv-preserve-userspace-htm-state-properly.patch queue-4.9/kvm-ppc-book3s-hv-context-switch-ebb-registers-properly.patch