kvm: arm64: Configure VTCR_EL2 per VM

Add support for setting the VTCR_EL2 per VM, rather than hard
coding a value at boot time per CPU. This would allow us to tune
the stage2 page table parameters per VM in the later changes.

We compute the VTCR fields based on the system wide sanitised
feature registers, except for the hardware management of Access
Flags (VTCR_EL2.HA). It is fine to run a system with a mix of
CPUs that may or may not update the page table Access Flags.
Since the bit is RES0 on CPUs that don't support it, the bit
should be ignored on them.

Suggested-by: Marc Zyngier <[email protected]>
Acked-by: Christoffer Dall <[email protected]>
Signed-off-by: Suzuki K Poulose <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>

arch/arm64/include/asm/kvm_arm.h

@@ -135,8 +135,7 @@
  * 40 bits wide (T0SZ = 24).  Systems with a PARange smaller than 40 bits are
  * not known to exist and will break with this configuration.
  *
- * VTCR_EL2.PS is extracted from ID_AA64MMFR0_EL1.PARange at boot time
- * (see hyp-init.S).
+ * The VTCR_EL2 is configured per VM and is initialised in kvm_arm_config_vm().
  *
  * Note that when using 4K pages, we concatenate two first level page tables
  * together. With 16K pages, we concatenate 16 first level page tables.

arch/arm64/include/asm/kvm_asm.h

@@ -72,8 +72,6 @@ extern void __vgic_v3_init_lrs(void);
 
 extern u32 __kvm_get_mdcr_el2(void);
 
-extern u32 __init_stage2_translation(void);
-
 /* Home-grown __this_cpu_{ptr,read} variants that always work at HYP */
 #define __hyp_this_cpu_ptr(sym)						\
 	({								\

arch/arm64/include/asm/kvm_host.h

@@ -61,11 +61,13 @@ struct kvm_arch {
 	u64    vmid_gen;
 	u32    vmid;
 
-	/* 1-level 2nd stage table, protected by kvm->mmu_lock */
+	/* stage2 entry level table */
 	pgd_t *pgd;
 
 	/* VTTBR value associated with above pgd and vmid */
 	u64    vttbr;
+	/* VTCR_EL2 value for this VM */
+	u64    vtcr;
 
 	/* The last vcpu id that ran on each physical CPU */
 	int __percpu *last_vcpu_ran;
@@ -442,10 +444,12 @@ int kvm_arm_vcpu_arch_has_attr(struct kvm_vcpu *vcpu,
 
 static inline void __cpu_init_stage2(void)
 {
-	u32 parange = kvm_call_hyp(__init_stage2_translation);
+	u32 ps;
 
-	WARN_ONCE(parange < 40,
-		  "PARange is %d bits, unsupported configuration!", parange);
+	/* Sanity check for minimum IPA size support */
+	ps = id_aa64mmfr0_parange_to_phys_shift(read_sysreg(id_aa64mmfr0_el1) & 0x7);
+	WARN_ONCE(ps < 40,
+		  "PARange is %d bits, unsupported configuration!", ps);
 }
 
 /* Guest/host FPSIMD coordination helpers */

arch/arm64/include/asm/kvm_hyp.h

@@ -161,6 +161,7 @@ void __noreturn __hyp_do_panic(unsigned long, ...);
  */
 static __always_inline void __hyp_text __load_guest_stage2(struct kvm *kvm)
 {
+	write_sysreg(kvm->arch.vtcr, vtcr_el2);
 	write_sysreg(kvm->arch.vttbr, vttbr_el2);
 }
 

arch/arm64/kvm/hyp/Makefile

@@ -19,7 +19,6 @@ obj-$(CONFIG_KVM_ARM_HOST) += switch.o
 obj-$(CONFIG_KVM_ARM_HOST) += fpsimd.o
 obj-$(CONFIG_KVM_ARM_HOST) += tlb.o
 obj-$(CONFIG_KVM_ARM_HOST) += hyp-entry.o
-obj-$(CONFIG_KVM_ARM_HOST) += s2-setup.o
 
 # KVM code is run at a different exception code with a different map, so
 # compiler instrumentation that inserts callbacks or checks into the code may

arch/arm64/kvm/reset.c

@@ -26,6 +26,7 @@
 
 #include <kvm/arm_arch_timer.h>
 
+#include <asm/cpufeature.h>
 #include <asm/cputype.h>
 #include <asm/ptrace.h>
 #include <asm/kvm_arm.h>
@@ -134,9 +135,43 @@ int kvm_reset_vcpu(struct kvm_vcpu *vcpu)
 	return kvm_timer_vcpu_reset(vcpu);
 }
 
+/*
+ * Configure the VTCR_EL2 for this VM. The VTCR value is common
+ * across all the physical CPUs on the system. We use system wide
+ * sanitised values to fill in different fields, except for Hardware
+ * Management of Access Flags. HA Flag is set unconditionally on
+ * all CPUs, as it is safe to run with or without the feature and
+ * the bit is RES0 on CPUs that don't support it.
+ */
 int kvm_arm_config_vm(struct kvm *kvm, unsigned long type)
 {
+	u64 vtcr = VTCR_EL2_FLAGS;
+	u32 parange, phys_shift;
+
 	if (type)
 		return -EINVAL;
+
+	parange = read_sanitised_ftr_reg(SYS_ID_AA64MMFR0_EL1) & 7;
+	if (parange > ID_AA64MMFR0_PARANGE_MAX)
+		parange = ID_AA64MMFR0_PARANGE_MAX;
+	vtcr |= parange << VTCR_EL2_PS_SHIFT;
+
+	phys_shift = id_aa64mmfr0_parange_to_phys_shift(parange);
+	if (phys_shift > KVM_PHYS_SHIFT)
+		phys_shift = KVM_PHYS_SHIFT;
+	vtcr |= VTCR_EL2_T0SZ(phys_shift);
+
+	/*
+	 * Enable the Hardware Access Flag management, unconditionally
+	 * on all CPUs. The features is RES0 on CPUs without the support
+	 * and must be ignored by the CPUs.
+	 */
+	vtcr |= VTCR_EL2_HA;
+
+	/* Set the vmid bits */
+	vtcr |= (kvm_get_vmid_bits() == 16) ?
+		VTCR_EL2_VS_16BIT :
+		VTCR_EL2_VS_8BIT;
+	kvm->arch.vtcr = vtcr;
 	return 0;
 }