LoongArch: KVM: Add Binary Translation extension support

Loongson Binary Translation (LBT) is used to accelerate binary translation,
which contains 4 scratch registers (scr0 to scr3), x86/ARM eflags (eflags)
and x87 fpu stack pointer (ftop).

Like FPU extension, here a lazy enabling method is used for LBT. the LBT
context is saved/restored on the vcpu context switch path.
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>

arch/loongarch/include/asm/kvm_host.h

@@ -133,8 +133,9 @@ enum emulation_result {
 #define KVM_LARCH_FPU		(0x1 << 0)
 #define KVM_LARCH_LSX		(0x1 << 1)
 #define KVM_LARCH_LASX		(0x1 << 2)
-#define KVM_LARCH_SWCSR_LATEST	(0x1 << 3)
-#define KVM_LARCH_HWCSR_USABLE	(0x1 << 4)
+#define KVM_LARCH_LBT		(0x1 << 3)
+#define KVM_LARCH_SWCSR_LATEST	(0x1 << 4)
+#define KVM_LARCH_HWCSR_USABLE	(0x1 << 5)
 
 struct kvm_vcpu_arch {
 	/*
@@ -168,6 +169,7 @@ struct kvm_vcpu_arch {
 
 	/* FPU state */
 	struct loongarch_fpu fpu FPU_ALIGN;
+	struct loongarch_lbt lbt;
 
 	/* CSR state */
 	struct loongarch_csrs *csr;
@@ -239,6 +241,11 @@ static inline bool kvm_guest_has_lasx(struct kvm_vcpu_arch *arch)
 	return arch->cpucfg[2] & CPUCFG2_LASX;
 }
 
+static inline bool kvm_guest_has_lbt(struct kvm_vcpu_arch *arch)
+{
+	return arch->cpucfg[2] & (CPUCFG2_X86BT | CPUCFG2_ARMBT | CPUCFG2_MIPSBT);
+}
+
 /* Debug: dump vcpu state */
 int kvm_arch_vcpu_dump_regs(struct kvm_vcpu *vcpu);
 

arch/loongarch/include/asm/kvm_vcpu.h

@@ -75,6 +75,12 @@ static inline void kvm_save_lasx(struct loongarch_fpu *fpu) { }
 static inline void kvm_restore_lasx(struct loongarch_fpu *fpu) { }
 #endif
 
+#ifdef CONFIG_CPU_HAS_LBT
+int kvm_own_lbt(struct kvm_vcpu *vcpu);
+#else
+static inline int kvm_own_lbt(struct kvm_vcpu *vcpu) { return -EINVAL; }
+#endif
+
 void kvm_init_timer(struct kvm_vcpu *vcpu, unsigned long hz);
 void kvm_save_timer(struct kvm_vcpu *vcpu);
 void kvm_restore_timer(struct kvm_vcpu *vcpu);

arch/loongarch/kvm/exit.c

@@ -748,6 +748,14 @@ static int kvm_handle_lasx_disabled(struct kvm_vcpu *vcpu)
 	return RESUME_GUEST;
 }
 
+static int kvm_handle_lbt_disabled(struct kvm_vcpu *vcpu)
+{
+	if (kvm_own_lbt(vcpu))
+		kvm_queue_exception(vcpu, EXCCODE_INE, 0);
+
+	return RESUME_GUEST;
+}
+
 static int kvm_send_pv_ipi(struct kvm_vcpu *vcpu)
 {
 	unsigned int min, cpu, i;
@@ -865,6 +873,7 @@ static exit_handle_fn kvm_fault_tables[EXCCODE_INT_START] = {
 	[EXCCODE_FPDIS]			= kvm_handle_fpu_disabled,
 	[EXCCODE_LSXDIS]		= kvm_handle_lsx_disabled,
 	[EXCCODE_LASXDIS]		= kvm_handle_lasx_disabled,
+	[EXCCODE_BTDIS]			= kvm_handle_lbt_disabled,
 	[EXCCODE_GSPR]			= kvm_handle_gspr,
 	[EXCCODE_HVC]			= kvm_handle_hypercall,
 };

arch/loongarch/kvm/vcpu.c

@@ -6,6 +6,7 @@
 #include <linux/kvm_host.h>
 #include <linux/entry-kvm.h>
 #include <asm/fpu.h>
+#include <asm/lbt.h>
 #include <asm/loongarch.h>
 #include <asm/setup.h>
 #include <asm/time.h>
@@ -983,12 +984,66 @@ int kvm_arch_vcpu_ioctl_set_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu)
 	return 0;
 }
 
+#ifdef CONFIG_CPU_HAS_LBT
+int kvm_own_lbt(struct kvm_vcpu *vcpu)
+{
+	if (!kvm_guest_has_lbt(&vcpu->arch))
+		return -EINVAL;
+
+	preempt_disable();
+	set_csr_euen(CSR_EUEN_LBTEN);
+	_restore_lbt(&vcpu->arch.lbt);
+	vcpu->arch.aux_inuse |= KVM_LARCH_LBT;
+	preempt_enable();
+
+	return 0;
+}
+
+static void kvm_lose_lbt(struct kvm_vcpu *vcpu)
+{
+	preempt_disable();
+	if (vcpu->arch.aux_inuse & KVM_LARCH_LBT) {
+		_save_lbt(&vcpu->arch.lbt);
+		clear_csr_euen(CSR_EUEN_LBTEN);
+		vcpu->arch.aux_inuse &= ~KVM_LARCH_LBT;
+	}
+	preempt_enable();
+}
+
+static void kvm_check_fcsr(struct kvm_vcpu *vcpu, unsigned long fcsr)
+{
+	/*
+	 * If TM is enabled, top register save/restore will
+	 * cause lbt exception, here enable lbt in advance
+	 */
+	if (fcsr & FPU_CSR_TM)
+		kvm_own_lbt(vcpu);
+}
+
+static void kvm_check_fcsr_alive(struct kvm_vcpu *vcpu)
+{
+	if (vcpu->arch.aux_inuse & KVM_LARCH_FPU) {
+		if (vcpu->arch.aux_inuse & KVM_LARCH_LBT)
+			return;
+		kvm_check_fcsr(vcpu, read_fcsr(LOONGARCH_FCSR0));
+	}
+}
+#else
+static inline void kvm_lose_lbt(struct kvm_vcpu *vcpu) { }
+static inline void kvm_check_fcsr(struct kvm_vcpu *vcpu, unsigned long fcsr) { }
+static inline void kvm_check_fcsr_alive(struct kvm_vcpu *vcpu) { }
+#endif
+
 /* Enable FPU and restore context */
 void kvm_own_fpu(struct kvm_vcpu *vcpu)
 {
 	preempt_disable();
 
-	/* Enable FPU */
+	/*
+	 * Enable FPU for guest
+	 * Set FR and FRE according to guest context
+	 */
+	kvm_check_fcsr(vcpu, vcpu->arch.fpu.fcsr);
 	set_csr_euen(CSR_EUEN_FPEN);
 
 	kvm_restore_fpu(&vcpu->arch.fpu);
@@ -1008,6 +1063,7 @@ int kvm_own_lsx(struct kvm_vcpu *vcpu)
 	preempt_disable();
 
 	/* Enable LSX for guest */
+	kvm_check_fcsr(vcpu, vcpu->arch.fpu.fcsr);
 	set_csr_euen(CSR_EUEN_LSXEN | CSR_EUEN_FPEN);
 	switch (vcpu->arch.aux_inuse & KVM_LARCH_FPU) {
 	case KVM_LARCH_FPU:
@@ -1042,6 +1098,7 @@ int kvm_own_lasx(struct kvm_vcpu *vcpu)
 
 	preempt_disable();
 
+	kvm_check_fcsr(vcpu, vcpu->arch.fpu.fcsr);
 	set_csr_euen(CSR_EUEN_FPEN | CSR_EUEN_LSXEN | CSR_EUEN_LASXEN);
 	switch (vcpu->arch.aux_inuse & (KVM_LARCH_FPU | KVM_LARCH_LSX)) {
 	case KVM_LARCH_LSX:
@@ -1073,6 +1130,7 @@ void kvm_lose_fpu(struct kvm_vcpu *vcpu)
 {
 	preempt_disable();
 
+	kvm_check_fcsr_alive(vcpu);
 	if (vcpu->arch.aux_inuse & KVM_LARCH_LASX) {
 		kvm_save_lasx(&vcpu->arch.fpu);
 		vcpu->arch.aux_inuse &= ~(KVM_LARCH_LSX | KVM_LARCH_FPU | KVM_LARCH_LASX);
@@ -1095,6 +1153,7 @@ void kvm_lose_fpu(struct kvm_vcpu *vcpu)
 		/* Disable FPU */
 		clear_csr_euen(CSR_EUEN_FPEN);
 	}
+	kvm_lose_lbt(vcpu);
 
 	preempt_enable();
 }