[PATCH 2/2] Provide fast path for "rep ins" emulation if possible.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



"rep ins" emulation is going through emulator now. This is slow because
emulator knows how to write back only one datum at a time. This patch
provides fast path for the instruction in certain conditions. The
conditions are: DF flag is not set, destination memory is RAM and single
datum does not cross page boundary. If fast path code fails it falls
back to emulation.

Signed-off-by: Gleb Natapov <gleb@xxxxxxxxxx>
---
 arch/x86/include/asm/kvm_host.h |    5 ++
 arch/x86/kvm/svm.c              |   20 ++++++--
 arch/x86/kvm/vmx.c              |   25 +++++++---
 arch/x86/kvm/x86.c              |   95 +++++++++++++++++++++++++++++++++++++++
 4 files changed, 133 insertions(+), 12 deletions(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 1bcd280..5129639 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -403,6 +403,9 @@ struct kvm_vcpu_arch {
 	/* emulate context */
 
 	struct x86_emulate_ctxt emulate_ctxt;
+	struct x86_fast_string_pio_ctxt {
+		unsigned long linear_addr;
+	} fast_string_pio_ctxt;
 	bool emulate_regs_need_sync_to_vcpu;
 	bool emulate_regs_need_sync_from_vcpu;
 	int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
@@ -763,6 +766,8 @@ int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
 struct x86_emulate_ctxt;
 
 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
+int kvm_fast_string_pio_in(struct kvm_vcpu *vcpu, int size, unsigned short port,
+		u8 addr_size);
 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index f75af40..20e3fb0 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -1887,21 +1887,31 @@ static int io_interception(struct vcpu_svm *svm)
 {
 	struct kvm_vcpu *vcpu = &svm->vcpu;
 	u32 io_info = svm->vmcb->control.exit_info_1; /* address size bug? */
-	int size, in, string;
+	int size, in, string, rep;
 	unsigned port;
 
 	++svm->vcpu.stat.io_exits;
 	string = (io_info & SVM_IOIO_STR_MASK) != 0;
+	rep = (io_info & SVM_IOIO_REP_MASK) != 0;
 	in = (io_info & SVM_IOIO_TYPE_MASK) != 0;
-	if (string || in)
-		return emulate_instruction(vcpu, 0) == EMULATE_DONE;
 
 	port = io_info >> 16;
 	size = (io_info & SVM_IOIO_SIZE_MASK) >> SVM_IOIO_SIZE_SHIFT;
 	svm->next_rip = svm->vmcb->control.exit_info_2;
-	skip_emulated_instruction(&svm->vcpu);
 
-	return kvm_fast_pio_out(vcpu, size, port);
+	if (!string && !in) {
+		skip_emulated_instruction(&svm->vcpu);
+		return kvm_fast_pio_out(vcpu, size, port);
+	} else if (string && in && rep) {
+		int addr_size = (io_info & SVM_IOIO_ASIZE_MASK) >>
+			SVM_IOIO_ASIZE_SHIFT;
+		int r = kvm_fast_string_pio_in(vcpu, size, port,
+				ffs(addr_size) - 1);
+		if (r != EMULATE_FAIL)
+			return r == EMULATE_DONE;
+	}
+
+	return emulate_instruction(vcpu, 0) == EMULATE_DONE;
 }
 
 static int nmi_interception(struct vcpu_svm *svm)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 32eb588..a2935f3 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -632,6 +632,7 @@ static unsigned long *vmx_msr_bitmap_longmode;
 
 static bool cpu_has_load_ia32_efer;
 static bool cpu_has_load_perf_global_ctrl;
+static bool cpu_has_ins_outs_inst_info;
 
 static DECLARE_BITMAP(vmx_vpid_bitmap, VMX_NR_VPIDS);
 static DEFINE_SPINLOCK(vmx_vpid_lock);
@@ -2510,6 +2511,8 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 	if (((vmx_msr_high >> 18) & 15) != 6)
 		return -EIO;
 
+	cpu_has_ins_outs_inst_info = vmx_msr_high & (1u << 22);
+
 	vmcs_conf->size = vmx_msr_high & 0x1fff;
 	vmcs_conf->order = get_order(vmcs_config.size);
 	vmcs_conf->revision_id = vmx_msr_low;
@@ -4333,23 +4336,31 @@ static int handle_triple_fault(struct kvm_vcpu *vcpu)
 static int handle_io(struct kvm_vcpu *vcpu)
 {
 	unsigned long exit_qualification;
-	int size, in, string;
+	int size, in, string, rep;
 	unsigned port;
 
 	exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
-	string = (exit_qualification & 16) != 0;
 	in = (exit_qualification & 8) != 0;
+	string = (exit_qualification & 16) != 0;
+	rep = (exit_qualification & 32) != 0;
 
 	++vcpu->stat.io_exits;
 
-	if (string || in)
-		return emulate_instruction(vcpu, 0) == EMULATE_DONE;
-
 	port = exit_qualification >> 16;
 	size = (exit_qualification & 7) + 1;
-	skip_emulated_instruction(vcpu);
 
-	return kvm_fast_pio_out(vcpu, size, port);
+	if (!string && !in) {
+		skip_emulated_instruction(vcpu);
+		return kvm_fast_pio_out(vcpu, size, port);
+	} else if (string && in && rep && cpu_has_ins_outs_inst_info) {
+		u32 inst_info = vmcs_read32(VMX_INSTRUCTION_INFO);
+		int r = kvm_fast_string_pio_in(vcpu, size, port,
+				(inst_info >> 7) & 7);
+		if (r != EMULATE_FAIL)
+			return r == EMULATE_DONE;
+	}
+
+	return emulate_instruction(vcpu, 0) == EMULATE_DONE;
 }
 
 static void
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 8361649..3bc7ad3 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -4660,6 +4660,101 @@ int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port)
 }
 EXPORT_SYMBOL_GPL(kvm_fast_pio_out);
 
+static int __kvm_fast_string_pio_in(struct kvm_vcpu *vcpu, int size,
+		                unsigned short port, unsigned long addr,
+				int count)
+{
+	struct page *page;
+	gpa_t gpa;
+	char *kaddr;
+	int ret;
+
+	gpa = kvm_mmu_gva_to_gpa_write(vcpu, addr, NULL);
+
+	if (gpa == UNMAPPED_GVA ||
+			(gpa & PAGE_MASK) == APIC_DEFAULT_PHYS_BASE)
+		return EMULATE_FAIL;
+
+	page = gfn_to_page(vcpu->kvm, gpa >> PAGE_SHIFT);
+	if (is_error_page(page)) {
+		kvm_release_page_clean(page);
+		return EMULATE_FAIL;
+	}
+
+	kaddr = kmap_atomic(page);
+	kaddr += offset_in_page(gpa);
+
+	ret = emulator_pio_in_emulated(&vcpu->arch.emulate_ctxt, size, port,
+			kaddr, count);
+
+	kunmap_atomic(kaddr);
+	if (ret) {
+		kvm_register_write(vcpu, VCPU_REGS_RCX,
+				kvm_register_read(vcpu, VCPU_REGS_RCX) - count);
+		kvm_register_write(vcpu, VCPU_REGS_RDI,
+				kvm_register_read(vcpu, VCPU_REGS_RDI) + count*size);
+		kvm_release_page_dirty(page);
+		return EMULATE_DONE;
+	}
+	kvm_release_page_clean(page);
+	return EMULATE_DO_MMIO;
+}
+
+static int complete_fast_string_pio(struct kvm_vcpu *vcpu)
+{
+	unsigned long linear_addr = vcpu->arch.fast_string_pio_ctxt.linear_addr;
+	int r;
+
+	BUG_ON(!vcpu->arch.pio.count);
+
+	vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
+	r = __kvm_fast_string_pio_in(vcpu, vcpu->arch.pio.size,
+			vcpu->arch.pio.port, linear_addr, vcpu->arch.pio.count);
+	BUG_ON(r == EMULATE_DO_MMIO);
+	if (r == EMULATE_FAIL) /* mem slot gone while we were not looking */
+		vcpu->arch.pio.count = 0; /* drop the pio data */
+	srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
+	return 1;
+}
+
+int kvm_fast_string_pio_in(struct kvm_vcpu *vcpu, int size,
+		unsigned short port, u8 addr_size)
+{
+	unsigned long masks[] = {0xffff, 0xffffffff, ~0};
+	unsigned long rdi = kvm_register_read(vcpu, VCPU_REGS_RDI);
+	unsigned long linear_addr = rdi + get_segment_base(vcpu, VCPU_SREG_ES);
+	unsigned long rcx = kvm_register_read(vcpu, VCPU_REGS_RCX), count;
+	int r;
+
+	if (rcx == 0) {
+		kvm_x86_ops->skip_emulated_instruction(vcpu);
+		return EMULATE_DONE;
+	}
+	if (kvm_get_rflags(vcpu) & X86_EFLAGS_DF)
+		return EMULATE_FAIL;
+	if (addr_size > 2)
+		return EMULATE_FAIL;
+
+	linear_addr &= masks[addr_size];
+
+	count = (PAGE_SIZE - offset_in_page(linear_addr))/size;
+
+	if (count == 0) /* 'in' crosses page boundry */
+		return EMULATE_FAIL;
+
+	count = min(count, rcx);
+
+	r = __kvm_fast_string_pio_in(vcpu, size, port, linear_addr, count);
+
+	if (r != EMULATE_DO_MMIO)
+		return r;
+
+	vcpu->arch.fast_string_pio_ctxt.linear_addr = linear_addr;
+	vcpu->arch.complete_userspace_io = complete_fast_string_pio;
+	return EMULATE_DO_MMIO;
+}
+EXPORT_SYMBOL_GPL(kvm_fast_string_pio_in);
+
 static void tsc_bad(void *info)
 {
 	__this_cpu_write(cpu_tsc_khz, 0);
-- 
1.7.7.3

--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[Index of Archives]     [KVM ARM]     [KVM ia64]     [KVM ppc]     [Virtualization Tools]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite Questions]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux