Skip to contentSkip to conversation

Emulator exception improvements

PATCH Applied Avi Kivity 2 participants 9 messages Nov 22, 2010 – Nov 30, 2010

Subsystems (3)
lore
Share & tools
lore

Avi Kivity

author
#0 lore raw

cover Emulator exception improvements

This boring patchset introduces a struct x86_exception to represent an exception, and uses it to communicate between the emulator and the rest of kvm. The primary benefit is that we can now pass a #GP from kvm into the emulator, not just #PFs.

I'd like to also fold vcpu->arch.fault into struct x86_exception, but that's more difficult, so deferred until later.

Avi Kivity (7):
KVM: x86 emulator: introduce struct x86_exception to communicate
faults
KVM: x86 emulator: make emulator memory callbacks return full
exception
KVM: x86 emulator: drop dead pf injection in emulate_popf()
KVM: x86 emulator: tighen up ->read_std() and ->write_std() error
checks
KVM: x86 emulator: simplify exception generation
KVM: Push struct x86_exception info the various gva_to_gpa variants
KVM: Push struct x86_exception into walk_addr()

 arch/x86/include/asm/kvm_emulate.h |   26 +++--
 arch/x86/include/asm/kvm_host.h    |   14 ++-
 arch/x86/kvm/emulate.c             |  243 ++++++++++++++----------------------
 arch/x86/kvm/mmu.c                 |   13 +-
 arch/x86/kvm/paging_tmpl.h         |   31 +++--
 arch/x86/kvm/x86.c                 |   97 ++++++++-------
 6 files changed, 195 insertions(+), 229 deletions(-)
author
#1 lore raw

PATCH 1/7 KVM: x86 emulator: introduce struct x86_exception to communicate faults

Introduce a structure that can contain an exception to be passed back to main kvm code.

---
 arch/x86/include/asm/kvm_emulate.h |   11 ++++++++---
 arch/x86/kvm/emulate.c             |   26 +++++++++++++++++++++++---
 arch/x86/kvm/x86.c                 |   13 +++++++------
 3 files changed, 38 insertions(+), 12 deletions(-)
arch/x86/include/asm/kvm_emulate.h +8 -3
index b48c133..b7c1127 100644
@@ -15,6 +15,12 @@
 
 struct x86_emulate_ctxt;
 
+struct x86_exception {
+	u8 vector;
+	bool error_code_valid;
+	u16 error_code;
+};
+
 /*
  * x86_emulate_ops:
  *
@@ -229,9 +235,8 @@ struct x86_emulate_ctxt {
 
 	bool perm_ok; /* do not check permissions if true */
 
-	int exception; /* exception that happens during emulation or -1 */
-	u32 error_code; /* error code for exception */
-	bool error_code_valid;
+	bool have_exception;
+	struct x86_exception exception;
 
 	/* decode cache */
 	struct decode_cache decode;
arch/x86/kvm/emulate.c +23 -3
index bdbbb18..18596e6 100644
@@ -469,9 +469,9 @@ static ulong linear(struct x86_emulate_ctxt *ctxt,
 static void emulate_exception(struct x86_emulate_ctxt *ctxt, int vec,
 				      u32 error, bool valid)
 {
-	ctxt->exception = vec;
-	ctxt->error_code = error;
-	ctxt->error_code_valid = valid;
+	ctxt->exception.vector = vec;
+	ctxt->exception.error_code = error;
+	ctxt->exception.error_code_valid = valid;
 }
 
 static void emulate_gp(struct x86_emulate_ctxt *ctxt, int err)
@@ -3015,23 +3015,27 @@ x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
 
 	if (ctxt->mode == X86EMUL_MODE_PROT64 && (c->d & No64)) {
 		emulate_ud(ctxt);
+		rc = X86EMUL_PROPAGATE_FAULT;
 		goto done;
 	}
 
 	/* LOCK prefix is allowed only with some instructions */
 	if (c->lock_prefix && (!(c->d & Lock) || c->dst.type != OP_MEM)) {
 		emulate_ud(ctxt);
+		rc = X86EMUL_PROPAGATE_FAULT;
 		goto done;
 	}
 
 	if ((c->d & SrcMask) == SrcMemFAddr && c->src.type != OP_MEM) {
 		emulate_ud(ctxt);
+		rc = X86EMUL_PROPAGATE_FAULT;
 		goto done;
 	}
 
 	/* Privileged instruction can be executed only in CPL=0 */
 	if ((c->d & Priv) && ops->cpl(ctxt->vcpu)) {
 		emulate_gp(ctxt, 0);
+		rc = X86EMUL_PROPAGATE_FAULT;
 		goto done;
 	}
 
@@ -3210,6 +3214,7 @@ special_insn:
 	case 0x8c:  /* mov r/m, sreg */
 		if (c->modrm_reg > VCPU_SREG_GS) {
 			emulate_ud(ctxt);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		c->dst.val = ops->get_segment_selector(c->modrm_reg, ctxt->vcpu);
@@ -3225,6 +3230,7 @@ special_insn:
 		if (c->modrm_reg == VCPU_SREG_CS ||
 		    c->modrm_reg > VCPU_SREG_GS) {
 			emulate_ud(ctxt);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 
@@ -3357,6 +3363,7 @@ special_insn:
 		c->dst.bytes = min(c->dst.bytes, 4u);
 		if (!emulator_io_permited(ctxt, ops, c->src.val, c->dst.bytes)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		if (!pio_in_emulated(ctxt, ops, c->dst.bytes, c->src.val,
@@ -3371,6 +3378,7 @@ special_insn:
 		if (!emulator_io_permited(ctxt, ops, c->dst.val,
 					  c->src.bytes)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		ops->pio_out_emulated(c->src.bytes, c->dst.val,
@@ -3396,6 +3404,7 @@ special_insn:
 	case 0xfa: /* cli */
 		if (emulator_bad_iopl(ctxt, ops)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		} else
 			ctxt->eflags &= ~X86_EFLAGS_IF;
@@ -3403,6 +3412,7 @@ special_insn:
 	case 0xfb: /* sti */
 		if (emulator_bad_iopl(ctxt, ops)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		} else {
 			ctxt->interruptibility = KVM_X86_SHADOW_INT_STI;
@@ -3475,6 +3485,8 @@ writeback:
 	ctxt->eip = c->eip;
 
 done:
+	if (rc == X86EMUL_PROPAGATE_FAULT)
+		ctxt->have_exception = true;
 	return (rc == X86EMUL_UNHANDLEABLE) ? EMULATION_FAILED : EMULATION_OK;
 
 twobyte_insn:
@@ -3537,6 +3549,7 @@ twobyte_insn:
 			break;
 		case 5: /* not defined */
 			emulate_ud(ctxt);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		case 7: /* invlpg*/
 			emulate_invlpg(ctxt->vcpu,
@@ -3567,6 +3580,7 @@ twobyte_insn:
 		case 5 ... 7:
 		case 9 ... 15:
 			emulate_ud(ctxt);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		c->dst.val = ops->get_cr(c->modrm_reg, ctxt->vcpu);
@@ -3575,6 +3589,7 @@ twobyte_insn:
 		if ((ops->get_cr(4, ctxt->vcpu) & X86_CR4_DE) &&
 		    (c->modrm_reg == 4 || c->modrm_reg == 5)) {
 			emulate_ud(ctxt);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		ops->get_dr(c->modrm_reg, &c->dst.val, ctxt->vcpu);
@@ -3582,6 +3597,7 @@ twobyte_insn:
 	case 0x22: /* mov reg, cr */
 		if (ops->set_cr(c->modrm_reg, c->src.val, ctxt->vcpu)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		c->dst.type = OP_NONE;
@@ -3590,6 +3606,7 @@ twobyte_insn:
 		if ((ops->get_cr(4, ctxt->vcpu) & X86_CR4_DE) &&
 		    (c->modrm_reg == 4 || c->modrm_reg == 5)) {
 			emulate_ud(ctxt);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 
@@ -3598,6 +3615,7 @@ twobyte_insn:
 				 ~0ULL : ~0U), ctxt->vcpu) < 0) {
 			/* #UD condition is already handled by the code above */
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 
@@ -3609,6 +3627,7 @@ twobyte_insn:
 			| ((u64)c->regs[VCPU_REGS_RDX] << 32);
 		if (ops->set_msr(ctxt->vcpu, c->regs[VCPU_REGS_RCX], msr_data)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		}
 		rc = X86EMUL_CONTINUE;
@@ -3617,6 +3636,7 @@ twobyte_insn:
 		/* rdmsr */
 		if (ops->get_msr(ctxt->vcpu, c->regs[VCPU_REGS_RCX], &msr_data)) {
 			emulate_gp(ctxt, 0);
+			rc = X86EMUL_PROPAGATE_FAULT;
 			goto done;
 		} else {
 			c->regs[VCPU_REGS_RAX] = (u32)msr_data;
arch/x86/kvm/x86.c +7 -6
index 410d2d1..ef72bb1 100644
@@ -4259,12 +4259,13 @@ static void toggle_interruptibility(struct kvm_vcpu *vcpu, u32 mask)
 static void inject_emulated_exception(struct kvm_vcpu *vcpu)
 {
 	struct x86_emulate_ctxt *ctxt = &vcpu->arch.emulate_ctxt;
-	if (ctxt->exception == PF_VECTOR)
+	if (ctxt->exception.vector == PF_VECTOR)
 		kvm_propagate_fault(vcpu);
-	else if (ctxt->error_code_valid)
-		kvm_queue_exception_e(vcpu, ctxt->exception, ctxt->error_code);
+	else if (ctxt->exception.error_code_valid)
+		kvm_queue_exception_e(vcpu, ctxt->exception.vector,
+				      ctxt->exception.error_code);
 	else
-		kvm_queue_exception(vcpu, ctxt->exception);
+		kvm_queue_exception(vcpu, ctxt->exception.vector);
 }
 
 static void init_emulate_ctxt(struct kvm_vcpu *vcpu)
@@ -4376,7 +4377,7 @@ int emulate_instruction(struct kvm_vcpu *vcpu,
 	if (!(emulation_type & EMULTYPE_NO_DECODE)) {
 		init_emulate_ctxt(vcpu);
 		vcpu->arch.emulate_ctxt.interruptibility = 0;
-		vcpu->arch.emulate_ctxt.exception = -1;
+		vcpu->arch.emulate_ctxt.have_exception = false;
 		vcpu->arch.emulate_ctxt.perm_ok = false;
 
 		r = x86_decode_insn(&vcpu->arch.emulate_ctxt);
@@ -4442,7 +4443,7 @@ restart:
 	}
 
 done:
-	if (vcpu->arch.emulate_ctxt.exception >= 0) {
+	if (vcpu->arch.emulate_ctxt.have_exception) {
 		inject_emulated_exception(vcpu);
 		r = EMULATE_DONE;
 	} else if (vcpu->arch.pio.count) {
Signature · 1 line
1.7.1
author
#2 lore raw

PATCH 2/7 KVM: x86 emulator: make emulator memory callbacks return full exception

In reply to Avi Kivity

This way, they can return #GP, not just #PF.

---
 arch/x86/include/asm/kvm_emulate.h |   15 ++++--
 arch/x86/kvm/emulate.c             |   89 ++++++++++++-----------------------
 arch/x86/kvm/x86.c                 |   76 ++++++++++++++++++-------------
 3 files changed, 84 insertions(+), 96 deletions(-)
arch/x86/include/asm/kvm_emulate.h +9 -6
index b7c1127..87d017e 100644
@@ -70,7 +70,8 @@ struct x86_emulate_ops {
 	 *  @bytes: [IN ] Number of bytes to read from memory.
 	 */
 	int (*read_std)(unsigned long addr, void *val,
-			unsigned int bytes, struct kvm_vcpu *vcpu, u32 *error);
+			unsigned int bytes, struct kvm_vcpu *vcpu,
+			struct x86_exception *fault);
 
 	/*
 	 * write_std: Write bytes of standard (non-emulated/special) memory.
@@ -80,7 +81,8 @@ struct x86_emulate_ops {
 	 *  @bytes: [IN ] Number of bytes to write to memory.
 	 */
 	int (*write_std)(unsigned long addr, void *val,
-			 unsigned int bytes, struct kvm_vcpu *vcpu, u32 *error);
+			 unsigned int bytes, struct kvm_vcpu *vcpu,
+			 struct x86_exception *fault);
 	/*
 	 * fetch: Read bytes of standard (non-emulated/special) memory.
 	 *        Used for instruction fetch.
@@ -89,7 +91,8 @@ struct x86_emulate_ops {
 	 *  @bytes: [IN ] Number of bytes to read from memory.
 	 */
 	int (*fetch)(unsigned long addr, void *val,
-			unsigned int bytes, struct kvm_vcpu *vcpu, u32 *error);
+		     unsigned int bytes, struct kvm_vcpu *vcpu,
+		     struct x86_exception *fault);
 
 	/*
 	 * read_emulated: Read bytes from emulated/special memory area.
@@ -100,7 +103,7 @@ struct x86_emulate_ops {
 	int (*read_emulated)(unsigned long addr,
 			     void *val,
 			     unsigned int bytes,
-			     unsigned int *error,
+			     struct x86_exception *fault,
 			     struct kvm_vcpu *vcpu);
 
 	/*
@@ -113,7 +116,7 @@ struct x86_emulate_ops {
 	int (*write_emulated)(unsigned long addr,
 			      const void *val,
 			      unsigned int bytes,
-			      unsigned int *error,
+			      struct x86_exception *fault,
 			      struct kvm_vcpu *vcpu);
 
 	/*
@@ -128,7 +131,7 @@ struct x86_emulate_ops {
 				const void *old,
 				const void *new,
 				unsigned int bytes,
-				unsigned int *error,
+				struct x86_exception *fault,
 				struct kvm_vcpu *vcpu);
 
 	int (*pio_in_emulated)(int size, unsigned short port, void *val,
arch/x86/kvm/emulate.c +31 -58

231 lines not on this page · raw message

arch/x86/kvm/x86.c +44 -32

199 lines not on this page · raw message

Signature · 1 line
1.7.1
author
#3 lore raw

PATCH 3/7 KVM: x86 emulator: drop dead pf injection in emulate_popf()

In reply to Avi Kivity

If rc == X86EMUL_PROPAGATE_FAULT, we would have returned earlier.

---
 arch/x86/kvm/emulate.c |    8 --------
 1 files changed, 0 insertions(+), 8 deletions(-)
arch/x86/kvm/emulate.c +0 -8
index 68d72db..b39df7b 100644
@@ -479,11 +479,6 @@ static void emulate_gp(struct x86_emulate_ctxt *ctxt, int err)
 	emulate_exception(ctxt, GP_VECTOR, err, true);
 }
 
-static void emulate_pf(struct x86_emulate_ctxt *ctxt)
-{
-	emulate_exception(ctxt, PF_VECTOR, 0, true);
-}
-
 static void emulate_ud(struct x86_emulate_ctxt *ctxt)
 {
 	emulate_exception(ctxt, UD_VECTOR, 0, false);
@@ -1184,9 +1179,6 @@ static int emulate_popf(struct x86_emulate_ctxt *ctxt,
 	*(unsigned long *)dest =
 		(ctxt->eflags & ~change_mask) | (val & change_mask);
 
-	if (rc == X86EMUL_PROPAGATE_FAULT)
-		emulate_pf(ctxt);
-
 	return rc;
 }
 
Signature · 1 line
1.7.1
author
#4 lore raw

PATCH 4/7 KVM: x86 emulator: tighen up ->read_std() and ->write_std() error checks

In reply to Avi Kivity

Instead of checking for X86EMUL_PROPAGATE_FAULT, check for any error, making the callers more reliable.

---
 arch/x86/kvm/emulate.c |   16 ++++++++--------
 1 files changed, 8 insertions(+), 8 deletions(-)
arch/x86/kvm/emulate.c +8 -8
index b39df7b..4b8f58b 100644
@@ -1915,7 +1915,7 @@ static int task_switch_16(struct x86_emulate_ctxt *ctxt,
 
 	ret = ops->read_std(old_tss_base, &tss_seg, sizeof tss_seg, ctxt->vcpu,
 			    &ctxt->exception);
-	if (ret == X86EMUL_PROPAGATE_FAULT)
+	if (ret != X86EMUL_CONTINUE)
 		/* FIXME: need to provide precise fault address */
 		return ret;
 
@@ -1923,13 +1923,13 @@ static int task_switch_16(struct x86_emulate_ctxt *ctxt,
 
 	ret = ops->write_std(old_tss_base, &tss_seg, sizeof tss_seg, ctxt->vcpu,
 			     &ctxt->exception);
-	if (ret == X86EMUL_PROPAGATE_FAULT)
+	if (ret != X86EMUL_CONTINUE)
 		/* FIXME: need to provide precise fault address */
 		return ret;
 
 	ret = ops->read_std(new_tss_base, &tss_seg, sizeof tss_seg, ctxt->vcpu,
 			    &ctxt->exception);
-	if (ret == X86EMUL_PROPAGATE_FAULT)
+	if (ret != X86EMUL_CONTINUE)
 		/* FIXME: need to provide precise fault address */
 		return ret;
 
@@ -1940,7 +1940,7 @@ static int task_switch_16(struct x86_emulate_ctxt *ctxt,
 				     &tss_seg.prev_task_link,
 				     sizeof tss_seg.prev_task_link,
 				     ctxt->vcpu, &ctxt->exception);
-		if (ret == X86EMUL_PROPAGATE_FAULT)
+		if (ret != X86EMUL_CONTINUE)
 			/* FIXME: need to provide precise fault address */
 			return ret;
 	}
@@ -2049,7 +2049,7 @@ static int task_switch_32(struct x86_emulate_ctxt *ctxt,
 
 	ret = ops->read_std(old_tss_base, &tss_seg, sizeof tss_seg, ctxt->vcpu,
 			    &ctxt->exception);
-	if (ret == X86EMUL_PROPAGATE_FAULT)
+	if (ret != X86EMUL_CONTINUE)
 		/* FIXME: need to provide precise fault address */
 		return ret;
 
@@ -2057,13 +2057,13 @@ static int task_switch_32(struct x86_emulate_ctxt *ctxt,
 
 	ret = ops->write_std(old_tss_base, &tss_seg, sizeof tss_seg, ctxt->vcpu,
 			     &ctxt->exception);
-	if (ret == X86EMUL_PROPAGATE_FAULT)
+	if (ret != X86EMUL_CONTINUE)
 		/* FIXME: need to provide precise fault address */
 		return ret;
 
 	ret = ops->read_std(new_tss_base, &tss_seg, sizeof tss_seg, ctxt->vcpu,
 			    &ctxt->exception);
-	if (ret == X86EMUL_PROPAGATE_FAULT)
+	if (ret != X86EMUL_CONTINUE)
 		/* FIXME: need to provide precise fault address */
 		return ret;
 
@@ -2074,7 +2074,7 @@ static int task_switch_32(struct x86_emulate_ctxt *ctxt,
 				     &tss_seg.prev_task_link,
 				     sizeof tss_seg.prev_task_link,
 				     ctxt->vcpu, &ctxt->exception);
-		if (ret == X86EMUL_PROPAGATE_FAULT)
+		if (ret != X86EMUL_CONTINUE)
 			/* FIXME: need to provide precise fault address */
 			return ret;
 	}
Signature · 1 line
1.7.1
author
#5 lore raw

PATCH 5/7 KVM: x86 emulator: simplify exception generation

In reply to Avi Kivity

Immediately after we generate an exception, we want a X86EMUL_PROPAGATE_FAULT constant, so return it from the generation functions.

---
 arch/x86/kvm/emulate.c |  140 +++++++++++++++++-------------------------------
 1 files changed, 50 insertions(+), 90 deletions(-)
arch/x86/kvm/emulate.c +50 -90
index 4b8f58b..6366735 100644
@@ -466,33 +466,33 @@ static ulong linear(struct x86_emulate_ctxt *ctxt,
 	return la;
 }
 
-static void emulate_exception(struct x86_emulate_ctxt *ctxt, int vec,
-				      u32 error, bool valid)
+static int emulate_exception(struct x86_emulate_ctxt *ctxt, int vec,
+			     u32 error, bool valid)
 {
 	ctxt->exception.vector = vec;
 	ctxt->exception.error_code = error;
 	ctxt->exception.error_code_valid = valid;
+	return X86EMUL_PROPAGATE_FAULT;
 }
 
-static void emulate_gp(struct x86_emulate_ctxt *ctxt, int err)
+static int emulate_gp(struct x86_emulate_ctxt *ctxt, int err)
 {
-	emulate_exception(ctxt, GP_VECTOR, err, true);
+	return emulate_exception(ctxt, GP_VECTOR, err, true);
 }
 
-static void emulate_ud(struct x86_emulate_ctxt *ctxt)
+static int emulate_ud(struct x86_emulate_ctxt *ctxt)
 {
-	emulate_exception(ctxt, UD_VECTOR, 0, false);
+	return emulate_exception(ctxt, UD_VECTOR, 0, false);
 }
 
-static void emulate_ts(struct x86_emulate_ctxt *ctxt, int err)
+static int emulate_ts(struct x86_emulate_ctxt *ctxt, int err)
 {
-	emulate_exception(ctxt, TS_VECTOR, err, true);
+	return emulate_exception(ctxt, TS_VECTOR, err, true);
 }
 
 static int emulate_de(struct x86_emulate_ctxt *ctxt)
 {
-	emulate_exception(ctxt, DE_VECTOR, 0, false);
-	return X86EMUL_PROPAGATE_FAULT;
+	return emulate_exception(ctxt, DE_VECTOR, 0, false);
 }
 
 static int do_fetch_insn_byte(struct x86_emulate_ctxt *ctxt,
@@ -898,10 +898,8 @@ static int read_segment_descriptor(struct x86_emulate_ctxt *ctxt,
 
 	get_descriptor_table_ptr(ctxt, ops, selector, &dt);
 
-	if (dt.size < index * 8 + 7) {
-		emulate_gp(ctxt, selector & 0xfffc);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (dt.size < index * 8 + 7)
+		return emulate_gp(ctxt, selector & 0xfffc);
 	addr = dt.address + index * 8;
 	ret = ops->read_std(addr, desc, sizeof *desc, ctxt->vcpu, 
 			    &ctxt->exception);
@@ -921,10 +919,8 @@ static int write_segment_descriptor(struct x86_emulate_ctxt *ctxt,
 
 	get_descriptor_table_ptr(ctxt, ops, selector, &dt);
 
-	if (dt.size < index * 8 + 7) {
-		emulate_gp(ctxt, selector & 0xfffc);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (dt.size < index * 8 + 7)
+		return emulate_gp(ctxt, selector & 0xfffc);
 
 	addr = dt.address + index * 8;
 	ret = ops->write_std(addr, desc, sizeof *desc, ctxt->vcpu,
@@ -1165,10 +1161,8 @@ static int emulate_popf(struct x86_emulate_ctxt *ctxt,
 			change_mask |= EFLG_IF;
 		break;
 	case X86EMUL_MODE_VM86:
-		if (iopl < 3) {
-			emulate_gp(ctxt, 0);
-			return X86EMUL_PROPAGATE_FAULT;
-		}
+		if (iopl < 3)
+			return emulate_gp(ctxt, 0);
 		change_mask |= EFLG_IF;
 		break;
 	default: /* real mode */
@@ -1347,10 +1341,8 @@ static int emulate_iret_real(struct x86_emulate_ctxt *ctxt,
 	if (rc != X86EMUL_CONTINUE)
 		return rc;
 
-	if (temp_eip & ~0xffff) {
-		emulate_gp(ctxt, 0);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (temp_eip & ~0xffff)
+		return emulate_gp(ctxt, 0);
 
 	rc = emulate_pop(ctxt, ops, &cs, c->op_bytes);
 
@@ -1601,10 +1593,8 @@ emulate_syscall(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 
 	/* syscall is not available in real mode */
 	if (ctxt->mode == X86EMUL_MODE_REAL ||
-	    ctxt->mode == X86EMUL_MODE_VM86) {
-		emulate_ud(ctxt);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	    ctxt->mode == X86EMUL_MODE_VM86)
+		return emulate_ud(ctxt);
 
 	setup_syscalls_segments(ctxt, ops, &cs, &ss);
 
@@ -1655,34 +1645,26 @@ emulate_sysenter(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 	u16 cs_sel, ss_sel;
 
 	/* inject #GP if in real mode */
-	if (ctxt->mode == X86EMUL_MODE_REAL) {
-		emulate_gp(ctxt, 0);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (ctxt->mode == X86EMUL_MODE_REAL)
+		return emulate_gp(ctxt, 0);
 
 	/* XXX sysenter/sysexit have not been tested in 64bit mode.
 	* Therefore, we inject an #UD.
 	*/
-	if (ctxt->mode == X86EMUL_MODE_PROT64) {
-		emulate_ud(ctxt);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (ctxt->mode == X86EMUL_MODE_PROT64)
+		return emulate_ud(ctxt);
 
 	setup_syscalls_segments(ctxt, ops, &cs, &ss);
 
 	ops->get_msr(ctxt->vcpu, MSR_IA32_SYSENTER_CS, &msr_data);
 	switch (ctxt->mode) {
 	case X86EMUL_MODE_PROT32:
-		if ((msr_data & 0xfffc) == 0x0) {
-			emulate_gp(ctxt, 0);
-			return X86EMUL_PROPAGATE_FAULT;
-		}
+		if ((msr_data & 0xfffc) == 0x0)
+			return emulate_gp(ctxt, 0);
 		break;
 	case X86EMUL_MODE_PROT64:
-		if (msr_data == 0x0) {
-			emulate_gp(ctxt, 0);
-			return X86EMUL_PROPAGATE_FAULT;
-		}
+		if (msr_data == 0x0)
+			return emulate_gp(ctxt, 0);
 		break;
 	}
 
@@ -1722,10 +1704,8 @@ emulate_sysexit(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 
 	/* inject #GP if in real mode or Virtual 8086 mode */
 	if (ctxt->mode == X86EMUL_MODE_REAL ||
-	    ctxt->mode == X86EMUL_MODE_VM86) {
-		emulate_gp(ctxt, 0);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	    ctxt->mode == X86EMUL_MODE_VM86)
+		return emulate_gp(ctxt, 0);
 
 	setup_syscalls_segments(ctxt, ops, &cs, &ss);
 
@@ -1740,18 +1720,14 @@ emulate_sysexit(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 	switch (usermode) {
 	case X86EMUL_MODE_PROT32:
 		cs_sel = (u16)(msr_data + 16);
-		if ((msr_data & 0xfffc) == 0x0) {
-			emulate_gp(ctxt, 0);
-			return X86EMUL_PROPAGATE_FAULT;
-		}
+		if ((msr_data & 0xfffc) == 0x0)
+			return emulate_gp(ctxt, 0);
 		ss_sel = (u16)(msr_data + 24);
 		break;
 	case X86EMUL_MODE_PROT64:
 		cs_sel = (u16)(msr_data + 32);
-		if (msr_data == 0x0) {
-			emulate_gp(ctxt, 0);
-			return X86EMUL_PROPAGATE_FAULT;
-		}
+		if (msr_data == 0x0)
+			return emulate_gp(ctxt, 0);
 		ss_sel = cs_sel + 8;
 		cs.d = 0;
 		cs.l = 1;
@@ -1982,10 +1958,8 @@ static int load_state_from_tss32(struct x86_emulate_ctxt *ctxt,
 	struct decode_cache *c = &ctxt->decode;
 	int ret;
 
-	if (ops->set_cr(3, tss->cr3, ctxt->vcpu)) {
-		emulate_gp(ctxt, 0);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (ops->set_cr(3, tss->cr3, ctxt->vcpu))
+		return emulate_gp(ctxt, 0);
 	c->eip = tss->eip;
 	ctxt->eflags = tss->eflags | 2;
 	c->regs[VCPU_REGS_RAX] = tss->eax;
@@ -2107,10 +2081,8 @@ static int emulator_do_task_switch(struct x86_emulate_ctxt *ctxt,
 
 	if (reason != TASK_SWITCH_IRET) {
 		if ((tss_selector & 3) > next_tss_desc.dpl ||
-		    ops->cpl(ctxt->vcpu) > next_tss_desc.dpl) {
-			emulate_gp(ctxt, 0);
-			return X86EMUL_PROPAGATE_FAULT;
-		}
+		    ops->cpl(ctxt->vcpu) > next_tss_desc.dpl)
+			return emulate_gp(ctxt, 0);
 	}
 
 	desc_limit = desc_limit_scaled(&next_tss_desc);
@@ -2331,10 +2303,8 @@ static int em_rdtsc(struct x86_emulate_ctxt *ctxt)
 	struct decode_cache *c = &ctxt->decode;
 	u64 tsc = 0;
 
-	if (cpl > 0 && (ctxt->ops->get_cr(4, ctxt->vcpu) & X86_CR4_TSD)) {
-		emulate_gp(ctxt, 0);
-		return X86EMUL_PROPAGATE_FAULT;
-	}
+	if (cpl > 0 && (ctxt->ops->get_cr(4, ctxt->vcpu) & X86_CR4_TSD))
+		return emulate_gp(ctxt, 0);
 	ctxt->ops->get_msr(ctxt->vcpu, MSR_IA32_TSC, &tsc);
 	c->regs[VCPU_REGS_RAX] = (u32)tsc;
 	c->regs[VCPU_REGS_RDX] = tsc >> 32;
@@ -2979,28 +2949,24 @@ x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
 	ctxt->decode.mem_read.pos = 0;
 
 	if (ctxt->mode == X86EMUL_MODE_PROT64 && (c->d & No64)) {
-		emulate_ud(ctxt);
-		rc = X86EMUL_PROPAGATE_FAULT;
+		rc = emulate_ud(ctxt);
 		goto done;
 	}
 
 	/* LOCK prefix is allowed only with some instructions */
 	if (c->lock_prefix && (!(c->d & Lock) || c->dst.type != OP_MEM)) {
-		emulate_ud(ctxt);
-		rc = X86EMUL_PROPAGATE_FAULT;
+		rc = emulate_ud(ctxt);
 		goto done;
 	}
 
 	if ((c->d & SrcMask) == SrcMemFAddr && c->src.type != OP_MEM) {
-		emulate_ud(ctxt);
-		rc = X86EMUL_PROPAGATE_FAULT;
+		rc = emulate_ud(ctxt);
 		goto done;
 	}
 
 	/* Privileged instruction can be executed only in CPL=0 */
 	if ((c->d & Priv) && ops->cpl(ctxt->vcpu)) {
-		emulate_gp(ctxt, 0);
-		rc = X86EMUL_PROPAGATE_FAULT;
+		rc = emulate_gp(ctxt, 0);
 		goto done;
 	}
 
@@ -3178,8 +3144,7 @@ special_insn:
 		break;
 	case 0x8c:  /* mov r/m, sreg */
 		if (c->modrm_reg > VCPU_SREG_GS) {
-			emulate_ud(ctxt);
-			rc = X86EMUL_PROPAGATE_FAULT;
+			rc = emulate_ud(ctxt);
 			goto done;
 		}
 		c->dst.val = ops->get_segment_selector(c->modrm_reg, ctxt->vcpu);
@@ -3194,8 +3159,7 @@ special_insn:
 
 		if (c->modrm_reg == VCPU_SREG_CS ||
 		    c->modrm_reg > VCPU_SREG_GS) {
-			emulate_ud(ctxt);
-			rc = X86EMUL_PROPAGATE_FAULT;
+			rc = emulate_ud(ctxt);
 			goto done;
 		}
 
@@ -3327,8 +3291,7 @@ special_insn:
 	do_io_in:
 		c->dst.bytes = min(c->dst.bytes, 4u);
 		if (!emulator_io_permited(ctxt, ops, c->src.val, c->dst.bytes)) {
-			emulate_gp(ctxt, 0);
-			rc = X86EMUL_PROPAGATE_FAULT;
+			rc = emulate_gp(ctxt, 0);
 			goto done;
 		}
 		if (!pio_in_emulated(ctxt, ops, c->dst.bytes, c->src.val,
@@ -3342,8 +3305,7 @@ special_insn:
 		c->src.bytes = min(c->src.bytes, 4u);
 		if (!emulator_io_permited(ctxt, ops, c->dst.val,
 					  c->src.bytes)) {
-			emulate_gp(ctxt, 0);
-			rc = X86EMUL_PROPAGATE_FAULT;
+			rc = emulate_gp(ctxt, 0);
 			goto done;
 		}
 		ops->pio_out_emulated(c->src.bytes, c->dst.val,
@@ -3368,16 +3330,14 @@ special_insn:
 		break;
 	case 0xfa: /* cli */
 		if (emulator_bad_iopl(ctxt, ops)) {
-			emulate_gp(ctxt, 0);
-			rc = X86EMUL_PROPAGATE_FAULT;
+			rc = emulate_gp(ctxt, 0);
 			goto done;
 		} else
 			ctxt->eflags &= ~X86_EFLAGS_IF;
 		break;
 	case 0xfb: /* sti */
 		if (emulator_bad_iopl(ctxt, ops)) {
-			emulate_gp(ctxt, 0);
-			rc = X86EMUL_PROPAGATE_FAULT;
+			rc = emulate_gp(ctxt, 0);
 			goto done;
 		} else {
 			ctxt->interruptibility = KVM_X86_SHADOW_INT_STI;
Signature · 1 line
1.7.1
author
#6 lore raw

PATCH 6/7 KVM: Push struct x86_exception info the various gva_to_gpa variants

In reply to Avi Kivity
---
 arch/x86/include/asm/kvm_host.h |   14 ++++++---
 arch/x86/kvm/mmu.c              |   13 +++++----
 arch/x86/kvm/paging_tmpl.h      |   19 +++++++++----
 arch/x86/kvm/x86.c              |   52 ++++++++++++++++----------------------
 4 files changed, 51 insertions(+), 47 deletions(-)
arch/x86/include/asm/kvm_host.h +9 -5
index b04c0fa..24bad0a 100644
@@ -245,7 +245,7 @@ struct kvm_mmu {
 	void (*inject_page_fault)(struct kvm_vcpu *vcpu);
 	void (*free)(struct kvm_vcpu *vcpu);
 	gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
-			    u32 *error);
+			    struct x86_exception *exception);
 	gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
 	void (*prefetch_page)(struct kvm_vcpu *vcpu,
 			      struct kvm_mmu_page *page);
@@ -707,10 +707,14 @@ void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
 int kvm_mmu_load(struct kvm_vcpu *vcpu);
 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
-gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
-gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
-gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
-gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva, u32 *error);
+gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
+			      struct x86_exception *exception);
+gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
+			       struct x86_exception *exception);
+gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
+			       struct x86_exception *exception);
+gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
+				struct x86_exception *exception);
 
 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
 
arch/x86/kvm/mmu.c +7 -6
index bdb9fa9..28ddc13 100644
@@ -2568,18 +2568,19 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
 }
 
 static gpa_t nonpaging_gva_to_gpa(struct kvm_vcpu *vcpu, gva_t vaddr,
-				  u32 access, u32 *error)
+				  u32 access, struct x86_exception *exception)
 {
-	if (error)
-		*error = 0;
+	if (exception)
+		exception->error_code = 0;
 	return vaddr;
 }
 
 static gpa_t nonpaging_gva_to_gpa_nested(struct kvm_vcpu *vcpu, gva_t vaddr,
-					 u32 access, u32 *error)
+					 u32 access,
+					 struct x86_exception *exception)
 {
-	if (error)
-		*error = 0;
+	if (exception)
+		exception->error_code = 0;
 	return vcpu->arch.nested_mmu.translate_gpa(vcpu, vaddr, access);
 }
 
arch/x86/kvm/paging_tmpl.h +13 -6
index 590bf12..f448da5 100644
@@ -670,7 +670,7 @@ static void FNAME(invlpg)(struct kvm_vcpu *vcpu, gva_t gva)
 }
 
 static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t vaddr, u32 access,
-			       u32 *error)
+			       struct x86_exception *exception)
 {
 	struct guest_walker walker;
 	gpa_t gpa = UNMAPPED_GVA;
@@ -681,14 +681,18 @@ static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t vaddr, u32 access,
 	if (r) {
 		gpa = gfn_to_gpa(walker.gfn);
 		gpa |= vaddr & ~PAGE_MASK;
-	} else if (error)
-		*error = walker.error_code;
+	} else if (exception) {
+		exception->vector = PF_VECTOR;
+		exception->error_code_valid = true;
+		exception->error_code = walker.error_code;
+	}
 
 	return gpa;
 }
 
 static gpa_t FNAME(gva_to_gpa_nested)(struct kvm_vcpu *vcpu, gva_t vaddr,
-				      u32 access, u32 *error)
+				      u32 access,
+				      struct x86_exception *exception)
 {
 	struct guest_walker walker;
 	gpa_t gpa = UNMAPPED_GVA;
@@ -699,8 +703,11 @@ static gpa_t FNAME(gva_to_gpa_nested)(struct kvm_vcpu *vcpu, gva_t vaddr,
 	if (r) {
 		gpa = gfn_to_gpa(walker.gfn);
 		gpa |= vaddr & ~PAGE_MASK;
-	} else if (error)
-		*error = walker.error_code;
+	} else if (exception) {
+		exception->vector = PF_VECTOR;
+		exception->error_code_valid = true;
+		exception->error_code = walker.error_code;
+	}
 
 	return gpa;
 }
arch/x86/kvm/x86.c +22 -30

137 lines not on this page · raw message

Signature · 1 line
1.7.1
author
#7 lore raw

PATCH 7/7 KVM: Push struct x86_exception into walk_addr()

In reply to Avi Kivity
---
 arch/x86/kvm/paging_tmpl.h |   32 ++++++++++++++------------------
 1 files changed, 14 insertions(+), 18 deletions(-)
arch/x86/kvm/paging_tmpl.h +14 -18
index f448da5..aaed914 100644
@@ -72,7 +72,7 @@ struct guest_walker {
 	unsigned pt_access;
 	unsigned pte_access;
 	gfn_t gfn;
-	u32 error_code;
+	struct x86_exception fault;
 };
 
 static gfn_t gpte_to_gfn_lvl(pt_element_t gpte, int lvl)
@@ -266,21 +266,23 @@ walk:
 	return 1;
 
 error:
-	walker->error_code = 0;
+	walker->fault.vector = PF_VECTOR;
+	walker->fault.error_code_valid = true;
+	walker->fault.error_code = 0;
 	if (present)
-		walker->error_code |= PFERR_PRESENT_MASK;
+		walker->fault.error_code |= PFERR_PRESENT_MASK;
 
-	walker->error_code |= write_fault | user_fault;
+	walker->fault.error_code |= write_fault | user_fault;
 
 	if (fetch_fault && mmu->nx)
-		walker->error_code |= PFERR_FETCH_MASK;
+		walker->fault.error_code |= PFERR_FETCH_MASK;
 	if (rsvd_fault)
-		walker->error_code |= PFERR_RSVD_MASK;
+		walker->fault.error_code |= PFERR_RSVD_MASK;
 
 	vcpu->arch.fault.address    = addr;
-	vcpu->arch.fault.error_code = walker->error_code;
+	vcpu->arch.fault.error_code = walker->fault.error_code;
 
-	trace_kvm_mmu_walker_error(walker->error_code);
+	trace_kvm_mmu_walker_error(walker->fault.error_code);
 	return 0;
 }
 
@@ -681,11 +683,8 @@ static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t vaddr, u32 access,
 	if (r) {
 		gpa = gfn_to_gpa(walker.gfn);
 		gpa |= vaddr & ~PAGE_MASK;
-	} else if (exception) {
-		exception->vector = PF_VECTOR;
-		exception->error_code_valid = true;
-		exception->error_code = walker.error_code;
-	}
+	} else if (exception)
+		*exception = walker.fault;
 
 	return gpa;
 }
@@ -703,11 +702,8 @@ static gpa_t FNAME(gva_to_gpa_nested)(struct kvm_vcpu *vcpu, gva_t vaddr,
 	if (r) {
 		gpa = gfn_to_gpa(walker.gfn);
 		gpa |= vaddr & ~PAGE_MASK;
-	} else if (exception) {
-		exception->vector = PF_VECTOR;
-		exception->error_code_valid = true;
-		exception->error_code = walker.error_code;
-	}
+	} else if (exception)
+		*exception = walker.fault;
 
 	return gpa;
 }
Signature · 1 line
1.7.1
#8 lore raw
In reply to Avi Kivity
Quoted 20 lines · Avi Kivity

This boring patchset introduces a struct x86_exception to represent an exception, and uses it to communicate between the emulator and the rest of kvm. The primary benefit is that we can now pass a #GP from kvm into the emulator, not just #PFs.

I'd like to also fold vcpu->arch.fault into struct x86_exception, but that's more difficult, so deferred until later.

Avi Kivity (7):
KVM: x86 emulator: introduce struct x86_exception to communicate
faults
KVM: x86 emulator: make emulator memory callbacks return full
exception
KVM: x86 emulator: drop dead pf injection in emulate_popf()
KVM: x86 emulator: tighen up ->read_std() and ->write_std() error
checks
KVM: x86 emulator: simplify exception generation
KVM: Push struct x86_exception info the various gva_to_gpa variants
KVM: Push struct x86_exception into walk_addr()

Quoted 7 lines
 arch/x86/include/asm/kvm_emulate.h |   26 +++--
 arch/x86/include/asm/kvm_host.h    |   14 ++-
 arch/x86/kvm/emulate.c             |  243 ++++++++++++++----------------------
 arch/x86/kvm/mmu.c                 |   13 +-
 arch/x86/kvm/paging_tmpl.h         |   31 +++--
 arch/x86/kvm/x86.c                 |   97 ++++++++-------
 6 files changed, 195 insertions(+), 229 deletions(-)

Applied, thanks.