[PATCH 00/22] Direct decode in emulator

[PATCH 00/22] Direct decode in emulator

[Avi Kivity]

Currently the emulator decoder is indirect in two ways: groups are looked up
in a separate table based on group number; and opcode execution is looked up
in a big switch (actually two) after the opcode flags lookup.

This patchset converts groups to direct decode and begins the work for opcode
decode.

The motivations are:
- reduce errors from discrepancies between the two tables (like we had for
  IRET)
- make the code easier to read, removing many switch () statements
- eventually remove the #UD mini-decoder in x86.c by adding subgroup support
  to the emulator

Avi Kivity (22):
  KVM: x86 emulator: drop parentheses in repreat macros
  KVM: x86 emulator: introduce 'struct opcode'
  KVM: x86 emulator: Add wrappers for easily defining opcodes
  KVM: x86 emulator: move group tables to top
  KVM: x86 emulator: reserve group code 0
  KVM: x86 emulator: allow specifying group directly in opcode
  KVM: x86 emulator: convert group 1 to new style
  KVM: x86 emulator: convert group 1A to new style
  KVM: x86 emulator: convert group 3 to new style
  KVM: x86 emulator: convert group 4 to new style
  KVM: x86 emulator: convert group 5 to new style
  KVM: x86 emulator: convert group 7 to new style
  KVM: x86 emulator: convert group 8 to new style
  KVM: x86 emulator: convert group 9 to new style
  KVM: x86 emulator: drop support for old-style groups
  KVM: x86 emulator: move ByteOp and Dst back to bits 0:3
  KVM: x86 emulator: store x86_emulate_ops in emulation context
  KVM: x86 emulator: allow storing emulator execution function in
    decode tables
  KVM: x86 emulator: move x86_decode_insn() downwards
  KVM: x86 emulator: move decode tables downwards
  KVM: x86 emulator: allow repeat macro arguments to contain commas
  KVM: x86 emulator: convert some push instructions to direct decode

 arch/x86/include/asm/kvm_emulate.h |   10 +-
 arch/x86/kvm/emulate.c             | 1253 +++++++++++++++++++-----------------
 arch/x86/kvm/x86.c                 |    7 +-
 3 files changed, 661 insertions(+), 609 deletions(-)

[PATCH 01/22] KVM: x86 emulator: drop parentheses in repreat macros

[Avi Kivity]

The parenthese make is impossible to use the macros with initializers that
require braces.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 3d88151..2f1b2c2 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -95,10 +95,10 @@
 #define Src2One     (3<<29)
 #define Src2Mask    (7<<29)
 
-#define X2(x) (x), (x)
-#define X3(x) X2(x), (x)
+#define X2(x) x, x
+#define X3(x) X2(x), x
 #define X4(x) X2(x), X2(x)
-#define X5(x) X4(x), (x)
+#define X5(x) X4(x), x
 #define X6(x) X4(x), X2(x)
 #define X7(x) X4(x), X3(x)
 #define X8(x) X4(x), X4(x)
-- 
1.7.1

[PATCH 02/22] KVM: x86 emulator: introduce 'struct opcode'

[Avi Kivity]

This will hold all the information known about the opcode.  Currently, this
is just the decode flags.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   20 ++++++++++++--------
 1 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 2f1b2c2..afe778a 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -108,7 +108,11 @@ enum {
 	Group1, Group1A, Group3, Group4, Group5, Group7, Group8, Group9,
 };
 
-static u32 opcode_table[256] = {
+struct opcode {
+	u32 flags;
+};
+
+static struct opcode opcode_table[256] = {
 	/* 0x00 - 0x07 */
 	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
 	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
@@ -222,7 +226,7 @@ static u32 opcode_table[256] = {
 	ImplicitOps, ImplicitOps, Group | Group4, Group | Group5,
 };
 
-static u32 twobyte_table[256] = {
+static struct opcode twobyte_table[256] = {
 	/* 0x00 - 0x0F */
 	0, Group | GroupDual | Group7, 0, 0,
 	0, ImplicitOps, ImplicitOps | Priv, 0,
@@ -284,7 +288,7 @@ static u32 twobyte_table[256] = {
 	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
 };
 
-static u32 group_table[] = {
+static struct opcode group_table[] = {
 	[Group1*8] =
 	X7(Lock), 0,
 	[Group1A*8] =
@@ -313,7 +317,7 @@ static u32 group_table[] = {
 	0, DstMem64 | ModRM | Lock, 0, 0, 0, 0, 0, 0,
 };
 
-static u32 group2_table[] = {
+static struct opcode group2_table[] = {
 	[Group7*8] =
 	SrcNone | ModRM | Priv, 0, 0, SrcNone | ModRM | Priv,
 	SrcNone | ModRM | DstMem | Mov, 0,
@@ -1005,13 +1009,13 @@ done_prefixes:
 			c->op_bytes = 8;	/* REX.W */
 
 	/* Opcode byte(s). */
-	c->d = opcode_table[c->b];
+	c->d = opcode_table[c->b].flags;
 	if (c->d == 0) {
 		/* Two-byte opcode? */
 		if (c->b == 0x0f) {
 			c->twobyte = 1;
 			c->b = insn_fetch(u8, 1, c->eip);
-			c->d = twobyte_table[c->b];
+			c->d = twobyte_table[c->b].flags;
 		}
 	}
 
@@ -1024,9 +1028,9 @@ done_prefixes:
 		group = (group << 3) + ((c->modrm >> 3) & 7);
 		c->d &= ~(Group | GroupDual | GroupMask);
 		if (dual && (c->modrm >> 6) == 3)
-			c->d |= group2_table[group];
+			c->d |= group2_table[group].flags;
 		else
-			c->d |= group_table[group];
+			c->d |= group_table[group].flags;
 	}
 
 	/* Unrecognised? */
-- 
1.7.1

[PATCH 03/22] KVM: x86 emulator: Add wrappers for easily defining opcodes

[Avi Kivity]

Once 'struct opcode' grows, its initializer will become more complicated.
Wrap the simple initializers in a D() macro, and replace the empty initializers
with an even simpler N macro.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |  294 ++++++++++++++++++++++++-----------------------
 1 files changed, 150 insertions(+), 144 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index afe778a..cd8a219 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -112,220 +112,226 @@ struct opcode {
 	u32 flags;
 };
 
+#define D(_y) { .flags = (_y) }
+#define N    D(0)
+
 static struct opcode opcode_table[256] = {
 	/* 0x00 - 0x07 */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImm, DstAcc | SrcImm,
-	ImplicitOps | Stack | No64, ImplicitOps | Stack | No64,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
 	/* 0x08 - 0x0F */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImm, DstAcc | SrcImm,
-	ImplicitOps | Stack | No64, 0,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), N,
 	/* 0x10 - 0x17 */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImm, DstAcc | SrcImm,
-	ImplicitOps | Stack | No64, ImplicitOps | Stack | No64,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
 	/* 0x18 - 0x1F */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImm, DstAcc | SrcImm,
-	ImplicitOps | Stack | No64, ImplicitOps | Stack | No64,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
 	/* 0x20 - 0x27 */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImmByte, DstAcc | SrcImm, 0, 0,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
 	/* 0x28 - 0x2F */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImmByte, DstAcc | SrcImm, 0, 0,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
 	/* 0x30 - 0x37 */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImmByte, DstAcc | SrcImm, 0, 0,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
 	/* 0x38 - 0x3F */
-	ByteOp | DstMem | SrcReg | ModRM, DstMem | SrcReg | ModRM,
-	ByteOp | DstReg | SrcMem | ModRM, DstReg | SrcMem | ModRM,
-	ByteOp | DstAcc | SrcImm, DstAcc | SrcImm,
-	0, 0,
+	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	N, N,
 	/* 0x40 - 0x4F */
-	X16(DstReg),
+	X16(D(DstReg)),
 	/* 0x50 - 0x57 */
-	X8(SrcReg | Stack),
+	X8(D(SrcReg | Stack)),
 	/* 0x58 - 0x5F */
-	X8(DstReg | Stack),
+	X8(D(DstReg | Stack)),
 	/* 0x60 - 0x67 */
-	ImplicitOps | Stack | No64, ImplicitOps | Stack | No64,
-	0, DstReg | SrcMem32 | ModRM | Mov /* movsxd (x86/64) */ ,
-	0, 0, 0, 0,
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
+	N, D(DstReg | SrcMem32 | ModRM | Mov) /* movsxd (x86/64) */ ,
+	N, N, N, N,
 	/* 0x68 - 0x6F */
-	SrcImm | Mov | Stack, 0, SrcImmByte | Mov | Stack, 0,
-	DstDI | ByteOp | Mov | String, DstDI | Mov | String, /* insb, insw/insd */
-	SrcSI | ByteOp | ImplicitOps | String, SrcSI | ImplicitOps | String, /* outsb, outsw/outsd */
+	D(SrcImm | Mov | Stack), N, D(SrcImmByte | Mov | Stack), N,
+	D(DstDI | ByteOp | Mov | String), D(DstDI | Mov | String), /* insb, insw/insd */
+	D(SrcSI | ByteOp | ImplicitOps | String), D(SrcSI | ImplicitOps | String), /* outsb, outsw/outsd */
 	/* 0x70 - 0x7F */
-	X16(SrcImmByte),
+	X16(D(SrcImmByte)),
 	/* 0x80 - 0x87 */
-	ByteOp | DstMem | SrcImm | ModRM | Group | Group1,
-	DstMem | SrcImm | ModRM | Group | Group1,
-	ByteOp | DstMem | SrcImm | ModRM | No64 | Group | Group1,
-	DstMem | SrcImmByte | ModRM | Group | Group1,
-	ByteOp | DstMem | SrcReg | ModRM, DstMem | SrcReg | ModRM,
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
+	D(ByteOp | DstMem | SrcImm | ModRM | Group | Group1),
+	D(DstMem | SrcImm | ModRM | Group | Group1),
+	D(ByteOp | DstMem | SrcImm | ModRM | No64 | Group | Group1),
+	D(DstMem | SrcImmByte | ModRM | Group | Group1),
+	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
 	/* 0x88 - 0x8F */
-	ByteOp | DstMem | SrcReg | ModRM | Mov, DstMem | SrcReg | ModRM | Mov,
-	ByteOp | DstReg | SrcMem | ModRM | Mov, DstReg | SrcMem | ModRM | Mov,
-	DstMem | SrcNone | ModRM | Mov, ModRM | DstReg,
-	ImplicitOps | SrcMem16 | ModRM, Group | Group1A,
+	D(ByteOp | DstMem | SrcReg | ModRM | Mov), D(DstMem | SrcReg | ModRM | Mov),
+	D(ByteOp | DstReg | SrcMem | ModRM | Mov), D(DstReg | SrcMem | ModRM | Mov),
+	D(DstMem | SrcNone | ModRM | Mov), D(ModRM | DstReg),
+	D(ImplicitOps | SrcMem16 | ModRM), D(Group | Group1A),
 	/* 0x90 - 0x97 */
-	DstReg, DstReg, DstReg, DstReg,	DstReg, DstReg, DstReg, DstReg,
+	D(DstReg), D(DstReg), D(DstReg), D(DstReg),	D(DstReg), D(DstReg), D(DstReg), D(DstReg),
 	/* 0x98 - 0x9F */
-	0, 0, SrcImmFAddr | No64, 0,
-	ImplicitOps | Stack, ImplicitOps | Stack, 0, 0,
+	N, N, D(SrcImmFAddr | No64), N,
+	D(ImplicitOps | Stack), D(ImplicitOps | Stack), N, N,
 	/* 0xA0 - 0xA7 */
-	ByteOp | DstAcc | SrcMem | Mov | MemAbs, DstAcc | SrcMem | Mov | MemAbs,
-	ByteOp | DstMem | SrcAcc | Mov | MemAbs, DstMem | SrcAcc | Mov | MemAbs,
-	ByteOp | SrcSI | DstDI | Mov | String, SrcSI | DstDI | Mov | String,
-	ByteOp | SrcSI | DstDI | String, SrcSI | DstDI | String,
+	D(ByteOp | DstAcc | SrcMem | Mov | MemAbs), D(DstAcc | SrcMem | Mov | MemAbs),
+	D(ByteOp | DstMem | SrcAcc | Mov | MemAbs), D(DstMem | SrcAcc | Mov | MemAbs),
+	D(ByteOp | SrcSI | DstDI | Mov | String), D(SrcSI | DstDI | Mov | String),
+	D(ByteOp | SrcSI | DstDI | String), D(SrcSI | DstDI | String),
 	/* 0xA8 - 0xAF */
-	DstAcc | SrcImmByte | ByteOp, DstAcc | SrcImm, ByteOp | DstDI | Mov | String, DstDI | Mov | String,
-	ByteOp | SrcSI | DstAcc | Mov | String, SrcSI | DstAcc | Mov | String,
-	ByteOp | DstDI | String, DstDI | String,
+	D(DstAcc | SrcImmByte | ByteOp), D(DstAcc | SrcImm), D(ByteOp | DstDI | Mov | String), D(DstDI | Mov | String),
+	D(ByteOp | SrcSI | DstAcc | Mov | String), D(SrcSI | DstAcc | Mov | String),
+	D(ByteOp | DstDI | String), D(DstDI | String),
 	/* 0xB0 - 0xB7 */
-	X8(ByteOp | DstReg | SrcImm | Mov),
+	X8(D(ByteOp | DstReg | SrcImm | Mov)),
 	/* 0xB8 - 0xBF */
-	X8(DstReg | SrcImm | Mov),
+	X8(D(DstReg | SrcImm | Mov)),
 	/* 0xC0 - 0xC7 */
-	ByteOp | DstMem | SrcImm | ModRM, DstMem | SrcImmByte | ModRM,
-	0, ImplicitOps | Stack, 0, 0,
-	ByteOp | DstMem | SrcImm | ModRM | Mov, DstMem | SrcImm | ModRM | Mov,
+	D(ByteOp | DstMem | SrcImm | ModRM), D(DstMem | SrcImmByte | ModRM),
+	N, D(ImplicitOps | Stack), N, N,
+	D(ByteOp | DstMem | SrcImm | ModRM | Mov), D(DstMem | SrcImm | ModRM | Mov),
 	/* 0xC8 - 0xCF */
-	0, 0, 0, ImplicitOps | Stack,
-	ImplicitOps, SrcImmByte, ImplicitOps | No64, ImplicitOps,
+	N, N, N, D(ImplicitOps | Stack),
+	D(ImplicitOps), D(SrcImmByte), D(ImplicitOps | No64), D(ImplicitOps),
 	/* 0xD0 - 0xD7 */
-	ByteOp | DstMem | SrcImplicit | ModRM, DstMem | SrcImplicit | ModRM,
-	ByteOp | DstMem | SrcImplicit | ModRM, DstMem | SrcImplicit | ModRM,
-	0, 0, 0, 0,
+	D(ByteOp | DstMem | SrcImplicit | ModRM), D(DstMem | SrcImplicit | ModRM),
+	D(ByteOp | DstMem | SrcImplicit | ModRM), D(DstMem | SrcImplicit | ModRM),
+	N, N, N, N,
 	/* 0xD8 - 0xDF */
-	0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N,
 	/* 0xE0 - 0xE7 */
-	0, 0, 0, 0,
-	ByteOp | SrcImmUByte | DstAcc, SrcImmUByte | DstAcc,
-	ByteOp | SrcImmUByte | DstAcc, SrcImmUByte | DstAcc,
+	N, N, N, N,
+	D(ByteOp | SrcImmUByte | DstAcc), D(SrcImmUByte | DstAcc),
+	D(ByteOp | SrcImmUByte | DstAcc), D(SrcImmUByte | DstAcc),
 	/* 0xE8 - 0xEF */
-	SrcImm | Stack, SrcImm | ImplicitOps,
-	SrcImmFAddr | No64, SrcImmByte | ImplicitOps,
-	SrcNone | ByteOp | DstAcc, SrcNone | DstAcc,
-	SrcNone | ByteOp | DstAcc, SrcNone | DstAcc,
+	D(SrcImm | Stack), D(SrcImm | ImplicitOps),
+	D(SrcImmFAddr | No64), D(SrcImmByte | ImplicitOps),
+	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
+	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
 	/* 0xF0 - 0xF7 */
-	0, 0, 0, 0,
-	ImplicitOps | Priv, ImplicitOps, ByteOp | Group | Group3, Group | Group3,
+	N, N, N, N,
+	D(ImplicitOps | Priv), D(ImplicitOps), D(ByteOp | Group | Group3), D(Group | Group3),
 	/* 0xF8 - 0xFF */
-	ImplicitOps, 0, ImplicitOps, ImplicitOps,
-	ImplicitOps, ImplicitOps, Group | Group4, Group | Group5,
+	D(ImplicitOps), N, D(ImplicitOps), D(ImplicitOps),
+	D(ImplicitOps), D(ImplicitOps), D(Group | Group4), D(Group | Group5),
 };
 
 static struct opcode twobyte_table[256] = {
 	/* 0x00 - 0x0F */
-	0, Group | GroupDual | Group7, 0, 0,
-	0, ImplicitOps, ImplicitOps | Priv, 0,
-	ImplicitOps | Priv, ImplicitOps | Priv, 0, 0,
-	0, ImplicitOps | ModRM, 0, 0,
+	N, D(Group | GroupDual | Group7), N, N,
+	N, D(ImplicitOps), D(ImplicitOps | Priv), N,
+	D(ImplicitOps | Priv), D(ImplicitOps | Priv), N, N,
+	N, D(ImplicitOps | ModRM), N, N,
 	/* 0x10 - 0x1F */
-	0, 0, 0, 0, 0, 0, 0, 0, ImplicitOps | ModRM, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, D(ImplicitOps | ModRM), N, N, N, N, N, N, N,
 	/* 0x20 - 0x2F */
-	ModRM | ImplicitOps | Priv, ModRM | Priv,
-	ModRM | ImplicitOps | Priv, ModRM | Priv,
-	0, 0, 0, 0,
-	0, 0, 0, 0, 0, 0, 0, 0,
+	D(ModRM | ImplicitOps | Priv), D(ModRM | Priv),
+	D(ModRM | ImplicitOps | Priv), D(ModRM | Priv),
+	N, N, N, N,
+	N, N, N, N, N, N, N, N,
 	/* 0x30 - 0x3F */
-	ImplicitOps | Priv, 0, ImplicitOps | Priv, 0,
-	ImplicitOps, ImplicitOps | Priv, 0, 0,
-	0, 0, 0, 0, 0, 0, 0, 0,
+	D(ImplicitOps | Priv), N, D(ImplicitOps | Priv), N,
+	D(ImplicitOps), D(ImplicitOps | Priv), N, N,
+	N, N, N, N, N, N, N, N,
 	/* 0x40 - 0x4F */
-	X16(DstReg | SrcMem | ModRM | Mov),
+	X16(D(DstReg | SrcMem | ModRM | Mov)),
 	/* 0x50 - 0x5F */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
 	/* 0x60 - 0x6F */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
 	/* 0x70 - 0x7F */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
 	/* 0x80 - 0x8F */
-	X16(SrcImm),
+	X16(D(SrcImm)),
 	/* 0x90 - 0x9F */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
 	/* 0xA0 - 0xA7 */
-	ImplicitOps | Stack, ImplicitOps | Stack,
-	0, DstMem | SrcReg | ModRM | BitOp,
-	DstMem | SrcReg | Src2ImmByte | ModRM,
-	DstMem | SrcReg | Src2CL | ModRM, 0, 0,
+	D(ImplicitOps | Stack), D(ImplicitOps | Stack),
+	N, D(DstMem | SrcReg | ModRM | BitOp),
+	D(DstMem | SrcReg | Src2ImmByte | ModRM),
+	D(DstMem | SrcReg | Src2CL | ModRM), N, N,
 	/* 0xA8 - 0xAF */
-	ImplicitOps | Stack, ImplicitOps | Stack,
-	0, DstMem | SrcReg | ModRM | BitOp | Lock,
-	DstMem | SrcReg | Src2ImmByte | ModRM,
-	DstMem | SrcReg | Src2CL | ModRM,
-	ModRM, 0,
+	D(ImplicitOps | Stack), D(ImplicitOps | Stack),
+	N, D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	D(DstMem | SrcReg | Src2ImmByte | ModRM),
+	D(DstMem | SrcReg | Src2CL | ModRM),
+	D(ModRM), N,
 	/* 0xB0 - 0xB7 */
-	ByteOp | DstMem | SrcReg | ModRM | Lock, DstMem | SrcReg | ModRM | Lock,
-	0, DstMem | SrcReg | ModRM | BitOp | Lock,
-	0, 0, ByteOp | DstReg | SrcMem | ModRM | Mov,
-	    DstReg | SrcMem16 | ModRM | Mov,
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	N, D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
+	    D(DstReg | SrcMem16 | ModRM | Mov),
 	/* 0xB8 - 0xBF */
-	0, 0,
-	Group | Group8, DstMem | SrcReg | ModRM | BitOp | Lock,
-	0, 0, ByteOp | DstReg | SrcMem | ModRM | Mov,
-	    DstReg | SrcMem16 | ModRM | Mov,
+	N, N,
+	D(Group | Group8), D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
+	    D(DstReg | SrcMem16 | ModRM | Mov),
 	/* 0xC0 - 0xCF */
-	0, 0, 0, DstMem | SrcReg | ModRM | Mov,
-	0, 0, 0, Group | GroupDual | Group9,
-	0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, D(DstMem | SrcReg | ModRM | Mov),
+	N, N, N, D(Group | GroupDual | Group9),
+	N, N, N, N, N, N, N, N,
 	/* 0xD0 - 0xDF */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
 	/* 0xE0 - 0xEF */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
 	/* 0xF0 - 0xFF */
-	0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N
 };
 
 static struct opcode group_table[] = {
 	[Group1*8] =
-	X7(Lock), 0,
+	X7(D(Lock)), N,
 	[Group1A*8] =
-	DstMem | SrcNone | ModRM | Mov | Stack, 0, 0, 0, 0, 0, 0, 0,
+	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
 	[Group3*8] =
-	DstMem | SrcImm | ModRM, DstMem | SrcImm | ModRM,
-	DstMem | SrcNone | ModRM | Lock, DstMem | SrcNone | ModRM | Lock,
-	X4(Undefined),
+	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
+	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
+	X4(D(Undefined)),
 	[Group4*8] =
-	ByteOp | DstMem | SrcNone | ModRM | Lock, ByteOp | DstMem | SrcNone | ModRM | Lock,
-	0, 0, 0, 0, 0, 0,
+	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
+	N, N, N, N, N, N,
 	[Group5*8] =
-	DstMem | SrcNone | ModRM | Lock, DstMem | SrcNone | ModRM | Lock,
-	SrcMem | ModRM | Stack, 0,
-	SrcMem | ModRM | Stack, SrcMemFAddr | ModRM | ImplicitOps,
-	SrcMem | ModRM | Stack, 0,
+	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
+	D(SrcMem | ModRM | Stack), N,
+	D(SrcMem | ModRM | Stack), D(SrcMemFAddr | ModRM | ImplicitOps),
+	D(SrcMem | ModRM | Stack), N,
 	[Group7*8] =
-	0, 0, ModRM | SrcMem | Priv, ModRM | SrcMem | Priv,
-	SrcNone | ModRM | DstMem | Mov, 0,
-	SrcMem16 | ModRM | Mov | Priv, SrcMem | ModRM | ByteOp | Priv,
+	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), D(SrcMem | ModRM | ByteOp | Priv),
 	[Group8*8] =
-	0, 0, 0, 0,
-	DstMem | SrcImmByte | ModRM, DstMem | SrcImmByte | ModRM | Lock,
-	DstMem | SrcImmByte | ModRM | Lock, DstMem | SrcImmByte | ModRM | Lock,
+	N, N, N, N,
+	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
+	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
 	[Group9*8] =
-	0, DstMem64 | ModRM | Lock, 0, 0, 0, 0, 0, 0,
+	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
 };
 
 static struct opcode group2_table[] = {
 	[Group7*8] =
-	SrcNone | ModRM | Priv, 0, 0, SrcNone | ModRM | Priv,
-	SrcNone | ModRM | DstMem | Mov, 0,
-	SrcMem16 | ModRM | Mov | Priv, 0,
+	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), N,
 	[Group9*8] =
-	0, 0, 0, 0, 0, 0, 0, 0,
+	N, N, N, N, N, N, N, N,
 };
 
+#undef D
+#undef N
+
 /* EFLAGS bit definitions. */
 #define EFLG_ID (1<<21)
 #define EFLG_VIP (1<<20)
-- 
1.7.1

[PATCH 04/22] KVM: x86 emulator: move group tables to top

[Avi Kivity]

No code changes.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   76 ++++++++++++++++++++++++------------------------
 1 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index cd8a219..91359b5 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -115,6 +115,44 @@ struct opcode {
 #define D(_y) { .flags = (_y) }
 #define N    D(0)
 
+static struct opcode group_table[] = {
+	[Group1*8] =
+	X7(D(Lock)), N,
+	[Group1A*8] =
+	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
+	[Group3*8] =
+	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
+	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
+	X4(D(Undefined)),
+	[Group4*8] =
+	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
+	N, N, N, N, N, N,
+	[Group5*8] =
+	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
+	D(SrcMem | ModRM | Stack), N,
+	D(SrcMem | ModRM | Stack), D(SrcMemFAddr | ModRM | ImplicitOps),
+	D(SrcMem | ModRM | Stack), N,
+	[Group7*8] =
+	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), D(SrcMem | ModRM | ByteOp | Priv),
+	[Group8*8] =
+	N, N, N, N,
+	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
+	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
+	[Group9*8] =
+	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
+};
+
+static struct opcode group2_table[] = {
+	[Group7*8] =
+	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), N,
+	[Group9*8] =
+	N, N, N, N, N, N, N, N,
+};
+
 static struct opcode opcode_table[256] = {
 	/* 0x00 - 0x07 */
 	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
@@ -291,44 +329,6 @@ static struct opcode twobyte_table[256] = {
 	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N
 };
 
-static struct opcode group_table[] = {
-	[Group1*8] =
-	X7(D(Lock)), N,
-	[Group1A*8] =
-	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
-	[Group3*8] =
-	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
-	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
-	X4(D(Undefined)),
-	[Group4*8] =
-	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
-	N, N, N, N, N, N,
-	[Group5*8] =
-	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
-	D(SrcMem | ModRM | Stack), N,
-	D(SrcMem | ModRM | Stack), D(SrcMemFAddr | ModRM | ImplicitOps),
-	D(SrcMem | ModRM | Stack), N,
-	[Group7*8] =
-	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
-	D(SrcNone | ModRM | DstMem | Mov), N,
-	D(SrcMem16 | ModRM | Mov | Priv), D(SrcMem | ModRM | ByteOp | Priv),
-	[Group8*8] =
-	N, N, N, N,
-	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
-	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
-	[Group9*8] =
-	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
-};
-
-static struct opcode group2_table[] = {
-	[Group7*8] =
-	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
-	D(SrcNone | ModRM | DstMem | Mov), N,
-	D(SrcMem16 | ModRM | Mov | Priv), N,
-	[Group9*8] =
-	N, N, N, N, N, N, N, N,
-};
-
 #undef D
 #undef N
 
-- 
1.7.1

[PATCH 05/22] KVM: x86 emulator: reserve group code 0

[Avi Kivity]

We'll be using that to distinguish between new-style and old-style groups.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 91359b5..b0100f6 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	Group1, Group1A, Group3, Group4, Group5, Group7, Group8, Group9,
+	NoGrp, Group1, Group1A, Group3, Group4, Group5, Group7, Group8, Group9,
 };
 
 struct opcode {
-- 
1.7.1

[PATCH 06/22] KVM: x86 emulator: allow specifying group directly in opcode

[Avi Kivity]

Instead of having a group number, store the group table pointer directly in
the opcode.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   47 ++++++++++++++++++++++++++++++++++++++---------
 1 files changed, 38 insertions(+), 9 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index b0100f6..ee5b38d 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -110,10 +110,21 @@ enum {
 
 struct opcode {
 	u32 flags;
+	union {
+		struct opcode *group;
+		struct group_dual *gdual;
+	} u;
+};
+
+struct group_dual {
+	struct opcode mod012[8];
+	struct opcode mod3[8];
 };
 
 #define D(_y) { .flags = (_y) }
 #define N    D(0)
+#define G(_f, _g) { .flags = ((_f) | Group), .u.group = (_g) }
+#define GD(_f, _g) { .flags = ((_f) | Group | GroupDual), .u.gdual = (_g) }
 
 static struct opcode group_table[] = {
 	[Group1*8] =
@@ -331,6 +342,8 @@ static struct opcode twobyte_table[256] = {
 
 #undef D
 #undef N
+#undef G
+#undef GD
 
 /* EFLAGS bit definitions. */
 #define EFLG_ID (1<<21)
@@ -927,8 +940,8 @@ x86_decode_insn(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 	struct decode_cache *c = &ctxt->decode;
 	int rc = X86EMUL_CONTINUE;
 	int mode = ctxt->mode;
-	int def_op_bytes, def_ad_bytes, group, dual;
-
+	int def_op_bytes, def_ad_bytes, group, dual, goffset;
+	struct opcode opcode, *g_mod012, *g_mod3;
 
 	/* we cannot decode insn before we complete previous rep insn */
 	WARN_ON(ctxt->restart);
@@ -1015,15 +1028,16 @@ done_prefixes:
 			c->op_bytes = 8;	/* REX.W */
 
 	/* Opcode byte(s). */
-	c->d = opcode_table[c->b].flags;
-	if (c->d == 0) {
+	opcode = opcode_table[c->b];
+	if (opcode.flags == 0) {
 		/* Two-byte opcode? */
 		if (c->b == 0x0f) {
 			c->twobyte = 1;
 			c->b = insn_fetch(u8, 1, c->eip);
-			c->d = twobyte_table[c->b].flags;
+			opcode = twobyte_table[c->b];
 		}
 	}
+	c->d = opcode.flags;
 
 	if (c->d & Group) {
 		group = c->d & GroupMask;
@@ -1031,12 +1045,27 @@ done_prefixes:
 		c->modrm = insn_fetch(u8, 1, c->eip);
 		--c->eip;
 
-		group = (group << 3) + ((c->modrm >> 3) & 7);
+		if (group) {
+			g_mod012 = g_mod3 = &group_table[group * 8];
+			if (c->d & GroupDual)
+				g_mod3 = &group2_table[group * 8];
+		} else {
+			if (c->d & GroupDual) {
+				g_mod012 = opcode.u.gdual->mod012;
+				g_mod3 = opcode.u.gdual->mod3;
+			} else
+				g_mod012 = g_mod3 = opcode.u.group;
+		}
+
 		c->d &= ~(Group | GroupDual | GroupMask);
-		if (dual && (c->modrm >> 6) == 3)
-			c->d |= group2_table[group].flags;
+
+		goffset = (c->modrm >> 3) & 7;
+
+		if ((c->modrm >> 6) == 3)
+			opcode = g_mod3[goffset];
 		else
-			c->d |= group_table[group].flags;
+			opcode = g_mod012[goffset];
+		c->d |= opcode.flags;
 	}
 
 	/* Unrecognised? */
-- 
1.7.1

[PATCH 07/22] KVM: x86 emulator: convert group 1 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   16 +++++++++-------
 1 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index ee5b38d..139a324 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group1, Group1A, Group3, Group4, Group5, Group7, Group8, Group9,
+	NoGrp, Group1A, Group3, Group4, Group5, Group7, Group8, Group9,
 };
 
 struct opcode {
@@ -126,9 +126,11 @@ struct group_dual {
 #define G(_f, _g) { .flags = ((_f) | Group), .u.group = (_g) }
 #define GD(_f, _g) { .flags = ((_f) | Group | GroupDual), .u.gdual = (_g) }
 
+static struct opcode group1[] = {
+	X7(D(Lock)), N
+};
+
 static struct opcode group_table[] = {
-	[Group1*8] =
-	X7(D(Lock)), N,
 	[Group1A*8] =
 	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
 	[Group3*8] =
@@ -219,10 +221,10 @@ static struct opcode opcode_table[256] = {
 	/* 0x70 - 0x7F */
 	X16(D(SrcImmByte)),
 	/* 0x80 - 0x87 */
-	D(ByteOp | DstMem | SrcImm | ModRM | Group | Group1),
-	D(DstMem | SrcImm | ModRM | Group | Group1),
-	D(ByteOp | DstMem | SrcImm | ModRM | No64 | Group | Group1),
-	D(DstMem | SrcImmByte | ModRM | Group | Group1),
+	G(ByteOp | DstMem | SrcImm | ModRM | Group, group1),
+	G(DstMem | SrcImm | ModRM | Group, group1),
+	G(ByteOp | DstMem | SrcImm | ModRM | No64 | Group, group1),
+	G(DstMem | SrcImmByte | ModRM | Group, group1),
 	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
 	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
 	/* 0x88 - 0x8F */
-- 
1.7.1

[PATCH 08/22] KVM: x86 emulator: convert group 1A to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 139a324..637f2c9 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group1A, Group3, Group4, Group5, Group7, Group8, Group9,
+	NoGrp, Group3, Group4, Group5, Group7, Group8, Group9,
 };
 
 struct opcode {
@@ -130,9 +130,11 @@ static struct opcode group1[] = {
 	X7(D(Lock)), N
 };
 
-static struct opcode group_table[] = {
-	[Group1A*8] =
+static struct opcode group1A[] = {
 	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
+};
+
+static struct opcode group_table[] = {
 	[Group3*8] =
 	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
 	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
@@ -231,7 +233,7 @@ static struct opcode opcode_table[256] = {
 	D(ByteOp | DstMem | SrcReg | ModRM | Mov), D(DstMem | SrcReg | ModRM | Mov),
 	D(ByteOp | DstReg | SrcMem | ModRM | Mov), D(DstReg | SrcMem | ModRM | Mov),
 	D(DstMem | SrcNone | ModRM | Mov), D(ModRM | DstReg),
-	D(ImplicitOps | SrcMem16 | ModRM), D(Group | Group1A),
+	D(ImplicitOps | SrcMem16 | ModRM), G(0, group1A),
 	/* 0x90 - 0x97 */
 	D(DstReg), D(DstReg), D(DstReg), D(DstReg),	D(DstReg), D(DstReg), D(DstReg), D(DstReg),
 	/* 0x98 - 0x9F */
-- 
1.7.1

[PATCH 09/22] KVM: x86 emulator: convert group 3 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 637f2c9..6fee257 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group3, Group4, Group5, Group7, Group8, Group9,
+	NoGrp, Group4, Group5, Group7, Group8, Group9,
 };
 
 struct opcode {
@@ -134,11 +134,13 @@ static struct opcode group1A[] = {
 	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
 };
 
-static struct opcode group_table[] = {
-	[Group3*8] =
+static struct opcode group3[] = {
 	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
 	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
 	X4(D(Undefined)),
+};
+
+static struct opcode group_table[] = {
 	[Group4*8] =
 	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
 	N, N, N, N, N, N,
@@ -276,7 +278,7 @@ static struct opcode opcode_table[256] = {
 	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
 	/* 0xF0 - 0xF7 */
 	N, N, N, N,
-	D(ImplicitOps | Priv), D(ImplicitOps), D(ByteOp | Group | Group3), D(Group | Group3),
+	D(ImplicitOps | Priv), D(ImplicitOps), G(ByteOp, group3), G(0, group3),
 	/* 0xF8 - 0xFF */
 	D(ImplicitOps), N, D(ImplicitOps), D(ImplicitOps),
 	D(ImplicitOps), D(ImplicitOps), D(Group | Group4), D(Group | Group5),
-- 
1.7.1

[PATCH 10/22] KVM: x86 emulator: convert group 4 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 6fee257..afdfd86 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group4, Group5, Group7, Group8, Group9,
+	NoGrp, Group5, Group7, Group8, Group9,
 };
 
 struct opcode {
@@ -140,10 +140,12 @@ static struct opcode group3[] = {
 	X4(D(Undefined)),
 };
 
-static struct opcode group_table[] = {
-	[Group4*8] =
+static struct opcode group4[] = {
 	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
 	N, N, N, N, N, N,
+};
+
+static struct opcode group_table[] = {
 	[Group5*8] =
 	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
 	D(SrcMem | ModRM | Stack), N,
@@ -281,7 +283,7 @@ static struct opcode opcode_table[256] = {
 	D(ImplicitOps | Priv), D(ImplicitOps), G(ByteOp, group3), G(0, group3),
 	/* 0xF8 - 0xFF */
 	D(ImplicitOps), N, D(ImplicitOps), D(ImplicitOps),
-	D(ImplicitOps), D(ImplicitOps), D(Group | Group4), D(Group | Group5),
+	D(ImplicitOps), D(ImplicitOps), G(0, group4), D(Group | Group5),
 };
 
 static struct opcode twobyte_table[256] = {
-- 
1.7.1

[PATCH 11/22] KVM: x86 emulator: convert group 5 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index afdfd86..39745e5 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group5, Group7, Group8, Group9,
+	NoGrp, Group7, Group8, Group9,
 };
 
 struct opcode {
@@ -145,12 +145,14 @@ static struct opcode group4[] = {
 	N, N, N, N, N, N,
 };
 
-static struct opcode group_table[] = {
-	[Group5*8] =
+static struct opcode group5[] = {
 	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
 	D(SrcMem | ModRM | Stack), N,
 	D(SrcMem | ModRM | Stack), D(SrcMemFAddr | ModRM | ImplicitOps),
 	D(SrcMem | ModRM | Stack), N,
+};
+
+static struct opcode group_table[] = {
 	[Group7*8] =
 	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
 	D(SrcNone | ModRM | DstMem | Mov), N,
@@ -283,7 +285,7 @@ static struct opcode opcode_table[256] = {
 	D(ImplicitOps | Priv), D(ImplicitOps), G(ByteOp, group3), G(0, group3),
 	/* 0xF8 - 0xFF */
 	D(ImplicitOps), N, D(ImplicitOps), D(ImplicitOps),
-	D(ImplicitOps), D(ImplicitOps), G(0, group4), D(Group | Group5),
+	D(ImplicitOps), D(ImplicitOps), G(0, group4), G(0, group5),
 };
 
 static struct opcode twobyte_table[256] = {
-- 
1.7.1

[PATCH 12/22] KVM: x86 emulator: convert group 7 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   18 ++++++++++--------
 1 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 39745e5..52995e3 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group7, Group8, Group9,
+	NoGrp, Group8, Group9,
 };
 
 struct opcode {
@@ -152,11 +152,17 @@ static struct opcode group5[] = {
 	D(SrcMem | ModRM | Stack), N,
 };
 
-static struct opcode group_table[] = {
-	[Group7*8] =
+static struct group_dual group7 = { {
 	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
 	D(SrcNone | ModRM | DstMem | Mov), N,
 	D(SrcMem16 | ModRM | Mov | Priv), D(SrcMem | ModRM | ByteOp | Priv),
+}, {
+	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), N,
+} };
+
+static struct opcode group_table[] = {
 	[Group8*8] =
 	N, N, N, N,
 	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
@@ -166,10 +172,6 @@ static struct opcode group_table[] = {
 };
 
 static struct opcode group2_table[] = {
-	[Group7*8] =
-	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
-	D(SrcNone | ModRM | DstMem | Mov), N,
-	D(SrcMem16 | ModRM | Mov | Priv), N,
 	[Group9*8] =
 	N, N, N, N, N, N, N, N,
 };
@@ -290,7 +292,7 @@ static struct opcode opcode_table[256] = {
 
 static struct opcode twobyte_table[256] = {
 	/* 0x00 - 0x0F */
-	N, D(Group | GroupDual | Group7), N, N,
+	N, GD(0, &group7), N, N,
 	N, D(ImplicitOps), D(ImplicitOps | Priv), N,
 	D(ImplicitOps | Priv), D(ImplicitOps | Priv), N, N,
 	N, D(ImplicitOps | ModRM), N, N,
-- 
1.7.1

[PATCH 13/22] KVM: x86 emulator: convert group 8 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 52995e3..adddf32 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group8, Group9,
+	NoGrp, Group9,
 };
 
 struct opcode {
@@ -162,11 +162,13 @@ static struct group_dual group7 = { {
 	D(SrcMem16 | ModRM | Mov | Priv), N,
 } };
 
-static struct opcode group_table[] = {
-	[Group8*8] =
+static struct opcode group8[] = {
 	N, N, N, N,
 	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
 	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
+};
+
+static struct opcode group_table[] = {
 	[Group9*8] =
 	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
 };
@@ -337,7 +339,7 @@ static struct opcode twobyte_table[256] = {
 	    D(DstReg | SrcMem16 | ModRM | Mov),
 	/* 0xB8 - 0xBF */
 	N, N,
-	D(Group | Group8), D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	G(0, group8), D(DstMem | SrcReg | ModRM | BitOp | Lock),
 	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
 	    D(DstReg | SrcMem16 | ModRM | Mov),
 	/* 0xC0 - 0xCF */
-- 
1.7.1

[PATCH 14/22] KVM: x86 emulator: convert group 9 to new style

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   14 ++++++++------
 1 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index adddf32..42bd255 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -105,7 +105,7 @@
 #define X16(x) X8(x), X8(x)
 
 enum {
-	NoGrp, Group9,
+	NoGrp,
 };
 
 struct opcode {
@@ -168,14 +168,16 @@ static struct opcode group8[] = {
 	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
 };
 
-static struct opcode group_table[] = {
-	[Group9*8] =
+static struct group_dual group9 = { {
 	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
+}, {
+	N, N, N, N, N, N, N, N,
+} };
+
+static struct opcode group_table[] = {
 };
 
 static struct opcode group2_table[] = {
-	[Group9*8] =
-	N, N, N, N, N, N, N, N,
 };
 
 static struct opcode opcode_table[256] = {
@@ -344,7 +346,7 @@ static struct opcode twobyte_table[256] = {
 	    D(DstReg | SrcMem16 | ModRM | Mov),
 	/* 0xC0 - 0xCF */
 	N, N, N, D(DstMem | SrcReg | ModRM | Mov),
-	N, N, N, D(Group | GroupDual | Group9),
+	N, N, N, GD(0, &group9),
 	N, N, N, N, N, N, N, N,
 	/* 0xD0 - 0xDF */
 	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-- 
1.7.1

[PATCH 15/22] KVM: x86 emulator: drop support for old-style groups

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   32 +++++++-------------------------
 1 files changed, 7 insertions(+), 25 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 42bd255..bb05687 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -82,7 +82,6 @@
 #define Stack       (1<<13)     /* Stack instruction (push/pop) */
 #define Group       (1<<14)     /* Bits 3:5 of modrm byte extend opcode */
 #define GroupDual   (1<<15)     /* Alternate decoding of mod == 3 */
-#define GroupMask   0x0f        /* Group number stored in bits 0:3 */
 /* Misc flags */
 #define Undefined   (1<<25) /* No Such Instruction */
 #define Lock        (1<<26) /* lock prefix is allowed for the instruction */
@@ -104,10 +103,6 @@
 #define X8(x) X4(x), X4(x)
 #define X16(x) X8(x), X8(x)
 
-enum {
-	NoGrp,
-};
-
 struct opcode {
 	u32 flags;
 	union {
@@ -174,12 +169,6 @@ static struct group_dual group9 = { {
 	N, N, N, N, N, N, N, N,
 } };
 
-static struct opcode group_table[] = {
-};
-
-static struct opcode group2_table[] = {
-};
-
 static struct opcode opcode_table[256] = {
 	/* 0x00 - 0x07 */
 	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
@@ -956,7 +945,7 @@ x86_decode_insn(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
 	struct decode_cache *c = &ctxt->decode;
 	int rc = X86EMUL_CONTINUE;
 	int mode = ctxt->mode;
-	int def_op_bytes, def_ad_bytes, group, dual, goffset;
+	int def_op_bytes, def_ad_bytes, dual, goffset;
 	struct opcode opcode, *g_mod012, *g_mod3;
 
 	/* we cannot decode insn before we complete previous rep insn */
@@ -1056,24 +1045,17 @@ done_prefixes:
 	c->d = opcode.flags;
 
 	if (c->d & Group) {
-		group = c->d & GroupMask;
 		dual = c->d & GroupDual;
 		c->modrm = insn_fetch(u8, 1, c->eip);
 		--c->eip;
 
-		if (group) {
-			g_mod012 = g_mod3 = &group_table[group * 8];
-			if (c->d & GroupDual)
-				g_mod3 = &group2_table[group * 8];
-		} else {
-			if (c->d & GroupDual) {
-				g_mod012 = opcode.u.gdual->mod012;
-				g_mod3 = opcode.u.gdual->mod3;
-			} else
-				g_mod012 = g_mod3 = opcode.u.group;
-		}
+		if (c->d & GroupDual) {
+			g_mod012 = opcode.u.gdual->mod012;
+			g_mod3 = opcode.u.gdual->mod3;
+		} else
+			g_mod012 = g_mod3 = opcode.u.group;
 
-		c->d &= ~(Group | GroupDual | GroupMask);
+		c->d &= ~(Group | GroupDual);
 
 		goffset = (c->modrm >> 3) & 7;
 
-- 
1.7.1

[PATCH 16/22] KVM: x86 emulator: move ByteOp and Dst back to bits 0:3

[Avi Kivity]

Now that the group index no longer exists, the space is free.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   16 ++++++++--------
 1 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index bb05687..b10f270 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -46,15 +46,15 @@
  */
 
 /* Operand sizes: 8-bit operands or specified/overridden size. */
-#define ByteOp      (1<<16)	/* 8-bit operands. */
+#define ByteOp      (1<<0)	/* 8-bit operands. */
 /* Destination operand type. */
-#define ImplicitOps (1<<17)	/* Implicit in opcode. No generic decode. */
-#define DstReg      (2<<17)	/* Register operand. */
-#define DstMem      (3<<17)	/* Memory operand. */
-#define DstAcc      (4<<17)	/* Destination Accumulator */
-#define DstDI       (5<<17)	/* Destination is in ES:(E)DI */
-#define DstMem64    (6<<17)	/* 64bit memory operand */
-#define DstMask     (7<<17)
+#define ImplicitOps (1<<1)	/* Implicit in opcode. No generic decode. */
+#define DstReg      (2<<1)	/* Register operand. */
+#define DstMem      (3<<1)	/* Memory operand. */
+#define DstAcc      (4<<1)	/* Destination Accumulator */
+#define DstDI       (5<<1)	/* Destination is in ES:(E)DI */
+#define DstMem64    (6<<1)	/* 64bit memory operand */
+#define DstMask     (7<<1)
 /* Source operand type. */
 #define SrcNone     (0<<4)	/* No source operand. */
 #define SrcImplicit (0<<4)	/* Source operand is implicit in the opcode. */
-- 
1.7.1

[PATCH 17/22] KVM: x86 emulator: store x86_emulate_ops in emulation context

[Avi Kivity]

It doesn't ever change, so we don't need to pass it around everywhere.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/include/asm/kvm_emulate.h |    9 ++++-----
 arch/x86/kvm/emulate.c             |    8 +++++---
 arch/x86/kvm/x86.c                 |    7 ++++---
 3 files changed, 13 insertions(+), 11 deletions(-)

diff --git a/arch/x86/include/asm/kvm_emulate.h b/arch/x86/include/asm/kvm_emulate.h
index 51cfd73..daf4477 100644
--- a/arch/x86/include/asm/kvm_emulate.h
+++ b/arch/x86/include/asm/kvm_emulate.h
@@ -203,6 +203,8 @@ struct decode_cache {
 };
 
 struct x86_emulate_ctxt {
+	struct x86_emulate_ops *ops;
+
 	/* Register state before/after emulation. */
 	struct kvm_vcpu *vcpu;
 
@@ -244,12 +246,9 @@ struct x86_emulate_ctxt {
 #define X86EMUL_MODE_HOST X86EMUL_MODE_PROT64
 #endif
 
-int x86_decode_insn(struct x86_emulate_ctxt *ctxt,
-		    struct x86_emulate_ops *ops);
-int x86_emulate_insn(struct x86_emulate_ctxt *ctxt,
-		     struct x86_emulate_ops *ops);
+int x86_decode_insn(struct x86_emulate_ctxt *ctxt);
+int x86_emulate_insn(struct x86_emulate_ctxt *ctxt);
 int emulator_task_switch(struct x86_emulate_ctxt *ctxt,
-			 struct x86_emulate_ops *ops,
 			 u16 tss_selector, int reason,
 			 bool has_error_code, u32 error_code);
 
diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index b10f270..f3f6fd9 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -940,8 +940,9 @@ done:
 }
 
 int
-x86_decode_insn(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
+x86_decode_insn(struct x86_emulate_ctxt *ctxt)
 {
+	struct x86_emulate_ops *ops = ctxt->ops;
 	struct decode_cache *c = &ctxt->decode;
 	int rc = X86EMUL_CONTINUE;
 	int mode = ctxt->mode;
@@ -2512,10 +2513,10 @@ static int emulator_do_task_switch(struct x86_emulate_ctxt *ctxt,
 }
 
 int emulator_task_switch(struct x86_emulate_ctxt *ctxt,
-			 struct x86_emulate_ops *ops,
 			 u16 tss_selector, int reason,
 			 bool has_error_code, u32 error_code)
 {
+	struct x86_emulate_ops *ops = ctxt->ops;
 	struct decode_cache *c = &ctxt->decode;
 	int rc;
 
@@ -2545,8 +2546,9 @@ static void string_addr_inc(struct x86_emulate_ctxt *ctxt, unsigned long base,
 }
 
 int
-x86_emulate_insn(struct x86_emulate_ctxt *ctxt, struct x86_emulate_ops *ops)
+x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
 {
+	struct x86_emulate_ops *ops = ctxt->ops;
 	u64 msr_data;
 	struct decode_cache *c = &ctxt->decode;
 	int rc = X86EMUL_CONTINUE;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 76fbc32..7308bb8 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3998,7 +3998,7 @@ int emulate_instruction(struct kvm_vcpu *vcpu,
 		vcpu->arch.emulate_ctxt.interruptibility = 0;
 		vcpu->arch.emulate_ctxt.exception = -1;
 
-		r = x86_decode_insn(&vcpu->arch.emulate_ctxt, &emulate_ops);
+		r = x86_decode_insn(&vcpu->arch.emulate_ctxt);
 		trace_kvm_emulate_insn_start(vcpu);
 
 		/* Only allow emulation of specific instructions on #UD
@@ -4048,7 +4048,7 @@ int emulate_instruction(struct kvm_vcpu *vcpu,
 	memcpy(c->regs, vcpu->arch.regs, sizeof c->regs);
 
 restart:
-	r = x86_emulate_insn(&vcpu->arch.emulate_ctxt, &emulate_ops);
+	r = x86_emulate_insn(&vcpu->arch.emulate_ctxt);
 
 	if (r) { /* emulation failed */
 		if (reexecute_instruction(vcpu, cr2))
@@ -5067,7 +5067,7 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason,
 	memset(c, 0, sizeof(struct decode_cache));
 	memcpy(c->regs, vcpu->arch.regs, sizeof c->regs);
 
-	ret = emulator_task_switch(&vcpu->arch.emulate_ctxt, &emulate_ops,
+	ret = emulator_task_switch(&vcpu->arch.emulate_ctxt,
 				   tss_selector, reason, has_error_code,
 				   error_code);
 
@@ -5424,6 +5424,7 @@ int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
 	BUG_ON(vcpu->kvm == NULL);
 	kvm = vcpu->kvm;
 
+	vcpu->arch.emulate_ctxt.ops = &emulate_ops;
 	vcpu->arch.mmu.root_hpa = INVALID_PAGE;
 	if (!irqchip_in_kernel(kvm) || kvm_vcpu_is_bsp(vcpu))
 		vcpu->arch.mp_state = KVM_MP_STATE_RUNNABLE;
-- 
1.7.1

[PATCH 18/22] KVM: x86 emulator: allow storing emulator execution function in decode tables

[Avi Kivity]

Instead of looking up the opcode twice (once for decode flags, once for
the big execution switch) look up both flags and function in the decode tables.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/include/asm/kvm_emulate.h |    1 +
 arch/x86/kvm/emulate.c             |   12 ++++++++++++
 2 files changed, 13 insertions(+), 0 deletions(-)

diff --git a/arch/x86/include/asm/kvm_emulate.h b/arch/x86/include/asm/kvm_emulate.h
index daf4477..f397b79 100644
--- a/arch/x86/include/asm/kvm_emulate.h
+++ b/arch/x86/include/asm/kvm_emulate.h
@@ -185,6 +185,7 @@ struct decode_cache {
 	bool has_seg_override;
 	u8 seg_override;
 	unsigned int d;
+	int (*execute)(struct x86_emulate_ctxt *ctxt);
 	unsigned long regs[NR_VCPU_REGS];
 	unsigned long eip;
 	/* modrm */
diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index f3f6fd9..8b19da2 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -106,6 +106,7 @@
 struct opcode {
 	u32 flags;
 	union {
+		int (*execute)(struct x86_emulate_ctxt *ctxt);
 		struct opcode *group;
 		struct group_dual *gdual;
 	} u;
@@ -120,6 +121,7 @@ struct group_dual {
 #define N    D(0)
 #define G(_f, _g) { .flags = ((_f) | Group), .u.group = (_g) }
 #define GD(_f, _g) { .flags = ((_f) | Group | GroupDual), .u.gdual = (_g) }
+#define I(_f, _e) { .flags = (_f), .u.execute = (_e) }
 
 static struct opcode group1[] = {
 	X7(D(Lock)), N
@@ -349,6 +351,7 @@ static struct opcode twobyte_table[256] = {
 #undef N
 #undef G
 #undef GD
+#undef I
 
 /* EFLAGS bit definitions. */
 #define EFLG_ID (1<<21)
@@ -1067,6 +1070,8 @@ done_prefixes:
 		c->d |= opcode.flags;
 	}
 
+	c->execute = opcode.u.execute;
+
 	/* Unrecognised? */
 	if (c->d == 0 || (c->d & Undefined)) {
 		DPRINTF("Cannot emulate %02x\n", c->b);
@@ -2631,6 +2636,13 @@ x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
 
 special_insn:
 
+	if (c->execute) {
+		rc = c->execute(ctxt);
+		if (rc != X86EMUL_CONTINUE)
+			goto done;
+		goto writeback;
+	}
+
 	if (c->twobyte)
 		goto twobyte_insn;
 
-- 
1.7.1

[PATCH 19/22] KVM: x86 emulator: move x86_decode_insn() downwards

[Avi Kivity]

No code changes.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |  744 ++++++++++++++++++++++++------------------------
 1 files changed, 372 insertions(+), 372 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 8b19da2..620525f 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -942,378 +942,6 @@ done:
 	return rc;
 }
 
-int
-x86_decode_insn(struct x86_emulate_ctxt *ctxt)
-{
-	struct x86_emulate_ops *ops = ctxt->ops;
-	struct decode_cache *c = &ctxt->decode;
-	int rc = X86EMUL_CONTINUE;
-	int mode = ctxt->mode;
-	int def_op_bytes, def_ad_bytes, dual, goffset;
-	struct opcode opcode, *g_mod012, *g_mod3;
-
-	/* we cannot decode insn before we complete previous rep insn */
-	WARN_ON(ctxt->restart);
-
-	c->eip = ctxt->eip;
-	c->fetch.start = c->fetch.end = c->eip;
-	ctxt->cs_base = seg_base(ctxt, ops, VCPU_SREG_CS);
-
-	switch (mode) {
-	case X86EMUL_MODE_REAL:
-	case X86EMUL_MODE_VM86:
-	case X86EMUL_MODE_PROT16:
-		def_op_bytes = def_ad_bytes = 2;
-		break;
-	case X86EMUL_MODE_PROT32:
-		def_op_bytes = def_ad_bytes = 4;
-		break;
-#ifdef CONFIG_X86_64
-	case X86EMUL_MODE_PROT64:
-		def_op_bytes = 4;
-		def_ad_bytes = 8;
-		break;
-#endif
-	default:
-		return -1;
-	}
-
-	c->op_bytes = def_op_bytes;
-	c->ad_bytes = def_ad_bytes;
-
-	/* Legacy prefixes. */
-	for (;;) {
-		switch (c->b = insn_fetch(u8, 1, c->eip)) {
-		case 0x66:	/* operand-size override */
-			/* switch between 2/4 bytes */
-			c->op_bytes = def_op_bytes ^ 6;
-			break;
-		case 0x67:	/* address-size override */
-			if (mode == X86EMUL_MODE_PROT64)
-				/* switch between 4/8 bytes */
-				c->ad_bytes = def_ad_bytes ^ 12;
-			else
-				/* switch between 2/4 bytes */
-				c->ad_bytes = def_ad_bytes ^ 6;
-			break;
-		case 0x26:	/* ES override */
-		case 0x2e:	/* CS override */
-		case 0x36:	/* SS override */
-		case 0x3e:	/* DS override */
-			set_seg_override(c, (c->b >> 3) & 3);
-			break;
-		case 0x64:	/* FS override */
-		case 0x65:	/* GS override */
-			set_seg_override(c, c->b & 7);
-			break;
-		case 0x40 ... 0x4f: /* REX */
-			if (mode != X86EMUL_MODE_PROT64)
-				goto done_prefixes;
-			c->rex_prefix = c->b;
-			continue;
-		case 0xf0:	/* LOCK */
-			c->lock_prefix = 1;
-			break;
-		case 0xf2:	/* REPNE/REPNZ */
-			c->rep_prefix = REPNE_PREFIX;
-			break;
-		case 0xf3:	/* REP/REPE/REPZ */
-			c->rep_prefix = REPE_PREFIX;
-			break;
-		default:
-			goto done_prefixes;
-		}
-
-		/* Any legacy prefix after a REX prefix nullifies its effect. */
-
-		c->rex_prefix = 0;
-	}
-
-done_prefixes:
-
-	/* REX prefix. */
-	if (c->rex_prefix)
-		if (c->rex_prefix & 8)
-			c->op_bytes = 8;	/* REX.W */
-
-	/* Opcode byte(s). */
-	opcode = opcode_table[c->b];
-	if (opcode.flags == 0) {
-		/* Two-byte opcode? */
-		if (c->b == 0x0f) {
-			c->twobyte = 1;
-			c->b = insn_fetch(u8, 1, c->eip);
-			opcode = twobyte_table[c->b];
-		}
-	}
-	c->d = opcode.flags;
-
-	if (c->d & Group) {
-		dual = c->d & GroupDual;
-		c->modrm = insn_fetch(u8, 1, c->eip);
-		--c->eip;
-
-		if (c->d & GroupDual) {
-			g_mod012 = opcode.u.gdual->mod012;
-			g_mod3 = opcode.u.gdual->mod3;
-		} else
-			g_mod012 = g_mod3 = opcode.u.group;
-
-		c->d &= ~(Group | GroupDual);
-
-		goffset = (c->modrm >> 3) & 7;
-
-		if ((c->modrm >> 6) == 3)
-			opcode = g_mod3[goffset];
-		else
-			opcode = g_mod012[goffset];
-		c->d |= opcode.flags;
-	}
-
-	c->execute = opcode.u.execute;
-
-	/* Unrecognised? */
-	if (c->d == 0 || (c->d & Undefined)) {
-		DPRINTF("Cannot emulate %02x\n", c->b);
-		return -1;
-	}
-
-	if (mode == X86EMUL_MODE_PROT64 && (c->d & Stack))
-		c->op_bytes = 8;
-
-	/* ModRM and SIB bytes. */
-	if (c->d & ModRM)
-		rc = decode_modrm(ctxt, ops);
-	else if (c->d & MemAbs)
-		rc = decode_abs(ctxt, ops);
-	if (rc != X86EMUL_CONTINUE)
-		goto done;
-
-	if (!c->has_seg_override)
-		set_seg_override(c, VCPU_SREG_DS);
-
-	if (!(!c->twobyte && c->b == 0x8d))
-		c->modrm_ea += seg_override_base(ctxt, ops, c);
-
-	if (c->ad_bytes != 8)
-		c->modrm_ea = (u32)c->modrm_ea;
-
-	if (c->rip_relative)
-		c->modrm_ea += c->eip;
-
-	/*
-	 * Decode and fetch the source operand: register, memory
-	 * or immediate.
-	 */
-	switch (c->d & SrcMask) {
-	case SrcNone:
-		break;
-	case SrcReg:
-		decode_register_operand(&c->src, c, 0);
-		break;
-	case SrcMem16:
-		c->src.bytes = 2;
-		goto srcmem_common;
-	case SrcMem32:
-		c->src.bytes = 4;
-		goto srcmem_common;
-	case SrcMem:
-		c->src.bytes = (c->d & ByteOp) ? 1 :
-							   c->op_bytes;
-		/* Don't fetch the address for invlpg: it could be unmapped. */
-		if (c->twobyte && c->b == 0x01 && c->modrm_reg == 7)
-			break;
-	srcmem_common:
-		/*
-		 * For instructions with a ModR/M byte, switch to register
-		 * access if Mod = 3.
-		 */
-		if ((c->d & ModRM) && c->modrm_mod == 3) {
-			c->src.type = OP_REG;
-			c->src.val = c->modrm_val;
-			c->src.ptr = c->modrm_ptr;
-			break;
-		}
-		c->src.type = OP_MEM;
-		c->src.ptr = (unsigned long *)c->modrm_ea;
-		c->src.val = 0;
-		break;
-	case SrcImm:
-	case SrcImmU:
-		c->src.type = OP_IMM;
-		c->src.ptr = (unsigned long *)c->eip;
-		c->src.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-		if (c->src.bytes == 8)
-			c->src.bytes = 4;
-		/* NB. Immediates are sign-extended as necessary. */
-		switch (c->src.bytes) {
-		case 1:
-			c->src.val = insn_fetch(s8, 1, c->eip);
-			break;
-		case 2:
-			c->src.val = insn_fetch(s16, 2, c->eip);
-			break;
-		case 4:
-			c->src.val = insn_fetch(s32, 4, c->eip);
-			break;
-		}
-		if ((c->d & SrcMask) == SrcImmU) {
-			switch (c->src.bytes) {
-			case 1:
-				c->src.val &= 0xff;
-				break;
-			case 2:
-				c->src.val &= 0xffff;
-				break;
-			case 4:
-				c->src.val &= 0xffffffff;
-				break;
-			}
-		}
-		break;
-	case SrcImmByte:
-	case SrcImmUByte:
-		c->src.type = OP_IMM;
-		c->src.ptr = (unsigned long *)c->eip;
-		c->src.bytes = 1;
-		if ((c->d & SrcMask) == SrcImmByte)
-			c->src.val = insn_fetch(s8, 1, c->eip);
-		else
-			c->src.val = insn_fetch(u8, 1, c->eip);
-		break;
-	case SrcAcc:
-		c->src.type = OP_REG;
-		c->src.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-		c->src.ptr = &c->regs[VCPU_REGS_RAX];
-		switch (c->src.bytes) {
-			case 1:
-				c->src.val = *(u8 *)c->src.ptr;
-				break;
-			case 2:
-				c->src.val = *(u16 *)c->src.ptr;
-				break;
-			case 4:
-				c->src.val = *(u32 *)c->src.ptr;
-				break;
-			case 8:
-				c->src.val = *(u64 *)c->src.ptr;
-				break;
-		}
-		break;
-	case SrcOne:
-		c->src.bytes = 1;
-		c->src.val = 1;
-		break;
-	case SrcSI:
-		c->src.type = OP_MEM;
-		c->src.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-		c->src.ptr = (unsigned long *)
-			register_address(c,  seg_override_base(ctxt, ops, c),
-					 c->regs[VCPU_REGS_RSI]);
-		c->src.val = 0;
-		break;
-	case SrcImmFAddr:
-		c->src.type = OP_IMM;
-		c->src.ptr = (unsigned long *)c->eip;
-		c->src.bytes = c->op_bytes + 2;
-		insn_fetch_arr(c->src.valptr, c->src.bytes, c->eip);
-		break;
-	case SrcMemFAddr:
-		c->src.type = OP_MEM;
-		c->src.ptr = (unsigned long *)c->modrm_ea;
-		c->src.bytes = c->op_bytes + 2;
-		break;
-	}
-
-	/*
-	 * Decode and fetch the second source operand: register, memory
-	 * or immediate.
-	 */
-	switch (c->d & Src2Mask) {
-	case Src2None:
-		break;
-	case Src2CL:
-		c->src2.bytes = 1;
-		c->src2.val = c->regs[VCPU_REGS_RCX] & 0x8;
-		break;
-	case Src2ImmByte:
-		c->src2.type = OP_IMM;
-		c->src2.ptr = (unsigned long *)c->eip;
-		c->src2.bytes = 1;
-		c->src2.val = insn_fetch(u8, 1, c->eip);
-		break;
-	case Src2One:
-		c->src2.bytes = 1;
-		c->src2.val = 1;
-		break;
-	}
-
-	/* Decode and fetch the destination operand: register or memory. */
-	switch (c->d & DstMask) {
-	case ImplicitOps:
-		/* Special instructions do their own operand decoding. */
-		return 0;
-	case DstReg:
-		decode_register_operand(&c->dst, c,
-			 c->twobyte && (c->b == 0xb6 || c->b == 0xb7));
-		break;
-	case DstMem:
-	case DstMem64:
-		if ((c->d & ModRM) && c->modrm_mod == 3) {
-			c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-			c->dst.type = OP_REG;
-			c->dst.val = c->dst.orig_val = c->modrm_val;
-			c->dst.ptr = c->modrm_ptr;
-			break;
-		}
-		c->dst.type = OP_MEM;
-		c->dst.ptr = (unsigned long *)c->modrm_ea;
-		if ((c->d & DstMask) == DstMem64)
-			c->dst.bytes = 8;
-		else
-			c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-		c->dst.val = 0;
-		if (c->d & BitOp) {
-			unsigned long mask = ~(c->dst.bytes * 8 - 1);
-
-			c->dst.ptr = (void *)c->dst.ptr +
-						   (c->src.val & mask) / 8;
-		}
-		break;
-	case DstAcc:
-		c->dst.type = OP_REG;
-		c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-		c->dst.ptr = &c->regs[VCPU_REGS_RAX];
-		switch (c->dst.bytes) {
-			case 1:
-				c->dst.val = *(u8 *)c->dst.ptr;
-				break;
-			case 2:
-				c->dst.val = *(u16 *)c->dst.ptr;
-				break;
-			case 4:
-				c->dst.val = *(u32 *)c->dst.ptr;
-				break;
-			case 8:
-				c->dst.val = *(u64 *)c->dst.ptr;
-				break;
-		}
-		c->dst.orig_val = c->dst.val;
-		break;
-	case DstDI:
-		c->dst.type = OP_MEM;
-		c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
-		c->dst.ptr = (unsigned long *)
-			register_address(c, es_base(ctxt, ops),
-					 c->regs[VCPU_REGS_RDI]);
-		c->dst.val = 0;
-		break;
-	}
-
-done:
-	return (rc == X86EMUL_UNHANDLEABLE) ? -1 : 0;
-}
-
 static int read_emulated(struct x86_emulate_ctxt *ctxt,
 			 struct x86_emulate_ops *ops,
 			 unsigned long addr, void *dest, unsigned size)
@@ -2551,6 +2179,378 @@ static void string_addr_inc(struct x86_emulate_ctxt *ctxt, unsigned long base,
 }
 
 int
+x86_decode_insn(struct x86_emulate_ctxt *ctxt)
+{
+	struct x86_emulate_ops *ops = ctxt->ops;
+	struct decode_cache *c = &ctxt->decode;
+	int rc = X86EMUL_CONTINUE;
+	int mode = ctxt->mode;
+	int def_op_bytes, def_ad_bytes, dual, goffset;
+	struct opcode opcode, *g_mod012, *g_mod3;
+
+	/* we cannot decode insn before we complete previous rep insn */
+	WARN_ON(ctxt->restart);
+
+	c->eip = ctxt->eip;
+	c->fetch.start = c->fetch.end = c->eip;
+	ctxt->cs_base = seg_base(ctxt, ops, VCPU_SREG_CS);
+
+	switch (mode) {
+	case X86EMUL_MODE_REAL:
+	case X86EMUL_MODE_VM86:
+	case X86EMUL_MODE_PROT16:
+		def_op_bytes = def_ad_bytes = 2;
+		break;
+	case X86EMUL_MODE_PROT32:
+		def_op_bytes = def_ad_bytes = 4;
+		break;
+#ifdef CONFIG_X86_64
+	case X86EMUL_MODE_PROT64:
+		def_op_bytes = 4;
+		def_ad_bytes = 8;
+		break;
+#endif
+	default:
+		return -1;
+	}
+
+	c->op_bytes = def_op_bytes;
+	c->ad_bytes = def_ad_bytes;
+
+	/* Legacy prefixes. */
+	for (;;) {
+		switch (c->b = insn_fetch(u8, 1, c->eip)) {
+		case 0x66:	/* operand-size override */
+			/* switch between 2/4 bytes */
+			c->op_bytes = def_op_bytes ^ 6;
+			break;
+		case 0x67:	/* address-size override */
+			if (mode == X86EMUL_MODE_PROT64)
+				/* switch between 4/8 bytes */
+				c->ad_bytes = def_ad_bytes ^ 12;
+			else
+				/* switch between 2/4 bytes */
+				c->ad_bytes = def_ad_bytes ^ 6;
+			break;
+		case 0x26:	/* ES override */
+		case 0x2e:	/* CS override */
+		case 0x36:	/* SS override */
+		case 0x3e:	/* DS override */
+			set_seg_override(c, (c->b >> 3) & 3);
+			break;
+		case 0x64:	/* FS override */
+		case 0x65:	/* GS override */
+			set_seg_override(c, c->b & 7);
+			break;
+		case 0x40 ... 0x4f: /* REX */
+			if (mode != X86EMUL_MODE_PROT64)
+				goto done_prefixes;
+			c->rex_prefix = c->b;
+			continue;
+		case 0xf0:	/* LOCK */
+			c->lock_prefix = 1;
+			break;
+		case 0xf2:	/* REPNE/REPNZ */
+			c->rep_prefix = REPNE_PREFIX;
+			break;
+		case 0xf3:	/* REP/REPE/REPZ */
+			c->rep_prefix = REPE_PREFIX;
+			break;
+		default:
+			goto done_prefixes;
+		}
+
+		/* Any legacy prefix after a REX prefix nullifies its effect. */
+
+		c->rex_prefix = 0;
+	}
+
+done_prefixes:
+
+	/* REX prefix. */
+	if (c->rex_prefix)
+		if (c->rex_prefix & 8)
+			c->op_bytes = 8;	/* REX.W */
+
+	/* Opcode byte(s). */
+	opcode = opcode_table[c->b];
+	if (opcode.flags == 0) {
+		/* Two-byte opcode? */
+		if (c->b == 0x0f) {
+			c->twobyte = 1;
+			c->b = insn_fetch(u8, 1, c->eip);
+			opcode = twobyte_table[c->b];
+		}
+	}
+	c->d = opcode.flags;
+
+	if (c->d & Group) {
+		dual = c->d & GroupDual;
+		c->modrm = insn_fetch(u8, 1, c->eip);
+		--c->eip;
+
+		if (c->d & GroupDual) {
+			g_mod012 = opcode.u.gdual->mod012;
+			g_mod3 = opcode.u.gdual->mod3;
+		} else
+			g_mod012 = g_mod3 = opcode.u.group;
+
+		c->d &= ~(Group | GroupDual);
+
+		goffset = (c->modrm >> 3) & 7;
+
+		if ((c->modrm >> 6) == 3)
+			opcode = g_mod3[goffset];
+		else
+			opcode = g_mod012[goffset];
+		c->d |= opcode.flags;
+	}
+
+	c->execute = opcode.u.execute;
+
+	/* Unrecognised? */
+	if (c->d == 0 || (c->d & Undefined)) {
+		DPRINTF("Cannot emulate %02x\n", c->b);
+		return -1;
+	}
+
+	if (mode == X86EMUL_MODE_PROT64 && (c->d & Stack))
+		c->op_bytes = 8;
+
+	/* ModRM and SIB bytes. */
+	if (c->d & ModRM)
+		rc = decode_modrm(ctxt, ops);
+	else if (c->d & MemAbs)
+		rc = decode_abs(ctxt, ops);
+	if (rc != X86EMUL_CONTINUE)
+		goto done;
+
+	if (!c->has_seg_override)
+		set_seg_override(c, VCPU_SREG_DS);
+
+	if (!(!c->twobyte && c->b == 0x8d))
+		c->modrm_ea += seg_override_base(ctxt, ops, c);
+
+	if (c->ad_bytes != 8)
+		c->modrm_ea = (u32)c->modrm_ea;
+
+	if (c->rip_relative)
+		c->modrm_ea += c->eip;
+
+	/*
+	 * Decode and fetch the source operand: register, memory
+	 * or immediate.
+	 */
+	switch (c->d & SrcMask) {
+	case SrcNone:
+		break;
+	case SrcReg:
+		decode_register_operand(&c->src, c, 0);
+		break;
+	case SrcMem16:
+		c->src.bytes = 2;
+		goto srcmem_common;
+	case SrcMem32:
+		c->src.bytes = 4;
+		goto srcmem_common;
+	case SrcMem:
+		c->src.bytes = (c->d & ByteOp) ? 1 :
+							   c->op_bytes;
+		/* Don't fetch the address for invlpg: it could be unmapped. */
+		if (c->twobyte && c->b == 0x01 && c->modrm_reg == 7)
+			break;
+	srcmem_common:
+		/*
+		 * For instructions with a ModR/M byte, switch to register
+		 * access if Mod = 3.
+		 */
+		if ((c->d & ModRM) && c->modrm_mod == 3) {
+			c->src.type = OP_REG;
+			c->src.val = c->modrm_val;
+			c->src.ptr = c->modrm_ptr;
+			break;
+		}
+		c->src.type = OP_MEM;
+		c->src.ptr = (unsigned long *)c->modrm_ea;
+		c->src.val = 0;
+		break;
+	case SrcImm:
+	case SrcImmU:
+		c->src.type = OP_IMM;
+		c->src.ptr = (unsigned long *)c->eip;
+		c->src.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+		if (c->src.bytes == 8)
+			c->src.bytes = 4;
+		/* NB. Immediates are sign-extended as necessary. */
+		switch (c->src.bytes) {
+		case 1:
+			c->src.val = insn_fetch(s8, 1, c->eip);
+			break;
+		case 2:
+			c->src.val = insn_fetch(s16, 2, c->eip);
+			break;
+		case 4:
+			c->src.val = insn_fetch(s32, 4, c->eip);
+			break;
+		}
+		if ((c->d & SrcMask) == SrcImmU) {
+			switch (c->src.bytes) {
+			case 1:
+				c->src.val &= 0xff;
+				break;
+			case 2:
+				c->src.val &= 0xffff;
+				break;
+			case 4:
+				c->src.val &= 0xffffffff;
+				break;
+			}
+		}
+		break;
+	case SrcImmByte:
+	case SrcImmUByte:
+		c->src.type = OP_IMM;
+		c->src.ptr = (unsigned long *)c->eip;
+		c->src.bytes = 1;
+		if ((c->d & SrcMask) == SrcImmByte)
+			c->src.val = insn_fetch(s8, 1, c->eip);
+		else
+			c->src.val = insn_fetch(u8, 1, c->eip);
+		break;
+	case SrcAcc:
+		c->src.type = OP_REG;
+		c->src.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+		c->src.ptr = &c->regs[VCPU_REGS_RAX];
+		switch (c->src.bytes) {
+			case 1:
+				c->src.val = *(u8 *)c->src.ptr;
+				break;
+			case 2:
+				c->src.val = *(u16 *)c->src.ptr;
+				break;
+			case 4:
+				c->src.val = *(u32 *)c->src.ptr;
+				break;
+			case 8:
+				c->src.val = *(u64 *)c->src.ptr;
+				break;
+		}
+		break;
+	case SrcOne:
+		c->src.bytes = 1;
+		c->src.val = 1;
+		break;
+	case SrcSI:
+		c->src.type = OP_MEM;
+		c->src.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+		c->src.ptr = (unsigned long *)
+			register_address(c,  seg_override_base(ctxt, ops, c),
+					 c->regs[VCPU_REGS_RSI]);
+		c->src.val = 0;
+		break;
+	case SrcImmFAddr:
+		c->src.type = OP_IMM;
+		c->src.ptr = (unsigned long *)c->eip;
+		c->src.bytes = c->op_bytes + 2;
+		insn_fetch_arr(c->src.valptr, c->src.bytes, c->eip);
+		break;
+	case SrcMemFAddr:
+		c->src.type = OP_MEM;
+		c->src.ptr = (unsigned long *)c->modrm_ea;
+		c->src.bytes = c->op_bytes + 2;
+		break;
+	}
+
+	/*
+	 * Decode and fetch the second source operand: register, memory
+	 * or immediate.
+	 */
+	switch (c->d & Src2Mask) {
+	case Src2None:
+		break;
+	case Src2CL:
+		c->src2.bytes = 1;
+		c->src2.val = c->regs[VCPU_REGS_RCX] & 0x8;
+		break;
+	case Src2ImmByte:
+		c->src2.type = OP_IMM;
+		c->src2.ptr = (unsigned long *)c->eip;
+		c->src2.bytes = 1;
+		c->src2.val = insn_fetch(u8, 1, c->eip);
+		break;
+	case Src2One:
+		c->src2.bytes = 1;
+		c->src2.val = 1;
+		break;
+	}
+
+	/* Decode and fetch the destination operand: register or memory. */
+	switch (c->d & DstMask) {
+	case ImplicitOps:
+		/* Special instructions do their own operand decoding. */
+		return 0;
+	case DstReg:
+		decode_register_operand(&c->dst, c,
+			 c->twobyte && (c->b == 0xb6 || c->b == 0xb7));
+		break;
+	case DstMem:
+	case DstMem64:
+		if ((c->d & ModRM) && c->modrm_mod == 3) {
+			c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+			c->dst.type = OP_REG;
+			c->dst.val = c->dst.orig_val = c->modrm_val;
+			c->dst.ptr = c->modrm_ptr;
+			break;
+		}
+		c->dst.type = OP_MEM;
+		c->dst.ptr = (unsigned long *)c->modrm_ea;
+		if ((c->d & DstMask) == DstMem64)
+			c->dst.bytes = 8;
+		else
+			c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+		c->dst.val = 0;
+		if (c->d & BitOp) {
+			unsigned long mask = ~(c->dst.bytes * 8 - 1);
+
+			c->dst.ptr = (void *)c->dst.ptr +
+						   (c->src.val & mask) / 8;
+		}
+		break;
+	case DstAcc:
+		c->dst.type = OP_REG;
+		c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+		c->dst.ptr = &c->regs[VCPU_REGS_RAX];
+		switch (c->dst.bytes) {
+			case 1:
+				c->dst.val = *(u8 *)c->dst.ptr;
+				break;
+			case 2:
+				c->dst.val = *(u16 *)c->dst.ptr;
+				break;
+			case 4:
+				c->dst.val = *(u32 *)c->dst.ptr;
+				break;
+			case 8:
+				c->dst.val = *(u64 *)c->dst.ptr;
+				break;
+		}
+		c->dst.orig_val = c->dst.val;
+		break;
+	case DstDI:
+		c->dst.type = OP_MEM;
+		c->dst.bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
+		c->dst.ptr = (unsigned long *)
+			register_address(c, es_base(ctxt, ops),
+					 c->regs[VCPU_REGS_RDI]);
+		c->dst.val = 0;
+		break;
+	}
+
+done:
+	return (rc == X86EMUL_UNHANDLEABLE) ? -1 : 0;
+}
+
+int
 x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
 {
 	struct x86_emulate_ops *ops = ctxt->ops;
-- 
1.7.1

[PATCH 20/22] KVM: x86 emulator: move decode tables downwards

[Avi Kivity]

So they can reference execution functions.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |  472 ++++++++++++++++++++++++------------------------
 1 files changed, 236 insertions(+), 236 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 620525f..11fd2c9 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -117,242 +117,6 @@ struct group_dual {
 	struct opcode mod3[8];
 };
 
-#define D(_y) { .flags = (_y) }
-#define N    D(0)
-#define G(_f, _g) { .flags = ((_f) | Group), .u.group = (_g) }
-#define GD(_f, _g) { .flags = ((_f) | Group | GroupDual), .u.gdual = (_g) }
-#define I(_f, _e) { .flags = (_f), .u.execute = (_e) }
-
-static struct opcode group1[] = {
-	X7(D(Lock)), N
-};
-
-static struct opcode group1A[] = {
-	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
-};
-
-static struct opcode group3[] = {
-	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
-	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
-	X4(D(Undefined)),
-};
-
-static struct opcode group4[] = {
-	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
-	N, N, N, N, N, N,
-};
-
-static struct opcode group5[] = {
-	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
-	D(SrcMem | ModRM | Stack), N,
-	D(SrcMem | ModRM | Stack), D(SrcMemFAddr | ModRM | ImplicitOps),
-	D(SrcMem | ModRM | Stack), N,
-};
-
-static struct group_dual group7 = { {
-	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
-	D(SrcNone | ModRM | DstMem | Mov), N,
-	D(SrcMem16 | ModRM | Mov | Priv), D(SrcMem | ModRM | ByteOp | Priv),
-}, {
-	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
-	D(SrcNone | ModRM | DstMem | Mov), N,
-	D(SrcMem16 | ModRM | Mov | Priv), N,
-} };
-
-static struct opcode group8[] = {
-	N, N, N, N,
-	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
-	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
-};
-
-static struct group_dual group9 = { {
-	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
-}, {
-	N, N, N, N, N, N, N, N,
-} };
-
-static struct opcode opcode_table[256] = {
-	/* 0x00 - 0x07 */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
-	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
-	/* 0x08 - 0x0F */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
-	D(ImplicitOps | Stack | No64), N,
-	/* 0x10 - 0x17 */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
-	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
-	/* 0x18 - 0x1F */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
-	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
-	/* 0x20 - 0x27 */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
-	/* 0x28 - 0x2F */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
-	/* 0x30 - 0x37 */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
-	/* 0x38 - 0x3F */
-	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
-	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
-	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
-	N, N,
-	/* 0x40 - 0x4F */
-	X16(D(DstReg)),
-	/* 0x50 - 0x57 */
-	X8(D(SrcReg | Stack)),
-	/* 0x58 - 0x5F */
-	X8(D(DstReg | Stack)),
-	/* 0x60 - 0x67 */
-	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
-	N, D(DstReg | SrcMem32 | ModRM | Mov) /* movsxd (x86/64) */ ,
-	N, N, N, N,
-	/* 0x68 - 0x6F */
-	D(SrcImm | Mov | Stack), N, D(SrcImmByte | Mov | Stack), N,
-	D(DstDI | ByteOp | Mov | String), D(DstDI | Mov | String), /* insb, insw/insd */
-	D(SrcSI | ByteOp | ImplicitOps | String), D(SrcSI | ImplicitOps | String), /* outsb, outsw/outsd */
-	/* 0x70 - 0x7F */
-	X16(D(SrcImmByte)),
-	/* 0x80 - 0x87 */
-	G(ByteOp | DstMem | SrcImm | ModRM | Group, group1),
-	G(DstMem | SrcImm | ModRM | Group, group1),
-	G(ByteOp | DstMem | SrcImm | ModRM | No64 | Group, group1),
-	G(DstMem | SrcImmByte | ModRM | Group, group1),
-	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	/* 0x88 - 0x8F */
-	D(ByteOp | DstMem | SrcReg | ModRM | Mov), D(DstMem | SrcReg | ModRM | Mov),
-	D(ByteOp | DstReg | SrcMem | ModRM | Mov), D(DstReg | SrcMem | ModRM | Mov),
-	D(DstMem | SrcNone | ModRM | Mov), D(ModRM | DstReg),
-	D(ImplicitOps | SrcMem16 | ModRM), G(0, group1A),
-	/* 0x90 - 0x97 */
-	D(DstReg), D(DstReg), D(DstReg), D(DstReg),	D(DstReg), D(DstReg), D(DstReg), D(DstReg),
-	/* 0x98 - 0x9F */
-	N, N, D(SrcImmFAddr | No64), N,
-	D(ImplicitOps | Stack), D(ImplicitOps | Stack), N, N,
-	/* 0xA0 - 0xA7 */
-	D(ByteOp | DstAcc | SrcMem | Mov | MemAbs), D(DstAcc | SrcMem | Mov | MemAbs),
-	D(ByteOp | DstMem | SrcAcc | Mov | MemAbs), D(DstMem | SrcAcc | Mov | MemAbs),
-	D(ByteOp | SrcSI | DstDI | Mov | String), D(SrcSI | DstDI | Mov | String),
-	D(ByteOp | SrcSI | DstDI | String), D(SrcSI | DstDI | String),
-	/* 0xA8 - 0xAF */
-	D(DstAcc | SrcImmByte | ByteOp), D(DstAcc | SrcImm), D(ByteOp | DstDI | Mov | String), D(DstDI | Mov | String),
-	D(ByteOp | SrcSI | DstAcc | Mov | String), D(SrcSI | DstAcc | Mov | String),
-	D(ByteOp | DstDI | String), D(DstDI | String),
-	/* 0xB0 - 0xB7 */
-	X8(D(ByteOp | DstReg | SrcImm | Mov)),
-	/* 0xB8 - 0xBF */
-	X8(D(DstReg | SrcImm | Mov)),
-	/* 0xC0 - 0xC7 */
-	D(ByteOp | DstMem | SrcImm | ModRM), D(DstMem | SrcImmByte | ModRM),
-	N, D(ImplicitOps | Stack), N, N,
-	D(ByteOp | DstMem | SrcImm | ModRM | Mov), D(DstMem | SrcImm | ModRM | Mov),
-	/* 0xC8 - 0xCF */
-	N, N, N, D(ImplicitOps | Stack),
-	D(ImplicitOps), D(SrcImmByte), D(ImplicitOps | No64), D(ImplicitOps),
-	/* 0xD0 - 0xD7 */
-	D(ByteOp | DstMem | SrcImplicit | ModRM), D(DstMem | SrcImplicit | ModRM),
-	D(ByteOp | DstMem | SrcImplicit | ModRM), D(DstMem | SrcImplicit | ModRM),
-	N, N, N, N,
-	/* 0xD8 - 0xDF */
-	N, N, N, N, N, N, N, N,
-	/* 0xE0 - 0xE7 */
-	N, N, N, N,
-	D(ByteOp | SrcImmUByte | DstAcc), D(SrcImmUByte | DstAcc),
-	D(ByteOp | SrcImmUByte | DstAcc), D(SrcImmUByte | DstAcc),
-	/* 0xE8 - 0xEF */
-	D(SrcImm | Stack), D(SrcImm | ImplicitOps),
-	D(SrcImmFAddr | No64), D(SrcImmByte | ImplicitOps),
-	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
-	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
-	/* 0xF0 - 0xF7 */
-	N, N, N, N,
-	D(ImplicitOps | Priv), D(ImplicitOps), G(ByteOp, group3), G(0, group3),
-	/* 0xF8 - 0xFF */
-	D(ImplicitOps), N, D(ImplicitOps), D(ImplicitOps),
-	D(ImplicitOps), D(ImplicitOps), G(0, group4), G(0, group5),
-};
-
-static struct opcode twobyte_table[256] = {
-	/* 0x00 - 0x0F */
-	N, GD(0, &group7), N, N,
-	N, D(ImplicitOps), D(ImplicitOps | Priv), N,
-	D(ImplicitOps | Priv), D(ImplicitOps | Priv), N, N,
-	N, D(ImplicitOps | ModRM), N, N,
-	/* 0x10 - 0x1F */
-	N, N, N, N, N, N, N, N, D(ImplicitOps | ModRM), N, N, N, N, N, N, N,
-	/* 0x20 - 0x2F */
-	D(ModRM | ImplicitOps | Priv), D(ModRM | Priv),
-	D(ModRM | ImplicitOps | Priv), D(ModRM | Priv),
-	N, N, N, N,
-	N, N, N, N, N, N, N, N,
-	/* 0x30 - 0x3F */
-	D(ImplicitOps | Priv), N, D(ImplicitOps | Priv), N,
-	D(ImplicitOps), D(ImplicitOps | Priv), N, N,
-	N, N, N, N, N, N, N, N,
-	/* 0x40 - 0x4F */
-	X16(D(DstReg | SrcMem | ModRM | Mov)),
-	/* 0x50 - 0x5F */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-	/* 0x60 - 0x6F */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-	/* 0x70 - 0x7F */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-	/* 0x80 - 0x8F */
-	X16(D(SrcImm)),
-	/* 0x90 - 0x9F */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-	/* 0xA0 - 0xA7 */
-	D(ImplicitOps | Stack), D(ImplicitOps | Stack),
-	N, D(DstMem | SrcReg | ModRM | BitOp),
-	D(DstMem | SrcReg | Src2ImmByte | ModRM),
-	D(DstMem | SrcReg | Src2CL | ModRM), N, N,
-	/* 0xA8 - 0xAF */
-	D(ImplicitOps | Stack), D(ImplicitOps | Stack),
-	N, D(DstMem | SrcReg | ModRM | BitOp | Lock),
-	D(DstMem | SrcReg | Src2ImmByte | ModRM),
-	D(DstMem | SrcReg | Src2CL | ModRM),
-	D(ModRM), N,
-	/* 0xB0 - 0xB7 */
-	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
-	N, D(DstMem | SrcReg | ModRM | BitOp | Lock),
-	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
-	    D(DstReg | SrcMem16 | ModRM | Mov),
-	/* 0xB8 - 0xBF */
-	N, N,
-	G(0, group8), D(DstMem | SrcReg | ModRM | BitOp | Lock),
-	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
-	    D(DstReg | SrcMem16 | ModRM | Mov),
-	/* 0xC0 - 0xCF */
-	N, N, N, D(DstMem | SrcReg | ModRM | Mov),
-	N, N, N, GD(0, &group9),
-	N, N, N, N, N, N, N, N,
-	/* 0xD0 - 0xDF */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-	/* 0xE0 - 0xEF */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
-	/* 0xF0 - 0xFF */
-	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N
-};
-
-#undef D
-#undef N
-#undef G
-#undef GD
-#undef I
-
 /* EFLAGS bit definitions. */
 #define EFLG_ID (1<<21)
 #define EFLG_VIP (1<<20)
@@ -2178,6 +1942,242 @@ static void string_addr_inc(struct x86_emulate_ctxt *ctxt, unsigned long base,
 	op->ptr = (unsigned long *)register_address(c,  base, c->regs[reg]);
 }
 
+#define D(_y) { .flags = (_y) }
+#define N    D(0)
+#define G(_f, _g) { .flags = ((_f) | Group), .u.group = (_g) }
+#define GD(_f, _g) { .flags = ((_f) | Group | GroupDual), .u.gdual = (_g) }
+#define I(_f, _e) { .flags = (_f), .u.execute = (_e) }
+
+static struct opcode group1[] = {
+	X7(D(Lock)), N
+};
+
+static struct opcode group1A[] = {
+	D(DstMem | SrcNone | ModRM | Mov | Stack), N, N, N, N, N, N, N,
+};
+
+static struct opcode group3[] = {
+	D(DstMem | SrcImm | ModRM), D(DstMem | SrcImm | ModRM),
+	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
+	X4(D(Undefined)),
+};
+
+static struct opcode group4[] = {
+	D(ByteOp | DstMem | SrcNone | ModRM | Lock), D(ByteOp | DstMem | SrcNone | ModRM | Lock),
+	N, N, N, N, N, N,
+};
+
+static struct opcode group5[] = {
+	D(DstMem | SrcNone | ModRM | Lock), D(DstMem | SrcNone | ModRM | Lock),
+	D(SrcMem | ModRM | Stack), N,
+	D(SrcMem | ModRM | Stack), D(SrcMemFAddr | ModRM | ImplicitOps),
+	D(SrcMem | ModRM | Stack), N,
+};
+
+static struct group_dual group7 = { {
+	N, N, D(ModRM | SrcMem | Priv), D(ModRM | SrcMem | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), D(SrcMem | ModRM | ByteOp | Priv),
+}, {
+	D(SrcNone | ModRM | Priv), N, N, D(SrcNone | ModRM | Priv),
+	D(SrcNone | ModRM | DstMem | Mov), N,
+	D(SrcMem16 | ModRM | Mov | Priv), N,
+} };
+
+static struct opcode group8[] = {
+	N, N, N, N,
+	D(DstMem | SrcImmByte | ModRM), D(DstMem | SrcImmByte | ModRM | Lock),
+	D(DstMem | SrcImmByte | ModRM | Lock), D(DstMem | SrcImmByte | ModRM | Lock),
+};
+
+static struct group_dual group9 = { {
+	N, D(DstMem64 | ModRM | Lock), N, N, N, N, N, N,
+}, {
+	N, N, N, N, N, N, N, N,
+} };
+
+static struct opcode opcode_table[256] = {
+	/* 0x00 - 0x07 */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
+	/* 0x08 - 0x0F */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), N,
+	/* 0x10 - 0x17 */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
+	/* 0x18 - 0x1F */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
+	/* 0x20 - 0x27 */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
+	/* 0x28 - 0x2F */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
+	/* 0x30 - 0x37 */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImmByte), D(DstAcc | SrcImm), N, N,
+	/* 0x38 - 0x3F */
+	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
+	D(ByteOp | DstReg | SrcMem | ModRM), D(DstReg | SrcMem | ModRM),
+	D(ByteOp | DstAcc | SrcImm), D(DstAcc | SrcImm),
+	N, N,
+	/* 0x40 - 0x4F */
+	X16(D(DstReg)),
+	/* 0x50 - 0x57 */
+	X8(D(SrcReg | Stack)),
+	/* 0x58 - 0x5F */
+	X8(D(DstReg | Stack)),
+	/* 0x60 - 0x67 */
+	D(ImplicitOps | Stack | No64), D(ImplicitOps | Stack | No64),
+	N, D(DstReg | SrcMem32 | ModRM | Mov) /* movsxd (x86/64) */ ,
+	N, N, N, N,
+	/* 0x68 - 0x6F */
+	D(SrcImm | Mov | Stack), N, D(SrcImmByte | Mov | Stack), N,
+	D(DstDI | ByteOp | Mov | String), D(DstDI | Mov | String), /* insb, insw/insd */
+	D(SrcSI | ByteOp | ImplicitOps | String), D(SrcSI | ImplicitOps | String), /* outsb, outsw/outsd */
+	/* 0x70 - 0x7F */
+	X16(D(SrcImmByte)),
+	/* 0x80 - 0x87 */
+	G(ByteOp | DstMem | SrcImm | ModRM | Group, group1),
+	G(DstMem | SrcImm | ModRM | Group, group1),
+	G(ByteOp | DstMem | SrcImm | ModRM | No64 | Group, group1),
+	G(DstMem | SrcImmByte | ModRM | Group, group1),
+	D(ByteOp | DstMem | SrcReg | ModRM), D(DstMem | SrcReg | ModRM),
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	/* 0x88 - 0x8F */
+	D(ByteOp | DstMem | SrcReg | ModRM | Mov), D(DstMem | SrcReg | ModRM | Mov),
+	D(ByteOp | DstReg | SrcMem | ModRM | Mov), D(DstReg | SrcMem | ModRM | Mov),
+	D(DstMem | SrcNone | ModRM | Mov), D(ModRM | DstReg),
+	D(ImplicitOps | SrcMem16 | ModRM), G(0, group1A),
+	/* 0x90 - 0x97 */
+	D(DstReg), D(DstReg), D(DstReg), D(DstReg),	D(DstReg), D(DstReg), D(DstReg), D(DstReg),
+	/* 0x98 - 0x9F */
+	N, N, D(SrcImmFAddr | No64), N,
+	D(ImplicitOps | Stack), D(ImplicitOps | Stack), N, N,
+	/* 0xA0 - 0xA7 */
+	D(ByteOp | DstAcc | SrcMem | Mov | MemAbs), D(DstAcc | SrcMem | Mov | MemAbs),
+	D(ByteOp | DstMem | SrcAcc | Mov | MemAbs), D(DstMem | SrcAcc | Mov | MemAbs),
+	D(ByteOp | SrcSI | DstDI | Mov | String), D(SrcSI | DstDI | Mov | String),
+	D(ByteOp | SrcSI | DstDI | String), D(SrcSI | DstDI | String),
+	/* 0xA8 - 0xAF */
+	D(DstAcc | SrcImmByte | ByteOp), D(DstAcc | SrcImm), D(ByteOp | DstDI | Mov | String), D(DstDI | Mov | String),
+	D(ByteOp | SrcSI | DstAcc | Mov | String), D(SrcSI | DstAcc | Mov | String),
+	D(ByteOp | DstDI | String), D(DstDI | String),
+	/* 0xB0 - 0xB7 */
+	X8(D(ByteOp | DstReg | SrcImm | Mov)),
+	/* 0xB8 - 0xBF */
+	X8(D(DstReg | SrcImm | Mov)),
+	/* 0xC0 - 0xC7 */
+	D(ByteOp | DstMem | SrcImm | ModRM), D(DstMem | SrcImmByte | ModRM),
+	N, D(ImplicitOps | Stack), N, N,
+	D(ByteOp | DstMem | SrcImm | ModRM | Mov), D(DstMem | SrcImm | ModRM | Mov),
+	/* 0xC8 - 0xCF */
+	N, N, N, D(ImplicitOps | Stack),
+	D(ImplicitOps), D(SrcImmByte), D(ImplicitOps | No64), D(ImplicitOps),
+	/* 0xD0 - 0xD7 */
+	D(ByteOp | DstMem | SrcImplicit | ModRM), D(DstMem | SrcImplicit | ModRM),
+	D(ByteOp | DstMem | SrcImplicit | ModRM), D(DstMem | SrcImplicit | ModRM),
+	N, N, N, N,
+	/* 0xD8 - 0xDF */
+	N, N, N, N, N, N, N, N,
+	/* 0xE0 - 0xE7 */
+	N, N, N, N,
+	D(ByteOp | SrcImmUByte | DstAcc), D(SrcImmUByte | DstAcc),
+	D(ByteOp | SrcImmUByte | DstAcc), D(SrcImmUByte | DstAcc),
+	/* 0xE8 - 0xEF */
+	D(SrcImm | Stack), D(SrcImm | ImplicitOps),
+	D(SrcImmFAddr | No64), D(SrcImmByte | ImplicitOps),
+	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
+	D(SrcNone | ByteOp | DstAcc), D(SrcNone | DstAcc),
+	/* 0xF0 - 0xF7 */
+	N, N, N, N,
+	D(ImplicitOps | Priv), D(ImplicitOps), G(ByteOp, group3), G(0, group3),
+	/* 0xF8 - 0xFF */
+	D(ImplicitOps), N, D(ImplicitOps), D(ImplicitOps),
+	D(ImplicitOps), D(ImplicitOps), G(0, group4), G(0, group5),
+};
+
+static struct opcode twobyte_table[256] = {
+	/* 0x00 - 0x0F */
+	N, GD(0, &group7), N, N,
+	N, D(ImplicitOps), D(ImplicitOps | Priv), N,
+	D(ImplicitOps | Priv), D(ImplicitOps | Priv), N, N,
+	N, D(ImplicitOps | ModRM), N, N,
+	/* 0x10 - 0x1F */
+	N, N, N, N, N, N, N, N, D(ImplicitOps | ModRM), N, N, N, N, N, N, N,
+	/* 0x20 - 0x2F */
+	D(ModRM | ImplicitOps | Priv), D(ModRM | Priv),
+	D(ModRM | ImplicitOps | Priv), D(ModRM | Priv),
+	N, N, N, N,
+	N, N, N, N, N, N, N, N,
+	/* 0x30 - 0x3F */
+	D(ImplicitOps | Priv), N, D(ImplicitOps | Priv), N,
+	D(ImplicitOps), D(ImplicitOps | Priv), N, N,
+	N, N, N, N, N, N, N, N,
+	/* 0x40 - 0x4F */
+	X16(D(DstReg | SrcMem | ModRM | Mov)),
+	/* 0x50 - 0x5F */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+	/* 0x60 - 0x6F */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+	/* 0x70 - 0x7F */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+	/* 0x80 - 0x8F */
+	X16(D(SrcImm)),
+	/* 0x90 - 0x9F */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+	/* 0xA0 - 0xA7 */
+	D(ImplicitOps | Stack), D(ImplicitOps | Stack),
+	N, D(DstMem | SrcReg | ModRM | BitOp),
+	D(DstMem | SrcReg | Src2ImmByte | ModRM),
+	D(DstMem | SrcReg | Src2CL | ModRM), N, N,
+	/* 0xA8 - 0xAF */
+	D(ImplicitOps | Stack), D(ImplicitOps | Stack),
+	N, D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	D(DstMem | SrcReg | Src2ImmByte | ModRM),
+	D(DstMem | SrcReg | Src2CL | ModRM),
+	D(ModRM), N,
+	/* 0xB0 - 0xB7 */
+	D(ByteOp | DstMem | SrcReg | ModRM | Lock), D(DstMem | SrcReg | ModRM | Lock),
+	N, D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
+	    D(DstReg | SrcMem16 | ModRM | Mov),
+	/* 0xB8 - 0xBF */
+	N, N,
+	G(0, group8), D(DstMem | SrcReg | ModRM | BitOp | Lock),
+	N, N, D(ByteOp | DstReg | SrcMem | ModRM | Mov),
+	    D(DstReg | SrcMem16 | ModRM | Mov),
+	/* 0xC0 - 0xCF */
+	N, N, N, D(DstMem | SrcReg | ModRM | Mov),
+	N, N, N, GD(0, &group9),
+	N, N, N, N, N, N, N, N,
+	/* 0xD0 - 0xDF */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+	/* 0xE0 - 0xEF */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N,
+	/* 0xF0 - 0xFF */
+	N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N
+};
+
+#undef D
+#undef N
+#undef G
+#undef GD
+#undef I
+
 int
 x86_decode_insn(struct x86_emulate_ctxt *ctxt)
 {
-- 
1.7.1

[PATCH 21/22] KVM: x86 emulator: allow repeat macro arguments to contain commas

[Avi Kivity]

Needed for repeating instructions with execution functions.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   16 ++++++++--------
 1 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 11fd2c9..3b8a316 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -94,14 +94,14 @@
 #define Src2One     (3<<29)
 #define Src2Mask    (7<<29)
 
-#define X2(x) x, x
-#define X3(x) X2(x), x
-#define X4(x) X2(x), X2(x)
-#define X5(x) X4(x), x
-#define X6(x) X4(x), X2(x)
-#define X7(x) X4(x), X3(x)
-#define X8(x) X4(x), X4(x)
-#define X16(x) X8(x), X8(x)
+#define X2(x...) x, x
+#define X3(x...) X2(x), x
+#define X4(x...) X2(x), X2(x)
+#define X5(x...) X4(x), x
+#define X6(x...) X4(x), X2(x)
+#define X7(x...) X4(x), X3(x)
+#define X8(x...) X4(x), X4(x)
+#define X16(x...) X8(x), X8(x)
 
 struct opcode {
 	u32 flags;
-- 
1.7.1

[PATCH 22/22] KVM: x86 emulator: convert some push instructions to direct decode

[Avi Kivity]

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 arch/x86/kvm/emulate.c |   18 +++++++++---------
 1 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
index 3b8a316..fedfc9c 100644
--- a/arch/x86/kvm/emulate.c
+++ b/arch/x86/kvm/emulate.c
@@ -1942,6 +1942,12 @@ static void string_addr_inc(struct x86_emulate_ctxt *ctxt, unsigned long base,
 	op->ptr = (unsigned long *)register_address(c,  base, c->regs[reg]);
 }
 
+static int em_push(struct x86_emulate_ctxt *ctxt)
+{
+	emulate_push(ctxt, ctxt->ops);
+	return X86EMUL_CONTINUE;
+}
+
 #define D(_y) { .flags = (_y) }
 #define N    D(0)
 #define G(_f, _g) { .flags = ((_f) | Group), .u.group = (_g) }
@@ -2037,7 +2043,7 @@ static struct opcode opcode_table[256] = {
 	/* 0x40 - 0x4F */
 	X16(D(DstReg)),
 	/* 0x50 - 0x57 */
-	X8(D(SrcReg | Stack)),
+	X8(I(SrcReg | Stack, em_push)),
 	/* 0x58 - 0x5F */
 	X8(D(DstReg | Stack)),
 	/* 0x60 - 0x67 */
@@ -2045,7 +2051,8 @@ static struct opcode opcode_table[256] = {
 	N, D(DstReg | SrcMem32 | ModRM | Mov) /* movsxd (x86/64) */ ,
 	N, N, N, N,
 	/* 0x68 - 0x6F */
-	D(SrcImm | Mov | Stack), N, D(SrcImmByte | Mov | Stack), N,
+	I(SrcImm | Mov | Stack, em_push), N,
+	I(SrcImmByte | Mov | Stack, em_push), N,
 	D(DstDI | ByteOp | Mov | String), D(DstDI | Mov | String), /* insb, insw/insd */
 	D(SrcSI | ByteOp | ImplicitOps | String), D(SrcSI | ImplicitOps | String), /* outsb, outsw/outsd */
 	/* 0x70 - 0x7F */
@@ -2712,9 +2719,6 @@ special_insn:
 	case 0x48 ... 0x4f: /* dec r16/r32 */
 		emulate_1op("dec", c->dst, ctxt->eflags);
 		break;
-	case 0x50 ... 0x57:  /* push reg */
-		emulate_push(ctxt, ops);
-		break;
 	case 0x58 ... 0x5f: /* pop reg */
 	pop_instruction:
 		rc = emulate_pop(ctxt, ops, &c->dst.val, c->op_bytes);
@@ -2736,10 +2740,6 @@ special_insn:
 			goto cannot_emulate;
 		c->dst.val = (s32) c->src.val;
 		break;
-	case 0x68: /* push imm */
-	case 0x6a: /* push imm8 */
-		emulate_push(ctxt, ops);
-		break;
 	case 0x6c:		/* insb */
 	case 0x6d:		/* insw/insd */
 		c->dst.bytes = min(c->dst.bytes, 4u);
-- 
1.7.1

Re: [PATCH 00/22] Direct decode in emulator

[Marcelo Tosatti]

On Thu, Jul 29, 2010 at 03:11:33PM +0300, Avi Kivity wrote:
> Currently the emulator decoder is indirect in two ways: groups are looked up
> in a separate table based on group number; and opcode execution is looked up
> in a big switch (actually two) after the opcode flags lookup.
> 
> This patchset converts groups to direct decode and begins the work for opcode
> decode.
> 
> The motivations are:
> - reduce errors from discrepancies between the two tables (like we had for
>   IRET)
> - make the code easier to read, removing many switch () statements
> - eventually remove the #UD mini-decoder in x86.c by adding subgroup support
>   to the emulator
> 
> Avi Kivity (22):
>   KVM: x86 emulator: drop parentheses in repreat macros
>   KVM: x86 emulator: introduce 'struct opcode'
>   KVM: x86 emulator: Add wrappers for easily defining opcodes
>   KVM: x86 emulator: move group tables to top
>   KVM: x86 emulator: reserve group code 0
>   KVM: x86 emulator: allow specifying group directly in opcode
>   KVM: x86 emulator: convert group 1 to new style
>   KVM: x86 emulator: convert group 1A to new style
>   KVM: x86 emulator: convert group 3 to new style
>   KVM: x86 emulator: convert group 4 to new style
>   KVM: x86 emulator: convert group 5 to new style
>   KVM: x86 emulator: convert group 7 to new style
>   KVM: x86 emulator: convert group 8 to new style
>   KVM: x86 emulator: convert group 9 to new style
>   KVM: x86 emulator: drop support for old-style groups
>   KVM: x86 emulator: move ByteOp and Dst back to bits 0:3
>   KVM: x86 emulator: store x86_emulate_ops in emulation context
>   KVM: x86 emulator: allow storing emulator execution function in
>     decode tables
>   KVM: x86 emulator: move x86_decode_insn() downwards
>   KVM: x86 emulator: move decode tables downwards
>   KVM: x86 emulator: allow repeat macro arguments to contain commas
>   KVM: x86 emulator: convert some push instructions to direct decode
> 
>  arch/x86/include/asm/kvm_emulate.h |   10 +-
>  arch/x86/kvm/emulate.c             | 1253 +++++++++++++++++++-----------------
>  arch/x86/kvm/x86.c                 |    7 +-
>  3 files changed, 661 insertions(+), 609 deletions(-)

Applied, thanks.