lk/linux-kernel-test
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

KVM: x86 emulator: SSE support

Browse Source 
Add support for marking an instruction as SSE, switching registers used
to the SSE register file.

Signed-off-by: Avi Kivity <avi@redhat.com>
This commit is contained in:
Avi Kivity
2011-03-29 11:41:27 +02:00
parent 0d7cdee83a
commit 1253791df9
2 changed files with 104 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
arch/x86/include/asm/kvm_emulate.h
View File

@@ -162,9 +162,11 @@ struct x86_emulate_ops {
void (*put_fpu)(struct x86_emulate_ctxt *ctxt); /* reenables preempt */ void (*put_fpu)(struct x86_emulate_ctxt *ctxt); /* reenables preempt */
}; };
typedef u32 __attribute__((vector_size(16))) sse128_t;
/* Type, address-of, and value of an instruction's operand. */ /* Type, address-of, and value of an instruction's operand. */
struct operand { struct operand {
enum { OP_REG, OP_MEM, OP_IMM, OP_NONE } type; enum { OP_REG, OP_MEM, OP_IMM, OP_XMM, OP_NONE } type;
unsigned int bytes; unsigned int bytes;
union { union {
unsigned long orig_val; unsigned long orig_val;
@@ -176,11 +178,13 @@ struct operand {
ulong ea; ulong ea;
unsigned seg; unsigned seg;
} mem; } mem;
unsigned xmm;
} addr; } addr;
union { union {
unsigned long val; unsigned long val;
u64 val64; u64 val64;
char valptr[sizeof(unsigned long) + 2]; char valptr[sizeof(unsigned long) + 2];
sse128_t vec_val;
}; };
}; };

102
arch/x86/kvm/emulate.c
View File

@@ -76,6 +76,7 @@
#define Group (1<<14) /* Bits 3:5 of modrm byte extend opcode */ #define Group (1<<14) /* Bits 3:5 of modrm byte extend opcode */
#define GroupDual (1<<15) /* Alternate decoding of mod == 3 */ #define GroupDual (1<<15) /* Alternate decoding of mod == 3 */
#define Prefix (1<<16) /* Instruction varies with 66/f2/f3 prefix */ #define Prefix (1<<16) /* Instruction varies with 66/f2/f3 prefix */
#define Sse (1<<17) /* SSE Vector instruction */
/* Misc flags */ /* Misc flags */
#define VendorSpecific (1<<22) /* Vendor specific instruction */ #define VendorSpecific (1<<22) /* Vendor specific instruction */
#define NoAccess (1<<23) /* Don't access memory (lea/invlpg/verr etc) */ #define NoAccess (1<<23) /* Don't access memory (lea/invlpg/verr etc) */
@@ -505,6 +506,11 @@ static int emulate_de(struct x86_emulate_ctxt *ctxt)
return emulate_exception(ctxt, DE_VECTOR, 0, false); return emulate_exception(ctxt, DE_VECTOR, 0, false);
} }
static int emulate_nm(struct x86_emulate_ctxt *ctxt)
{
return emulate_exception(ctxt, NM_VECTOR, 0, false);
}
static int do_fetch_insn_byte(struct x86_emulate_ctxt *ctxt, static int do_fetch_insn_byte(struct x86_emulate_ctxt *ctxt,
struct x86_emulate_ops *ops, struct x86_emulate_ops *ops,
unsigned long eip, u8 *dest) unsigned long eip, u8 *dest)
@@ -632,7 +638,63 @@ static void fetch_register_operand(struct operand *op)
} }
} }
static void decode_register_operand(struct operand *op, static void read_sse_reg(struct x86_emulate_ctxt *ctxt, sse128_t *data, int reg)
{
ctxt->ops->get_fpu(ctxt);
switch (reg) {
case 0: asm("movdqu %%xmm0, %0" : "=m"(*data)); break;
case 1: asm("movdqu %%xmm1, %0" : "=m"(*data)); break;
case 2: asm("movdqu %%xmm2, %0" : "=m"(*data)); break;
case 3: asm("movdqu %%xmm3, %0" : "=m"(*data)); break;
case 4: asm("movdqu %%xmm4, %0" : "=m"(*data)); break;
case 5: asm("movdqu %%xmm5, %0" : "=m"(*data)); break;
case 6: asm("movdqu %%xmm6, %0" : "=m"(*data)); break;
case 7: asm("movdqu %%xmm7, %0" : "=m"(*data)); break;
#ifdef CONFIG_X86_64
case 8: asm("movdqu %%xmm8, %0" : "=m"(*data)); break;
case 9: asm("movdqu %%xmm9, %0" : "=m"(*data)); break;
case 10: asm("movdqu %%xmm10, %0" : "=m"(*data)); break;
case 11: asm("movdqu %%xmm11, %0" : "=m"(*data)); break;
case 12: asm("movdqu %%xmm12, %0" : "=m"(*data)); break;
case 13: asm("movdqu %%xmm13, %0" : "=m"(*data)); break;
case 14: asm("movdqu %%xmm14, %0" : "=m"(*data)); break;
case 15: asm("movdqu %%xmm15, %0" : "=m"(*data)); break;
#endif
default: BUG();
}
ctxt->ops->put_fpu(ctxt);
}
static void write_sse_reg(struct x86_emulate_ctxt *ctxt, sse128_t *data,
int reg)
{
ctxt->ops->get_fpu(ctxt);
switch (reg) {
case 0: asm("movdqu %0, %%xmm0" : : "m"(*data)); break;
case 1: asm("movdqu %0, %%xmm1" : : "m"(*data)); break;
case 2: asm("movdqu %0, %%xmm2" : : "m"(*data)); break;
case 3: asm("movdqu %0, %%xmm3" : : "m"(*data)); break;
case 4: asm("movdqu %0, %%xmm4" : : "m"(*data)); break;
case 5: asm("movdqu %0, %%xmm5" : : "m"(*data)); break;
case 6: asm("movdqu %0, %%xmm6" : : "m"(*data)); break;
case 7: asm("movdqu %0, %%xmm7" : : "m"(*data)); break;
#ifdef CONFIG_X86_64
case 8: asm("movdqu %0, %%xmm8" : : "m"(*data)); break;
case 9: asm("movdqu %0, %%xmm9" : : "m"(*data)); break;
case 10: asm("movdqu %0, %%xmm10" : : "m"(*data)); break;
case 11: asm("movdqu %0, %%xmm11" : : "m"(*data)); break;
case 12: asm("movdqu %0, %%xmm12" : : "m"(*data)); break;
case 13: asm("movdqu %0, %%xmm13" : : "m"(*data)); break;
case 14: asm("movdqu %0, %%xmm14" : : "m"(*data)); break;
case 15: asm("movdqu %0, %%xmm15" : : "m"(*data)); break;
#endif
default: BUG();
}
ctxt->ops->put_fpu(ctxt);
}
static void decode_register_operand(struct x86_emulate_ctxt *ctxt,
struct operand *op,
struct decode_cache *c, struct decode_cache *c,
int inhibit_bytereg) int inhibit_bytereg)
{ {
@@ -641,6 +703,15 @@ static void decode_register_operand(struct operand *op,
if (!(c->d & ModRM)) if (!(c->d & ModRM))
reg = (c->b & 7) | ((c->rex_prefix & 1) << 3); reg = (c->b & 7) | ((c->rex_prefix & 1) << 3);
if (c->d & Sse) {
op->type = OP_XMM;
op->bytes = 16;
op->addr.xmm = reg;
read_sse_reg(ctxt, &op->vec_val, reg);
return;
}
op->type = OP_REG; op->type = OP_REG;
if ((c->d & ByteOp) && !inhibit_bytereg) { if ((c->d & ByteOp) && !inhibit_bytereg) {
op->addr.reg = decode_register(reg, c->regs, highbyte_regs); op->addr.reg = decode_register(reg, c->regs, highbyte_regs);
@@ -680,6 +751,13 @@ static int decode_modrm(struct x86_emulate_ctxt *ctxt,
op->bytes = (c->d & ByteOp) ? 1 : c->op_bytes; op->bytes = (c->d & ByteOp) ? 1 : c->op_bytes;
op->addr.reg = decode_register(c->modrm_rm, op->addr.reg = decode_register(c->modrm_rm,
c->regs, c->d & ByteOp); c->regs, c->d & ByteOp);
if (c->d & Sse) {
op->type = OP_XMM;
op->bytes = 16;
op->addr.xmm = c->modrm_rm;
read_sse_reg(ctxt, &op->vec_val, c->modrm_rm);
return rc;
}
fetch_register_operand(op); fetch_register_operand(op);
return rc; return rc;
} }
@@ -1107,6 +1185,9 @@ static inline int writeback(struct x86_emulate_ctxt *ctxt,
if (rc != X86EMUL_CONTINUE) if (rc != X86EMUL_CONTINUE)
return rc; return rc;
break; break;
case OP_XMM:
write_sse_reg(ctxt, &c->dst.vec_val, c->dst.addr.xmm);
break;
case OP_NONE: case OP_NONE:
/* no writeback */ /* no writeback */
break; break;
@@ -2785,6 +2866,9 @@ done_prefixes:
c->op_bytes = 4; c->op_bytes = 4;
} }
if (c->d & Sse)
c->op_bytes = 16;
/* ModRM and SIB bytes. */ /* ModRM and SIB bytes. */
if (c->d & ModRM) { if (c->d & ModRM) {
rc = decode_modrm(ctxt, ops, &memop); rc = decode_modrm(ctxt, ops, &memop);
@@ -2814,7 +2898,7 @@ done_prefixes:
case SrcNone: case SrcNone:
break; break;
case SrcReg: case SrcReg:
decode_register_operand(&c->src, c, 0); decode_register_operand(ctxt, &c->src, c, 0);
break; break;
case SrcMem16: case SrcMem16:
memop.bytes = 2; memop.bytes = 2;
@@ -2905,7 +2989,7 @@ done_prefixes:
/* Decode and fetch the destination operand: register or memory. */ /* Decode and fetch the destination operand: register or memory. */
switch (c->d & DstMask) { switch (c->d & DstMask) {
case DstReg: case DstReg:
decode_register_operand(&c->dst, c, decode_register_operand(ctxt, &c->dst, c,
c->twobyte && (c->b == 0xb6 || c->b == 0xb7)); c->twobyte && (c->b == 0xb6 || c->b == 0xb7));
break; break;
case DstImmUByte: case DstImmUByte:
@@ -3001,6 +3085,18 @@ x86_emulate_insn(struct x86_emulate_ctxt *ctxt)
goto done; goto done;
} }
if ((c->d & Sse)
&& ((ops->get_cr(0, ctxt->vcpu) & X86_CR0_EM)
|| !(ops->get_cr(4, ctxt->vcpu) & X86_CR4_OSFXSR))) {
rc = emulate_ud(ctxt);
goto done;
}
if ((c->d & Sse) && (ops->get_cr(0, ctxt->vcpu) & X86_CR0_TS)) {
rc = emulate_nm(ctxt);
goto done;
}
/* Privileged instruction can be executed only in CPL=0 */ /* Privileged instruction can be executed only in CPL=0 */
if ((c->d & Priv) && ops->cpl(ctxt->vcpu)) { if ((c->d & Priv) && ops->cpl(ctxt->vcpu)) {
rc = emulate_gp(ctxt, 0); rc = emulate_gp(ctxt, 0);