Foundations of jit, no actual compiling yet

emu/cpu.h

@@ -20,6 +20,7 @@ union xmm_reg {
 
 struct cpu_state {
     struct mem *mem;
+    struct jit *jit;
 
     // assumes little endian (as does literally everything)
 #define _REG(n) \

emu/decode.h

@@ -12,18 +12,17 @@
 __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
     DECLARE_LOCALS;
 
-    dword_t addr_offset = 0;
     byte_t insn;
     uint64_t imm;
+    struct modrm modrm;
 #define READIMM_(name, size) _READIMM(name, size); TRACE("imm %llx ", (long long) name)
-#define READADDR READIMM_(addr_offset, 32); addr += addr_offset
 #define READINSN _READIMM(insn, 8); TRACE("%02x ", insn)
 #define READIMM READIMM_(imm, OP_SIZE)
 #define READIMM8 READIMM_(imm, 8)
 #define READIMM16 READIMM_(imm, 16)
 
 restart:
-    TRACE("%d %08x\t", current->pid, cpu->eip);
+    TRACEIP();
     READINSN;
     switch (insn) {
 #define MAKE_OP(x, OP, op) \
@@ -168,10 +167,7 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
                 case 0x9f: TRACEI("setnle\t");
                            READMODRM; SET(!LE, modrm_val); break;
 
-                case 0xa2:
-                    TRACEI("cpuid");
-                    do_cpuid(&cpu->eax, &cpu->ebx, &cpu->ecx, &cpu->edx);
-                    break;
+                case 0xa2: TRACEI("cpuid"); CPUID(); break;
 
                 case 0xa3: TRACEI("bt reg, modrm");
                            READMODRM; BT(modrm_reg, modrm_val,); break;
@@ -310,16 +306,15 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
         case 0x5e: TRACEI("pop osi"); POP(osi); break;
         case 0x5f: TRACEI("pop odi"); POP(odi); break;
 
-        case 0x65: TRACELN("segment gs");
-                   addr += cpu->tls_ptr; goto restart;
+        case 0x65: TRACELN("segment gs"); SEG_GS(); goto restart;
 
         case 0x66:
 #if OP_SIZE == 32
             TRACELN("entering 16 bit mode");
-            return glue(DECODER_NAME, 16)(DECODER_PASS_ARGS);
+            RETURN(glue(DECODER_NAME, 16)(DECODER_PASS_ARGS));
 #else
             TRACELN("entering 32 bit mode");
-            return glue(DECODER_NAME, 32)(DECODER_PASS_ARGS);
+            RETURN(glue(DECODER_NAME, 32)(DECODER_PASS_ARGS));
 #endif
 
         case 0x67: TRACEI("address size prefix (ignored)"); goto restart;
@@ -725,5 +720,5 @@ __no_instrument DECODER_RET glue(DECODER_NAME, OP_SIZE)(DECODER_ARGS) {
             UNDEFINED;
     }
     TRACELN("");
-    FINISH;
+    RETURN(-1); // everything is ok.
 }

emu/gadgets-x86/gadgets.S

@@ -0,0 +1,44 @@
+#include "gadgets.h"
+#include "cpu-offsets.h"
+
+.global jit_enter
+.type jit_enter,function
+jit_enter:
+    push %rbp
+    push %rbx
+    leaq JIT_BLOCK_code(%rsi), %ip
+    movq %rdi, %cpu
+    movl CPU_eax(%cpu), %eax
+    movl CPU_ebx(%cpu), %ebx
+    movl CPU_ecx(%cpu), %ecx
+    movl CPU_edx(%cpu), %edx
+    movl CPU_esi(%cpu), %esi
+    movl CPU_edi(%cpu), %edi
+    movl CPU_ebp(%cpu), %ebp
+    movl CPU_esp(%cpu), %xsp
+    # TODO more of those
+    endgadget
+
+.global jit_exit
+jit_exit:
+    movl %eax, CPU_eax(%cpu)
+    movl %ebx, CPU_ebx(%cpu)
+    movl %ecx, CPU_ecx(%cpu)
+    movl %edx, CPU_edx(%cpu)
+    movl %esi, CPU_esi(%cpu)
+    movl %edi, CPU_edi(%cpu)
+    movl %ebp, CPU_ebp(%cpu)
+    movl %xsp, CPU_esp(%cpu)
+    # TODO more of those
+    pop %rbx
+    pop %rbp
+    mov %tmp, %eax
+    ret
+
+gadget interrupt
+    op %tmp
+    jmp jit_exit
+
+gadget exit
+    movl $-1, %tmp
+    jmp jit_exit

emu/gadgets-x86/gadgets.h

@@ -0,0 +1,19 @@
+// register assignments
+#define ip r8
+#define tmp r9d
+#define xsp r10d
+#define cpu r11
+
+.macro gadget name
+.global gadget_\()\name
+gadget_\()\name :
+.endmacro
+.macro endgadget
+    lea 8(%ip), %ip
+    jmp *-8(%ip)
+.endmacro
+
+.macro op reg
+    mov (%ip), \reg
+    lea 8(%ip), %ip
+.endmacro

emu/gadgets-x86/offsets.c

@@ -0,0 +1,20 @@
+#include "emu/cpu.h"
+
+void cpu() {
+    OFFSET(CPU_eax, struct cpu_state, eax);
+    OFFSET(CPU_ebx, struct cpu_state, ebx);
+    OFFSET(CPU_ecx, struct cpu_state, ecx);
+    OFFSET(CPU_edx, struct cpu_state, edx);
+    OFFSET(CPU_esi, struct cpu_state, esi);
+    OFFSET(CPU_edi, struct cpu_state, edi);
+    OFFSET(CPU_ebp, struct cpu_state, ebp);
+    OFFSET(CPU_esp, struct cpu_state, esp);
+    OFFSET(CPU_ax, struct cpu_state, ax);
+    OFFSET(CPU_bx, struct cpu_state, bx);
+    OFFSET(CPU_cx, struct cpu_state, cx);
+    OFFSET(CPU_dx, struct cpu_state, dx);
+    OFFSET(CPU_si, struct cpu_state, si);
+    OFFSET(CPU_di, struct cpu_state, di);
+    OFFSET(CPU_bp, struct cpu_state, bp);
+    OFFSET(CPU_sp, struct cpu_state, sp);
+}

emu/gadgets.h

@@ -0,0 +1,7 @@
+#ifndef EMU_GADGETS_H
+#define EMU_GADGETS_H
+
+void gadget_exit();
+void gadget_interrupt();
+
+#endif

emu/gen.c

@@ -0,0 +1,157 @@
+static void gen_step32(struct gen_state *state, addr_t ip, struct tlb *tlb);
+static void gen_step16(struct gen_state *state, addr_t ip, struct tlb *tlb);
+
+#define DECLARE_LOCALS \
+    dword_t addr_offset = 0;
+#define RETURN(thing) (void) (thing)
+
+#define TRACEIP() TRACE("%d %08x\t", current->pid, ip);
+
+#define _READIMM(name, size) \
+    tlb_read(tlb, ip, &name, size/8); \
+    ip += size/8
+
+#define READMODRM modrm_decode32(&ip, tlb, &modrm)
+#define READADDR _READIMM(addr_offset, 32)
+#define SEG_GS() UNDEFINED
+
+#define UNDEFINED GG(interrupt, INT_UNDEFINED); return;
+
+#define ADD(src, dst,z) UNDEFINED
+#define OR(src, dst,z) UNDEFINED
+#define ADC(src, dst,z) UNDEFINED
+#define SBB(src, dst,z) UNDEFINED
+#define AND(src, dst,z) UNDEFINED
+#define SUB(src, dst,z) UNDEFINED
+#define XOR(src, dst,z) UNDEFINED
+#define CMP(src, dst,z) UNDEFINED
+#define TEST(src, dst,z) UNDEFINED
+#define NOT(val,z) UNDEFINED
+#define NEG(val,z) UNDEFINED
+
+#define INC(val,z) UNDEFINED
+#define DEC(val,z) UNDEFINED
+
+#define JMP(loc) UNDEFINED
+#define JMP_REL(off) UNDEFINED
+#define JCXZ_REL(off) UNDEFINED
+#define J_REL(cc, off) UNDEFINED
+#define CALL(loc) UNDEFINED
+#define CALL_REL(off) UNDEFINED
+#define SET(cc, dst) UNDEFINED
+#define RET_NEAR_IMM(imm) UNDEFINED
+#define RET_NEAR() UNDEFINED
+#define INT(code) UNDEFINED
+
+#define PUSHF() UNDEFINED
+#define POPF() UNDEFINED
+#define SAHF UNDEFINED
+#define CLD UNDEFINED
+#define STD UNDEFINED
+
+#define MOV(src, dst,z) UNDEFINED
+#define MOVZX(src, dst,zs,zd) UNDEFINED
+#define MOVSX(src, dst,zs,zd) UNDEFINED
+#define XCHG(src, dst,z) UNDEFINED
+#define CMOV(cc, src, dst,z) UNDEFINED
+
+#define POP(thing) UNDEFINED
+#define PUSH(thing) UNDEFINED
+
+#define MUL18(val) UNDEFINED
+#define MUL1(val,z) UNDEFINED
+#define IMUL1(val,z) UNDEFINED
+#define MUL2(val, reg) UNDEFINED
+#define IMUL2(val, reg,z) UNDEFINED
+#define MUL3(imm, src, dst) UNDEFINED
+#define IMUL3(imm, src, dst,z) UNDEFINED
+#define DIV(reg, val, rem,z) UNDEFINED
+#define IDIV(reg, val, rem,z) UNDEFINED
+
+#define CVT UNDEFINED
+#define CVTE UNDEFINED
+
+#define ROL(count, val,z) UNDEFINED
+#define ROR(count, val,z) UNDEFINED
+#define SHL(count, val,z) UNDEFINED
+#define SHR(count, val,z) UNDEFINED
+#define SAR(count, val,z) UNDEFINED
+
+#define SHLD(count, extra, dst,z) UNDEFINED
+#define SHRD(count, extra, dst,z) UNDEFINED
+
+#define BT(bit, val,z) UNDEFINED
+#define BTC(bit, val,z) UNDEFINED
+#define BTS(bit, val,z) UNDEFINED
+#define BTR(bit, val,z) UNDEFINED
+#define BSF(src, dst,z) UNDEFINED
+#define BSR(src, dst,z) UNDEFINED
+
+#define BSWAP(dst) UNDEFINED
+
+#define SCAS(z) UNDEFINED
+#define MOVS(z) UNDEFINED
+#define LODS(z) UNDEFINED
+#define STOS(z) UNDEFINED
+#define CMPS(z) UNDEFINED
+#define REP(op) UNDEFINED
+#define REPZ(op) UNDEFINED
+#define REPNZ(op) UNDEFINED
+
+#define CMPXCHG(src, dst,z) UNDEFINED
+#define XADD(src, dst,z) UNDEFINED
+
+#define RDTSC UNDEFINED
+#define CPUID() UNDEFINED
+
+// sse
+#define XORP(src, dst) UNDEFINED
+#define PSRLQ(src, dst) UNDEFINED
+#define PCMPEQD(src, dst) UNDEFINED
+#define PADD(src, dst) UNDEFINED
+#define PSUB(src, dst) UNDEFINED
+#define MOVQ(src, dst) UNDEFINED
+#define MOVD(src, dst) UNDEFINED
+#define CVTTSD2SI(src, dst) UNDEFINED
+
+// fpu
+#define FLD() UNDEFINED
+#define FILD(val,z) UNDEFINED
+#define FLDM(val,z) UNDEFINED
+#define FSTM(dst,z) UNDEFINED
+#define FIST(dst,z) UNDEFINED
+#define FXCH() UNDEFINED
+#define FUCOM() UNDEFINED
+#define FUCOMI() UNDEFINED
+#define FST() UNDEFINED
+#define FCHS() UNDEFINED
+#define FABS() UNDEFINED
+#define FLDC(what) UNDEFINED
+#define FPREM() UNDEFINED
+#define FSTSW(dst) UNDEFINED
+#define FSTCW(dst) UNDEFINED
+#define FLDCW(dst) UNDEFINED
+#define FPOP UNDEFINED
+#define FADD(src, dst) UNDEFINED
+#define FIADD(val,z) UNDEFINED
+#define FADDM(val,z) UNDEFINED
+#define FSUB(src, dst) UNDEFINED
+#define FSUBM(val,z) UNDEFINED
+#define FISUB(val,z) UNDEFINED
+#define FMUL(src, dst) UNDEFINED
+#define FIMUL(val,z) UNDEFINED
+#define FMULM(val,z) UNDEFINED
+#define FDIV(src, dst) UNDEFINED
+#define FIDIV(val,z) UNDEFINED
+#define FDIVM(val,z) UNDEFINED
+
+#define DECODER_RET static void
+#define DECODER_NAME gen_step
+#define DECODER_ARGS struct gen_state *state, addr_t ip, struct tlb *tlb
+#define DECODER_PASS_ARGS state, ip, tlb
+#define OP_SIZE 32
+#include "emu/decode.h"
+#undef OP_SIZE
+#define OP_SIZE 16
+#include "emu/decode.h"
+#undef OP_SIZE

emu/interp.c

@@ -8,8 +8,8 @@
 #pragma GCC diagnostic ignored "-Wtautological-constant-out-of-range-compare"
 
 #define DECLARE_LOCALS \
+    dword_t addr_offset = 0; \
     dword_t saved_ip = cpu->eip; \
-    struct modrm modrm; \
     struct regptr modrm_regptr, modrm_base; \
     dword_t addr = 0; \
     \
@@ -18,8 +18,7 @@
     \
     extFloat80_t ftmp;
 
-#define FINISH \
-    return -1 // everything is ok.
+#define RETURN(thing) return (thing)
 
 #define UNDEFINED { cpu->eip = saved_ip; return INT_UNDEFINED; }
 
@@ -31,11 +30,16 @@ static bool modrm_compute(struct cpu_state *cpu, struct tlb *tlb, addr_t *addr_o
         cpu->eip = saved_ip; \
         return INT_GPF; \
     }
+#define READADDR READIMM_(addr_offset, 32); addr += addr_offset
 
 #define _READIMM(name,size) \
     name = mem_read(cpu->eip, size); \
     cpu->eip += size/8
 
+#define TRACEIP() TRACE("%d %08x\t", current->pid, cpu->eip);
+
+#define SEG_GS() addr += cpu->tls_ptr
+
 // this is a completely insane way to turn empty into OP_SIZE and any other size into itself
 #define sz(x) sz_##x
 #define sz_ OP_SIZE
@@ -590,6 +594,9 @@ static bool modrm_compute(struct cpu_state *cpu, struct tlb *tlb, addr_t *addr_o
     cpu->eax = imm & 0xffffffff; \
     cpu->edx = imm >> 32
 
+#define CPUID() \
+    do_cpuid(&cpu->eax, &cpu->ebx, &cpu->ecx, &cpu->edx)
+
 #include "emu/interp/sse.h"
 #include "emu/interp/fpu.h"