diff options
author | bellard <bellard@c046a42c-6fe2-441c-8c8c-71466251a162> | 2003-09-30 20:34:21 +0000 |
---|---|---|
committer | bellard <bellard@c046a42c-6fe2-441c-8c8c-71466251a162> | 2003-09-30 20:34:21 +0000 |
commit | 2c0262afa75d7d6cb68217f9a7587170ed3cd5b3 (patch) | |
tree | 4f8130ebaf041129d5aabede251eb87f3a1fa559 /target-i386 | |
parent | 196ad10903ee4c2148f296600f537d7028ea845d (diff) |
new directory structure
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@385 c046a42c-6fe2-441c-8c8c-71466251a162
Diffstat (limited to 'target-i386')
-rw-r--r-- | target-i386/cpu.h | 401 | ||||
-rw-r--r-- | target-i386/exec.h | 416 | ||||
-rw-r--r-- | target-i386/helper.c | 1803 | ||||
-rw-r--r-- | target-i386/helper2.c | 390 | ||||
-rw-r--r-- | target-i386/op.c | 2054 | ||||
-rw-r--r-- | target-i386/opreg_template.h | 134 | ||||
-rw-r--r-- | target-i386/ops_mem.h | 66 | ||||
-rw-r--r-- | target-i386/ops_template.h | 617 | ||||
-rw-r--r-- | target-i386/ops_template_mem.h | 429 | ||||
-rw-r--r-- | target-i386/translate.c | 4487 |
10 files changed, 10797 insertions, 0 deletions
diff --git a/target-i386/cpu.h b/target-i386/cpu.h new file mode 100644 index 0000000000..ffc4654d34 --- /dev/null +++ b/target-i386/cpu.h @@ -0,0 +1,401 @@ +/* + * i386 virtual CPU header + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#ifndef CPU_I386_H +#define CPU_I386_H + +#include "cpu-defs.h" + +#define R_EAX 0 +#define R_ECX 1 +#define R_EDX 2 +#define R_EBX 3 +#define R_ESP 4 +#define R_EBP 5 +#define R_ESI 6 +#define R_EDI 7 + +#define R_AL 0 +#define R_CL 1 +#define R_DL 2 +#define R_BL 3 +#define R_AH 4 +#define R_CH 5 +#define R_DH 6 +#define R_BH 7 + +#define R_ES 0 +#define R_CS 1 +#define R_SS 2 +#define R_DS 3 +#define R_FS 4 +#define R_GS 5 + +/* segment descriptor fields */ +#define DESC_G_MASK (1 << 23) +#define DESC_B_SHIFT 22 +#define DESC_B_MASK (1 << DESC_B_SHIFT) +#define DESC_AVL_MASK (1 << 20) +#define DESC_P_MASK (1 << 15) +#define DESC_DPL_SHIFT 13 +#define DESC_S_MASK (1 << 12) +#define DESC_TYPE_SHIFT 8 +#define DESC_A_MASK (1 << 8) + +#define DESC_CS_MASK (1 << 11) +#define DESC_C_MASK (1 << 10) +#define DESC_R_MASK (1 << 9) + +#define DESC_E_MASK (1 << 10) +#define DESC_W_MASK (1 << 9) + +/* eflags masks */ +#define CC_C 0x0001 +#define CC_P 0x0004 +#define CC_A 0x0010 +#define CC_Z 0x0040 +#define CC_S 0x0080 +#define CC_O 0x0800 + +#define TF_SHIFT 8 +#define IOPL_SHIFT 12 +#define VM_SHIFT 17 + +#define TF_MASK 0x00000100 +#define IF_MASK 0x00000200 +#define DF_MASK 0x00000400 +#define IOPL_MASK 0x00003000 +#define NT_MASK 0x00004000 +#define RF_MASK 0x00010000 +#define VM_MASK 0x00020000 +#define AC_MASK 0x00040000 +#define VIF_MASK 0x00080000 +#define VIP_MASK 0x00100000 +#define ID_MASK 0x00200000 + +/* hidden flags - used internally by qemu to represent additionnal cpu + states. Only the CPL and INHIBIT_IRQ are not redundant. We avoid + using the IOPL_MASK, TF_MASK and VM_MASK bit position to ease oring + with eflags. */ +/* current cpl */ +#define HF_CPL_SHIFT 0 +/* true if soft mmu is being used */ +#define HF_SOFTMMU_SHIFT 2 +/* true if hardware interrupts must be disabled for next instruction */ +#define HF_INHIBIT_IRQ_SHIFT 3 +/* 16 or 32 segments */ +#define HF_CS32_SHIFT 4 +#define HF_SS32_SHIFT 5 +/* zero base for DS, ES and SS */ +#define HF_ADDSEG_SHIFT 6 + +#define HF_CPL_MASK (3 << HF_CPL_SHIFT) +#define HF_SOFTMMU_MASK (1 << HF_SOFTMMU_SHIFT) +#define HF_INHIBIT_IRQ_MASK (1 << HF_INHIBIT_IRQ_SHIFT) +#define HF_CS32_MASK (1 << HF_CS32_SHIFT) +#define HF_SS32_MASK (1 << HF_SS32_SHIFT) +#define HF_ADDSEG_MASK (1 << HF_ADDSEG_SHIFT) + +#define CR0_PE_MASK (1 << 0) +#define CR0_TS_MASK (1 << 3) +#define CR0_WP_MASK (1 << 16) +#define CR0_AM_MASK (1 << 18) +#define CR0_PG_MASK (1 << 31) + +#define CR4_VME_MASK (1 << 0) +#define CR4_PVI_MASK (1 << 1) +#define CR4_TSD_MASK (1 << 2) +#define CR4_DE_MASK (1 << 3) +#define CR4_PSE_MASK (1 << 4) + +#define PG_PRESENT_BIT 0 +#define PG_RW_BIT 1 +#define PG_USER_BIT 2 +#define PG_PWT_BIT 3 +#define PG_PCD_BIT 4 +#define PG_ACCESSED_BIT 5 +#define PG_DIRTY_BIT 6 +#define PG_PSE_BIT 7 +#define PG_GLOBAL_BIT 8 + +#define PG_PRESENT_MASK (1 << PG_PRESENT_BIT) +#define PG_RW_MASK (1 << PG_RW_BIT) +#define PG_USER_MASK (1 << PG_USER_BIT) +#define PG_PWT_MASK (1 << PG_PWT_BIT) +#define PG_PCD_MASK (1 << PG_PCD_BIT) +#define PG_ACCESSED_MASK (1 << PG_ACCESSED_BIT) +#define PG_DIRTY_MASK (1 << PG_DIRTY_BIT) +#define PG_PSE_MASK (1 << PG_PSE_BIT) +#define PG_GLOBAL_MASK (1 << PG_GLOBAL_BIT) + +#define PG_ERROR_W_BIT 1 + +#define PG_ERROR_P_MASK 0x01 +#define PG_ERROR_W_MASK (1 << PG_ERROR_W_BIT) +#define PG_ERROR_U_MASK 0x04 +#define PG_ERROR_RSVD_MASK 0x08 + +#define MSR_IA32_APICBASE 0x1b +#define MSR_IA32_APICBASE_BSP (1<<8) +#define MSR_IA32_APICBASE_ENABLE (1<<11) +#define MSR_IA32_APICBASE_BASE (0xfffff<<12) + +#define MSR_IA32_SYSENTER_CS 0x174 +#define MSR_IA32_SYSENTER_ESP 0x175 +#define MSR_IA32_SYSENTER_EIP 0x176 + +#define EXCP00_DIVZ 0 +#define EXCP01_SSTP 1 +#define EXCP02_NMI 2 +#define EXCP03_INT3 3 +#define EXCP04_INTO 4 +#define EXCP05_BOUND 5 +#define EXCP06_ILLOP 6 +#define EXCP07_PREX 7 +#define EXCP08_DBLE 8 +#define EXCP09_XERR 9 +#define EXCP0A_TSS 10 +#define EXCP0B_NOSEG 11 +#define EXCP0C_STACK 12 +#define EXCP0D_GPF 13 +#define EXCP0E_PAGE 14 +#define EXCP10_COPR 16 +#define EXCP11_ALGN 17 +#define EXCP12_MCHK 18 + +enum { + CC_OP_DYNAMIC, /* must use dynamic code to get cc_op */ + CC_OP_EFLAGS, /* all cc are explicitely computed, CC_SRC = flags */ + CC_OP_MUL, /* modify all flags, C, O = (CC_SRC != 0) */ + + CC_OP_ADDB, /* modify all flags, CC_DST = res, CC_SRC = src1 */ + CC_OP_ADDW, + CC_OP_ADDL, + + CC_OP_ADCB, /* modify all flags, CC_DST = res, CC_SRC = src1 */ + CC_OP_ADCW, + CC_OP_ADCL, + + CC_OP_SUBB, /* modify all flags, CC_DST = res, CC_SRC = src1 */ + CC_OP_SUBW, + CC_OP_SUBL, + + CC_OP_SBBB, /* modify all flags, CC_DST = res, CC_SRC = src1 */ + CC_OP_SBBW, + CC_OP_SBBL, + + CC_OP_LOGICB, /* modify all flags, CC_DST = res */ + CC_OP_LOGICW, + CC_OP_LOGICL, + + CC_OP_INCB, /* modify all flags except, CC_DST = res, CC_SRC = C */ + CC_OP_INCW, + CC_OP_INCL, + + CC_OP_DECB, /* modify all flags except, CC_DST = res, CC_SRC = C */ + CC_OP_DECW, + CC_OP_DECL, + + CC_OP_SHLB, /* modify all flags, CC_DST = res, CC_SRC.lsb = C */ + CC_OP_SHLW, + CC_OP_SHLL, + + CC_OP_SARB, /* modify all flags, CC_DST = res, CC_SRC.lsb = C */ + CC_OP_SARW, + CC_OP_SARL, + + CC_OP_NB, +}; + +#ifdef __i386__ +#define USE_X86LDOUBLE +#endif + +#ifdef USE_X86LDOUBLE +typedef long double CPU86_LDouble; +#else +typedef double CPU86_LDouble; +#endif + +typedef struct SegmentCache { + uint32_t selector; + uint8_t *base; + uint32_t limit; + uint32_t flags; +} SegmentCache; + +typedef struct CPUX86State { + /* standard registers */ + uint32_t regs[8]; + uint32_t eip; + uint32_t eflags; /* eflags register. During CPU emulation, CC + flags and DF are set to zero because they are + stored elsewhere */ + + /* emulator internal eflags handling */ + uint32_t cc_src; + uint32_t cc_dst; + uint32_t cc_op; + int32_t df; /* D flag : 1 if D = 0, -1 if D = 1 */ + uint32_t hflags; /* hidden flags, see HF_xxx constants */ + + /* FPU state */ + unsigned int fpstt; /* top of stack index */ + unsigned int fpus; + unsigned int fpuc; + uint8_t fptags[8]; /* 0 = valid, 1 = empty */ + CPU86_LDouble fpregs[8]; + + /* emulator internal variables */ + CPU86_LDouble ft0; + union { + float f; + double d; + int i32; + int64_t i64; + } fp_convert; + + /* segments */ + SegmentCache segs[6]; /* selector values */ + SegmentCache ldt; + SegmentCache tr; + SegmentCache gdt; /* only base and limit are used */ + SegmentCache idt; /* only base and limit are used */ + + /* sysenter registers */ + uint32_t sysenter_cs; + uint32_t sysenter_esp; + uint32_t sysenter_eip; + + /* exception/interrupt handling */ + jmp_buf jmp_env; + int exception_index; + int error_code; + int exception_is_int; + int exception_next_eip; + struct TranslationBlock *current_tb; /* currently executing TB */ + uint32_t cr[5]; /* NOTE: cr1 is unused */ + uint32_t dr[8]; /* debug registers */ + int interrupt_request; + int user_mode_only; /* user mode only simulation */ + + /* soft mmu support */ + /* 0 = kernel, 1 = user */ + CPUTLBEntry tlb_read[2][CPU_TLB_SIZE]; + CPUTLBEntry tlb_write[2][CPU_TLB_SIZE]; + + /* ice debug support */ + uint32_t breakpoints[MAX_BREAKPOINTS]; + int nb_breakpoints; + int singlestep_enabled; + + /* user data */ + void *opaque; +} CPUX86State; + +#ifndef IN_OP_I386 +void cpu_x86_outb(CPUX86State *env, int addr, int val); +void cpu_x86_outw(CPUX86State *env, int addr, int val); +void cpu_x86_outl(CPUX86State *env, int addr, int val); +int cpu_x86_inb(CPUX86State *env, int addr); +int cpu_x86_inw(CPUX86State *env, int addr); +int cpu_x86_inl(CPUX86State *env, int addr); +#endif + +CPUX86State *cpu_x86_init(void); +int cpu_x86_exec(CPUX86State *s); +void cpu_x86_close(CPUX86State *s); +int cpu_x86_get_pic_interrupt(CPUX86State *s); + +/* this function must always be used to load data in the segment + cache: it synchronizes the hflags with the segment cache values */ +static inline void cpu_x86_load_seg_cache(CPUX86State *env, + int seg_reg, unsigned int selector, + uint8_t *base, unsigned int limit, + unsigned int flags) +{ + SegmentCache *sc; + unsigned int new_hflags; + + sc = &env->segs[seg_reg]; + sc->selector = selector; + sc->base = base; + sc->limit = limit; + sc->flags = flags; + + /* update the hidden flags */ + new_hflags = (env->segs[R_CS].flags & DESC_B_MASK) + >> (DESC_B_SHIFT - HF_CS32_SHIFT); + new_hflags |= (env->segs[R_SS].flags & DESC_B_MASK) + >> (DESC_B_SHIFT - HF_SS32_SHIFT); + if (!(env->cr[0] & CR0_PE_MASK) || (env->eflags & VM_MASK)) { + /* XXX: try to avoid this test. The problem comes from the + fact that is real mode or vm86 mode we only modify the + 'base' and 'selector' fields of the segment cache to go + faster. A solution may be to force addseg to one in + translate-i386.c. */ + new_hflags |= HF_ADDSEG_MASK; + } else { + new_hflags |= (((unsigned long)env->segs[R_DS].base | + (unsigned long)env->segs[R_ES].base | + (unsigned long)env->segs[R_SS].base) != 0) << + HF_ADDSEG_SHIFT; + } + env->hflags = (env->hflags & + ~(HF_CS32_MASK | HF_SS32_MASK | HF_ADDSEG_MASK)) | new_hflags; +} + +/* wrapper, just in case memory mappings must be changed */ +static inline void cpu_x86_set_cpl(CPUX86State *s, int cpl) +{ +#if HF_CPL_MASK == 3 + s->hflags = (s->hflags & ~HF_CPL_MASK) | cpl; +#else +#error HF_CPL_MASK is hardcoded +#endif +} + +/* the following helpers are only usable in user mode simulation as + they can trigger unexpected exceptions */ +void cpu_x86_load_seg(CPUX86State *s, int seg_reg, int selector); +void cpu_x86_fsave(CPUX86State *s, uint8_t *ptr, int data32); +void cpu_x86_frstor(CPUX86State *s, uint8_t *ptr, int data32); + +/* you can call this signal handler from your SIGBUS and SIGSEGV + signal handlers to inform the virtual CPU of exceptions. non zero + is returned if the signal was handled by the virtual CPU. */ +struct siginfo; +int cpu_x86_signal_handler(int host_signum, struct siginfo *info, + void *puc); + +/* MMU defines */ +void cpu_x86_init_mmu(CPUX86State *env); +extern int phys_ram_size; +extern int phys_ram_fd; +extern uint8_t *phys_ram_base; + +/* used to debug */ +#define X86_DUMP_FPU 0x0001 /* dump FPU state too */ +#define X86_DUMP_CCOP 0x0002 /* dump qemu flag cache */ +void cpu_x86_dump_state(CPUX86State *env, FILE *f, int flags); + +#define TARGET_PAGE_BITS 12 +#include "cpu-all.h" + +#endif /* CPU_I386_H */ diff --git a/target-i386/exec.h b/target-i386/exec.h new file mode 100644 index 0000000000..b53928c98e --- /dev/null +++ b/target-i386/exec.h @@ -0,0 +1,416 @@ +/* + * i386 execution defines + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#include "dyngen-exec.h" + +/* at least 4 register variables are defines */ +register struct CPUX86State *env asm(AREG0); +register uint32_t T0 asm(AREG1); +register uint32_t T1 asm(AREG2); +register uint32_t T2 asm(AREG3); + +#define A0 T2 + +/* if more registers are available, we define some registers too */ +#ifdef AREG4 +register uint32_t EAX asm(AREG4); +#define reg_EAX +#endif + +#ifdef AREG5 +register uint32_t ESP asm(AREG5); +#define reg_ESP +#endif + +#ifdef AREG6 +register uint32_t EBP asm(AREG6); +#define reg_EBP +#endif + +#ifdef AREG7 +register uint32_t ECX asm(AREG7); +#define reg_ECX +#endif + +#ifdef AREG8 +register uint32_t EDX asm(AREG8); +#define reg_EDX +#endif + +#ifdef AREG9 +register uint32_t EBX asm(AREG9); +#define reg_EBX +#endif + +#ifdef AREG10 +register uint32_t ESI asm(AREG10); +#define reg_ESI +#endif + +#ifdef AREG11 +register uint32_t EDI asm(AREG11); +#define reg_EDI +#endif + +extern FILE *logfile; +extern int loglevel; + +#ifndef reg_EAX +#define EAX (env->regs[R_EAX]) +#endif +#ifndef reg_ECX +#define ECX (env->regs[R_ECX]) +#endif +#ifndef reg_EDX +#define EDX (env->regs[R_EDX]) +#endif +#ifndef reg_EBX +#define EBX (env->regs[R_EBX]) +#endif +#ifndef reg_ESP +#define ESP (env->regs[R_ESP]) +#endif +#ifndef reg_EBP +#define EBP (env->regs[R_EBP]) +#endif +#ifndef reg_ESI +#define ESI (env->regs[R_ESI]) +#endif +#ifndef reg_EDI +#define EDI (env->regs[R_EDI]) +#endif +#define EIP (env->eip) +#define DF (env->df) + +#define CC_SRC (env->cc_src) +#define CC_DST (env->cc_dst) +#define CC_OP (env->cc_op) + +/* float macros */ +#define FT0 (env->ft0) +#define ST0 (env->fpregs[env->fpstt]) +#define ST(n) (env->fpregs[(env->fpstt + (n)) & 7]) +#define ST1 ST(1) + +#ifdef USE_FP_CONVERT +#define FP_CONVERT (env->fp_convert) +#endif + +#include "cpu.h" +#include "exec-all.h" + +typedef struct CCTable { + int (*compute_all)(void); /* return all the flags */ + int (*compute_c)(void); /* return the C flag */ +} CCTable; + +extern CCTable cc_table[]; + +void load_seg(int seg_reg, int selector, unsigned cur_eip); +void helper_ljmp_protected_T0_T1(void); +void helper_lcall_real_T0_T1(int shift, int next_eip); +void helper_lcall_protected_T0_T1(int shift, int next_eip); +void helper_iret_real(int shift); +void helper_iret_protected(int shift); +void helper_lret_protected(int shift, int addend); +void helper_lldt_T0(void); +void helper_ltr_T0(void); +void helper_movl_crN_T0(int reg); +void helper_movl_drN_T0(int reg); +void helper_invlpg(unsigned int addr); +void cpu_x86_update_cr0(CPUX86State *env); +void cpu_x86_update_cr3(CPUX86State *env); +void cpu_x86_flush_tlb(CPUX86State *env, uint32_t addr); +int cpu_x86_handle_mmu_fault(CPUX86State *env, uint32_t addr, int is_write); +void tlb_fill(unsigned long addr, int is_write, void *retaddr); +void __hidden cpu_lock(void); +void __hidden cpu_unlock(void); +void do_interrupt(int intno, int is_int, int error_code, + unsigned int next_eip, int is_hw); +void do_interrupt_user(int intno, int is_int, int error_code, + unsigned int next_eip); +void raise_interrupt(int intno, int is_int, int error_code, + unsigned int next_eip); +void raise_exception_err(int exception_index, int error_code); +void raise_exception(int exception_index); +void __hidden cpu_loop_exit(void); +void helper_fsave(uint8_t *ptr, int data32); +void helper_frstor(uint8_t *ptr, int data32); + +void OPPROTO op_movl_eflags_T0(void); +void OPPROTO op_movl_T0_eflags(void); +void raise_interrupt(int intno, int is_int, int error_code, + unsigned int next_eip); +void raise_exception_err(int exception_index, int error_code); +void raise_exception(int exception_index); +void helper_divl_EAX_T0(uint32_t eip); +void helper_idivl_EAX_T0(uint32_t eip); +void helper_cmpxchg8b(void); +void helper_cpuid(void); +void helper_rdtsc(void); +void helper_rdmsr(void); +void helper_wrmsr(void); +void helper_lsl(void); +void helper_lar(void); + +#ifdef USE_X86LDOUBLE +/* use long double functions */ +#define lrint lrintl +#define llrint llrintl +#define fabs fabsl +#define sin sinl +#define cos cosl +#define sqrt sqrtl +#define pow powl +#define log logl +#define tan tanl +#define atan2 atan2l +#define floor floorl +#define ceil ceill +#define rint rintl +#endif + +extern int lrint(CPU86_LDouble x); +extern int64_t llrint(CPU86_LDouble x); +extern CPU86_LDouble fabs(CPU86_LDouble x); +extern CPU86_LDouble sin(CPU86_LDouble x); +extern CPU86_LDouble cos(CPU86_LDouble x); +extern CPU86_LDouble sqrt(CPU86_LDouble x); +extern CPU86_LDouble pow(CPU86_LDouble, CPU86_LDouble); +extern CPU86_LDouble log(CPU86_LDouble x); +extern CPU86_LDouble tan(CPU86_LDouble x); +extern CPU86_LDouble atan2(CPU86_LDouble, CPU86_LDouble); +extern CPU86_LDouble floor(CPU86_LDouble x); +extern CPU86_LDouble ceil(CPU86_LDouble x); +extern CPU86_LDouble rint(CPU86_LDouble x); + +#define RC_MASK 0xc00 +#define RC_NEAR 0x000 +#define RC_DOWN 0x400 +#define RC_UP 0x800 +#define RC_CHOP 0xc00 + +#define MAXTAN 9223372036854775808.0 + +#ifdef __arm__ +/* we have no way to do correct rounding - a FPU emulator is needed */ +#define FE_DOWNWARD FE_TONEAREST +#define FE_UPWARD FE_TONEAREST +#define FE_TOWARDZERO FE_TONEAREST +#endif + +#ifdef USE_X86LDOUBLE + +/* only for x86 */ +typedef union { + long double d; + struct { + unsigned long long lower; + unsigned short upper; + } l; +} CPU86_LDoubleU; + +/* the following deal with x86 long double-precision numbers */ +#define MAXEXPD 0x7fff +#define EXPBIAS 16383 +#define EXPD(fp) (fp.l.upper & 0x7fff) +#define SIGND(fp) ((fp.l.upper) & 0x8000) +#define MANTD(fp) (fp.l.lower) +#define BIASEXPONENT(fp) fp.l.upper = (fp.l.upper & ~(0x7fff)) | EXPBIAS + +#else + +/* NOTE: arm is horrible as double 32 bit words are stored in big endian ! */ +typedef union { + double d; +#if !defined(WORDS_BIGENDIAN) && !defined(__arm__) + struct { + uint32_t lower; + int32_t upper; + } l; +#else + struct { + int32_t upper; + uint32_t lower; + } l; +#endif +#ifndef __arm__ + int64_t ll; +#endif +} CPU86_LDoubleU; + +/* the following deal with IEEE double-precision numbers */ +#define MAXEXPD 0x7ff +#define EXPBIAS 1023 +#define EXPD(fp) (((fp.l.upper) >> 20) & 0x7FF) +#define SIGND(fp) ((fp.l.upper) & 0x80000000) +#ifdef __arm__ +#define MANTD(fp) (fp.l.lower | ((uint64_t)(fp.l.upper & ((1 << 20) - 1)) << 32)) +#else +#define MANTD(fp) (fp.ll & ((1LL << 52) - 1)) +#endif +#define BIASEXPONENT(fp) fp.l.upper = (fp.l.upper & ~(0x7ff << 20)) | (EXPBIAS << 20) +#endif + +static inline void fpush(void) +{ + env->fpstt = (env->fpstt - 1) & 7; + env->fptags[env->fpstt] = 0; /* validate stack entry */ +} + +static inline void fpop(void) +{ + env->fptags[env->fpstt] = 1; /* invvalidate stack entry */ + env->fpstt = (env->fpstt + 1) & 7; +} + +#ifndef USE_X86LDOUBLE +static inline CPU86_LDouble helper_fldt(uint8_t *ptr) +{ + CPU86_LDoubleU temp; + int upper, e; + uint64_t ll; + + /* mantissa */ + upper = lduw(ptr + 8); + /* XXX: handle overflow ? */ + e = (upper & 0x7fff) - 16383 + EXPBIAS; /* exponent */ + e |= (upper >> 4) & 0x800; /* sign */ + ll = (ldq(ptr) >> 11) & ((1LL << 52) - 1); +#ifdef __arm__ + temp.l.upper = (e << 20) | (ll >> 32); + temp.l.lower = ll; +#else + temp.ll = ll | ((uint64_t)e << 52); +#endif + return temp.d; +} + +static inline void helper_fstt(CPU86_LDouble f, uint8_t *ptr) +{ + CPU86_LDoubleU temp; + int e; + + temp.d = f; + /* mantissa */ + stq(ptr, (MANTD(temp) << 11) | (1LL << 63)); + /* exponent + sign */ + e = EXPD(temp) - EXPBIAS + 16383; + e |= SIGND(temp) >> 16; + stw(ptr + 8, e); +} +#endif + +const CPU86_LDouble f15rk[7]; + +void helper_fldt_ST0_A0(void); +void helper_fstt_ST0_A0(void); +void helper_fbld_ST0_A0(void); +void helper_fbst_ST0_A0(void); +void helper_f2xm1(void); +void helper_fyl2x(void); +void helper_fptan(void); +void helper_fpatan(void); +void helper_fxtract(void); +void helper_fprem1(void); +void helper_fprem(void); +void helper_fyl2xp1(void); +void helper_fsqrt(void); +void helper_fsincos(void); +void helper_frndint(void); +void helper_fscale(void); +void helper_fsin(void); +void helper_fcos(void); +void helper_fxam_ST0(void); +void helper_fstenv(uint8_t *ptr, int data32); +void helper_fldenv(uint8_t *ptr, int data32); +void helper_fsave(uint8_t *ptr, int data32); +void helper_frstor(uint8_t *ptr, int data32); + +const uint8_t parity_table[256]; +const uint8_t rclw_table[32]; +const uint8_t rclb_table[32]; + +static inline uint32_t compute_eflags(void) +{ + return env->eflags | cc_table[CC_OP].compute_all() | (DF & DF_MASK); +} + +#define FL_UPDATE_MASK32 (TF_MASK | AC_MASK | ID_MASK) + +#define FL_UPDATE_CPL0_MASK (TF_MASK | IF_MASK | IOPL_MASK | NT_MASK | \ + RF_MASK | AC_MASK | ID_MASK) + +/* NOTE: CC_OP must be modified manually to CC_OP_EFLAGS */ +static inline void load_eflags(int eflags, int update_mask) +{ + CC_SRC = eflags & (CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C); + DF = 1 - (2 * ((eflags >> 10) & 1)); + env->eflags = (env->eflags & ~update_mask) | + (eflags & update_mask); +} + +/* memory access macros */ + +#define ldul ldl +#define lduq ldq +#define ldul_user ldl_user +#define ldul_kernel ldl_kernel + +#define ldub_raw ldub +#define ldsb_raw ldsb +#define lduw_raw lduw +#define ldsw_raw ldsw +#define ldl_raw ldl +#define ldq_raw ldq + +#define stb_raw stb +#define stw_raw stw +#define stl_raw stl +#define stq_raw stq + +#define MEMUSER 0 +#define DATA_SIZE 1 +#include "softmmu_header.h" + +#define DATA_SIZE 2 +#include "softmmu_header.h" + +#define DATA_SIZE 4 +#include "softmmu_header.h" + +#define DATA_SIZE 8 +#include "softmmu_header.h" + +#undef MEMUSER +#define MEMUSER 1 +#define DATA_SIZE 1 +#include "softmmu_header.h" + +#define DATA_SIZE 2 +#include "softmmu_header.h" + +#define DATA_SIZE 4 +#include "softmmu_header.h" + +#define DATA_SIZE 8 +#include "softmmu_header.h" + +#undef MEMUSER + diff --git a/target-i386/helper.c b/target-i386/helper.c new file mode 100644 index 0000000000..075a99f255 --- /dev/null +++ b/target-i386/helper.c @@ -0,0 +1,1803 @@ +/* + * i386 helpers + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#include "exec.h" + +const uint8_t parity_table[256] = { + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + CC_P, 0, 0, CC_P, 0, CC_P, CC_P, 0, + 0, CC_P, CC_P, 0, CC_P, 0, 0, CC_P, +}; + +/* modulo 17 table */ +const uint8_t rclw_table[32] = { + 0, 1, 2, 3, 4, 5, 6, 7, + 8, 9,10,11,12,13,14,15, + 16, 0, 1, 2, 3, 4, 5, 6, + 7, 8, 9,10,11,12,13,14, +}; + +/* modulo 9 table */ +const uint8_t rclb_table[32] = { + 0, 1, 2, 3, 4, 5, 6, 7, + 8, 0, 1, 2, 3, 4, 5, 6, + 7, 8, 0, 1, 2, 3, 4, 5, + 6, 7, 8, 0, 1, 2, 3, 4, +}; + +const CPU86_LDouble f15rk[7] = +{ + 0.00000000000000000000L, + 1.00000000000000000000L, + 3.14159265358979323851L, /*pi*/ + 0.30102999566398119523L, /*lg2*/ + 0.69314718055994530943L, /*ln2*/ + 1.44269504088896340739L, /*l2e*/ + 3.32192809488736234781L, /*l2t*/ +}; + +/* thread support */ + +spinlock_t global_cpu_lock = SPIN_LOCK_UNLOCKED; + +void cpu_lock(void) +{ + spin_lock(&global_cpu_lock); +} + +void cpu_unlock(void) +{ + spin_unlock(&global_cpu_lock); +} + +void cpu_loop_exit(void) +{ + /* NOTE: the register at this point must be saved by hand because + longjmp restore them */ +#ifdef reg_EAX + env->regs[R_EAX] = EAX; +#endif +#ifdef reg_ECX + env->regs[R_ECX] = ECX; +#endif +#ifdef reg_EDX + env->regs[R_EDX] = EDX; +#endif +#ifdef reg_EBX + env->regs[R_EBX] = EBX; +#endif +#ifdef reg_ESP + env->regs[R_ESP] = ESP; +#endif +#ifdef reg_EBP + env->regs[R_EBP] = EBP; +#endif +#ifdef reg_ESI + env->regs[R_ESI] = ESI; +#endif +#ifdef reg_EDI + env->regs[R_EDI] = EDI; +#endif + longjmp(env->jmp_env, 1); +} + +static inline void get_ss_esp_from_tss(uint32_t *ss_ptr, + uint32_t *esp_ptr, int dpl) +{ + int type, index, shift; + +#if 0 + { + int i; + printf("TR: base=%p limit=%x\n", env->tr.base, env->tr.limit); + for(i=0;i<env->tr.limit;i++) { + printf("%02x ", env->tr.base[i]); + if ((i & 7) == 7) printf("\n"); + } + printf("\n"); + } +#endif + + if (!(env->tr.flags & DESC_P_MASK)) + cpu_abort(env, "invalid tss"); + type = (env->tr.flags >> DESC_TYPE_SHIFT) & 0xf; + if ((type & 7) != 1) + cpu_abort(env, "invalid tss type"); + shift = type >> 3; + index = (dpl * 4 + 2) << shift; + if (index + (4 << shift) - 1 > env->tr.limit) + raise_exception_err(EXCP0A_TSS, env->tr.selector & 0xfffc); + if (shift == 0) { + *esp_ptr = lduw(env->tr.base + index); + *ss_ptr = lduw(env->tr.base + index + 2); + } else { + *esp_ptr = ldl(env->tr.base + index); + *ss_ptr = lduw(env->tr.base + index + 4); + } +} + +/* return non zero if error */ +static inline int load_segment(uint32_t *e1_ptr, uint32_t *e2_ptr, + int selector) +{ + SegmentCache *dt; + int index; + uint8_t *ptr; + + if (selector & 0x4) + dt = &env->ldt; + else + dt = &env->gdt; + index = selector & ~7; + if ((index + 7) > dt->limit) + return -1; + ptr = dt->base + index; + *e1_ptr = ldl(ptr); + *e2_ptr = ldl(ptr + 4); + return 0; +} + +static inline unsigned int get_seg_limit(uint32_t e1, uint32_t e2) +{ + unsigned int limit; + limit = (e1 & 0xffff) | (e2 & 0x000f0000); + if (e2 & DESC_G_MASK) + limit = (limit << 12) | 0xfff; + return limit; +} + +static inline uint8_t *get_seg_base(uint32_t e1, uint32_t e2) +{ + return (uint8_t *)((e1 >> 16) | ((e2 & 0xff) << 16) | (e2 & 0xff000000)); +} + +static inline void load_seg_cache_raw_dt(SegmentCache *sc, uint32_t e1, uint32_t e2) +{ + sc->base = get_seg_base(e1, e2); + sc->limit = get_seg_limit(e1, e2); + sc->flags = e2; +} + +/* init the segment cache in vm86 mode. */ +static inline void load_seg_vm(int seg, int selector) +{ + selector &= 0xffff; + cpu_x86_load_seg_cache(env, seg, selector, + (uint8_t *)(selector << 4), 0xffff, 0); +} + +/* protected mode interrupt */ +static void do_interrupt_protected(int intno, int is_int, int error_code, + unsigned int next_eip, int is_hw) +{ + SegmentCache *dt; + uint8_t *ptr, *ssp; + int type, dpl, selector, ss_dpl, cpl; + int has_error_code, new_stack, shift; + uint32_t e1, e2, offset, ss, esp, ss_e1, ss_e2, push_size; + uint32_t old_cs, old_ss, old_esp, old_eip; + + dt = &env->idt; + if (intno * 8 + 7 > dt->limit) + raise_exception_err(EXCP0D_GPF, intno * 8 + 2); + ptr = dt->base + intno * 8; + e1 = ldl(ptr); + e2 = ldl(ptr + 4); + /* check gate type */ + type = (e2 >> DESC_TYPE_SHIFT) & 0x1f; + switch(type) { + case 5: /* task gate */ + cpu_abort(env, "task gate not supported"); + break; + case 6: /* 286 interrupt gate */ + case 7: /* 286 trap gate */ + case 14: /* 386 interrupt gate */ + case 15: /* 386 trap gate */ + break; + default: + raise_exception_err(EXCP0D_GPF, intno * 8 + 2); + break; + } + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + cpl = env->hflags & HF_CPL_MASK; + /* check privledge if software int */ + if (is_int && dpl < cpl) + raise_exception_err(EXCP0D_GPF, intno * 8 + 2); + /* check valid bit */ + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, intno * 8 + 2); + selector = e1 >> 16; + offset = (e2 & 0xffff0000) | (e1 & 0x0000ffff); + if ((selector & 0xfffc) == 0) + raise_exception_err(EXCP0D_GPF, 0); + + if (load_segment(&e1, &e2, selector) != 0) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + if (!(e2 & DESC_S_MASK) || !(e2 & (DESC_CS_MASK))) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + if (dpl > cpl) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, selector & 0xfffc); + if (!(e2 & DESC_C_MASK) && dpl < cpl) { + /* to inner priviledge */ + get_ss_esp_from_tss(&ss, &esp, dpl); + if ((ss & 0xfffc) == 0) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if ((ss & 3) != dpl) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if (load_segment(&ss_e1, &ss_e2, ss) != 0) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + ss_dpl = (ss_e2 >> DESC_DPL_SHIFT) & 3; + if (ss_dpl != dpl) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if (!(ss_e2 & DESC_S_MASK) || + (ss_e2 & DESC_CS_MASK) || + !(ss_e2 & DESC_W_MASK)) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if (!(ss_e2 & DESC_P_MASK)) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + new_stack = 1; + } else if ((e2 & DESC_C_MASK) || dpl == cpl) { + /* to same priviledge */ + new_stack = 0; + } else { + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + new_stack = 0; /* avoid warning */ + } + + shift = type >> 3; + has_error_code = 0; + if (!is_int && !is_hw) { + switch(intno) { + case 8: + case 10: + case 11: + case 12: + case 13: + case 14: + case 17: + has_error_code = 1; + break; + } + } + push_size = 6 + (new_stack << 2) + (has_error_code << 1); + if (env->eflags & VM_MASK) + push_size += 8; + push_size <<= shift; + + /* XXX: check that enough room is available */ + if (new_stack) { + old_esp = ESP; + old_ss = env->segs[R_SS].selector; + ss = (ss & ~3) | dpl; + cpu_x86_load_seg_cache(env, R_SS, ss, + get_seg_base(ss_e1, ss_e2), + get_seg_limit(ss_e1, ss_e2), + ss_e2); + } else { + old_esp = 0; + old_ss = 0; + esp = ESP; + } + if (is_int) + old_eip = next_eip; + else + old_eip = env->eip; + old_cs = env->segs[R_CS].selector; + selector = (selector & ~3) | dpl; + cpu_x86_load_seg_cache(env, R_CS, selector, + get_seg_base(e1, e2), + get_seg_limit(e1, e2), + e2); + cpu_x86_set_cpl(env, dpl); + env->eip = offset; + ESP = esp - push_size; + ssp = env->segs[R_SS].base + esp; + if (shift == 1) { + int old_eflags; + if (env->eflags & VM_MASK) { + ssp -= 4; + stl(ssp, env->segs[R_GS].selector); + ssp -= 4; + stl(ssp, env->segs[R_FS].selector); + ssp -= 4; + stl(ssp, env->segs[R_DS].selector); + ssp -= 4; + stl(ssp, env->segs[R_ES].selector); + } + if (new_stack) { + ssp -= 4; + stl(ssp, old_ss); + ssp -= 4; + stl(ssp, old_esp); + } + ssp -= 4; + old_eflags = compute_eflags(); + stl(ssp, old_eflags); + ssp -= 4; + stl(ssp, old_cs); + ssp -= 4; + stl(ssp, old_eip); + if (has_error_code) { + ssp -= 4; + stl(ssp, error_code); + } + } else { + if (new_stack) { + ssp -= 2; + stw(ssp, old_ss); + ssp -= 2; + stw(ssp, old_esp); + } + ssp -= 2; + stw(ssp, compute_eflags()); + ssp -= 2; + stw(ssp, old_cs); + ssp -= 2; + stw(ssp, old_eip); + if (has_error_code) { + ssp -= 2; + stw(ssp, error_code); + } + } + + /* interrupt gate clear IF mask */ + if ((type & 1) == 0) { + env->eflags &= ~IF_MASK; + } + env->eflags &= ~(TF_MASK | VM_MASK | RF_MASK | NT_MASK); +} + +/* real mode interrupt */ +static void do_interrupt_real(int intno, int is_int, int error_code, + unsigned int next_eip) +{ + SegmentCache *dt; + uint8_t *ptr, *ssp; + int selector; + uint32_t offset, esp; + uint32_t old_cs, old_eip; + + /* real mode (simpler !) */ + dt = &env->idt; + if (intno * 4 + 3 > dt->limit) + raise_exception_err(EXCP0D_GPF, intno * 8 + 2); + ptr = dt->base + intno * 4; + offset = lduw(ptr); + selector = lduw(ptr + 2); + esp = ESP; + ssp = env->segs[R_SS].base; + if (is_int) + old_eip = next_eip; + else + old_eip = env->eip; + old_cs = env->segs[R_CS].selector; + esp -= 2; + stw(ssp + (esp & 0xffff), compute_eflags()); + esp -= 2; + stw(ssp + (esp & 0xffff), old_cs); + esp -= 2; + stw(ssp + (esp & 0xffff), old_eip); + + /* update processor state */ + ESP = (ESP & ~0xffff) | (esp & 0xffff); + env->eip = offset; + env->segs[R_CS].selector = selector; + env->segs[R_CS].base = (uint8_t *)(selector << 4); + env->eflags &= ~(IF_MASK | TF_MASK | AC_MASK | RF_MASK); +} + +/* fake user mode interrupt */ +void do_interrupt_user(int intno, int is_int, int error_code, + unsigned int next_eip) +{ + SegmentCache *dt; + uint8_t *ptr; + int dpl, cpl; + uint32_t e2; + + dt = &env->idt; + ptr = dt->base + (intno * 8); + e2 = ldl(ptr + 4); + + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + cpl = env->hflags & HF_CPL_MASK; + /* check privledge if software int */ + if (is_int && dpl < cpl) + raise_exception_err(EXCP0D_GPF, intno * 8 + 2); + + /* Since we emulate only user space, we cannot do more than + exiting the emulation with the suitable exception and error + code */ + if (is_int) + EIP = next_eip; +} + +/* + * Begin excution of an interruption. is_int is TRUE if coming from + * the int instruction. next_eip is the EIP value AFTER the interrupt + * instruction. It is only relevant if is_int is TRUE. + */ +void do_interrupt(int intno, int is_int, int error_code, + unsigned int next_eip, int is_hw) +{ + if (env->cr[0] & CR0_PE_MASK) { + do_interrupt_protected(intno, is_int, error_code, next_eip, is_hw); + } else { + do_interrupt_real(intno, is_int, error_code, next_eip); + } +} + +/* + * Signal an interruption. It is executed in the main CPU loop. + * is_int is TRUE if coming from the int instruction. next_eip is the + * EIP value AFTER the interrupt instruction. It is only relevant if + * is_int is TRUE. + */ +void raise_interrupt(int intno, int is_int, int error_code, + unsigned int next_eip) +{ + env->exception_index = intno; + env->error_code = error_code; + env->exception_is_int = is_int; + env->exception_next_eip = next_eip; + cpu_loop_exit(); +} + +/* shortcuts to generate exceptions */ +void raise_exception_err(int exception_index, int error_code) +{ + raise_interrupt(exception_index, 0, error_code, 0); +} + +void raise_exception(int exception_index) +{ + raise_interrupt(exception_index, 0, 0, 0); +} + +#ifdef BUGGY_GCC_DIV64 +/* gcc 2.95.4 on PowerPC does not seem to like using __udivdi3, so we + call it from another function */ +uint32_t div64(uint32_t *q_ptr, uint64_t num, uint32_t den) +{ + *q_ptr = num / den; + return num % den; +} + +int32_t idiv64(int32_t *q_ptr, int64_t num, int32_t den) +{ + *q_ptr = num / den; + return num % den; +} +#endif + +void helper_divl_EAX_T0(uint32_t eip) +{ + unsigned int den, q, r; + uint64_t num; + + num = EAX | ((uint64_t)EDX << 32); + den = T0; + if (den == 0) { + EIP = eip; + raise_exception(EXCP00_DIVZ); + } +#ifdef BUGGY_GCC_DIV64 + r = div64(&q, num, den); +#else + q = (num / den); + r = (num % den); +#endif + EAX = q; + EDX = r; +} + +void helper_idivl_EAX_T0(uint32_t eip) +{ + int den, q, r; + int64_t num; + + num = EAX | ((uint64_t)EDX << 32); + den = T0; + if (den == 0) { + EIP = eip; + raise_exception(EXCP00_DIVZ); + } +#ifdef BUGGY_GCC_DIV64 + r = idiv64(&q, num, den); +#else + q = (num / den); + r = (num % den); +#endif + EAX = q; + EDX = r; +} + +void helper_cmpxchg8b(void) +{ + uint64_t d; + int eflags; + + eflags = cc_table[CC_OP].compute_all(); + d = ldq((uint8_t *)A0); + if (d == (((uint64_t)EDX << 32) | EAX)) { + stq((uint8_t *)A0, ((uint64_t)ECX << 32) | EBX); + eflags |= CC_Z; + } else { + EDX = d >> 32; + EAX = d; + eflags &= ~CC_Z; + } + CC_SRC = eflags; +} + +/* We simulate a pre-MMX pentium as in valgrind */ +#define CPUID_FP87 (1 << 0) +#define CPUID_VME (1 << 1) +#define CPUID_DE (1 << 2) +#define CPUID_PSE (1 << 3) +#define CPUID_TSC (1 << 4) +#define CPUID_MSR (1 << 5) +#define CPUID_PAE (1 << 6) +#define CPUID_MCE (1 << 7) +#define CPUID_CX8 (1 << 8) +#define CPUID_APIC (1 << 9) +#define CPUID_SEP (1 << 11) /* sysenter/sysexit */ +#define CPUID_MTRR (1 << 12) +#define CPUID_PGE (1 << 13) +#define CPUID_MCA (1 << 14) +#define CPUID_CMOV (1 << 15) +/* ... */ +#define CPUID_MMX (1 << 23) +#define CPUID_FXSR (1 << 24) +#define CPUID_SSE (1 << 25) +#define CPUID_SSE2 (1 << 26) + +void helper_cpuid(void) +{ + if (EAX == 0) { + EAX = 1; /* max EAX index supported */ + EBX = 0x756e6547; + ECX = 0x6c65746e; + EDX = 0x49656e69; + } else if (EAX == 1) { + int family, model, stepping; + /* EAX = 1 info */ +#if 0 + /* pentium 75-200 */ + family = 5; + model = 2; + stepping = 11; +#else + /* pentium pro */ + family = 6; + model = 1; + stepping = 3; +#endif + EAX = (family << 8) | (model << 4) | stepping; + EBX = 0; + ECX = 0; + EDX = CPUID_FP87 | CPUID_DE | CPUID_PSE | + CPUID_TSC | CPUID_MSR | CPUID_MCE | + CPUID_CX8 | CPUID_PGE | CPUID_CMOV; + } +} + +void helper_lldt_T0(void) +{ + int selector; + SegmentCache *dt; + uint32_t e1, e2; + int index; + uint8_t *ptr; + + selector = T0 & 0xffff; + if ((selector & 0xfffc) == 0) { + /* XXX: NULL selector case: invalid LDT */ + env->ldt.base = NULL; + env->ldt.limit = 0; + } else { + if (selector & 0x4) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + dt = &env->gdt; + index = selector & ~7; + if ((index + 7) > dt->limit) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + ptr = dt->base + index; + e1 = ldl(ptr); + e2 = ldl(ptr + 4); + if ((e2 & DESC_S_MASK) || ((e2 >> DESC_TYPE_SHIFT) & 0xf) != 2) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, selector & 0xfffc); + load_seg_cache_raw_dt(&env->ldt, e1, e2); + } + env->ldt.selector = selector; +} + +void helper_ltr_T0(void) +{ + int selector; + SegmentCache *dt; + uint32_t e1, e2; + int index, type; + uint8_t *ptr; + + selector = T0 & 0xffff; + if ((selector & 0xfffc) == 0) { + /* NULL selector case: invalid LDT */ + env->tr.base = NULL; + env->tr.limit = 0; + env->tr.flags = 0; + } else { + if (selector & 0x4) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + dt = &env->gdt; + index = selector & ~7; + if ((index + 7) > dt->limit) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + ptr = dt->base + index; + e1 = ldl(ptr); + e2 = ldl(ptr + 4); + type = (e2 >> DESC_TYPE_SHIFT) & 0xf; + if ((e2 & DESC_S_MASK) || + (type != 2 && type != 9)) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, selector & 0xfffc); + load_seg_cache_raw_dt(&env->tr, e1, e2); + e2 |= 0x00000200; /* set the busy bit */ + stl(ptr + 4, e2); + } + env->tr.selector = selector; +} + +/* only works if protected mode and not VM86. Calling load_seg with + seg_reg == R_CS is discouraged */ +void load_seg(int seg_reg, int selector, unsigned int cur_eip) +{ + uint32_t e1, e2; + + if ((selector & 0xfffc) == 0) { + /* null selector case */ + if (seg_reg == R_SS) { + EIP = cur_eip; + raise_exception_err(EXCP0D_GPF, 0); + } else { + cpu_x86_load_seg_cache(env, seg_reg, selector, NULL, 0, 0); + } + } else { + if (load_segment(&e1, &e2, selector) != 0) { + EIP = cur_eip; + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + } + if (!(e2 & DESC_S_MASK) || + (e2 & (DESC_CS_MASK | DESC_R_MASK)) == DESC_CS_MASK) { + EIP = cur_eip; + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + } + + if (seg_reg == R_SS) { + if ((e2 & (DESC_CS_MASK | DESC_W_MASK)) == 0) { + EIP = cur_eip; + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + } + } else { + if ((e2 & (DESC_CS_MASK | DESC_R_MASK)) == DESC_CS_MASK) { + EIP = cur_eip; + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + } + } + + if (!(e2 & DESC_P_MASK)) { + EIP = cur_eip; + if (seg_reg == R_SS) + raise_exception_err(EXCP0C_STACK, selector & 0xfffc); + else + raise_exception_err(EXCP0B_NOSEG, selector & 0xfffc); + } + cpu_x86_load_seg_cache(env, seg_reg, selector, + get_seg_base(e1, e2), + get_seg_limit(e1, e2), + e2); +#if 0 + fprintf(logfile, "load_seg: sel=0x%04x base=0x%08lx limit=0x%08lx flags=%08x\n", + selector, (unsigned long)sc->base, sc->limit, sc->flags); +#endif + } +} + +/* protected mode jump */ +void helper_ljmp_protected_T0_T1(void) +{ + int new_cs, new_eip; + uint32_t e1, e2, cpl, dpl, rpl, limit; + + new_cs = T0; + new_eip = T1; + if ((new_cs & 0xfffc) == 0) + raise_exception_err(EXCP0D_GPF, 0); + if (load_segment(&e1, &e2, new_cs) != 0) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + cpl = env->hflags & HF_CPL_MASK; + if (e2 & DESC_S_MASK) { + if (!(e2 & DESC_CS_MASK)) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + if (e2 & DESC_CS_MASK) { + /* conforming code segment */ + if (dpl > cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + } else { + /* non conforming code segment */ + rpl = new_cs & 3; + if (rpl > cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + if (dpl != cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + } + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, new_cs & 0xfffc); + limit = get_seg_limit(e1, e2); + if (new_eip > limit) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + cpu_x86_load_seg_cache(env, R_CS, (new_cs & 0xfffc) | cpl, + get_seg_base(e1, e2), limit, e2); + EIP = new_eip; + } else { + cpu_abort(env, "jmp to call/task gate not supported 0x%04x:0x%08x", + new_cs, new_eip); + } +} + +/* real mode call */ +void helper_lcall_real_T0_T1(int shift, int next_eip) +{ + int new_cs, new_eip; + uint32_t esp, esp_mask; + uint8_t *ssp; + + new_cs = T0; + new_eip = T1; + esp = ESP; + esp_mask = 0xffffffff; + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + esp_mask = 0xffff; + ssp = env->segs[R_SS].base; + if (shift) { + esp -= 4; + stl(ssp + (esp & esp_mask), env->segs[R_CS].selector); + esp -= 4; + stl(ssp + (esp & esp_mask), next_eip); + } else { + esp -= 2; + stw(ssp + (esp & esp_mask), env->segs[R_CS].selector); + esp -= 2; + stw(ssp + (esp & esp_mask), next_eip); + } + + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + ESP = (ESP & ~0xffff) | (esp & 0xffff); + else + ESP = esp; + env->eip = new_eip; + env->segs[R_CS].selector = new_cs; + env->segs[R_CS].base = (uint8_t *)(new_cs << 4); +} + +/* protected mode call */ +void helper_lcall_protected_T0_T1(int shift, int next_eip) +{ + int new_cs, new_eip; + uint32_t e1, e2, cpl, dpl, rpl, selector, offset, param_count; + uint32_t ss, ss_e1, ss_e2, push_size, sp, type, ss_dpl; + uint32_t old_ss, old_esp, val, i, limit; + uint8_t *ssp, *old_ssp; + + new_cs = T0; + new_eip = T1; + if ((new_cs & 0xfffc) == 0) + raise_exception_err(EXCP0D_GPF, 0); + if (load_segment(&e1, &e2, new_cs) != 0) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + cpl = env->hflags & HF_CPL_MASK; + if (e2 & DESC_S_MASK) { + if (!(e2 & DESC_CS_MASK)) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + if (e2 & DESC_CS_MASK) { + /* conforming code segment */ + if (dpl > cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + } else { + /* non conforming code segment */ + rpl = new_cs & 3; + if (rpl > cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + if (dpl != cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + } + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, new_cs & 0xfffc); + + sp = ESP; + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + sp &= 0xffff; + ssp = env->segs[R_SS].base + sp; + if (shift) { + ssp -= 4; + stl(ssp, env->segs[R_CS].selector); + ssp -= 4; + stl(ssp, next_eip); + } else { + ssp -= 2; + stw(ssp, env->segs[R_CS].selector); + ssp -= 2; + stw(ssp, next_eip); + } + sp -= (4 << shift); + + limit = get_seg_limit(e1, e2); + if (new_eip > limit) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + /* from this point, not restartable */ + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + ESP = (ESP & 0xffff0000) | (sp & 0xffff); + else + ESP = sp; + cpu_x86_load_seg_cache(env, R_CS, (new_cs & 0xfffc) | cpl, + get_seg_base(e1, e2), limit, e2); + EIP = new_eip; + } else { + /* check gate type */ + type = (e2 >> DESC_TYPE_SHIFT) & 0x1f; + switch(type) { + case 1: /* available 286 TSS */ + case 9: /* available 386 TSS */ + case 5: /* task gate */ + cpu_abort(env, "task gate not supported"); + break; + case 4: /* 286 call gate */ + case 12: /* 386 call gate */ + break; + default: + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + break; + } + shift = type >> 3; + + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + rpl = new_cs & 3; + if (dpl < cpl || dpl < rpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + /* check valid bit */ + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, new_cs & 0xfffc); + selector = e1 >> 16; + offset = (e2 & 0xffff0000) | (e1 & 0x0000ffff); + if ((selector & 0xfffc) == 0) + raise_exception_err(EXCP0D_GPF, 0); + + if (load_segment(&e1, &e2, selector) != 0) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + if (!(e2 & DESC_S_MASK) || !(e2 & (DESC_CS_MASK))) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + if (dpl > cpl) + raise_exception_err(EXCP0D_GPF, selector & 0xfffc); + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, selector & 0xfffc); + + if (!(e2 & DESC_C_MASK) && dpl < cpl) { + /* to inner priviledge */ + get_ss_esp_from_tss(&ss, &sp, dpl); + if ((ss & 0xfffc) == 0) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if ((ss & 3) != dpl) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if (load_segment(&ss_e1, &ss_e2, ss) != 0) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + ss_dpl = (ss_e2 >> DESC_DPL_SHIFT) & 3; + if (ss_dpl != dpl) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if (!(ss_e2 & DESC_S_MASK) || + (ss_e2 & DESC_CS_MASK) || + !(ss_e2 & DESC_W_MASK)) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + if (!(ss_e2 & DESC_P_MASK)) + raise_exception_err(EXCP0A_TSS, ss & 0xfffc); + + param_count = e2 & 0x1f; + push_size = ((param_count * 2) + 8) << shift; + + old_esp = ESP; + old_ss = env->segs[R_SS].selector; + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + old_esp &= 0xffff; + old_ssp = env->segs[R_SS].base + old_esp; + + /* XXX: from this point not restartable */ + ss = (ss & ~3) | dpl; + cpu_x86_load_seg_cache(env, R_SS, ss, + get_seg_base(ss_e1, ss_e2), + get_seg_limit(ss_e1, ss_e2), + ss_e2); + + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + sp &= 0xffff; + ssp = env->segs[R_SS].base + sp; + if (shift) { + ssp -= 4; + stl(ssp, old_ss); + ssp -= 4; + stl(ssp, old_esp); + ssp -= 4 * param_count; + for(i = 0; i < param_count; i++) { + val = ldl(old_ssp + i * 4); + stl(ssp + i * 4, val); + } + } else { + ssp -= 2; + stw(ssp, old_ss); + ssp -= 2; + stw(ssp, old_esp); + ssp -= 2 * param_count; + for(i = 0; i < param_count; i++) { + val = lduw(old_ssp + i * 2); + stw(ssp + i * 2, val); + } + } + } else { + /* to same priviledge */ + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + sp &= 0xffff; + ssp = env->segs[R_SS].base + sp; + push_size = (4 << shift); + } + + if (shift) { + ssp -= 4; + stl(ssp, env->segs[R_CS].selector); + ssp -= 4; + stl(ssp, next_eip); + } else { + ssp -= 2; + stw(ssp, env->segs[R_CS].selector); + ssp -= 2; + stw(ssp, next_eip); + } + + sp -= push_size; + selector = (selector & ~3) | dpl; + cpu_x86_load_seg_cache(env, R_CS, selector, + get_seg_base(e1, e2), + get_seg_limit(e1, e2), + e2); + cpu_x86_set_cpl(env, dpl); + + /* from this point, not restartable if same priviledge */ + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + ESP = (ESP & 0xffff0000) | (sp & 0xffff); + else + ESP = sp; + EIP = offset; + } +} + +/* real mode iret */ +void helper_iret_real(int shift) +{ + uint32_t sp, new_cs, new_eip, new_eflags, new_esp; + uint8_t *ssp; + int eflags_mask; + + sp = ESP & 0xffff; + ssp = env->segs[R_SS].base + sp; + if (shift == 1) { + /* 32 bits */ + new_eflags = ldl(ssp + 8); + new_cs = ldl(ssp + 4) & 0xffff; + new_eip = ldl(ssp) & 0xffff; + } else { + /* 16 bits */ + new_eflags = lduw(ssp + 4); + new_cs = lduw(ssp + 2); + new_eip = lduw(ssp); + } + new_esp = sp + (6 << shift); + ESP = (ESP & 0xffff0000) | + (new_esp & 0xffff); + load_seg_vm(R_CS, new_cs); + env->eip = new_eip; + eflags_mask = FL_UPDATE_CPL0_MASK; + if (shift == 0) + eflags_mask &= 0xffff; + load_eflags(new_eflags, eflags_mask); +} + +/* protected mode iret */ +static inline void helper_ret_protected(int shift, int is_iret, int addend) +{ + uint32_t sp, new_cs, new_eip, new_eflags, new_esp, new_ss; + uint32_t new_es, new_ds, new_fs, new_gs; + uint32_t e1, e2, ss_e1, ss_e2; + int cpl, dpl, rpl, eflags_mask; + uint8_t *ssp; + + sp = ESP; + if (!(env->segs[R_SS].flags & DESC_B_MASK)) + sp &= 0xffff; + ssp = env->segs[R_SS].base + sp; + if (shift == 1) { + /* 32 bits */ + if (is_iret) + new_eflags = ldl(ssp + 8); + new_cs = ldl(ssp + 4) & 0xffff; + new_eip = ldl(ssp); + if (is_iret && (new_eflags & VM_MASK)) + goto return_to_vm86; + } else { + /* 16 bits */ + if (is_iret) + new_eflags = lduw(ssp + 4); + new_cs = lduw(ssp + 2); + new_eip = lduw(ssp); + } + if ((new_cs & 0xfffc) == 0) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + if (load_segment(&e1, &e2, new_cs) != 0) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + if (!(e2 & DESC_S_MASK) || + !(e2 & DESC_CS_MASK)) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + cpl = env->hflags & HF_CPL_MASK; + rpl = new_cs & 3; + if (rpl < cpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + dpl = (e2 >> DESC_DPL_SHIFT) & 3; + if (e2 & DESC_CS_MASK) { + if (dpl > rpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + } else { + if (dpl != rpl) + raise_exception_err(EXCP0D_GPF, new_cs & 0xfffc); + } + if (!(e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, new_cs & 0xfffc); + + if (rpl == cpl) { + /* return to same priledge level */ + cpu_x86_load_seg_cache(env, R_CS, new_cs, + get_seg_base(e1, e2), + get_seg_limit(e1, e2), + e2); + new_esp = sp + (4 << shift) + ((2 * is_iret) << shift) + addend; + } else { + /* return to different priviledge level */ + ssp += (4 << shift) + ((2 * is_iret) << shift) + addend; + if (shift == 1) { + /* 32 bits */ + new_esp = ldl(ssp); + new_ss = ldl(ssp + 4) & 0xffff; + } else { + /* 16 bits */ + new_esp = lduw(ssp); + new_ss = lduw(ssp + 2); + } + + if ((new_ss & 3) != rpl) + raise_exception_err(EXCP0D_GPF, new_ss & 0xfffc); + if (load_segment(&ss_e1, &ss_e2, new_ss) != 0) + raise_exception_err(EXCP0D_GPF, new_ss & 0xfffc); + if (!(ss_e2 & DESC_S_MASK) || + (ss_e2 & DESC_CS_MASK) || + !(ss_e2 & DESC_W_MASK)) + raise_exception_err(EXCP0D_GPF, new_ss & 0xfffc); + dpl = (ss_e2 >> DESC_DPL_SHIFT) & 3; + if (dpl != rpl) + raise_exception_err(EXCP0D_GPF, new_ss & 0xfffc); + if (!(ss_e2 & DESC_P_MASK)) + raise_exception_err(EXCP0B_NOSEG, new_ss & 0xfffc); + + cpu_x86_load_seg_cache(env, R_CS, new_cs, + get_seg_base(e1, e2), + get_seg_limit(e1, e2), + e2); + cpu_x86_load_seg_cache(env, R_SS, new_ss, + get_seg_base(ss_e1, ss_e2), + get_seg_limit(ss_e1, ss_e2), + ss_e2); + cpu_x86_set_cpl(env, rpl); + } + if (env->segs[R_SS].flags & DESC_B_MASK) + ESP = new_esp; + else + ESP = (ESP & 0xffff0000) | + (new_esp & 0xffff); + env->eip = new_eip; + if (is_iret) { + /* NOTE: 'cpl' can be different from the current CPL */ + if (cpl == 0) + eflags_mask = FL_UPDATE_CPL0_MASK; + else + eflags_mask = FL_UPDATE_MASK32; + if (shift == 0) + eflags_mask &= 0xffff; + load_eflags(new_eflags, eflags_mask); + } + return; + + return_to_vm86: + new_esp = ldl(ssp + 12); + new_ss = ldl(ssp + 16); + new_es = ldl(ssp + 20); + new_ds = ldl(ssp + 24); + new_fs = ldl(ssp + 28); + new_gs = ldl(ssp + 32); + + /* modify processor state */ + load_eflags(new_eflags, FL_UPDATE_CPL0_MASK | VM_MASK | VIF_MASK | VIP_MASK); + load_seg_vm(R_CS, new_cs); + cpu_x86_set_cpl(env, 3); + load_seg_vm(R_SS, new_ss); + load_seg_vm(R_ES, new_es); + load_seg_vm(R_DS, new_ds); + load_seg_vm(R_FS, new_fs); + load_seg_vm(R_GS, new_gs); + + env->eip = new_eip; + ESP = new_esp; +} + +void helper_iret_protected(int shift) +{ + helper_ret_protected(shift, 1, 0); +} + +void helper_lret_protected(int shift, int addend) +{ + helper_ret_protected(shift, 0, addend); +} + +void helper_movl_crN_T0(int reg) +{ + env->cr[reg] = T0; + switch(reg) { + case 0: + cpu_x86_update_cr0(env); + break; + case 3: + cpu_x86_update_cr3(env); + break; + } +} + +/* XXX: do more */ +void helper_movl_drN_T0(int reg) +{ + env->dr[reg] = T0; +} + +void helper_invlpg(unsigned int addr) +{ + cpu_x86_flush_tlb(env, addr); +} + +/* rdtsc */ +#ifndef __i386__ +uint64_t emu_time; +#endif + +void helper_rdtsc(void) +{ + uint64_t val; +#ifdef __i386__ + asm("rdtsc" : "=A" (val)); +#else + /* better than nothing: the time increases */ + val = emu_time++; +#endif + EAX = val; + EDX = val >> 32; +} + +void helper_wrmsr(void) +{ + switch(ECX) { + case MSR_IA32_SYSENTER_CS: + env->sysenter_cs = EAX & 0xffff; + break; + case MSR_IA32_SYSENTER_ESP: + env->sysenter_esp = EAX; + break; + case MSR_IA32_SYSENTER_EIP: + env->sysenter_eip = EAX; + break; + default: + /* XXX: exception ? */ + break; + } +} + +void helper_rdmsr(void) +{ + switch(ECX) { + case MSR_IA32_SYSENTER_CS: + EAX = env->sysenter_cs; + EDX = 0; + break; + case MSR_IA32_SYSENTER_ESP: + EAX = env->sysenter_esp; + EDX = 0; + break; + case MSR_IA32_SYSENTER_EIP: + EAX = env->sysenter_eip; + EDX = 0; + break; + default: + /* XXX: exception ? */ + break; + } +} + +void helper_lsl(void) +{ + unsigned int selector, limit; + uint32_t e1, e2; + + CC_SRC = cc_table[CC_OP].compute_all() & ~CC_Z; + selector = T0 & 0xffff; + if (load_segment(&e1, &e2, selector) != 0) + return; + limit = (e1 & 0xffff) | (e2 & 0x000f0000); + if (e2 & (1 << 23)) + limit = (limit << 12) | 0xfff; + T1 = limit; + CC_SRC |= CC_Z; +} + +void helper_lar(void) +{ + unsigned int selector; + uint32_t e1, e2; + + CC_SRC = cc_table[CC_OP].compute_all() & ~CC_Z; + selector = T0 & 0xffff; + if (load_segment(&e1, &e2, selector) != 0) + return; + T1 = e2 & 0x00f0ff00; + CC_SRC |= CC_Z; +} + +/* FPU helpers */ + +#ifndef USE_X86LDOUBLE +void helper_fldt_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; + env->fpregs[new_fpstt] = helper_fldt((uint8_t *)A0); + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void helper_fstt_ST0_A0(void) +{ + helper_fstt(ST0, (uint8_t *)A0); +} +#endif + +/* BCD ops */ + +#define MUL10(iv) ( iv + iv + (iv << 3) ) + +void helper_fbld_ST0_A0(void) +{ + CPU86_LDouble tmp; + uint64_t val; + unsigned int v; + int i; + + val = 0; + for(i = 8; i >= 0; i--) { + v = ldub((uint8_t *)A0 + i); + val = (val * 100) + ((v >> 4) * 10) + (v & 0xf); + } + tmp = val; + if (ldub((uint8_t *)A0 + 9) & 0x80) + tmp = -tmp; + fpush(); + ST0 = tmp; +} + +void helper_fbst_ST0_A0(void) +{ + CPU86_LDouble tmp; + int v; + uint8_t *mem_ref, *mem_end; + int64_t val; + + tmp = rint(ST0); + val = (int64_t)tmp; + mem_ref = (uint8_t *)A0; + mem_end = mem_ref + 9; + if (val < 0) { + stb(mem_end, 0x80); + val = -val; + } else { + stb(mem_end, 0x00); + } + while (mem_ref < mem_end) { + if (val == 0) + break; + v = val % 100; + val = val / 100; + v = ((v / 10) << 4) | (v % 10); + stb(mem_ref++, v); + } + while (mem_ref < mem_end) { + stb(mem_ref++, 0); + } +} + +void helper_f2xm1(void) +{ + ST0 = pow(2.0,ST0) - 1.0; +} + +void helper_fyl2x(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if (fptemp>0.0){ + fptemp = log(fptemp)/log(2.0); /* log2(ST) */ + ST1 *= fptemp; + fpop(); + } else { + env->fpus &= (~0x4700); + env->fpus |= 0x400; + } +} + +void helper_fptan(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if((fptemp > MAXTAN)||(fptemp < -MAXTAN)) { + env->fpus |= 0x400; + } else { + ST0 = tan(fptemp); + fpush(); + ST0 = 1.0; + env->fpus &= (~0x400); /* C2 <-- 0 */ + /* the above code is for |arg| < 2**52 only */ + } +} + +void helper_fpatan(void) +{ + CPU86_LDouble fptemp, fpsrcop; + + fpsrcop = ST1; + fptemp = ST0; + ST1 = atan2(fpsrcop,fptemp); + fpop(); +} + +void helper_fxtract(void) +{ + CPU86_LDoubleU temp; + unsigned int expdif; + + temp.d = ST0; + expdif = EXPD(temp) - EXPBIAS; + /*DP exponent bias*/ + ST0 = expdif; + fpush(); + BIASEXPONENT(temp); + ST0 = temp.d; +} + +void helper_fprem1(void) +{ + CPU86_LDouble dblq, fpsrcop, fptemp; + CPU86_LDoubleU fpsrcop1, fptemp1; + int expdif; + int q; + + fpsrcop = ST0; + fptemp = ST1; + fpsrcop1.d = fpsrcop; + fptemp1.d = fptemp; + expdif = EXPD(fpsrcop1) - EXPD(fptemp1); + if (expdif < 53) { + dblq = fpsrcop / fptemp; + dblq = (dblq < 0.0)? ceil(dblq): floor(dblq); + ST0 = fpsrcop - fptemp*dblq; + q = (int)dblq; /* cutting off top bits is assumed here */ + env->fpus &= (~0x4700); /* (C3,C2,C1,C0) <-- 0000 */ + /* (C0,C1,C3) <-- (q2,q1,q0) */ + env->fpus |= (q&0x4) << 6; /* (C0) <-- q2 */ + env->fpus |= (q&0x2) << 8; /* (C1) <-- q1 */ + env->fpus |= (q&0x1) << 14; /* (C3) <-- q0 */ + } else { + env->fpus |= 0x400; /* C2 <-- 1 */ + fptemp = pow(2.0, expdif-50); + fpsrcop = (ST0 / ST1) / fptemp; + /* fpsrcop = integer obtained by rounding to the nearest */ + fpsrcop = (fpsrcop-floor(fpsrcop) < ceil(fpsrcop)-fpsrcop)? + floor(fpsrcop): ceil(fpsrcop); + ST0 -= (ST1 * fpsrcop * fptemp); + } +} + +void helper_fprem(void) +{ + CPU86_LDouble dblq, fpsrcop, fptemp; + CPU86_LDoubleU fpsrcop1, fptemp1; + int expdif; + int q; + + fpsrcop = ST0; + fptemp = ST1; + fpsrcop1.d = fpsrcop; + fptemp1.d = fptemp; + expdif = EXPD(fpsrcop1) - EXPD(fptemp1); + if ( expdif < 53 ) { + dblq = fpsrcop / fptemp; + dblq = (dblq < 0.0)? ceil(dblq): floor(dblq); + ST0 = fpsrcop - fptemp*dblq; + q = (int)dblq; /* cutting off top bits is assumed here */ + env->fpus &= (~0x4700); /* (C3,C2,C1,C0) <-- 0000 */ + /* (C0,C1,C3) <-- (q2,q1,q0) */ + env->fpus |= (q&0x4) << 6; /* (C0) <-- q2 */ + env->fpus |= (q&0x2) << 8; /* (C1) <-- q1 */ + env->fpus |= (q&0x1) << 14; /* (C3) <-- q0 */ + } else { + env->fpus |= 0x400; /* C2 <-- 1 */ + fptemp = pow(2.0, expdif-50); + fpsrcop = (ST0 / ST1) / fptemp; + /* fpsrcop = integer obtained by chopping */ + fpsrcop = (fpsrcop < 0.0)? + -(floor(fabs(fpsrcop))): floor(fpsrcop); + ST0 -= (ST1 * fpsrcop * fptemp); + } +} + +void helper_fyl2xp1(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if ((fptemp+1.0)>0.0) { + fptemp = log(fptemp+1.0) / log(2.0); /* log2(ST+1.0) */ + ST1 *= fptemp; + fpop(); + } else { + env->fpus &= (~0x4700); + env->fpus |= 0x400; + } +} + +void helper_fsqrt(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if (fptemp<0.0) { + env->fpus &= (~0x4700); /* (C3,C2,C1,C0) <-- 0000 */ + env->fpus |= 0x400; + } + ST0 = sqrt(fptemp); +} + +void helper_fsincos(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if ((fptemp > MAXTAN)||(fptemp < -MAXTAN)) { + env->fpus |= 0x400; + } else { + ST0 = sin(fptemp); + fpush(); + ST0 = cos(fptemp); + env->fpus &= (~0x400); /* C2 <-- 0 */ + /* the above code is for |arg| < 2**63 only */ + } +} + +void helper_frndint(void) +{ + CPU86_LDouble a; + + a = ST0; +#ifdef __arm__ + switch(env->fpuc & RC_MASK) { + default: + case RC_NEAR: + asm("rndd %0, %1" : "=f" (a) : "f"(a)); + break; + case RC_DOWN: + asm("rnddm %0, %1" : "=f" (a) : "f"(a)); + break; + case RC_UP: + asm("rnddp %0, %1" : "=f" (a) : "f"(a)); + break; + case RC_CHOP: + asm("rnddz %0, %1" : "=f" (a) : "f"(a)); + break; + } +#else + a = rint(a); +#endif + ST0 = a; +} + +void helper_fscale(void) +{ + CPU86_LDouble fpsrcop, fptemp; + + fpsrcop = 2.0; + fptemp = pow(fpsrcop,ST1); + ST0 *= fptemp; +} + +void helper_fsin(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if ((fptemp > MAXTAN)||(fptemp < -MAXTAN)) { + env->fpus |= 0x400; + } else { + ST0 = sin(fptemp); + env->fpus &= (~0x400); /* C2 <-- 0 */ + /* the above code is for |arg| < 2**53 only */ + } +} + +void helper_fcos(void) +{ + CPU86_LDouble fptemp; + + fptemp = ST0; + if((fptemp > MAXTAN)||(fptemp < -MAXTAN)) { + env->fpus |= 0x400; + } else { + ST0 = cos(fptemp); + env->fpus &= (~0x400); /* C2 <-- 0 */ + /* the above code is for |arg5 < 2**63 only */ + } +} + +void helper_fxam_ST0(void) +{ + CPU86_LDoubleU temp; + int expdif; + + temp.d = ST0; + + env->fpus &= (~0x4700); /* (C3,C2,C1,C0) <-- 0000 */ + if (SIGND(temp)) + env->fpus |= 0x200; /* C1 <-- 1 */ + + expdif = EXPD(temp); + if (expdif == MAXEXPD) { + if (MANTD(temp) == 0) + env->fpus |= 0x500 /*Infinity*/; + else + env->fpus |= 0x100 /*NaN*/; + } else if (expdif == 0) { + if (MANTD(temp) == 0) + env->fpus |= 0x4000 /*Zero*/; + else + env->fpus |= 0x4400 /*Denormal*/; + } else { + env->fpus |= 0x400; + } +} + +void helper_fstenv(uint8_t *ptr, int data32) +{ + int fpus, fptag, exp, i; + uint64_t mant; + CPU86_LDoubleU tmp; + + fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11; + fptag = 0; + for (i=7; i>=0; i--) { + fptag <<= 2; + if (env->fptags[i]) { + fptag |= 3; + } else { + tmp.d = env->fpregs[i]; + exp = EXPD(tmp); + mant = MANTD(tmp); + if (exp == 0 && mant == 0) { + /* zero */ + fptag |= 1; + } else if (exp == 0 || exp == MAXEXPD +#ifdef USE_X86LDOUBLE + || (mant & (1LL << 63)) == 0 +#endif + ) { + /* NaNs, infinity, denormal */ + fptag |= 2; + } + } + } + if (data32) { + /* 32 bit */ + stl(ptr, env->fpuc); + stl(ptr + 4, fpus); + stl(ptr + 8, fptag); + stl(ptr + 12, 0); + stl(ptr + 16, 0); + stl(ptr + 20, 0); + stl(ptr + 24, 0); + } else { + /* 16 bit */ + stw(ptr, env->fpuc); + stw(ptr + 2, fpus); + stw(ptr + 4, fptag); + stw(ptr + 6, 0); + stw(ptr + 8, 0); + stw(ptr + 10, 0); + stw(ptr + 12, 0); + } +} + +void helper_fldenv(uint8_t *ptr, int data32) +{ + int i, fpus, fptag; + + if (data32) { + env->fpuc = lduw(ptr); + fpus = lduw(ptr + 4); + fptag = lduw(ptr + 8); + } + else { + env->fpuc = lduw(ptr); + fpus = lduw(ptr + 2); + fptag = lduw(ptr + 4); + } + env->fpstt = (fpus >> 11) & 7; + env->fpus = fpus & ~0x3800; + for(i = 0;i < 7; i++) { + env->fptags[i] = ((fptag & 3) == 3); + fptag >>= 2; + } +} + +void helper_fsave(uint8_t *ptr, int data32) +{ + CPU86_LDouble tmp; + int i; + + helper_fstenv(ptr, data32); + + ptr += (14 << data32); + for(i = 0;i < 8; i++) { + tmp = ST(i); +#ifdef USE_X86LDOUBLE + *(long double *)ptr = tmp; +#else + helper_fstt(tmp, ptr); +#endif + ptr += 10; + } + + /* fninit */ + env->fpus = 0; + env->fpstt = 0; + env->fpuc = 0x37f; + env->fptags[0] = 1; + env->fptags[1] = 1; + env->fptags[2] = 1; + env->fptags[3] = 1; + env->fptags[4] = 1; + env->fptags[5] = 1; + env->fptags[6] = 1; + env->fptags[7] = 1; +} + +void helper_frstor(uint8_t *ptr, int data32) +{ + CPU86_LDouble tmp; + int i; + + helper_fldenv(ptr, data32); + ptr += (14 << data32); + + for(i = 0;i < 8; i++) { +#ifdef USE_X86LDOUBLE + tmp = *(long double *)ptr; +#else + tmp = helper_fldt(ptr); +#endif + ST(i) = tmp; + ptr += 10; + } +} + +#define SHIFT 0 +#include "softmmu_template.h" + +#define SHIFT 1 +#include "softmmu_template.h" + +#define SHIFT 2 +#include "softmmu_template.h" + +#define SHIFT 3 +#include "softmmu_template.h" + +/* try to fill the TLB and return an exception if error */ +void tlb_fill(unsigned long addr, int is_write, void *retaddr) +{ + TranslationBlock *tb; + int ret; + unsigned long pc; + ret = cpu_x86_handle_mmu_fault(env, addr, is_write); + if (ret) { + /* now we have a real cpu fault */ + pc = (unsigned long)retaddr; + tb = tb_find_pc(pc); + if (tb) { + /* the PC is inside the translated code. It means that we have + a virtual CPU fault */ + cpu_restore_state(tb, env, pc); + } + raise_exception_err(EXCP0E_PAGE, env->error_code); + } +} diff --git a/target-i386/helper2.c b/target-i386/helper2.c new file mode 100644 index 0000000000..bc0d426bc1 --- /dev/null +++ b/target-i386/helper2.c @@ -0,0 +1,390 @@ +/* + * i386 helpers (without register variable usage) + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#include <stdarg.h> +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <inttypes.h> +#include <signal.h> +#include <assert.h> +#include <sys/mman.h> + +#include "cpu.h" +#include "exec-all.h" + +//#define DEBUG_MMU + +CPUX86State *cpu_x86_init(void) +{ + CPUX86State *env; + int i; + static int inited; + + cpu_exec_init(); + + env = malloc(sizeof(CPUX86State)); + if (!env) + return NULL; + memset(env, 0, sizeof(CPUX86State)); + /* basic FPU init */ + for(i = 0;i < 8; i++) + env->fptags[i] = 1; + env->fpuc = 0x37f; + /* flags setup : we activate the IRQs by default as in user mode */ + env->eflags = 0x2 | IF_MASK; + + tlb_flush(env); +#ifdef CONFIG_SOFTMMU + env->hflags |= HF_SOFTMMU_MASK; +#endif + /* init various static tables */ + if (!inited) { + inited = 1; + optimize_flags_init(); + } + return env; +} + +void cpu_x86_close(CPUX86State *env) +{ + free(env); +} + +/***********************************************************/ +/* x86 debug */ + +static const char *cc_op_str[] = { + "DYNAMIC", + "EFLAGS", + "MUL", + "ADDB", + "ADDW", + "ADDL", + "ADCB", + "ADCW", + "ADCL", + "SUBB", + "SUBW", + "SUBL", + "SBBB", + "SBBW", + "SBBL", + "LOGICB", + "LOGICW", + "LOGICL", + "INCB", + "INCW", + "INCL", + "DECB", + "DECW", + "DECL", + "SHLB", + "SHLW", + "SHLL", + "SARB", + "SARW", + "SARL", +}; + +void cpu_x86_dump_state(CPUX86State *env, FILE *f, int flags) +{ + int eflags; + char cc_op_name[32]; + + eflags = env->eflags; + fprintf(f, "EAX=%08x EBX=%08x ECX=%08x EDX=%08x\n" + "ESI=%08x EDI=%08x EBP=%08x ESP=%08x\n" + "EIP=%08x EFL=%08x [%c%c%c%c%c%c%c]\n", + env->regs[R_EAX], env->regs[R_EBX], env->regs[R_ECX], env->regs[R_EDX], + env->regs[R_ESI], env->regs[R_EDI], env->regs[R_EBP], env->regs[R_ESP], + env->eip, eflags, + eflags & DF_MASK ? 'D' : '-', + eflags & CC_O ? 'O' : '-', + eflags & CC_S ? 'S' : '-', + eflags & CC_Z ? 'Z' : '-', + eflags & CC_A ? 'A' : '-', + eflags & CC_P ? 'P' : '-', + eflags & CC_C ? 'C' : '-'); + fprintf(f, "CS=%04x SS=%04x DS=%04x ES=%04x FS=%04x GS=%04x\n", + env->segs[R_CS].selector, + env->segs[R_SS].selector, + env->segs[R_DS].selector, + env->segs[R_ES].selector, + env->segs[R_FS].selector, + env->segs[R_GS].selector); + if (flags & X86_DUMP_CCOP) { + if ((unsigned)env->cc_op < CC_OP_NB) + strcpy(cc_op_name, cc_op_str[env->cc_op]); + else + snprintf(cc_op_name, sizeof(cc_op_name), "[%d]", env->cc_op); + fprintf(f, "CCS=%08x CCD=%08x CCO=%-8s\n", + env->cc_src, env->cc_dst, cc_op_name); + } + if (flags & X86_DUMP_FPU) { + fprintf(f, "ST0=%f ST1=%f ST2=%f ST3=%f\n", + (double)env->fpregs[0], + (double)env->fpregs[1], + (double)env->fpregs[2], + (double)env->fpregs[3]); + fprintf(f, "ST4=%f ST5=%f ST6=%f ST7=%f\n", + (double)env->fpregs[4], + (double)env->fpregs[5], + (double)env->fpregs[7], + (double)env->fpregs[8]); + } +} + +/***********************************************************/ +/* x86 mmu */ +/* XXX: add PGE support */ + +/* called when cr3 or PG bit are modified */ +static int last_pg_state = -1; +static int last_pe_state = 0; +int phys_ram_size; +int phys_ram_fd; +uint8_t *phys_ram_base; + +void cpu_x86_update_cr0(CPUX86State *env) +{ + int pg_state, pe_state; + +#ifdef DEBUG_MMU + printf("CR0 update: CR0=0x%08x\n", env->cr[0]); +#endif + pg_state = env->cr[0] & CR0_PG_MASK; + if (pg_state != last_pg_state) { + page_unmap(); + tlb_flush(env); + last_pg_state = pg_state; + } + pe_state = env->cr[0] & CR0_PE_MASK; + if (last_pe_state != pe_state) { + tb_flush(); + last_pe_state = pe_state; + } +} + +void cpu_x86_update_cr3(CPUX86State *env) +{ + if (env->cr[0] & CR0_PG_MASK) { +#if defined(DEBUG_MMU) + printf("CR3 update: CR3=%08x\n", env->cr[3]); +#endif + page_unmap(); + tlb_flush(env); + } +} + +void cpu_x86_init_mmu(CPUX86State *env) +{ + last_pg_state = -1; + cpu_x86_update_cr0(env); +} + +/* XXX: also flush 4MB pages */ +void cpu_x86_flush_tlb(CPUX86State *env, uint32_t addr) +{ + int flags; + unsigned long virt_addr; + + tlb_flush_page(env, addr); + + flags = page_get_flags(addr); + if (flags & PAGE_VALID) { + virt_addr = addr & ~0xfff; + munmap((void *)virt_addr, 4096); + page_set_flags(virt_addr, virt_addr + 4096, 0); + } +} + +/* return value: + -1 = cannot handle fault + 0 = nothing more to do + 1 = generate PF fault + 2 = soft MMU activation required for this block +*/ +int cpu_x86_handle_mmu_fault(CPUX86State *env, uint32_t addr, int is_write) +{ + uint8_t *pde_ptr, *pte_ptr; + uint32_t pde, pte, virt_addr; + int cpl, error_code, is_dirty, is_user, prot, page_size, ret; + unsigned long pd; + + cpl = env->hflags & HF_CPL_MASK; + is_user = (cpl == 3); + +#ifdef DEBUG_MMU + printf("MMU fault: addr=0x%08x w=%d u=%d eip=%08x\n", + addr, is_write, is_user, env->eip); +#endif + + if (env->user_mode_only) { + /* user mode only emulation */ + error_code = 0; + goto do_fault; + } + + if (!(env->cr[0] & CR0_PG_MASK)) { + pte = addr; + virt_addr = addr & ~0xfff; + prot = PROT_READ | PROT_WRITE; + page_size = 4096; + goto do_mapping; + } + + /* page directory entry */ + pde_ptr = phys_ram_base + ((env->cr[3] & ~0xfff) + ((addr >> 20) & ~3)); + pde = ldl(pde_ptr); + if (!(pde & PG_PRESENT_MASK)) { + error_code = 0; + goto do_fault; + } + if (is_user) { + if (!(pde & PG_USER_MASK)) + goto do_fault_protect; + if (is_write && !(pde & PG_RW_MASK)) + goto do_fault_protect; + } else { + if ((env->cr[0] & CR0_WP_MASK) && (pde & PG_USER_MASK) && + is_write && !(pde & PG_RW_MASK)) + goto do_fault_protect; + } + /* if PSE bit is set, then we use a 4MB page */ + if ((pde & PG_PSE_MASK) && (env->cr[4] & CR4_PSE_MASK)) { + is_dirty = is_write && !(pde & PG_DIRTY_MASK); + if (!(pde & PG_ACCESSED_MASK)) { + pde |= PG_ACCESSED_MASK; + if (is_dirty) + pde |= PG_DIRTY_MASK; + stl(pde_ptr, pde); + } + + pte = pde & ~0x003ff000; /* align to 4MB */ + page_size = 4096 * 1024; + virt_addr = addr & ~0x003fffff; + } else { + if (!(pde & PG_ACCESSED_MASK)) { + pde |= PG_ACCESSED_MASK; + stl(pde_ptr, pde); + } + + /* page directory entry */ + pte_ptr = phys_ram_base + ((pde & ~0xfff) + ((addr >> 10) & 0xffc)); + pte = ldl(pte_ptr); + if (!(pte & PG_PRESENT_MASK)) { + error_code = 0; + goto do_fault; + } + if (is_user) { + if (!(pte & PG_USER_MASK)) + goto do_fault_protect; + if (is_write && !(pte & PG_RW_MASK)) + goto do_fault_protect; + } else { + if ((env->cr[0] & CR0_WP_MASK) && (pte & PG_USER_MASK) && + is_write && !(pte & PG_RW_MASK)) + goto do_fault_protect; + } + is_dirty = is_write && !(pte & PG_DIRTY_MASK); + if (!(pte & PG_ACCESSED_MASK) || is_dirty) { + pte |= PG_ACCESSED_MASK; + if (is_dirty) + pte |= PG_DIRTY_MASK; + stl(pte_ptr, pte); + } + page_size = 4096; + virt_addr = addr & ~0xfff; + } + /* the page can be put in the TLB */ + prot = PROT_READ; + if (is_user) { + if (pte & PG_RW_MASK) + prot |= PROT_WRITE; + } else { + if (!(env->cr[0] & CR0_WP_MASK) || !(pte & PG_USER_MASK) || + (pte & PG_RW_MASK)) + prot |= PROT_WRITE; + } + + do_mapping: + if (env->hflags & HF_SOFTMMU_MASK) { + unsigned long paddr, vaddr, address, addend, page_offset; + int index; + + /* software MMU case. Even if 4MB pages, we map only one 4KB + page in the cache to avoid filling it too fast */ + page_offset = (addr & ~0xfff) & (page_size - 1); + paddr = (pte & ~0xfff) + page_offset; + vaddr = virt_addr + page_offset; + index = (addr >> 12) & (CPU_TLB_SIZE - 1); + pd = physpage_find(paddr); + if (pd & 0xfff) { + /* IO memory case */ + address = vaddr | pd; + addend = paddr; + } else { + /* standard memory */ + address = vaddr; + addend = (unsigned long)phys_ram_base + pd; + } + addend -= vaddr; + env->tlb_read[is_user][index].address = address; + env->tlb_read[is_user][index].addend = addend; + if (prot & PROT_WRITE) { + env->tlb_write[is_user][index].address = address; + env->tlb_write[is_user][index].addend = addend; + } + } + ret = 0; + /* XXX: incorrect for 4MB pages */ + pd = physpage_find(pte & ~0xfff); + if ((pd & 0xfff) != 0) { + /* IO access: no mapping is done as it will be handled by the + soft MMU */ + if (!(env->hflags & HF_SOFTMMU_MASK)) + ret = 2; + } else { + void *map_addr; + map_addr = mmap((void *)virt_addr, page_size, prot, + MAP_SHARED | MAP_FIXED, phys_ram_fd, pd); + if (map_addr == MAP_FAILED) { + fprintf(stderr, + "mmap failed when mapped physical address 0x%08x to virtual address 0x%08x\n", + pte & ~0xfff, virt_addr); + exit(1); + } +#ifdef DEBUG_MMU + printf("mmaping 0x%08x to virt 0x%08x pse=%d\n", + pte & ~0xfff, virt_addr, (page_size != 4096)); +#endif + page_set_flags(virt_addr, virt_addr + page_size, + PAGE_VALID | PAGE_EXEC | prot); + } + return ret; + do_fault_protect: + error_code = PG_ERROR_P_MASK; + do_fault: + env->cr[2] = addr; + env->error_code = (is_write << PG_ERROR_W_BIT) | error_code; + if (is_user) + env->error_code |= PG_ERROR_U_MASK; + return 1; +} diff --git a/target-i386/op.c b/target-i386/op.c new file mode 100644 index 0000000000..68191f9bc4 --- /dev/null +++ b/target-i386/op.c @@ -0,0 +1,2054 @@ +/* + * i386 micro operations + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#include "exec.h" + +/* n must be a constant to be efficient */ +static inline int lshift(int x, int n) +{ + if (n >= 0) + return x << n; + else + return x >> (-n); +} + +/* we define the various pieces of code used by the JIT */ + +#define REG EAX +#define REGNAME _EAX +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG ECX +#define REGNAME _ECX +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG EDX +#define REGNAME _EDX +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG EBX +#define REGNAME _EBX +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG ESP +#define REGNAME _ESP +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG EBP +#define REGNAME _EBP +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG ESI +#define REGNAME _ESI +#include "opreg_template.h" +#undef REG +#undef REGNAME + +#define REG EDI +#define REGNAME _EDI +#include "opreg_template.h" +#undef REG +#undef REGNAME + +/* operations with flags */ + +/* update flags with T0 and T1 (add/sub case) */ +void OPPROTO op_update2_cc(void) +{ + CC_SRC = T1; + CC_DST = T0; +} + +/* update flags with T0 (logic operation case) */ +void OPPROTO op_update1_cc(void) +{ + CC_DST = T0; +} + +void OPPROTO op_update_neg_cc(void) +{ + CC_SRC = -T0; + CC_DST = T0; +} + +void OPPROTO op_cmpl_T0_T1_cc(void) +{ + CC_SRC = T1; + CC_DST = T0 - T1; +} + +void OPPROTO op_update_inc_cc(void) +{ + CC_SRC = cc_table[CC_OP].compute_c(); + CC_DST = T0; +} + +void OPPROTO op_testl_T0_T1_cc(void) +{ + CC_DST = T0 & T1; +} + +/* operations without flags */ + +void OPPROTO op_addl_T0_T1(void) +{ + T0 += T1; +} + +void OPPROTO op_orl_T0_T1(void) +{ + T0 |= T1; +} + +void OPPROTO op_andl_T0_T1(void) +{ + T0 &= T1; +} + +void OPPROTO op_subl_T0_T1(void) +{ + T0 -= T1; +} + +void OPPROTO op_xorl_T0_T1(void) +{ + T0 ^= T1; +} + +void OPPROTO op_negl_T0(void) +{ + T0 = -T0; +} + +void OPPROTO op_incl_T0(void) +{ + T0++; +} + +void OPPROTO op_decl_T0(void) +{ + T0--; +} + +void OPPROTO op_notl_T0(void) +{ + T0 = ~T0; +} + +void OPPROTO op_bswapl_T0(void) +{ + T0 = bswap32(T0); +} + +/* multiply/divide */ +void OPPROTO op_mulb_AL_T0(void) +{ + unsigned int res; + res = (uint8_t)EAX * (uint8_t)T0; + EAX = (EAX & 0xffff0000) | res; + CC_SRC = (res & 0xff00); +} + +void OPPROTO op_imulb_AL_T0(void) +{ + int res; + res = (int8_t)EAX * (int8_t)T0; + EAX = (EAX & 0xffff0000) | (res & 0xffff); + CC_SRC = (res != (int8_t)res); +} + +void OPPROTO op_mulw_AX_T0(void) +{ + unsigned int res; + res = (uint16_t)EAX * (uint16_t)T0; + EAX = (EAX & 0xffff0000) | (res & 0xffff); + EDX = (EDX & 0xffff0000) | ((res >> 16) & 0xffff); + CC_SRC = res >> 16; +} + +void OPPROTO op_imulw_AX_T0(void) +{ + int res; + res = (int16_t)EAX * (int16_t)T0; + EAX = (EAX & 0xffff0000) | (res & 0xffff); + EDX = (EDX & 0xffff0000) | ((res >> 16) & 0xffff); + CC_SRC = (res != (int16_t)res); +} + +void OPPROTO op_mull_EAX_T0(void) +{ + uint64_t res; + res = (uint64_t)((uint32_t)EAX) * (uint64_t)((uint32_t)T0); + EAX = res; + EDX = res >> 32; + CC_SRC = res >> 32; +} + +void OPPROTO op_imull_EAX_T0(void) +{ + int64_t res; + res = (int64_t)((int32_t)EAX) * (int64_t)((int32_t)T0); + EAX = res; + EDX = res >> 32; + CC_SRC = (res != (int32_t)res); +} + +void OPPROTO op_imulw_T0_T1(void) +{ + int res; + res = (int16_t)T0 * (int16_t)T1; + T0 = res; + CC_SRC = (res != (int16_t)res); +} + +void OPPROTO op_imull_T0_T1(void) +{ + int64_t res; + res = (int64_t)((int32_t)T0) * (int64_t)((int32_t)T1); + T0 = res; + CC_SRC = (res != (int32_t)res); +} + +/* division, flags are undefined */ +/* XXX: add exceptions for overflow */ + +void OPPROTO op_divb_AL_T0(void) +{ + unsigned int num, den, q, r; + + num = (EAX & 0xffff); + den = (T0 & 0xff); + if (den == 0) { + EIP = PARAM1; + raise_exception(EXCP00_DIVZ); + } + q = (num / den) & 0xff; + r = (num % den) & 0xff; + EAX = (EAX & 0xffff0000) | (r << 8) | q; +} + +void OPPROTO op_idivb_AL_T0(void) +{ + int num, den, q, r; + + num = (int16_t)EAX; + den = (int8_t)T0; + if (den == 0) { + EIP = PARAM1; + raise_exception(EXCP00_DIVZ); + } + q = (num / den) & 0xff; + r = (num % den) & 0xff; + EAX = (EAX & 0xffff0000) | (r << 8) | q; +} + +void OPPROTO op_divw_AX_T0(void) +{ + unsigned int num, den, q, r; + + num = (EAX & 0xffff) | ((EDX & 0xffff) << 16); + den = (T0 & 0xffff); + if (den == 0) { + EIP = PARAM1; + raise_exception(EXCP00_DIVZ); + } + q = (num / den) & 0xffff; + r = (num % den) & 0xffff; + EAX = (EAX & 0xffff0000) | q; + EDX = (EDX & 0xffff0000) | r; +} + +void OPPROTO op_idivw_AX_T0(void) +{ + int num, den, q, r; + + num = (EAX & 0xffff) | ((EDX & 0xffff) << 16); + den = (int16_t)T0; + if (den == 0) { + EIP = PARAM1; + raise_exception(EXCP00_DIVZ); + } + q = (num / den) & 0xffff; + r = (num % den) & 0xffff; + EAX = (EAX & 0xffff0000) | q; + EDX = (EDX & 0xffff0000) | r; +} + +void OPPROTO op_divl_EAX_T0(void) +{ + helper_divl_EAX_T0(PARAM1); +} + +void OPPROTO op_idivl_EAX_T0(void) +{ + helper_idivl_EAX_T0(PARAM1); +} + +/* constant load & misc op */ + +void OPPROTO op_movl_T0_im(void) +{ + T0 = PARAM1; +} + +void OPPROTO op_addl_T0_im(void) +{ + T0 += PARAM1; +} + +void OPPROTO op_andl_T0_ffff(void) +{ + T0 = T0 & 0xffff; +} + +void OPPROTO op_andl_T0_im(void) +{ + T0 = T0 & PARAM1; +} + +void OPPROTO op_movl_T0_T1(void) +{ + T0 = T1; +} + +void OPPROTO op_movl_T1_im(void) +{ + T1 = PARAM1; +} + +void OPPROTO op_addl_T1_im(void) +{ + T1 += PARAM1; +} + +void OPPROTO op_movl_T1_A0(void) +{ + T1 = A0; +} + +void OPPROTO op_movl_A0_im(void) +{ + A0 = PARAM1; +} + +void OPPROTO op_addl_A0_im(void) +{ + A0 += PARAM1; +} + +void OPPROTO op_addl_A0_AL(void) +{ + A0 += (EAX & 0xff); +} + +void OPPROTO op_andl_A0_ffff(void) +{ + A0 = A0 & 0xffff; +} + +/* memory access */ + +#define MEMSUFFIX +#include "ops_mem.h" + +#define MEMSUFFIX _user +#include "ops_mem.h" + +#define MEMSUFFIX _kernel +#include "ops_mem.h" + +/* used for bit operations */ + +void OPPROTO op_add_bitw_A0_T1(void) +{ + A0 += ((int32_t)T1 >> 4) << 1; +} + +void OPPROTO op_add_bitl_A0_T1(void) +{ + A0 += ((int32_t)T1 >> 5) << 2; +} + +/* indirect jump */ + +void OPPROTO op_jmp_T0(void) +{ + EIP = T0; +} + +void OPPROTO op_jmp_im(void) +{ + EIP = PARAM1; +} + +void OPPROTO op_hlt(void) +{ + env->exception_index = EXCP_HLT; + cpu_loop_exit(); +} + +void OPPROTO op_debug(void) +{ + env->exception_index = EXCP_DEBUG; + cpu_loop_exit(); +} + +void OPPROTO op_raise_interrupt(void) +{ + int intno; + unsigned int next_eip; + intno = PARAM1; + next_eip = PARAM2; + raise_interrupt(intno, 1, 0, next_eip); +} + +void OPPROTO op_raise_exception(void) +{ + int exception_index; + exception_index = PARAM1; + raise_exception(exception_index); +} + +void OPPROTO op_into(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + if (eflags & CC_O) { + raise_interrupt(EXCP04_INTO, 1, 0, PARAM1); + } + FORCE_RET(); +} + +void OPPROTO op_cli(void) +{ + env->eflags &= ~IF_MASK; +} + +void OPPROTO op_sti(void) +{ + env->eflags |= IF_MASK; +} + +void OPPROTO op_set_inhibit_irq(void) +{ + env->hflags |= HF_INHIBIT_IRQ_MASK; +} + +void OPPROTO op_reset_inhibit_irq(void) +{ + env->hflags &= ~HF_INHIBIT_IRQ_MASK; +} + +#if 0 +/* vm86plus instructions */ +void OPPROTO op_cli_vm(void) +{ + env->eflags &= ~VIF_MASK; +} + +void OPPROTO op_sti_vm(void) +{ + env->eflags |= VIF_MASK; + if (env->eflags & VIP_MASK) { + EIP = PARAM1; + raise_exception(EXCP0D_GPF); + } + FORCE_RET(); +} +#endif + +void OPPROTO op_boundw(void) +{ + int low, high, v; + low = ldsw((uint8_t *)A0); + high = ldsw((uint8_t *)A0 + 2); + v = (int16_t)T0; + if (v < low || v > high) { + EIP = PARAM1; + raise_exception(EXCP05_BOUND); + } + FORCE_RET(); +} + +void OPPROTO op_boundl(void) +{ + int low, high, v; + low = ldl((uint8_t *)A0); + high = ldl((uint8_t *)A0 + 4); + v = T0; + if (v < low || v > high) { + EIP = PARAM1; + raise_exception(EXCP05_BOUND); + } + FORCE_RET(); +} + +void OPPROTO op_cmpxchg8b(void) +{ + helper_cmpxchg8b(); +} + +void OPPROTO op_jmp(void) +{ + JUMP_TB(op_jmp, PARAM1, 0, PARAM2); +} + +void OPPROTO op_movl_T0_0(void) +{ + T0 = 0; +} + +void OPPROTO op_exit_tb(void) +{ + EXIT_TB(); +} + +/* multiple size ops */ + +#define ldul ldl + +#define SHIFT 0 +#include "ops_template.h" +#undef SHIFT + +#define SHIFT 1 +#include "ops_template.h" +#undef SHIFT + +#define SHIFT 2 +#include "ops_template.h" +#undef SHIFT + +/* sign extend */ + +void OPPROTO op_movsbl_T0_T0(void) +{ + T0 = (int8_t)T0; +} + +void OPPROTO op_movzbl_T0_T0(void) +{ + T0 = (uint8_t)T0; +} + +void OPPROTO op_movswl_T0_T0(void) +{ + T0 = (int16_t)T0; +} + +void OPPROTO op_movzwl_T0_T0(void) +{ + T0 = (uint16_t)T0; +} + +void OPPROTO op_movswl_EAX_AX(void) +{ + EAX = (int16_t)EAX; +} + +void OPPROTO op_movsbw_AX_AL(void) +{ + EAX = (EAX & 0xffff0000) | ((int8_t)EAX & 0xffff); +} + +void OPPROTO op_movslq_EDX_EAX(void) +{ + EDX = (int32_t)EAX >> 31; +} + +void OPPROTO op_movswl_DX_AX(void) +{ + EDX = (EDX & 0xffff0000) | (((int16_t)EAX >> 15) & 0xffff); +} + +/* string ops helpers */ + +void OPPROTO op_addl_ESI_T0(void) +{ + ESI += T0; +} + +void OPPROTO op_addw_ESI_T0(void) +{ + ESI = (ESI & ~0xffff) | ((ESI + T0) & 0xffff); +} + +void OPPROTO op_addl_EDI_T0(void) +{ + EDI += T0; +} + +void OPPROTO op_addw_EDI_T0(void) +{ + EDI = (EDI & ~0xffff) | ((EDI + T0) & 0xffff); +} + +void OPPROTO op_decl_ECX(void) +{ + ECX--; +} + +void OPPROTO op_decw_ECX(void) +{ + ECX = (ECX & ~0xffff) | ((ECX - 1) & 0xffff); +} + +/* push/pop */ + +void op_pushl_T0(void) +{ + uint32_t offset; + offset = ESP - 4; + stl((void *)offset, T0); + /* modify ESP after to handle exceptions correctly */ + ESP = offset; +} + +void op_pushw_T0(void) +{ + uint32_t offset; + offset = ESP - 2; + stw((void *)offset, T0); + /* modify ESP after to handle exceptions correctly */ + ESP = offset; +} + +void op_pushl_ss32_T0(void) +{ + uint32_t offset; + offset = ESP - 4; + stl(env->segs[R_SS].base + offset, T0); + /* modify ESP after to handle exceptions correctly */ + ESP = offset; +} + +void op_pushw_ss32_T0(void) +{ + uint32_t offset; + offset = ESP - 2; + stw(env->segs[R_SS].base + offset, T0); + /* modify ESP after to handle exceptions correctly */ + ESP = offset; +} + +void op_pushl_ss16_T0(void) +{ + uint32_t offset; + offset = (ESP - 4) & 0xffff; + stl(env->segs[R_SS].base + offset, T0); + /* modify ESP after to handle exceptions correctly */ + ESP = (ESP & ~0xffff) | offset; +} + +void op_pushw_ss16_T0(void) +{ + uint32_t offset; + offset = (ESP - 2) & 0xffff; + stw(env->segs[R_SS].base + offset, T0); + /* modify ESP after to handle exceptions correctly */ + ESP = (ESP & ~0xffff) | offset; +} + +/* NOTE: ESP update is done after */ +void op_popl_T0(void) +{ + T0 = ldl((void *)ESP); +} + +void op_popw_T0(void) +{ + T0 = lduw((void *)ESP); +} + +void op_popl_ss32_T0(void) +{ + T0 = ldl(env->segs[R_SS].base + ESP); +} + +void op_popw_ss32_T0(void) +{ + T0 = lduw(env->segs[R_SS].base + ESP); +} + +void op_popl_ss16_T0(void) +{ + T0 = ldl(env->segs[R_SS].base + (ESP & 0xffff)); +} + +void op_popw_ss16_T0(void) +{ + T0 = lduw(env->segs[R_SS].base + (ESP & 0xffff)); +} + +void op_addl_ESP_4(void) +{ + ESP += 4; +} + +void op_addl_ESP_2(void) +{ + ESP += 2; +} + +void op_addw_ESP_4(void) +{ + ESP = (ESP & ~0xffff) | ((ESP + 4) & 0xffff); +} + +void op_addw_ESP_2(void) +{ + ESP = (ESP & ~0xffff) | ((ESP + 2) & 0xffff); +} + +void op_addl_ESP_im(void) +{ + ESP += PARAM1; +} + +void op_addw_ESP_im(void) +{ + ESP = (ESP & ~0xffff) | ((ESP + PARAM1) & 0xffff); +} + +void OPPROTO op_rdtsc(void) +{ + helper_rdtsc(); +} + +void OPPROTO op_cpuid(void) +{ + helper_cpuid(); +} + +void OPPROTO op_rdmsr(void) +{ + helper_rdmsr(); +} + +void OPPROTO op_wrmsr(void) +{ + helper_wrmsr(); +} + +/* bcd */ + +/* XXX: exception */ +void OPPROTO op_aam(void) +{ + int base = PARAM1; + int al, ah; + al = EAX & 0xff; + ah = al / base; + al = al % base; + EAX = (EAX & ~0xffff) | al | (ah << 8); + CC_DST = al; +} + +void OPPROTO op_aad(void) +{ + int base = PARAM1; + int al, ah; + al = EAX & 0xff; + ah = (EAX >> 8) & 0xff; + al = ((ah * base) + al) & 0xff; + EAX = (EAX & ~0xffff) | al; + CC_DST = al; +} + +void OPPROTO op_aaa(void) +{ + int icarry; + int al, ah, af; + int eflags; + + eflags = cc_table[CC_OP].compute_all(); + af = eflags & CC_A; + al = EAX & 0xff; + ah = (EAX >> 8) & 0xff; + + icarry = (al > 0xf9); + if (((al & 0x0f) > 9 ) || af) { + al = (al + 6) & 0x0f; + ah = (ah + 1 + icarry) & 0xff; + eflags |= CC_C | CC_A; + } else { + eflags &= ~(CC_C | CC_A); + al &= 0x0f; + } + EAX = (EAX & ~0xffff) | al | (ah << 8); + CC_SRC = eflags; +} + +void OPPROTO op_aas(void) +{ + int icarry; + int al, ah, af; + int eflags; + + eflags = cc_table[CC_OP].compute_all(); + af = eflags & CC_A; + al = EAX & 0xff; + ah = (EAX >> 8) & 0xff; + + icarry = (al < 6); + if (((al & 0x0f) > 9 ) || af) { + al = (al - 6) & 0x0f; + ah = (ah - 1 - icarry) & 0xff; + eflags |= CC_C | CC_A; + } else { + eflags &= ~(CC_C | CC_A); + al &= 0x0f; + } + EAX = (EAX & ~0xffff) | al | (ah << 8); + CC_SRC = eflags; +} + +void OPPROTO op_daa(void) +{ + int al, af, cf; + int eflags; + + eflags = cc_table[CC_OP].compute_all(); + cf = eflags & CC_C; + af = eflags & CC_A; + al = EAX & 0xff; + + eflags = 0; + if (((al & 0x0f) > 9 ) || af) { + al = (al + 6) & 0xff; + eflags |= CC_A; + } + if ((al > 0x9f) || cf) { + al = (al + 0x60) & 0xff; + eflags |= CC_C; + } + EAX = (EAX & ~0xff) | al; + /* well, speed is not an issue here, so we compute the flags by hand */ + eflags |= (al == 0) << 6; /* zf */ + eflags |= parity_table[al]; /* pf */ + eflags |= (al & 0x80); /* sf */ + CC_SRC = eflags; +} + +void OPPROTO op_das(void) +{ + int al, al1, af, cf; + int eflags; + + eflags = cc_table[CC_OP].compute_all(); + cf = eflags & CC_C; + af = eflags & CC_A; + al = EAX & 0xff; + + eflags = 0; + al1 = al; + if (((al & 0x0f) > 9 ) || af) { + eflags |= CC_A; + if (al < 6 || cf) + eflags |= CC_C; + al = (al - 6) & 0xff; + } + if ((al1 > 0x99) || cf) { + al = (al - 0x60) & 0xff; + eflags |= CC_C; + } + EAX = (EAX & ~0xff) | al; + /* well, speed is not an issue here, so we compute the flags by hand */ + eflags |= (al == 0) << 6; /* zf */ + eflags |= parity_table[al]; /* pf */ + eflags |= (al & 0x80); /* sf */ + CC_SRC = eflags; +} + +/* segment handling */ + +/* never use it with R_CS */ +void OPPROTO op_movl_seg_T0(void) +{ + load_seg(PARAM1, T0 & 0xffff, PARAM2); +} + +/* faster VM86 version */ +void OPPROTO op_movl_seg_T0_vm(void) +{ + int selector; + SegmentCache *sc; + + selector = T0 & 0xffff; + /* env->segs[] access */ + sc = (SegmentCache *)((char *)env + PARAM1); + sc->selector = selector; + sc->base = (void *)(selector << 4); +} + +void OPPROTO op_movl_T0_seg(void) +{ + T0 = env->segs[PARAM1].selector; +} + +void OPPROTO op_movl_A0_seg(void) +{ + A0 = *(unsigned long *)((char *)env + PARAM1); +} + +void OPPROTO op_addl_A0_seg(void) +{ + A0 += *(unsigned long *)((char *)env + PARAM1); +} + +void OPPROTO op_lsl(void) +{ + helper_lsl(); +} + +void OPPROTO op_lar(void) +{ + helper_lar(); +} + +/* T0: segment, T1:eip */ +void OPPROTO op_ljmp_protected_T0_T1(void) +{ + helper_ljmp_protected_T0_T1(); +} + +void OPPROTO op_lcall_real_T0_T1(void) +{ + helper_lcall_real_T0_T1(PARAM1, PARAM2); +} + +void OPPROTO op_lcall_protected_T0_T1(void) +{ + helper_lcall_protected_T0_T1(PARAM1, PARAM2); +} + +void OPPROTO op_iret_real(void) +{ + helper_iret_real(PARAM1); +} + +void OPPROTO op_iret_protected(void) +{ + helper_iret_protected(PARAM1); +} + +void OPPROTO op_lret_protected(void) +{ + helper_lret_protected(PARAM1, PARAM2); +} + +void OPPROTO op_lldt_T0(void) +{ + helper_lldt_T0(); +} + +void OPPROTO op_ltr_T0(void) +{ + helper_ltr_T0(); +} + +/* CR registers access */ +void OPPROTO op_movl_crN_T0(void) +{ + helper_movl_crN_T0(PARAM1); +} + +/* DR registers access */ +void OPPROTO op_movl_drN_T0(void) +{ + helper_movl_drN_T0(PARAM1); +} + +void OPPROTO op_lmsw_T0(void) +{ + /* only 4 lower bits of CR0 are modified */ + T0 = (env->cr[0] & ~0xf) | (T0 & 0xf); + helper_movl_crN_T0(0); +} + +void OPPROTO op_invlpg_A0(void) +{ + helper_invlpg(A0); +} + +void OPPROTO op_movl_T0_env(void) +{ + T0 = *(uint32_t *)((char *)env + PARAM1); +} + +void OPPROTO op_movl_env_T0(void) +{ + *(uint32_t *)((char *)env + PARAM1) = T0; +} + +void OPPROTO op_movl_env_T1(void) +{ + *(uint32_t *)((char *)env + PARAM1) = T1; +} + +void OPPROTO op_clts(void) +{ + env->cr[0] &= ~CR0_TS_MASK; +} + +/* flags handling */ + +/* slow jumps cases : in order to avoid calling a function with a + pointer (which can generate a stack frame on PowerPC), we use + op_setcc to set T0 and then call op_jcc. */ +void OPPROTO op_jcc(void) +{ + if (T0) + JUMP_TB(op_jcc, PARAM1, 0, PARAM2); + else + JUMP_TB(op_jcc, PARAM1, 1, PARAM3); + FORCE_RET(); +} + +void OPPROTO op_jcc_im(void) +{ + if (T0) + EIP = PARAM1; + else + EIP = PARAM2; + FORCE_RET(); +} + +/* slow set cases (compute x86 flags) */ +void OPPROTO op_seto_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = (eflags >> 11) & 1; +} + +void OPPROTO op_setb_T0_cc(void) +{ + T0 = cc_table[CC_OP].compute_c(); +} + +void OPPROTO op_setz_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = (eflags >> 6) & 1; +} + +void OPPROTO op_setbe_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = (eflags & (CC_Z | CC_C)) != 0; +} + +void OPPROTO op_sets_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = (eflags >> 7) & 1; +} + +void OPPROTO op_setp_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = (eflags >> 2) & 1; +} + +void OPPROTO op_setl_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = ((eflags ^ (eflags >> 4)) >> 7) & 1; +} + +void OPPROTO op_setle_T0_cc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + T0 = (((eflags ^ (eflags >> 4)) & 0x80) || (eflags & CC_Z)) != 0; +} + +void OPPROTO op_xor_T0_1(void) +{ + T0 ^= 1; +} + +void OPPROTO op_set_cc_op(void) +{ + CC_OP = PARAM1; +} + +#define FL_UPDATE_MASK16 (FL_UPDATE_MASK32 & 0xffff) + +void OPPROTO op_movl_eflags_T0(void) +{ + int eflags; + eflags = T0; + CC_SRC = eflags & (CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C); + DF = 1 - (2 * ((eflags >> 10) & 1)); + /* we also update some system flags as in user mode */ + env->eflags = (env->eflags & ~FL_UPDATE_MASK32) | + (eflags & FL_UPDATE_MASK32); +} + +void OPPROTO op_movw_eflags_T0(void) +{ + int eflags; + eflags = T0; + CC_SRC = eflags & (CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C); + DF = 1 - (2 * ((eflags >> 10) & 1)); + /* we also update some system flags as in user mode */ + env->eflags = (env->eflags & ~FL_UPDATE_MASK16) | + (eflags & FL_UPDATE_MASK16); +} + +void OPPROTO op_movl_eflags_T0_cpl0(void) +{ + load_eflags(T0, FL_UPDATE_CPL0_MASK); +} + +void OPPROTO op_movw_eflags_T0_cpl0(void) +{ + load_eflags(T0, FL_UPDATE_CPL0_MASK & 0xffff); +} + +#if 0 +/* vm86plus version */ +void OPPROTO op_movw_eflags_T0_vm(void) +{ + int eflags; + eflags = T0; + CC_SRC = eflags & (CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C); + DF = 1 - (2 * ((eflags >> 10) & 1)); + /* we also update some system flags as in user mode */ + env->eflags = (env->eflags & ~(FL_UPDATE_MASK16 | VIF_MASK)) | + (eflags & FL_UPDATE_MASK16); + if (eflags & IF_MASK) { + env->eflags |= VIF_MASK; + if (env->eflags & VIP_MASK) { + EIP = PARAM1; + raise_exception(EXCP0D_GPF); + } + } + FORCE_RET(); +} + +void OPPROTO op_movl_eflags_T0_vm(void) +{ + int eflags; + eflags = T0; + CC_SRC = eflags & (CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C); + DF = 1 - (2 * ((eflags >> 10) & 1)); + /* we also update some system flags as in user mode */ + env->eflags = (env->eflags & ~(FL_UPDATE_MASK32 | VIF_MASK)) | + (eflags & FL_UPDATE_MASK32); + if (eflags & IF_MASK) { + env->eflags |= VIF_MASK; + if (env->eflags & VIP_MASK) { + EIP = PARAM1; + raise_exception(EXCP0D_GPF); + } + } + FORCE_RET(); +} +#endif + +/* XXX: compute only O flag */ +void OPPROTO op_movb_eflags_T0(void) +{ + int of; + of = cc_table[CC_OP].compute_all() & CC_O; + CC_SRC = (T0 & (CC_S | CC_Z | CC_A | CC_P | CC_C)) | of; +} + +void OPPROTO op_movl_T0_eflags(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags |= (DF & DF_MASK); + eflags |= env->eflags & ~(VM_MASK | RF_MASK); + T0 = eflags; +} + +/* vm86plus version */ +#if 0 +void OPPROTO op_movl_T0_eflags_vm(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags |= (DF & DF_MASK); + eflags |= env->eflags & ~(VM_MASK | RF_MASK | IF_MASK); + if (env->eflags & VIF_MASK) + eflags |= IF_MASK; + T0 = eflags; +} +#endif + +void OPPROTO op_cld(void) +{ + DF = 1; +} + +void OPPROTO op_std(void) +{ + DF = -1; +} + +void OPPROTO op_clc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags &= ~CC_C; + CC_SRC = eflags; +} + +void OPPROTO op_stc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags |= CC_C; + CC_SRC = eflags; +} + +void OPPROTO op_cmc(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags ^= CC_C; + CC_SRC = eflags; +} + +void OPPROTO op_salc(void) +{ + int cf; + cf = cc_table[CC_OP].compute_c(); + EAX = (EAX & ~0xff) | ((-cf) & 0xff); +} + +static int compute_all_eflags(void) +{ + return CC_SRC; +} + +static int compute_c_eflags(void) +{ + return CC_SRC & CC_C; +} + +static int compute_c_mul(void) +{ + int cf; + cf = (CC_SRC != 0); + return cf; +} + +static int compute_all_mul(void) +{ + int cf, pf, af, zf, sf, of; + cf = (CC_SRC != 0); + pf = 0; /* undefined */ + af = 0; /* undefined */ + zf = 0; /* undefined */ + sf = 0; /* undefined */ + of = cf << 11; + return cf | pf | af | zf | sf | of; +} + +CCTable cc_table[CC_OP_NB] = { + [CC_OP_DYNAMIC] = { /* should never happen */ }, + + [CC_OP_EFLAGS] = { compute_all_eflags, compute_c_eflags }, + + [CC_OP_MUL] = { compute_all_mul, compute_c_mul }, + + [CC_OP_ADDB] = { compute_all_addb, compute_c_addb }, + [CC_OP_ADDW] = { compute_all_addw, compute_c_addw }, + [CC_OP_ADDL] = { compute_all_addl, compute_c_addl }, + + [CC_OP_ADCB] = { compute_all_adcb, compute_c_adcb }, + [CC_OP_ADCW] = { compute_all_adcw, compute_c_adcw }, + [CC_OP_ADCL] = { compute_all_adcl, compute_c_adcl }, + + [CC_OP_SUBB] = { compute_all_subb, compute_c_subb }, + [CC_OP_SUBW] = { compute_all_subw, compute_c_subw }, + [CC_OP_SUBL] = { compute_all_subl, compute_c_subl }, + + [CC_OP_SBBB] = { compute_all_sbbb, compute_c_sbbb }, + [CC_OP_SBBW] = { compute_all_sbbw, compute_c_sbbw }, + [CC_OP_SBBL] = { compute_all_sbbl, compute_c_sbbl }, + + [CC_OP_LOGICB] = { compute_all_logicb, compute_c_logicb }, + [CC_OP_LOGICW] = { compute_all_logicw, compute_c_logicw }, + [CC_OP_LOGICL] = { compute_all_logicl, compute_c_logicl }, + + [CC_OP_INCB] = { compute_all_incb, compute_c_incl }, + [CC_OP_INCW] = { compute_all_incw, compute_c_incl }, + [CC_OP_INCL] = { compute_all_incl, compute_c_incl }, + + [CC_OP_DECB] = { compute_all_decb, compute_c_incl }, + [CC_OP_DECW] = { compute_all_decw, compute_c_incl }, + [CC_OP_DECL] = { compute_all_decl, compute_c_incl }, + + [CC_OP_SHLB] = { compute_all_shlb, compute_c_shlb }, + [CC_OP_SHLW] = { compute_all_shlw, compute_c_shlw }, + [CC_OP_SHLL] = { compute_all_shll, compute_c_shll }, + + [CC_OP_SARB] = { compute_all_sarb, compute_c_sarl }, + [CC_OP_SARW] = { compute_all_sarw, compute_c_sarl }, + [CC_OP_SARL] = { compute_all_sarl, compute_c_sarl }, +}; + +/* floating point support. Some of the code for complicated x87 + functions comes from the LGPL'ed x86 emulator found in the Willows + TWIN windows emulator. */ + +#if defined(__powerpc__) +extern CPU86_LDouble copysign(CPU86_LDouble, CPU86_LDouble); + +/* correct (but slow) PowerPC rint() (glibc version is incorrect) */ +double qemu_rint(double x) +{ + double y = 4503599627370496.0; + if (fabs(x) >= y) + return x; + if (x < 0) + y = -y; + y = (x + y) - y; + if (y == 0.0) + y = copysign(y, x); + return y; +} + +#define rint qemu_rint +#endif + +/* fp load FT0 */ + +void OPPROTO op_flds_FT0_A0(void) +{ +#ifdef USE_FP_CONVERT + FP_CONVERT.i32 = ldl((void *)A0); + FT0 = FP_CONVERT.f; +#else + FT0 = ldfl((void *)A0); +#endif +} + +void OPPROTO op_fldl_FT0_A0(void) +{ +#ifdef USE_FP_CONVERT + FP_CONVERT.i64 = ldq((void *)A0); + FT0 = FP_CONVERT.d; +#else + FT0 = ldfq((void *)A0); +#endif +} + +/* helpers are needed to avoid static constant reference. XXX: find a better way */ +#ifdef USE_INT_TO_FLOAT_HELPERS + +void helper_fild_FT0_A0(void) +{ + FT0 = (CPU86_LDouble)ldsw((void *)A0); +} + +void helper_fildl_FT0_A0(void) +{ + FT0 = (CPU86_LDouble)((int32_t)ldl((void *)A0)); +} + +void helper_fildll_FT0_A0(void) +{ + FT0 = (CPU86_LDouble)((int64_t)ldq((void *)A0)); +} + +void OPPROTO op_fild_FT0_A0(void) +{ + helper_fild_FT0_A0(); +} + +void OPPROTO op_fildl_FT0_A0(void) +{ + helper_fildl_FT0_A0(); +} + +void OPPROTO op_fildll_FT0_A0(void) +{ + helper_fildll_FT0_A0(); +} + +#else + +void OPPROTO op_fild_FT0_A0(void) +{ +#ifdef USE_FP_CONVERT + FP_CONVERT.i32 = ldsw((void *)A0); + FT0 = (CPU86_LDouble)FP_CONVERT.i32; +#else + FT0 = (CPU86_LDouble)ldsw((void *)A0); +#endif +} + +void OPPROTO op_fildl_FT0_A0(void) +{ +#ifdef USE_FP_CONVERT + FP_CONVERT.i32 = (int32_t) ldl((void *)A0); + FT0 = (CPU86_LDouble)FP_CONVERT.i32; +#else + FT0 = (CPU86_LDouble)((int32_t)ldl((void *)A0)); +#endif +} + +void OPPROTO op_fildll_FT0_A0(void) +{ +#ifdef USE_FP_CONVERT + FP_CONVERT.i64 = (int64_t) ldq((void *)A0); + FT0 = (CPU86_LDouble)FP_CONVERT.i64; +#else + FT0 = (CPU86_LDouble)((int64_t)ldq((void *)A0)); +#endif +} +#endif + +/* fp load ST0 */ + +void OPPROTO op_flds_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; +#ifdef USE_FP_CONVERT + FP_CONVERT.i32 = ldl((void *)A0); + env->fpregs[new_fpstt] = FP_CONVERT.f; +#else + env->fpregs[new_fpstt] = ldfl((void *)A0); +#endif + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void OPPROTO op_fldl_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; +#ifdef USE_FP_CONVERT + FP_CONVERT.i64 = ldq((void *)A0); + env->fpregs[new_fpstt] = FP_CONVERT.d; +#else + env->fpregs[new_fpstt] = ldfq((void *)A0); +#endif + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +#ifdef USE_X86LDOUBLE +void OPPROTO op_fldt_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; + env->fpregs[new_fpstt] = *(long double *)A0; + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} +#else +void OPPROTO op_fldt_ST0_A0(void) +{ + helper_fldt_ST0_A0(); +} +#endif + +/* helpers are needed to avoid static constant reference. XXX: find a better way */ +#ifdef USE_INT_TO_FLOAT_HELPERS + +void helper_fild_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; + env->fpregs[new_fpstt] = (CPU86_LDouble)ldsw((void *)A0); + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void helper_fildl_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; + env->fpregs[new_fpstt] = (CPU86_LDouble)((int32_t)ldl((void *)A0)); + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void helper_fildll_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; + env->fpregs[new_fpstt] = (CPU86_LDouble)((int64_t)ldq((void *)A0)); + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void OPPROTO op_fild_ST0_A0(void) +{ + helper_fild_ST0_A0(); +} + +void OPPROTO op_fildl_ST0_A0(void) +{ + helper_fildl_ST0_A0(); +} + +void OPPROTO op_fildll_ST0_A0(void) +{ + helper_fildll_ST0_A0(); +} + +#else + +void OPPROTO op_fild_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; +#ifdef USE_FP_CONVERT + FP_CONVERT.i32 = ldsw((void *)A0); + env->fpregs[new_fpstt] = (CPU86_LDouble)FP_CONVERT.i32; +#else + env->fpregs[new_fpstt] = (CPU86_LDouble)ldsw((void *)A0); +#endif + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void OPPROTO op_fildl_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; +#ifdef USE_FP_CONVERT + FP_CONVERT.i32 = (int32_t) ldl((void *)A0); + env->fpregs[new_fpstt] = (CPU86_LDouble)FP_CONVERT.i32; +#else + env->fpregs[new_fpstt] = (CPU86_LDouble)((int32_t)ldl((void *)A0)); +#endif + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +void OPPROTO op_fildll_ST0_A0(void) +{ + int new_fpstt; + new_fpstt = (env->fpstt - 1) & 7; +#ifdef USE_FP_CONVERT + FP_CONVERT.i64 = (int64_t) ldq((void *)A0); + env->fpregs[new_fpstt] = (CPU86_LDouble)FP_CONVERT.i64; +#else + env->fpregs[new_fpstt] = (CPU86_LDouble)((int64_t)ldq((void *)A0)); +#endif + env->fpstt = new_fpstt; + env->fptags[new_fpstt] = 0; /* validate stack entry */ +} + +#endif + +/* fp store */ + +void OPPROTO op_fsts_ST0_A0(void) +{ +#ifdef USE_FP_CONVERT + FP_CONVERT.f = (float)ST0; + stfl((void *)A0, FP_CONVERT.f); +#else + stfl((void *)A0, (float)ST0); +#endif +} + +void OPPROTO op_fstl_ST0_A0(void) +{ + stfq((void *)A0, (double)ST0); +} + +#ifdef USE_X86LDOUBLE +void OPPROTO op_fstt_ST0_A0(void) +{ + *(long double *)A0 = ST0; +} +#else +void OPPROTO op_fstt_ST0_A0(void) +{ + helper_fstt_ST0_A0(); +} +#endif + +void OPPROTO op_fist_ST0_A0(void) +{ +#if defined(__sparc__) && !defined(__sparc_v9__) + register CPU86_LDouble d asm("o0"); +#else + CPU86_LDouble d; +#endif + int val; + + d = ST0; + val = lrint(d); + if (val != (int16_t)val) + val = -32768; + stw((void *)A0, val); +} + +void OPPROTO op_fistl_ST0_A0(void) +{ +#if defined(__sparc__) && !defined(__sparc_v9__) + register CPU86_LDouble d asm("o0"); +#else + CPU86_LDouble d; +#endif + int val; + + d = ST0; + val = lrint(d); + stl((void *)A0, val); +} + +void OPPROTO op_fistll_ST0_A0(void) +{ +#if defined(__sparc__) && !defined(__sparc_v9__) + register CPU86_LDouble d asm("o0"); +#else + CPU86_LDouble d; +#endif + int64_t val; + + d = ST0; + val = llrint(d); + stq((void *)A0, val); +} + +void OPPROTO op_fbld_ST0_A0(void) +{ + helper_fbld_ST0_A0(); +} + +void OPPROTO op_fbst_ST0_A0(void) +{ + helper_fbst_ST0_A0(); +} + +/* FPU move */ + +void OPPROTO op_fpush(void) +{ + fpush(); +} + +void OPPROTO op_fpop(void) +{ + fpop(); +} + +void OPPROTO op_fdecstp(void) +{ + env->fpstt = (env->fpstt - 1) & 7; + env->fpus &= (~0x4700); +} + +void OPPROTO op_fincstp(void) +{ + env->fpstt = (env->fpstt + 1) & 7; + env->fpus &= (~0x4700); +} + +void OPPROTO op_fmov_ST0_FT0(void) +{ + ST0 = FT0; +} + +void OPPROTO op_fmov_FT0_STN(void) +{ + FT0 = ST(PARAM1); +} + +void OPPROTO op_fmov_ST0_STN(void) +{ + ST0 = ST(PARAM1); +} + +void OPPROTO op_fmov_STN_ST0(void) +{ + ST(PARAM1) = ST0; +} + +void OPPROTO op_fxchg_ST0_STN(void) +{ + CPU86_LDouble tmp; + tmp = ST(PARAM1); + ST(PARAM1) = ST0; + ST0 = tmp; +} + +/* FPU operations */ + +/* XXX: handle nans */ +void OPPROTO op_fcom_ST0_FT0(void) +{ + env->fpus &= (~0x4500); /* (C3,C2,C0) <-- 000 */ + if (ST0 < FT0) + env->fpus |= 0x100; /* (C3,C2,C0) <-- 001 */ + else if (ST0 == FT0) + env->fpus |= 0x4000; /* (C3,C2,C0) <-- 100 */ + FORCE_RET(); +} + +/* XXX: handle nans */ +void OPPROTO op_fucom_ST0_FT0(void) +{ + env->fpus &= (~0x4500); /* (C3,C2,C0) <-- 000 */ + if (ST0 < FT0) + env->fpus |= 0x100; /* (C3,C2,C0) <-- 001 */ + else if (ST0 == FT0) + env->fpus |= 0x4000; /* (C3,C2,C0) <-- 100 */ + FORCE_RET(); +} + +/* XXX: handle nans */ +void OPPROTO op_fcomi_ST0_FT0(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags &= ~(CC_Z | CC_P | CC_C); + if (ST0 < FT0) + eflags |= CC_C; + else if (ST0 == FT0) + eflags |= CC_Z; + CC_SRC = eflags; + FORCE_RET(); +} + +/* XXX: handle nans */ +void OPPROTO op_fucomi_ST0_FT0(void) +{ + int eflags; + eflags = cc_table[CC_OP].compute_all(); + eflags &= ~(CC_Z | CC_P | CC_C); + if (ST0 < FT0) + eflags |= CC_C; + else if (ST0 == FT0) + eflags |= CC_Z; + CC_SRC = eflags; + FORCE_RET(); +} + +void OPPROTO op_fadd_ST0_FT0(void) +{ + ST0 += FT0; +} + +void OPPROTO op_fmul_ST0_FT0(void) +{ + ST0 *= FT0; +} + +void OPPROTO op_fsub_ST0_FT0(void) +{ + ST0 -= FT0; +} + +void OPPROTO op_fsubr_ST0_FT0(void) +{ + ST0 = FT0 - ST0; +} + +void OPPROTO op_fdiv_ST0_FT0(void) +{ + ST0 /= FT0; +} + +void OPPROTO op_fdivr_ST0_FT0(void) +{ + ST0 = FT0 / ST0; +} + +/* fp operations between STN and ST0 */ + +void OPPROTO op_fadd_STN_ST0(void) +{ + ST(PARAM1) += ST0; +} + +void OPPROTO op_fmul_STN_ST0(void) +{ + ST(PARAM1) *= ST0; +} + +void OPPROTO op_fsub_STN_ST0(void) +{ + ST(PARAM1) -= ST0; +} + +void OPPROTO op_fsubr_STN_ST0(void) +{ + CPU86_LDouble *p; + p = &ST(PARAM1); + *p = ST0 - *p; +} + +void OPPROTO op_fdiv_STN_ST0(void) +{ + ST(PARAM1) /= ST0; +} + +void OPPROTO op_fdivr_STN_ST0(void) +{ + CPU86_LDouble *p; + p = &ST(PARAM1); + *p = ST0 / *p; +} + +/* misc FPU operations */ +void OPPROTO op_fchs_ST0(void) +{ + ST0 = -ST0; +} + +void OPPROTO op_fabs_ST0(void) +{ + ST0 = fabs(ST0); +} + +void OPPROTO op_fxam_ST0(void) +{ + helper_fxam_ST0(); +} + +void OPPROTO op_fld1_ST0(void) +{ + ST0 = f15rk[1]; +} + +void OPPROTO op_fldl2t_ST0(void) +{ + ST0 = f15rk[6]; +} + +void OPPROTO op_fldl2e_ST0(void) +{ + ST0 = f15rk[5]; +} + +void OPPROTO op_fldpi_ST0(void) +{ + ST0 = f15rk[2]; +} + +void OPPROTO op_fldlg2_ST0(void) +{ + ST0 = f15rk[3]; +} + +void OPPROTO op_fldln2_ST0(void) +{ + ST0 = f15rk[4]; +} + +void OPPROTO op_fldz_ST0(void) +{ + ST0 = f15rk[0]; +} + +void OPPROTO op_fldz_FT0(void) +{ + ST0 = f15rk[0]; +} + +/* associated heplers to reduce generated code length and to simplify + relocation (FP constants are usually stored in .rodata section) */ + +void OPPROTO op_f2xm1(void) +{ + helper_f2xm1(); +} + +void OPPROTO op_fyl2x(void) +{ + helper_fyl2x(); +} + +void OPPROTO op_fptan(void) +{ + helper_fptan(); +} + +void OPPROTO op_fpatan(void) +{ + helper_fpatan(); +} + +void OPPROTO op_fxtract(void) +{ + helper_fxtract(); +} + +void OPPROTO op_fprem1(void) +{ + helper_fprem1(); +} + + +void OPPROTO op_fprem(void) +{ + helper_fprem(); +} + +void OPPROTO op_fyl2xp1(void) +{ + helper_fyl2xp1(); +} + +void OPPROTO op_fsqrt(void) +{ + helper_fsqrt(); +} + +void OPPROTO op_fsincos(void) +{ + helper_fsincos(); +} + +void OPPROTO op_frndint(void) +{ + helper_frndint(); +} + +void OPPROTO op_fscale(void) +{ + helper_fscale(); +} + +void OPPROTO op_fsin(void) +{ + helper_fsin(); +} + +void OPPROTO op_fcos(void) +{ + helper_fcos(); +} + +void OPPROTO op_fnstsw_A0(void) +{ + int fpus; + fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11; + stw((void *)A0, fpus); +} + +void OPPROTO op_fnstsw_EAX(void) +{ + int fpus; + fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11; + EAX = (EAX & 0xffff0000) | fpus; +} + +void OPPROTO op_fnstcw_A0(void) +{ + stw((void *)A0, env->fpuc); +} + +void OPPROTO op_fldcw_A0(void) +{ + int rnd_type; + env->fpuc = lduw((void *)A0); + /* set rounding mode */ + switch(env->fpuc & RC_MASK) { + default: + case RC_NEAR: + rnd_type = FE_TONEAREST; + break; + case RC_DOWN: + rnd_type = FE_DOWNWARD; + break; + case RC_UP: + rnd_type = FE_UPWARD; + break; + case RC_CHOP: + rnd_type = FE_TOWARDZERO; + break; + } + fesetround(rnd_type); +} + +void OPPROTO op_fclex(void) +{ + env->fpus &= 0x7f00; +} + +void OPPROTO op_fninit(void) +{ + env->fpus = 0; + env->fpstt = 0; + env->fpuc = 0x37f; + env->fptags[0] = 1; + env->fptags[1] = 1; + env->fptags[2] = 1; + env->fptags[3] = 1; + env->fptags[4] = 1; + env->fptags[5] = 1; + env->fptags[6] = 1; + env->fptags[7] = 1; +} + +void OPPROTO op_fnstenv_A0(void) +{ + helper_fstenv((uint8_t *)A0, PARAM1); +} + +void OPPROTO op_fldenv_A0(void) +{ + helper_fldenv((uint8_t *)A0, PARAM1); +} + +void OPPROTO op_fnsave_A0(void) +{ + helper_fsave((uint8_t *)A0, PARAM1); +} + +void OPPROTO op_frstor_A0(void) +{ + helper_frstor((uint8_t *)A0, PARAM1); +} + +/* threading support */ +void OPPROTO op_lock(void) +{ + cpu_lock(); +} + +void OPPROTO op_unlock(void) +{ + cpu_unlock(); +} + diff --git a/target-i386/opreg_template.h b/target-i386/opreg_template.h new file mode 100644 index 0000000000..f35a1bbc95 --- /dev/null +++ b/target-i386/opreg_template.h @@ -0,0 +1,134 @@ +/* + * i386 micro operations (templates for various register related + * operations) + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +void OPPROTO glue(op_movl_A0,REGNAME)(void) +{ + A0 = REG; +} + +void OPPROTO glue(op_addl_A0,REGNAME)(void) +{ + A0 += REG; +} + +void OPPROTO glue(glue(op_addl_A0,REGNAME),_s1)(void) +{ + A0 += REG << 1; +} + +void OPPROTO glue(glue(op_addl_A0,REGNAME),_s2)(void) +{ + A0 += REG << 2; +} + +void OPPROTO glue(glue(op_addl_A0,REGNAME),_s3)(void) +{ + A0 += REG << 3; +} + +void OPPROTO glue(op_movl_T0,REGNAME)(void) +{ + T0 = REG; +} + +void OPPROTO glue(op_movl_T1,REGNAME)(void) +{ + T1 = REG; +} + +void OPPROTO glue(op_movh_T0,REGNAME)(void) +{ + T0 = REG >> 8; +} + +void OPPROTO glue(op_movh_T1,REGNAME)(void) +{ + T1 = REG >> 8; +} + +void OPPROTO glue(glue(op_movl,REGNAME),_T0)(void) +{ + REG = T0; +} + +void OPPROTO glue(glue(op_movl,REGNAME),_T1)(void) +{ + REG = T1; +} + +void OPPROTO glue(glue(op_movl,REGNAME),_A0)(void) +{ + REG = A0; +} + +/* mov T1 to REG if T0 is true */ +void OPPROTO glue(glue(op_cmovw,REGNAME),_T1_T0)(void) +{ + if (T0) + REG = (REG & 0xffff0000) | (T1 & 0xffff); +} + +void OPPROTO glue(glue(op_cmovl,REGNAME),_T1_T0)(void) +{ + if (T0) + REG = T1; +} + +/* NOTE: T0 high order bits are ignored */ +void OPPROTO glue(glue(op_movw,REGNAME),_T0)(void) +{ + REG = (REG & 0xffff0000) | (T0 & 0xffff); +} + +/* NOTE: T0 high order bits are ignored */ +void OPPROTO glue(glue(op_movw,REGNAME),_T1)(void) +{ + REG = (REG & 0xffff0000) | (T1 & 0xffff); +} + +/* NOTE: A0 high order bits are ignored */ +void OPPROTO glue(glue(op_movw,REGNAME),_A0)(void) +{ + REG = (REG & 0xffff0000) | (A0 & 0xffff); +} + +/* NOTE: T0 high order bits are ignored */ +void OPPROTO glue(glue(op_movb,REGNAME),_T0)(void) +{ + REG = (REG & 0xffffff00) | (T0 & 0xff); +} + +/* NOTE: T0 high order bits are ignored */ +void OPPROTO glue(glue(op_movh,REGNAME),_T0)(void) +{ + REG = (REG & 0xffff00ff) | ((T0 & 0xff) << 8); +} + +/* NOTE: T1 high order bits are ignored */ +void OPPROTO glue(glue(op_movb,REGNAME),_T1)(void) +{ + REG = (REG & 0xffffff00) | (T1 & 0xff); +} + +/* NOTE: T1 high order bits are ignored */ +void OPPROTO glue(glue(op_movh,REGNAME),_T1)(void) +{ + REG = (REG & 0xffff00ff) | ((T1 & 0xff) << 8); +} diff --git a/target-i386/ops_mem.h b/target-i386/ops_mem.h new file mode 100644 index 0000000000..e1c8277bbc --- /dev/null +++ b/target-i386/ops_mem.h @@ -0,0 +1,66 @@ +void OPPROTO glue(glue(op_ldub, MEMSUFFIX), _T0_A0)(void) +{ + T0 = glue(ldub, MEMSUFFIX)((uint8_t *)A0); +} + +void OPPROTO glue(glue(op_ldsb, MEMSUFFIX), _T0_A0)(void) +{ + T0 = glue(ldsb, MEMSUFFIX)((int8_t *)A0); +} + +void OPPROTO glue(glue(op_lduw, MEMSUFFIX), _T0_A0)(void) +{ + T0 = glue(lduw, MEMSUFFIX)((uint8_t *)A0); +} + +void OPPROTO glue(glue(op_ldsw, MEMSUFFIX), _T0_A0)(void) +{ + T0 = glue(ldsw, MEMSUFFIX)((int8_t *)A0); +} + +void OPPROTO glue(glue(op_ldl, MEMSUFFIX), _T0_A0)(void) +{ + T0 = glue(ldl, MEMSUFFIX)((uint8_t *)A0); +} + +void OPPROTO glue(glue(op_ldub, MEMSUFFIX), _T1_A0)(void) +{ + T1 = glue(ldub, MEMSUFFIX)((uint8_t *)A0); +} + +void OPPROTO glue(glue(op_ldsb, MEMSUFFIX), _T1_A0)(void) +{ + T1 = glue(ldsb, MEMSUFFIX)((int8_t *)A0); +} + +void OPPROTO glue(glue(op_lduw, MEMSUFFIX), _T1_A0)(void) +{ + T1 = glue(lduw, MEMSUFFIX)((uint8_t *)A0); +} + +void OPPROTO glue(glue(op_ldsw, MEMSUFFIX), _T1_A0)(void) +{ + T1 = glue(ldsw, MEMSUFFIX)((int8_t *)A0); +} + +void OPPROTO glue(glue(op_ldl, MEMSUFFIX), _T1_A0)(void) +{ + T1 = glue(ldl, MEMSUFFIX)((uint8_t *)A0); +} + +void OPPROTO glue(glue(op_stb, MEMSUFFIX), _T0_A0)(void) +{ + glue(stb, MEMSUFFIX)((uint8_t *)A0, T0); +} + +void OPPROTO glue(glue(op_stw, MEMSUFFIX), _T0_A0)(void) +{ + glue(stw, MEMSUFFIX)((uint8_t *)A0, T0); +} + +void OPPROTO glue(glue(op_stl, MEMSUFFIX), _T0_A0)(void) +{ + glue(stl, MEMSUFFIX)((uint8_t *)A0, T0); +} + +#undef MEMSUFFIX diff --git a/target-i386/ops_template.h b/target-i386/ops_template.h new file mode 100644 index 0000000000..182296743b --- /dev/null +++ b/target-i386/ops_template.h @@ -0,0 +1,617 @@ +/* + * i386 micro operations (included several times to generate + * different operand sizes) + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#define DATA_BITS (1 << (3 + SHIFT)) +#define SHIFT_MASK (DATA_BITS - 1) +#define SIGN_MASK (1 << (DATA_BITS - 1)) + +#if DATA_BITS == 8 +#define SUFFIX b +#define DATA_TYPE uint8_t +#define DATA_STYPE int8_t +#define DATA_MASK 0xff +#elif DATA_BITS == 16 +#define SUFFIX w +#define DATA_TYPE uint16_t +#define DATA_STYPE int16_t +#define DATA_MASK 0xffff +#elif DATA_BITS == 32 +#define SUFFIX l +#define DATA_TYPE uint32_t +#define DATA_STYPE int32_t +#define DATA_MASK 0xffffffff +#else +#error unhandled operand size +#endif + +/* dynamic flags computation */ + +static int glue(compute_all_add, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + int src1, src2; + src1 = CC_SRC; + src2 = CC_DST - CC_SRC; + cf = (DATA_TYPE)CC_DST < (DATA_TYPE)src1; + pf = parity_table[(uint8_t)CC_DST]; + af = (CC_DST ^ src1 ^ src2) & 0x10; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = lshift((src1 ^ src2 ^ -1) & (src1 ^ CC_DST), 12 - DATA_BITS) & CC_O; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_c_add, SUFFIX)(void) +{ + int src1, cf; + src1 = CC_SRC; + cf = (DATA_TYPE)CC_DST < (DATA_TYPE)src1; + return cf; +} + +static int glue(compute_all_adc, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + int src1, src2; + src1 = CC_SRC; + src2 = CC_DST - CC_SRC - 1; + cf = (DATA_TYPE)CC_DST <= (DATA_TYPE)src1; + pf = parity_table[(uint8_t)CC_DST]; + af = (CC_DST ^ src1 ^ src2) & 0x10; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = lshift((src1 ^ src2 ^ -1) & (src1 ^ CC_DST), 12 - DATA_BITS) & CC_O; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_c_adc, SUFFIX)(void) +{ + int src1, cf; + src1 = CC_SRC; + cf = (DATA_TYPE)CC_DST <= (DATA_TYPE)src1; + return cf; +} + +static int glue(compute_all_sub, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + cf = (DATA_TYPE)src1 < (DATA_TYPE)src2; + pf = parity_table[(uint8_t)CC_DST]; + af = (CC_DST ^ src1 ^ src2) & 0x10; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = lshift((src1 ^ src2) & (src1 ^ CC_DST), 12 - DATA_BITS) & CC_O; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_c_sub, SUFFIX)(void) +{ + int src1, src2, cf; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + cf = (DATA_TYPE)src1 < (DATA_TYPE)src2; + return cf; +} + +static int glue(compute_all_sbb, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + int src1, src2; + src1 = CC_DST + CC_SRC + 1; + src2 = CC_SRC; + cf = (DATA_TYPE)src1 <= (DATA_TYPE)src2; + pf = parity_table[(uint8_t)CC_DST]; + af = (CC_DST ^ src1 ^ src2) & 0x10; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = lshift((src1 ^ src2) & (src1 ^ CC_DST), 12 - DATA_BITS) & CC_O; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_c_sbb, SUFFIX)(void) +{ + int src1, src2, cf; + src1 = CC_DST + CC_SRC + 1; + src2 = CC_SRC; + cf = (DATA_TYPE)src1 <= (DATA_TYPE)src2; + return cf; +} + +static int glue(compute_all_logic, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + cf = 0; + pf = parity_table[(uint8_t)CC_DST]; + af = 0; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = 0; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_c_logic, SUFFIX)(void) +{ + return 0; +} + +static int glue(compute_all_inc, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + int src1, src2; + src1 = CC_DST - 1; + src2 = 1; + cf = CC_SRC; + pf = parity_table[(uint8_t)CC_DST]; + af = (CC_DST ^ src1 ^ src2) & 0x10; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = ((CC_DST & DATA_MASK) == SIGN_MASK) << 11; + return cf | pf | af | zf | sf | of; +} + +#if DATA_BITS == 32 +static int glue(compute_c_inc, SUFFIX)(void) +{ + return CC_SRC; +} +#endif + +static int glue(compute_all_dec, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + int src1, src2; + src1 = CC_DST + 1; + src2 = 1; + cf = CC_SRC; + pf = parity_table[(uint8_t)CC_DST]; + af = (CC_DST ^ src1 ^ src2) & 0x10; + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + of = ((CC_DST & DATA_MASK) == ((uint32_t)SIGN_MASK - 1)) << 11; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_all_shl, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + cf = (CC_SRC >> (DATA_BITS - 1)) & CC_C; + pf = parity_table[(uint8_t)CC_DST]; + af = 0; /* undefined */ + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + /* of is defined if shift count == 1 */ + of = lshift(CC_SRC ^ CC_DST, 12 - DATA_BITS) & CC_O; + return cf | pf | af | zf | sf | of; +} + +static int glue(compute_c_shl, SUFFIX)(void) +{ + return (CC_SRC >> (DATA_BITS - 1)) & CC_C; +} + +#if DATA_BITS == 32 +static int glue(compute_c_sar, SUFFIX)(void) +{ + return CC_SRC & 1; +} +#endif + +static int glue(compute_all_sar, SUFFIX)(void) +{ + int cf, pf, af, zf, sf, of; + cf = CC_SRC & 1; + pf = parity_table[(uint8_t)CC_DST]; + af = 0; /* undefined */ + zf = ((DATA_TYPE)CC_DST == 0) << 6; + sf = lshift(CC_DST, 8 - DATA_BITS) & 0x80; + /* of is defined if shift count == 1 */ + of = lshift(CC_SRC ^ CC_DST, 12 - DATA_BITS) & CC_O; + return cf | pf | af | zf | sf | of; +} + +/* various optimized jumps cases */ + +void OPPROTO glue(op_jb_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + if ((DATA_TYPE)src1 < (DATA_TYPE)src2) + JUMP_TB(glue(op_jb_sub, SUFFIX), PARAM1, 0, PARAM2); + else + JUMP_TB(glue(op_jb_sub, SUFFIX), PARAM1, 1, PARAM3); + FORCE_RET(); +} + +void OPPROTO glue(op_jz_sub, SUFFIX)(void) +{ + if ((DATA_TYPE)CC_DST == 0) + JUMP_TB(glue(op_jz_sub, SUFFIX), PARAM1, 0, PARAM2); + else + JUMP_TB(glue(op_jz_sub, SUFFIX), PARAM1, 1, PARAM3); + FORCE_RET(); +} + +void OPPROTO glue(op_jbe_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + if ((DATA_TYPE)src1 <= (DATA_TYPE)src2) + JUMP_TB(glue(op_jbe_sub, SUFFIX), PARAM1, 0, PARAM2); + else + JUMP_TB(glue(op_jbe_sub, SUFFIX), PARAM1, 1, PARAM3); + FORCE_RET(); +} + +void OPPROTO glue(op_js_sub, SUFFIX)(void) +{ + if (CC_DST & SIGN_MASK) + JUMP_TB(glue(op_js_sub, SUFFIX), PARAM1, 0, PARAM2); + else + JUMP_TB(glue(op_js_sub, SUFFIX), PARAM1, 1, PARAM3); + FORCE_RET(); +} + +void OPPROTO glue(op_jl_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + if ((DATA_STYPE)src1 < (DATA_STYPE)src2) + JUMP_TB(glue(op_jl_sub, SUFFIX), PARAM1, 0, PARAM2); + else + JUMP_TB(glue(op_jl_sub, SUFFIX), PARAM1, 1, PARAM3); + FORCE_RET(); +} + +void OPPROTO glue(op_jle_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + if ((DATA_STYPE)src1 <= (DATA_STYPE)src2) + JUMP_TB(glue(op_jle_sub, SUFFIX), PARAM1, 0, PARAM2); + else + JUMP_TB(glue(op_jle_sub, SUFFIX), PARAM1, 1, PARAM3); + FORCE_RET(); +} + +/* oldies */ + +#if DATA_BITS >= 16 + +void OPPROTO glue(op_loopnz, SUFFIX)(void) +{ + unsigned int tmp; + int eflags; + eflags = cc_table[CC_OP].compute_all(); + tmp = (ECX - 1) & DATA_MASK; + ECX = (ECX & ~DATA_MASK) | tmp; + if (tmp != 0 && !(eflags & CC_Z)) + EIP = PARAM1; + else + EIP = PARAM2; + FORCE_RET(); +} + +void OPPROTO glue(op_loopz, SUFFIX)(void) +{ + unsigned int tmp; + int eflags; + eflags = cc_table[CC_OP].compute_all(); + tmp = (ECX - 1) & DATA_MASK; + ECX = (ECX & ~DATA_MASK) | tmp; + if (tmp != 0 && (eflags & CC_Z)) + EIP = PARAM1; + else + EIP = PARAM2; + FORCE_RET(); +} + +void OPPROTO glue(op_loop, SUFFIX)(void) +{ + unsigned int tmp; + tmp = (ECX - 1) & DATA_MASK; + ECX = (ECX & ~DATA_MASK) | tmp; + if (tmp != 0) + EIP = PARAM1; + else + EIP = PARAM2; + FORCE_RET(); +} + +void OPPROTO glue(op_jecxz, SUFFIX)(void) +{ + if ((DATA_TYPE)ECX == 0) + EIP = PARAM1; + else + EIP = PARAM2; + FORCE_RET(); +} + +#endif + +/* various optimized set cases */ + +void OPPROTO glue(op_setb_T0_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + T0 = ((DATA_TYPE)src1 < (DATA_TYPE)src2); +} + +void OPPROTO glue(op_setz_T0_sub, SUFFIX)(void) +{ + T0 = ((DATA_TYPE)CC_DST == 0); +} + +void OPPROTO glue(op_setbe_T0_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + T0 = ((DATA_TYPE)src1 <= (DATA_TYPE)src2); +} + +void OPPROTO glue(op_sets_T0_sub, SUFFIX)(void) +{ + T0 = lshift(CC_DST, -(DATA_BITS - 1)) & 1; +} + +void OPPROTO glue(op_setl_T0_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + T0 = ((DATA_STYPE)src1 < (DATA_STYPE)src2); +} + +void OPPROTO glue(op_setle_T0_sub, SUFFIX)(void) +{ + int src1, src2; + src1 = CC_DST + CC_SRC; + src2 = CC_SRC; + + T0 = ((DATA_STYPE)src1 <= (DATA_STYPE)src2); +} + +/* shifts */ + +void OPPROTO glue(glue(op_shl, SUFFIX), _T0_T1)(void) +{ + int count; + count = T1 & 0x1f; + T0 = T0 << count; + FORCE_RET(); +} + +void OPPROTO glue(glue(op_shr, SUFFIX), _T0_T1)(void) +{ + int count; + count = T1 & 0x1f; + T0 &= DATA_MASK; + T0 = T0 >> count; + FORCE_RET(); +} + +void OPPROTO glue(glue(op_sar, SUFFIX), _T0_T1)(void) +{ + int count, src; + count = T1 & 0x1f; + src = (DATA_STYPE)T0; + T0 = src >> count; + FORCE_RET(); +} + +#undef MEM_WRITE +#include "ops_template_mem.h" + +#define MEM_WRITE +#include "ops_template_mem.h" + +/* bit operations */ +#if DATA_BITS >= 16 + +void OPPROTO glue(glue(op_bt, SUFFIX), _T0_T1_cc)(void) +{ + int count; + count = T1 & SHIFT_MASK; + CC_SRC = T0 >> count; +} + +void OPPROTO glue(glue(op_bts, SUFFIX), _T0_T1_cc)(void) +{ + int count; + count = T1 & SHIFT_MASK; + T1 = T0 >> count; + T0 |= (1 << count); +} + +void OPPROTO glue(glue(op_btr, SUFFIX), _T0_T1_cc)(void) +{ + int count; + count = T1 & SHIFT_MASK; + T1 = T0 >> count; + T0 &= ~(1 << count); +} + +void OPPROTO glue(glue(op_btc, SUFFIX), _T0_T1_cc)(void) +{ + int count; + count = T1 & SHIFT_MASK; + T1 = T0 >> count; + T0 ^= (1 << count); +} + +void OPPROTO glue(glue(op_bsf, SUFFIX), _T0_cc)(void) +{ + int res, count; + res = T0 & DATA_MASK; + if (res != 0) { + count = 0; + while ((res & 1) == 0) { + count++; + res >>= 1; + } + T0 = count; + CC_DST = 1; /* ZF = 1 */ + } else { + CC_DST = 0; /* ZF = 1 */ + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_bsr, SUFFIX), _T0_cc)(void) +{ + int res, count; + res = T0 & DATA_MASK; + if (res != 0) { + count = DATA_BITS - 1; + while ((res & SIGN_MASK) == 0) { + count--; + res <<= 1; + } + T0 = count; + CC_DST = 1; /* ZF = 1 */ + } else { + CC_DST = 0; /* ZF = 1 */ + } + FORCE_RET(); +} + +#endif + +#if DATA_BITS == 32 +void OPPROTO op_update_bt_cc(void) +{ + CC_SRC = T1; +} +#endif + +/* string operations */ + +void OPPROTO glue(op_movl_T0_Dshift, SUFFIX)(void) +{ + T0 = DF << SHIFT; +} + +void OPPROTO glue(op_string_jz_sub, SUFFIX)(void) +{ + if ((DATA_TYPE)CC_DST == 0) + JUMP_TB2(glue(op_string_jz_sub, SUFFIX), PARAM1, 1); + FORCE_RET(); +} + +void OPPROTO glue(op_string_jnz_sub, SUFFIX)(void) +{ + if ((DATA_TYPE)CC_DST != 0) + JUMP_TB2(glue(op_string_jnz_sub, SUFFIX), PARAM1, 1); + FORCE_RET(); +} + +void OPPROTO glue(glue(op_string_jz_sub, SUFFIX), _im)(void) +{ + if ((DATA_TYPE)CC_DST == 0) { + EIP = PARAM1; + if (env->eflags & TF_MASK) { + raise_exception(EXCP01_SSTP); + } + T0 = 0; + EXIT_TB(); + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_string_jnz_sub, SUFFIX), _im)(void) +{ + if ((DATA_TYPE)CC_DST != 0) { + EIP = PARAM1; + if (env->eflags & TF_MASK) { + raise_exception(EXCP01_SSTP); + } + T0 = 0; + EXIT_TB(); + } + FORCE_RET(); +} + +#if DATA_BITS >= 16 +void OPPROTO glue(op_jz_ecx, SUFFIX)(void) +{ + if ((DATA_TYPE)ECX == 0) + JUMP_TB(glue(op_jz_ecx, SUFFIX), PARAM1, 1, PARAM2); + FORCE_RET(); +} + +void OPPROTO glue(glue(op_jz_ecx, SUFFIX), _im)(void) +{ + if ((DATA_TYPE)ECX == 0) { + EIP = PARAM1; + if (env->eflags & TF_MASK) { + raise_exception(EXCP01_SSTP); + } + T0 = 0; + EXIT_TB(); + } + FORCE_RET(); +} +#endif + +/* port I/O */ + +void OPPROTO glue(glue(op_out, SUFFIX), _T0_T1)(void) +{ + glue(cpu_x86_out, SUFFIX)(env, T0 & 0xffff, T1 & DATA_MASK); +} + +void OPPROTO glue(glue(op_in, SUFFIX), _T0_T1)(void) +{ + T1 = glue(cpu_x86_in, SUFFIX)(env, T0 & 0xffff); +} + +void OPPROTO glue(glue(op_in, SUFFIX), _DX_T0)(void) +{ + T0 = glue(cpu_x86_in, SUFFIX)(env, EDX & 0xffff); +} + +void OPPROTO glue(glue(op_out, SUFFIX), _DX_T0)(void) +{ + glue(cpu_x86_out, SUFFIX)(env, EDX & 0xffff, T0); +} + +#undef DATA_BITS +#undef SHIFT_MASK +#undef SIGN_MASK +#undef DATA_TYPE +#undef DATA_STYPE +#undef DATA_MASK +#undef SUFFIX diff --git a/target-i386/ops_template_mem.h b/target-i386/ops_template_mem.h new file mode 100644 index 0000000000..eacaabba5c --- /dev/null +++ b/target-i386/ops_template_mem.h @@ -0,0 +1,429 @@ +/* + * i386 micro operations (included several times to generate + * different operand sizes) + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#ifdef MEM_WRITE + +#if DATA_BITS == 8 +#define MEM_SUFFIX b_mem +#elif DATA_BITS == 16 +#define MEM_SUFFIX w_mem +#elif DATA_BITS == 32 +#define MEM_SUFFIX l_mem +#endif + +#else + +#define MEM_SUFFIX SUFFIX + +#endif + +void OPPROTO glue(glue(op_rol, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, src; + count = T1 & SHIFT_MASK; + if (count) { + src = T0; + T0 &= DATA_MASK; + T0 = (T0 << count) | (T0 >> (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#else + /* gcc 3.2 workaround. This is really a bug in gcc. */ + asm volatile("" : : "r" (T0)); +#endif + CC_SRC = (cc_table[CC_OP].compute_all() & ~(CC_O | CC_C)) | + (lshift(src ^ T0, 11 - (DATA_BITS - 1)) & CC_O) | + (T0 & CC_C); + CC_OP = CC_OP_EFLAGS; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_ror, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, src; + count = T1 & SHIFT_MASK; + if (count) { + src = T0; + T0 &= DATA_MASK; + T0 = (T0 >> count) | (T0 << (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#else + /* gcc 3.2 workaround. This is really a bug in gcc. */ + asm volatile("" : : "r" (T0)); +#endif + CC_SRC = (cc_table[CC_OP].compute_all() & ~(CC_O | CC_C)) | + (lshift(src ^ T0, 11 - (DATA_BITS - 1)) & CC_O) | + ((T0 >> (DATA_BITS - 1)) & CC_C); + CC_OP = CC_OP_EFLAGS; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_rol, MEM_SUFFIX), _T0_T1)(void) +{ + int count; + count = T1 & SHIFT_MASK; + if (count) { + T0 &= DATA_MASK; + T0 = (T0 << count) | (T0 >> (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_ror, MEM_SUFFIX), _T0_T1)(void) +{ + int count; + count = T1 & SHIFT_MASK; + if (count) { + T0 &= DATA_MASK; + T0 = (T0 >> count) | (T0 << (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_rcl, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, res, eflags; + unsigned int src; + + count = T1 & 0x1f; +#if DATA_BITS == 16 + count = rclw_table[count]; +#elif DATA_BITS == 8 + count = rclb_table[count]; +#endif + if (count) { + eflags = cc_table[CC_OP].compute_all(); + T0 &= DATA_MASK; + src = T0; + res = (T0 << count) | ((eflags & CC_C) << (count - 1)); + if (count > 1) + res |= T0 >> (DATA_BITS + 1 - count); + T0 = res; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = (eflags & ~(CC_C | CC_O)) | + (lshift(src ^ T0, 11 - (DATA_BITS - 1)) & CC_O) | + ((src >> (DATA_BITS - count)) & CC_C); + CC_OP = CC_OP_EFLAGS; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_rcr, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, res, eflags; + unsigned int src; + + count = T1 & 0x1f; +#if DATA_BITS == 16 + count = rclw_table[count]; +#elif DATA_BITS == 8 + count = rclb_table[count]; +#endif + if (count) { + eflags = cc_table[CC_OP].compute_all(); + T0 &= DATA_MASK; + src = T0; + res = (T0 >> count) | ((eflags & CC_C) << (DATA_BITS - count)); + if (count > 1) + res |= T0 << (DATA_BITS + 1 - count); + T0 = res; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = (eflags & ~(CC_C | CC_O)) | + (lshift(src ^ T0, 11 - (DATA_BITS - 1)) & CC_O) | + ((src >> (count - 1)) & CC_C); + CC_OP = CC_OP_EFLAGS; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_shl, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, src; + count = T1 & 0x1f; + if (count) { + src = (DATA_TYPE)T0 << (count - 1); + T0 = T0 << count; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = src; + CC_DST = T0; + CC_OP = CC_OP_SHLB + SHIFT; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_shr, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, src; + count = T1 & 0x1f; + if (count) { + T0 &= DATA_MASK; + src = T0 >> (count - 1); + T0 = T0 >> count; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = src; + CC_DST = T0; + CC_OP = CC_OP_SARB + SHIFT; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_sar, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int count, src; + count = T1 & 0x1f; + if (count) { + src = (DATA_STYPE)T0; + T0 = src >> count; + src = src >> (count - 1); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = src; + CC_DST = T0; + CC_OP = CC_OP_SARB + SHIFT; + } + FORCE_RET(); +} + +#if DATA_BITS == 16 +/* XXX: overflow flag might be incorrect in some cases in shldw */ +void OPPROTO glue(glue(op_shld, MEM_SUFFIX), _T0_T1_im_cc)(void) +{ + int count; + unsigned int res, tmp; + count = PARAM1; + T1 &= 0xffff; + res = T1 | (T0 << 16); + tmp = res >> (32 - count); + res <<= count; + if (count > 16) + res |= T1 << (count - 16); + T0 = res >> 16; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; +} + +void OPPROTO glue(glue(op_shld, MEM_SUFFIX), _T0_T1_ECX_cc)(void) +{ + int count; + unsigned int res, tmp; + count = ECX & 0x1f; + if (count) { + T1 &= 0xffff; + res = T1 | (T0 << 16); + tmp = res >> (32 - count); + res <<= count; + if (count > 16) + res |= T1 << (count - 16); + T0 = res >> 16; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; + CC_OP = CC_OP_SARB + SHIFT; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_shrd, MEM_SUFFIX), _T0_T1_im_cc)(void) +{ + int count; + unsigned int res, tmp; + + count = PARAM1; + res = (T0 & 0xffff) | (T1 << 16); + tmp = res >> (count - 1); + res >>= count; + if (count > 16) + res |= T1 << (32 - count); + T0 = res; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; +} + + +void OPPROTO glue(glue(op_shrd, MEM_SUFFIX), _T0_T1_ECX_cc)(void) +{ + int count; + unsigned int res, tmp; + + count = ECX & 0x1f; + if (count) { + res = (T0 & 0xffff) | (T1 << 16); + tmp = res >> (count - 1); + res >>= count; + if (count > 16) + res |= T1 << (32 - count); + T0 = res; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; + CC_OP = CC_OP_SARB + SHIFT; + } + FORCE_RET(); +} +#endif + +#if DATA_BITS == 32 +void OPPROTO glue(glue(op_shld, MEM_SUFFIX), _T0_T1_im_cc)(void) +{ + int count, tmp; + count = PARAM1; + T0 &= DATA_MASK; + T1 &= DATA_MASK; + tmp = T0 << (count - 1); + T0 = (T0 << count) | (T1 >> (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; +} + +void OPPROTO glue(glue(op_shld, MEM_SUFFIX), _T0_T1_ECX_cc)(void) +{ + int count, tmp; + count = ECX & 0x1f; + if (count) { + T0 &= DATA_MASK; + T1 &= DATA_MASK; + tmp = T0 << (count - 1); + T0 = (T0 << count) | (T1 >> (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; + CC_OP = CC_OP_SHLB + SHIFT; + } + FORCE_RET(); +} + +void OPPROTO glue(glue(op_shrd, MEM_SUFFIX), _T0_T1_im_cc)(void) +{ + int count, tmp; + count = PARAM1; + T0 &= DATA_MASK; + T1 &= DATA_MASK; + tmp = T0 >> (count - 1); + T0 = (T0 >> count) | (T1 << (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; +} + + +void OPPROTO glue(glue(op_shrd, MEM_SUFFIX), _T0_T1_ECX_cc)(void) +{ + int count, tmp; + count = ECX & 0x1f; + if (count) { + T0 &= DATA_MASK; + T1 &= DATA_MASK; + tmp = T0 >> (count - 1); + T0 = (T0 >> count) | (T1 << (DATA_BITS - count)); +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = tmp; + CC_DST = T0; + CC_OP = CC_OP_SARB + SHIFT; + } + FORCE_RET(); +} +#endif + +/* carry add/sub (we only need to set CC_OP differently) */ + +void OPPROTO glue(glue(op_adc, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int cf; + cf = cc_table[CC_OP].compute_c(); + T0 = T0 + T1 + cf; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = T1; + CC_DST = T0; + CC_OP = CC_OP_ADDB + SHIFT + cf * 3; +} + +void OPPROTO glue(glue(op_sbb, MEM_SUFFIX), _T0_T1_cc)(void) +{ + int cf; + cf = cc_table[CC_OP].compute_c(); + T0 = T0 - T1 - cf; +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = T1; + CC_DST = T0; + CC_OP = CC_OP_SUBB + SHIFT + cf * 3; +} + +void OPPROTO glue(glue(op_cmpxchg, MEM_SUFFIX), _T0_T1_EAX_cc)(void) +{ + unsigned int src, dst; + + src = T0; + dst = EAX - T0; + if ((DATA_TYPE)dst == 0) { + T0 = T1; + } else { + EAX = (EAX & ~DATA_MASK) | (T0 & DATA_MASK); + } +#ifdef MEM_WRITE + glue(st, SUFFIX)((uint8_t *)A0, T0); +#endif + CC_SRC = src; + CC_DST = dst; + FORCE_RET(); +} + +#undef MEM_SUFFIX +#undef MEM_WRITE diff --git a/target-i386/translate.c b/target-i386/translate.c new file mode 100644 index 0000000000..4d89ccbba9 --- /dev/null +++ b/target-i386/translate.c @@ -0,0 +1,4487 @@ +/* + * i386 translation + * + * Copyright (c) 2003 Fabrice Bellard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#include <stdarg.h> +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <inttypes.h> +#include <signal.h> +#include <assert.h> +#include <sys/mman.h> + +#include "cpu.h" +#include "exec-all.h" +#include "disas.h" + +/* XXX: move that elsewhere */ +static uint16_t *gen_opc_ptr; +static uint32_t *gen_opparam_ptr; + +#define PREFIX_REPZ 0x01 +#define PREFIX_REPNZ 0x02 +#define PREFIX_LOCK 0x04 +#define PREFIX_DATA 0x08 +#define PREFIX_ADR 0x10 + +typedef struct DisasContext { + /* current insn context */ + int override; /* -1 if no override */ + int prefix; + int aflag, dflag; + uint8_t *pc; /* pc = eip + cs_base */ + int is_jmp; /* 1 = means jump (stop translation), 2 means CPU + static state change (stop translation) */ + /* current block context */ + uint8_t *cs_base; /* base of CS segment */ + int pe; /* protected mode */ + int code32; /* 32 bit code segment */ + int ss32; /* 32 bit stack segment */ + int cc_op; /* current CC operation */ + int addseg; /* non zero if either DS/ES/SS have a non zero base */ + int f_st; /* currently unused */ + int vm86; /* vm86 mode */ + int cpl; + int iopl; + int tf; /* TF cpu flag */ + int jmp_opt; /* use direct block chaining for direct jumps */ + int mem_index; /* select memory access functions */ + struct TranslationBlock *tb; + int popl_esp_hack; /* for correct popl with esp base handling */ +} DisasContext; + +static void gen_eob(DisasContext *s); +static void gen_jmp(DisasContext *s, unsigned int eip); + +/* i386 arith/logic operations */ +enum { + OP_ADDL, + OP_ORL, + OP_ADCL, + OP_SBBL, + OP_ANDL, + OP_SUBL, + OP_XORL, + OP_CMPL, +}; + +/* i386 shift ops */ +enum { + OP_ROL, + OP_ROR, + OP_RCL, + OP_RCR, + OP_SHL, + OP_SHR, + OP_SHL1, /* undocumented */ + OP_SAR = 7, +}; + +enum { +#define DEF(s, n, copy_size) INDEX_op_ ## s, +#include "opc.h" +#undef DEF + NB_OPS, +}; + +#include "gen-op.h" + +/* operand size */ +enum { + OT_BYTE = 0, + OT_WORD, + OT_LONG, + OT_QUAD, +}; + +enum { + /* I386 int registers */ + OR_EAX, /* MUST be even numbered */ + OR_ECX, + OR_EDX, + OR_EBX, + OR_ESP, + OR_EBP, + OR_ESI, + OR_EDI, + OR_TMP0, /* temporary operand register */ + OR_TMP1, + OR_A0, /* temporary register used when doing address evaluation */ + OR_ZERO, /* fixed zero register */ + NB_OREGS, +}; + +typedef void (GenOpFunc)(void); +typedef void (GenOpFunc1)(long); +typedef void (GenOpFunc2)(long, long); +typedef void (GenOpFunc3)(long, long, long); + +static GenOpFunc *gen_op_mov_reg_T0[3][8] = { + [OT_BYTE] = { + gen_op_movb_EAX_T0, + gen_op_movb_ECX_T0, + gen_op_movb_EDX_T0, + gen_op_movb_EBX_T0, + gen_op_movh_EAX_T0, + gen_op_movh_ECX_T0, + gen_op_movh_EDX_T0, + gen_op_movh_EBX_T0, + }, + [OT_WORD] = { + gen_op_movw_EAX_T0, + gen_op_movw_ECX_T0, + gen_op_movw_EDX_T0, + gen_op_movw_EBX_T0, + gen_op_movw_ESP_T0, + gen_op_movw_EBP_T0, + gen_op_movw_ESI_T0, + gen_op_movw_EDI_T0, + }, + [OT_LONG] = { + gen_op_movl_EAX_T0, + gen_op_movl_ECX_T0, + gen_op_movl_EDX_T0, + gen_op_movl_EBX_T0, + gen_op_movl_ESP_T0, + gen_op_movl_EBP_T0, + gen_op_movl_ESI_T0, + gen_op_movl_EDI_T0, + }, +}; + +static GenOpFunc *gen_op_mov_reg_T1[3][8] = { + [OT_BYTE] = { + gen_op_movb_EAX_T1, + gen_op_movb_ECX_T1, + gen_op_movb_EDX_T1, + gen_op_movb_EBX_T1, + gen_op_movh_EAX_T1, + gen_op_movh_ECX_T1, + gen_op_movh_EDX_T1, + gen_op_movh_EBX_T1, + }, + [OT_WORD] = { + gen_op_movw_EAX_T1, + gen_op_movw_ECX_T1, + gen_op_movw_EDX_T1, + gen_op_movw_EBX_T1, + gen_op_movw_ESP_T1, + gen_op_movw_EBP_T1, + gen_op_movw_ESI_T1, + gen_op_movw_EDI_T1, + }, + [OT_LONG] = { + gen_op_movl_EAX_T1, + gen_op_movl_ECX_T1, + gen_op_movl_EDX_T1, + gen_op_movl_EBX_T1, + gen_op_movl_ESP_T1, + gen_op_movl_EBP_T1, + gen_op_movl_ESI_T1, + gen_op_movl_EDI_T1, + }, +}; + +static GenOpFunc *gen_op_mov_reg_A0[2][8] = { + [0] = { + gen_op_movw_EAX_A0, + gen_op_movw_ECX_A0, + gen_op_movw_EDX_A0, + gen_op_movw_EBX_A0, + gen_op_movw_ESP_A0, + gen_op_movw_EBP_A0, + gen_op_movw_ESI_A0, + gen_op_movw_EDI_A0, + }, + [1] = { + gen_op_movl_EAX_A0, + gen_op_movl_ECX_A0, + gen_op_movl_EDX_A0, + gen_op_movl_EBX_A0, + gen_op_movl_ESP_A0, + gen_op_movl_EBP_A0, + gen_op_movl_ESI_A0, + gen_op_movl_EDI_A0, + }, +}; + +static GenOpFunc *gen_op_mov_TN_reg[3][2][8] = +{ + [OT_BYTE] = { + { + gen_op_movl_T0_EAX, + gen_op_movl_T0_ECX, + gen_op_movl_T0_EDX, + gen_op_movl_T0_EBX, + gen_op_movh_T0_EAX, + gen_op_movh_T0_ECX, + gen_op_movh_T0_EDX, + gen_op_movh_T0_EBX, + }, + { + gen_op_movl_T1_EAX, + gen_op_movl_T1_ECX, + gen_op_movl_T1_EDX, + gen_op_movl_T1_EBX, + gen_op_movh_T1_EAX, + gen_op_movh_T1_ECX, + gen_op_movh_T1_EDX, + gen_op_movh_T1_EBX, + }, + }, + [OT_WORD] = { + { + gen_op_movl_T0_EAX, + gen_op_movl_T0_ECX, + gen_op_movl_T0_EDX, + gen_op_movl_T0_EBX, + gen_op_movl_T0_ESP, + gen_op_movl_T0_EBP, + gen_op_movl_T0_ESI, + gen_op_movl_T0_EDI, + }, + { + gen_op_movl_T1_EAX, + gen_op_movl_T1_ECX, + gen_op_movl_T1_EDX, + gen_op_movl_T1_EBX, + gen_op_movl_T1_ESP, + gen_op_movl_T1_EBP, + gen_op_movl_T1_ESI, + gen_op_movl_T1_EDI, + }, + }, + [OT_LONG] = { + { + gen_op_movl_T0_EAX, + gen_op_movl_T0_ECX, + gen_op_movl_T0_EDX, + gen_op_movl_T0_EBX, + gen_op_movl_T0_ESP, + gen_op_movl_T0_EBP, + gen_op_movl_T0_ESI, + gen_op_movl_T0_EDI, + }, + { + gen_op_movl_T1_EAX, + gen_op_movl_T1_ECX, + gen_op_movl_T1_EDX, + gen_op_movl_T1_EBX, + gen_op_movl_T1_ESP, + gen_op_movl_T1_EBP, + gen_op_movl_T1_ESI, + gen_op_movl_T1_EDI, + }, + }, +}; + +static GenOpFunc *gen_op_movl_A0_reg[8] = { + gen_op_movl_A0_EAX, + gen_op_movl_A0_ECX, + gen_op_movl_A0_EDX, + gen_op_movl_A0_EBX, + gen_op_movl_A0_ESP, + gen_op_movl_A0_EBP, + gen_op_movl_A0_ESI, + gen_op_movl_A0_EDI, +}; + +static GenOpFunc *gen_op_addl_A0_reg_sN[4][8] = { + [0] = { + gen_op_addl_A0_EAX, + gen_op_addl_A0_ECX, + gen_op_addl_A0_EDX, + gen_op_addl_A0_EBX, + gen_op_addl_A0_ESP, + gen_op_addl_A0_EBP, + gen_op_addl_A0_ESI, + gen_op_addl_A0_EDI, + }, + [1] = { + gen_op_addl_A0_EAX_s1, + gen_op_addl_A0_ECX_s1, + gen_op_addl_A0_EDX_s1, + gen_op_addl_A0_EBX_s1, + gen_op_addl_A0_ESP_s1, + gen_op_addl_A0_EBP_s1, + gen_op_addl_A0_ESI_s1, + gen_op_addl_A0_EDI_s1, + }, + [2] = { + gen_op_addl_A0_EAX_s2, + gen_op_addl_A0_ECX_s2, + gen_op_addl_A0_EDX_s2, + gen_op_addl_A0_EBX_s2, + gen_op_addl_A0_ESP_s2, + gen_op_addl_A0_EBP_s2, + gen_op_addl_A0_ESI_s2, + gen_op_addl_A0_EDI_s2, + }, + [3] = { + gen_op_addl_A0_EAX_s3, + gen_op_addl_A0_ECX_s3, + gen_op_addl_A0_EDX_s3, + gen_op_addl_A0_EBX_s3, + gen_op_addl_A0_ESP_s3, + gen_op_addl_A0_EBP_s3, + gen_op_addl_A0_ESI_s3, + gen_op_addl_A0_EDI_s3, + }, +}; + +static GenOpFunc *gen_op_cmov_reg_T1_T0[2][8] = { + [0] = { + gen_op_cmovw_EAX_T1_T0, + gen_op_cmovw_ECX_T1_T0, + gen_op_cmovw_EDX_T1_T0, + gen_op_cmovw_EBX_T1_T0, + gen_op_cmovw_ESP_T1_T0, + gen_op_cmovw_EBP_T1_T0, + gen_op_cmovw_ESI_T1_T0, + gen_op_cmovw_EDI_T1_T0, + }, + [1] = { + gen_op_cmovl_EAX_T1_T0, + gen_op_cmovl_ECX_T1_T0, + gen_op_cmovl_EDX_T1_T0, + gen_op_cmovl_EBX_T1_T0, + gen_op_cmovl_ESP_T1_T0, + gen_op_cmovl_EBP_T1_T0, + gen_op_cmovl_ESI_T1_T0, + gen_op_cmovl_EDI_T1_T0, + }, +}; + +static GenOpFunc *gen_op_arith_T0_T1_cc[8] = { + NULL, + gen_op_orl_T0_T1, + NULL, + NULL, + gen_op_andl_T0_T1, + NULL, + gen_op_xorl_T0_T1, + NULL, +}; + +static GenOpFunc *gen_op_arithc_T0_T1_cc[3][2] = { + [OT_BYTE] = { + gen_op_adcb_T0_T1_cc, + gen_op_sbbb_T0_T1_cc, + }, + [OT_WORD] = { + gen_op_adcw_T0_T1_cc, + gen_op_sbbw_T0_T1_cc, + }, + [OT_LONG] = { + gen_op_adcl_T0_T1_cc, + gen_op_sbbl_T0_T1_cc, + }, +}; + +static GenOpFunc *gen_op_arithc_mem_T0_T1_cc[3][2] = { + [OT_BYTE] = { + gen_op_adcb_mem_T0_T1_cc, + gen_op_sbbb_mem_T0_T1_cc, + }, + [OT_WORD] = { + gen_op_adcw_mem_T0_T1_cc, + gen_op_sbbw_mem_T0_T1_cc, + }, + [OT_LONG] = { + gen_op_adcl_mem_T0_T1_cc, + gen_op_sbbl_mem_T0_T1_cc, + }, +}; + +static const int cc_op_arithb[8] = { + CC_OP_ADDB, + CC_OP_LOGICB, + CC_OP_ADDB, + CC_OP_SUBB, + CC_OP_LOGICB, + CC_OP_SUBB, + CC_OP_LOGICB, + CC_OP_SUBB, +}; + +static GenOpFunc *gen_op_cmpxchg_T0_T1_EAX_cc[3] = { + gen_op_cmpxchgb_T0_T1_EAX_cc, + gen_op_cmpxchgw_T0_T1_EAX_cc, + gen_op_cmpxchgl_T0_T1_EAX_cc, +}; + +static GenOpFunc *gen_op_cmpxchg_mem_T0_T1_EAX_cc[3] = { + gen_op_cmpxchgb_mem_T0_T1_EAX_cc, + gen_op_cmpxchgw_mem_T0_T1_EAX_cc, + gen_op_cmpxchgl_mem_T0_T1_EAX_cc, +}; + +static GenOpFunc *gen_op_shift_T0_T1_cc[3][8] = { + [OT_BYTE] = { + gen_op_rolb_T0_T1_cc, + gen_op_rorb_T0_T1_cc, + gen_op_rclb_T0_T1_cc, + gen_op_rcrb_T0_T1_cc, + gen_op_shlb_T0_T1_cc, + gen_op_shrb_T0_T1_cc, + gen_op_shlb_T0_T1_cc, + gen_op_sarb_T0_T1_cc, + }, + [OT_WORD] = { + gen_op_rolw_T0_T1_cc, + gen_op_rorw_T0_T1_cc, + gen_op_rclw_T0_T1_cc, + gen_op_rcrw_T0_T1_cc, + gen_op_shlw_T0_T1_cc, + gen_op_shrw_T0_T1_cc, + gen_op_shlw_T0_T1_cc, + gen_op_sarw_T0_T1_cc, + }, + [OT_LONG] = { + gen_op_roll_T0_T1_cc, + gen_op_rorl_T0_T1_cc, + gen_op_rcll_T0_T1_cc, + gen_op_rcrl_T0_T1_cc, + gen_op_shll_T0_T1_cc, + gen_op_shrl_T0_T1_cc, + gen_op_shll_T0_T1_cc, + gen_op_sarl_T0_T1_cc, + }, +}; + +static GenOpFunc *gen_op_shift_mem_T0_T1_cc[3][8] = { + [OT_BYTE] = { + gen_op_rolb_mem_T0_T1_cc, + gen_op_rorb_mem_T0_T1_cc, + gen_op_rclb_mem_T0_T1_cc, + gen_op_rcrb_mem_T0_T1_cc, + gen_op_shlb_mem_T0_T1_cc, + gen_op_shrb_mem_T0_T1_cc, + gen_op_shlb_mem_T0_T1_cc, + gen_op_sarb_mem_T0_T1_cc, + }, + [OT_WORD] = { + gen_op_rolw_mem_T0_T1_cc, + gen_op_rorw_mem_T0_T1_cc, + gen_op_rclw_mem_T0_T1_cc, + gen_op_rcrw_mem_T0_T1_cc, + gen_op_shlw_mem_T0_T1_cc, + gen_op_shrw_mem_T0_T1_cc, + gen_op_shlw_mem_T0_T1_cc, + gen_op_sarw_mem_T0_T1_cc, + }, + [OT_LONG] = { + gen_op_roll_mem_T0_T1_cc, + gen_op_rorl_mem_T0_T1_cc, + gen_op_rcll_mem_T0_T1_cc, + gen_op_rcrl_mem_T0_T1_cc, + gen_op_shll_mem_T0_T1_cc, + gen_op_shrl_mem_T0_T1_cc, + gen_op_shll_mem_T0_T1_cc, + gen_op_sarl_mem_T0_T1_cc, + }, +}; + +static GenOpFunc1 *gen_op_shiftd_T0_T1_im_cc[2][2] = { + [0] = { + gen_op_shldw_T0_T1_im_cc, + gen_op_shrdw_T0_T1_im_cc, + }, + [1] = { + gen_op_shldl_T0_T1_im_cc, + gen_op_shrdl_T0_T1_im_cc, + }, +}; + +static GenOpFunc *gen_op_shiftd_T0_T1_ECX_cc[2][2] = { + [0] = { + gen_op_shldw_T0_T1_ECX_cc, + gen_op_shrdw_T0_T1_ECX_cc, + }, + [1] = { + gen_op_shldl_T0_T1_ECX_cc, + gen_op_shrdl_T0_T1_ECX_cc, + }, +}; + +static GenOpFunc1 *gen_op_shiftd_mem_T0_T1_im_cc[2][2] = { + [0] = { + gen_op_shldw_mem_T0_T1_im_cc, + gen_op_shrdw_mem_T0_T1_im_cc, + }, + [1] = { + gen_op_shldl_mem_T0_T1_im_cc, + gen_op_shrdl_mem_T0_T1_im_cc, + }, +}; + +static GenOpFunc *gen_op_shiftd_mem_T0_T1_ECX_cc[2][2] = { + [0] = { + gen_op_shldw_mem_T0_T1_ECX_cc, + gen_op_shrdw_mem_T0_T1_ECX_cc, + }, + [1] = { + gen_op_shldl_mem_T0_T1_ECX_cc, + gen_op_shrdl_mem_T0_T1_ECX_cc, + }, +}; + +static GenOpFunc *gen_op_btx_T0_T1_cc[2][4] = { + [0] = { + gen_op_btw_T0_T1_cc, + gen_op_btsw_T0_T1_cc, + gen_op_btrw_T0_T1_cc, + gen_op_btcw_T0_T1_cc, + }, + [1] = { + gen_op_btl_T0_T1_cc, + gen_op_btsl_T0_T1_cc, + gen_op_btrl_T0_T1_cc, + gen_op_btcl_T0_T1_cc, + }, +}; + +static GenOpFunc *gen_op_bsx_T0_cc[2][2] = { + [0] = { + gen_op_bsfw_T0_cc, + gen_op_bsrw_T0_cc, + }, + [1] = { + gen_op_bsfl_T0_cc, + gen_op_bsrl_T0_cc, + }, +}; + +static GenOpFunc *gen_op_lds_T0_A0[3 * 3] = { + gen_op_ldsb_T0_A0, + gen_op_ldsw_T0_A0, + NULL, + + gen_op_ldsb_kernel_T0_A0, + gen_op_ldsw_kernel_T0_A0, + NULL, + + gen_op_ldsb_user_T0_A0, + gen_op_ldsw_user_T0_A0, + NULL, +}; + +static GenOpFunc *gen_op_ldu_T0_A0[3 * 3] = { + gen_op_ldub_T0_A0, + gen_op_lduw_T0_A0, + NULL, + + gen_op_ldub_kernel_T0_A0, + gen_op_lduw_kernel_T0_A0, + NULL, + + gen_op_ldub_user_T0_A0, + gen_op_lduw_user_T0_A0, + NULL, +}; + +/* sign does not matter, except for lidt/lgdt call (TODO: fix it) */ +static GenOpFunc *gen_op_ld_T0_A0[3 * 3] = { + gen_op_ldub_T0_A0, + gen_op_lduw_T0_A0, + gen_op_ldl_T0_A0, + + gen_op_ldub_kernel_T0_A0, + gen_op_lduw_kernel_T0_A0, + gen_op_ldl_kernel_T0_A0, + + gen_op_ldub_user_T0_A0, + gen_op_lduw_user_T0_A0, + gen_op_ldl_user_T0_A0, +}; + +static GenOpFunc *gen_op_ld_T1_A0[3 * 3] = { + gen_op_ldub_T1_A0, + gen_op_lduw_T1_A0, + gen_op_ldl_T1_A0, + + gen_op_ldub_kernel_T1_A0, + gen_op_lduw_kernel_T1_A0, + gen_op_ldl_kernel_T1_A0, + + gen_op_ldub_user_T1_A0, + gen_op_lduw_user_T1_A0, + gen_op_ldl_user_T1_A0, +}; + +static GenOpFunc *gen_op_st_T0_A0[3 * 3] = { + gen_op_stb_T0_A0, + gen_op_stw_T0_A0, + gen_op_stl_T0_A0, + + gen_op_stb_kernel_T0_A0, + gen_op_stw_kernel_T0_A0, + gen_op_stl_kernel_T0_A0, + + gen_op_stb_user_T0_A0, + gen_op_stw_user_T0_A0, + gen_op_stl_user_T0_A0, +}; + +static inline void gen_string_movl_A0_ESI(DisasContext *s) +{ + int override; + + override = s->override; + if (s->aflag) { + /* 32 bit address */ + if (s->addseg && override < 0) + override = R_DS; + if (override >= 0) { + gen_op_movl_A0_seg(offsetof(CPUX86State,segs[override].base)); + gen_op_addl_A0_reg_sN[0][R_ESI](); + } else { + gen_op_movl_A0_reg[R_ESI](); + } + } else { + /* 16 address, always override */ + if (override < 0) + override = R_DS; + gen_op_movl_A0_reg[R_ESI](); + gen_op_andl_A0_ffff(); + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[override].base)); + } +} + +static inline void gen_string_movl_A0_EDI(DisasContext *s) +{ + if (s->aflag) { + if (s->addseg) { + gen_op_movl_A0_seg(offsetof(CPUX86State,segs[R_ES].base)); + gen_op_addl_A0_reg_sN[0][R_EDI](); + } else { + gen_op_movl_A0_reg[R_EDI](); + } + } else { + gen_op_movl_A0_reg[R_EDI](); + gen_op_andl_A0_ffff(); + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[R_ES].base)); + } +} + +static GenOpFunc *gen_op_movl_T0_Dshift[3] = { + gen_op_movl_T0_Dshiftb, + gen_op_movl_T0_Dshiftw, + gen_op_movl_T0_Dshiftl, +}; + +static GenOpFunc2 *gen_op_jz_ecx[2] = { + gen_op_jz_ecxw, + gen_op_jz_ecxl, +}; + +static GenOpFunc1 *gen_op_jz_ecx_im[2] = { + gen_op_jz_ecxw_im, + gen_op_jz_ecxl_im, +}; + +static GenOpFunc *gen_op_dec_ECX[2] = { + gen_op_decw_ECX, + gen_op_decl_ECX, +}; + +static GenOpFunc1 *gen_op_string_jnz_sub[2][3] = { + { + gen_op_string_jnz_subb, + gen_op_string_jnz_subw, + gen_op_string_jnz_subl, + }, + { + gen_op_string_jz_subb, + gen_op_string_jz_subw, + gen_op_string_jz_subl, + }, +}; + +static GenOpFunc1 *gen_op_string_jnz_sub_im[2][3] = { + { + gen_op_string_jnz_subb_im, + gen_op_string_jnz_subw_im, + gen_op_string_jnz_subl_im, + }, + { + gen_op_string_jz_subb_im, + gen_op_string_jz_subw_im, + gen_op_string_jz_subl_im, + }, +}; + +static GenOpFunc *gen_op_in_DX_T0[3] = { + gen_op_inb_DX_T0, + gen_op_inw_DX_T0, + gen_op_inl_DX_T0, +}; + +static GenOpFunc *gen_op_out_DX_T0[3] = { + gen_op_outb_DX_T0, + gen_op_outw_DX_T0, + gen_op_outl_DX_T0, +}; + +static inline void gen_movs(DisasContext *s, int ot) +{ + gen_string_movl_A0_ESI(s); + gen_op_ld_T0_A0[ot + s->mem_index](); + gen_string_movl_A0_EDI(s); + gen_op_st_T0_A0[ot + s->mem_index](); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_ESI_T0(); + gen_op_addl_EDI_T0(); + } else { + gen_op_addw_ESI_T0(); + gen_op_addw_EDI_T0(); + } +} + +static inline void gen_update_cc_op(DisasContext *s) +{ + if (s->cc_op != CC_OP_DYNAMIC) { + gen_op_set_cc_op(s->cc_op); + s->cc_op = CC_OP_DYNAMIC; + } +} + +static inline void gen_jz_ecx_string(DisasContext *s, unsigned int next_eip) +{ + if (s->jmp_opt) { + gen_op_jz_ecx[s->aflag]((long)s->tb, next_eip); + } else { + /* XXX: does not work with gdbstub "ice" single step - not a + serious problem */ + gen_op_jz_ecx_im[s->aflag](next_eip); + } +} + +static inline void gen_stos(DisasContext *s, int ot) +{ + gen_op_mov_TN_reg[OT_LONG][0][R_EAX](); + gen_string_movl_A0_EDI(s); + gen_op_st_T0_A0[ot + s->mem_index](); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_EDI_T0(); + } else { + gen_op_addw_EDI_T0(); + } +} + +static inline void gen_lods(DisasContext *s, int ot) +{ + gen_string_movl_A0_ESI(s); + gen_op_ld_T0_A0[ot + s->mem_index](); + gen_op_mov_reg_T0[ot][R_EAX](); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_ESI_T0(); + } else { + gen_op_addw_ESI_T0(); + } +} + +static inline void gen_scas(DisasContext *s, int ot) +{ + gen_op_mov_TN_reg[OT_LONG][0][R_EAX](); + gen_string_movl_A0_EDI(s); + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_cmpl_T0_T1_cc(); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_EDI_T0(); + } else { + gen_op_addw_EDI_T0(); + } +} + +static inline void gen_cmps(DisasContext *s, int ot) +{ + gen_string_movl_A0_ESI(s); + gen_op_ld_T0_A0[ot + s->mem_index](); + gen_string_movl_A0_EDI(s); + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_cmpl_T0_T1_cc(); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_ESI_T0(); + gen_op_addl_EDI_T0(); + } else { + gen_op_addw_ESI_T0(); + gen_op_addw_EDI_T0(); + } +} + +static inline void gen_ins(DisasContext *s, int ot) +{ + gen_op_in_DX_T0[ot](); + gen_string_movl_A0_EDI(s); + gen_op_st_T0_A0[ot + s->mem_index](); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_EDI_T0(); + } else { + gen_op_addw_EDI_T0(); + } +} + +static inline void gen_outs(DisasContext *s, int ot) +{ + gen_string_movl_A0_ESI(s); + gen_op_ld_T0_A0[ot + s->mem_index](); + gen_op_out_DX_T0[ot](); + gen_op_movl_T0_Dshift[ot](); + if (s->aflag) { + gen_op_addl_ESI_T0(); + } else { + gen_op_addw_ESI_T0(); + } +} + +/* same method as Valgrind : we generate jumps to current or next + instruction */ +#define GEN_REPZ(op) \ +static inline void gen_repz_ ## op(DisasContext *s, int ot, \ + unsigned int cur_eip, unsigned int next_eip) \ +{ \ + gen_update_cc_op(s); \ + gen_jz_ecx_string(s, next_eip); \ + gen_ ## op(s, ot); \ + gen_op_dec_ECX[s->aflag](); \ + /* a loop would cause two single step exceptions if ECX = 1 \ + before rep string_insn */ \ + if (!s->jmp_opt) \ + gen_op_jz_ecx_im[s->aflag](next_eip); \ + gen_jmp(s, cur_eip); \ +} + +#define GEN_REPZ2(op) \ +static inline void gen_repz_ ## op(DisasContext *s, int ot, \ + unsigned int cur_eip, \ + unsigned int next_eip, \ + int nz) \ +{ \ + gen_update_cc_op(s); \ + gen_jz_ecx_string(s, next_eip); \ + gen_ ## op(s, ot); \ + gen_op_dec_ECX[s->aflag](); \ + gen_op_set_cc_op(CC_OP_SUBB + ot); \ + if (!s->jmp_opt) \ + gen_op_string_jnz_sub_im[nz][ot](next_eip); \ + else \ + gen_op_string_jnz_sub[nz][ot]((long)s->tb); \ + if (!s->jmp_opt) \ + gen_op_jz_ecx_im[s->aflag](next_eip); \ + gen_jmp(s, cur_eip); \ +} + +GEN_REPZ(movs) +GEN_REPZ(stos) +GEN_REPZ(lods) +GEN_REPZ(ins) +GEN_REPZ(outs) +GEN_REPZ2(scas) +GEN_REPZ2(cmps) + +static GenOpFunc *gen_op_in[3] = { + gen_op_inb_T0_T1, + gen_op_inw_T0_T1, + gen_op_inl_T0_T1, +}; + +static GenOpFunc *gen_op_out[3] = { + gen_op_outb_T0_T1, + gen_op_outw_T0_T1, + gen_op_outl_T0_T1, +}; + +enum { + JCC_O, + JCC_B, + JCC_Z, + JCC_BE, + JCC_S, + JCC_P, + JCC_L, + JCC_LE, +}; + +static GenOpFunc3 *gen_jcc_sub[3][8] = { + [OT_BYTE] = { + NULL, + gen_op_jb_subb, + gen_op_jz_subb, + gen_op_jbe_subb, + gen_op_js_subb, + NULL, + gen_op_jl_subb, + gen_op_jle_subb, + }, + [OT_WORD] = { + NULL, + gen_op_jb_subw, + gen_op_jz_subw, + gen_op_jbe_subw, + gen_op_js_subw, + NULL, + gen_op_jl_subw, + gen_op_jle_subw, + }, + [OT_LONG] = { + NULL, + gen_op_jb_subl, + gen_op_jz_subl, + gen_op_jbe_subl, + gen_op_js_subl, + NULL, + gen_op_jl_subl, + gen_op_jle_subl, + }, +}; +static GenOpFunc2 *gen_op_loop[2][4] = { + [0] = { + gen_op_loopnzw, + gen_op_loopzw, + gen_op_loopw, + gen_op_jecxzw, + }, + [1] = { + gen_op_loopnzl, + gen_op_loopzl, + gen_op_loopl, + gen_op_jecxzl, + }, +}; + +static GenOpFunc *gen_setcc_slow[8] = { + gen_op_seto_T0_cc, + gen_op_setb_T0_cc, + gen_op_setz_T0_cc, + gen_op_setbe_T0_cc, + gen_op_sets_T0_cc, + gen_op_setp_T0_cc, + gen_op_setl_T0_cc, + gen_op_setle_T0_cc, +}; + +static GenOpFunc *gen_setcc_sub[3][8] = { + [OT_BYTE] = { + NULL, + gen_op_setb_T0_subb, + gen_op_setz_T0_subb, + gen_op_setbe_T0_subb, + gen_op_sets_T0_subb, + NULL, + gen_op_setl_T0_subb, + gen_op_setle_T0_subb, + }, + [OT_WORD] = { + NULL, + gen_op_setb_T0_subw, + gen_op_setz_T0_subw, + gen_op_setbe_T0_subw, + gen_op_sets_T0_subw, + NULL, + gen_op_setl_T0_subw, + gen_op_setle_T0_subw, + }, + [OT_LONG] = { + NULL, + gen_op_setb_T0_subl, + gen_op_setz_T0_subl, + gen_op_setbe_T0_subl, + gen_op_sets_T0_subl, + NULL, + gen_op_setl_T0_subl, + gen_op_setle_T0_subl, + }, +}; + +static GenOpFunc *gen_op_fp_arith_ST0_FT0[8] = { + gen_op_fadd_ST0_FT0, + gen_op_fmul_ST0_FT0, + gen_op_fcom_ST0_FT0, + gen_op_fcom_ST0_FT0, + gen_op_fsub_ST0_FT0, + gen_op_fsubr_ST0_FT0, + gen_op_fdiv_ST0_FT0, + gen_op_fdivr_ST0_FT0, +}; + +/* NOTE the exception in "r" op ordering */ +static GenOpFunc1 *gen_op_fp_arith_STN_ST0[8] = { + gen_op_fadd_STN_ST0, + gen_op_fmul_STN_ST0, + NULL, + NULL, + gen_op_fsubr_STN_ST0, + gen_op_fsub_STN_ST0, + gen_op_fdivr_STN_ST0, + gen_op_fdiv_STN_ST0, +}; + +/* if d == OR_TMP0, it means memory operand (address in A0) */ +static void gen_op(DisasContext *s1, int op, int ot, int d) +{ + GenOpFunc *gen_update_cc; + + if (d != OR_TMP0) { + gen_op_mov_TN_reg[ot][0][d](); + } else { + gen_op_ld_T0_A0[ot + s1->mem_index](); + } + switch(op) { + case OP_ADCL: + case OP_SBBL: + if (s1->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s1->cc_op); + if (d != OR_TMP0) { + gen_op_arithc_T0_T1_cc[ot][op - OP_ADCL](); + gen_op_mov_reg_T0[ot][d](); + } else { + gen_op_arithc_mem_T0_T1_cc[ot][op - OP_ADCL](); + } + s1->cc_op = CC_OP_DYNAMIC; + goto the_end; + case OP_ADDL: + gen_op_addl_T0_T1(); + s1->cc_op = CC_OP_ADDB + ot; + gen_update_cc = gen_op_update2_cc; + break; + case OP_SUBL: + gen_op_subl_T0_T1(); + s1->cc_op = CC_OP_SUBB + ot; + gen_update_cc = gen_op_update2_cc; + break; + default: + case OP_ANDL: + case OP_ORL: + case OP_XORL: + gen_op_arith_T0_T1_cc[op](); + s1->cc_op = CC_OP_LOGICB + ot; + gen_update_cc = gen_op_update1_cc; + break; + case OP_CMPL: + gen_op_cmpl_T0_T1_cc(); + s1->cc_op = CC_OP_SUBB + ot; + gen_update_cc = NULL; + break; + } + if (op != OP_CMPL) { + if (d != OR_TMP0) + gen_op_mov_reg_T0[ot][d](); + else + gen_op_st_T0_A0[ot + s1->mem_index](); + } + /* the flags update must happen after the memory write (precise + exception support) */ + if (gen_update_cc) + gen_update_cc(); + the_end: ; +} + +/* if d == OR_TMP0, it means memory operand (address in A0) */ +static void gen_inc(DisasContext *s1, int ot, int d, int c) +{ + if (d != OR_TMP0) + gen_op_mov_TN_reg[ot][0][d](); + else + gen_op_ld_T0_A0[ot + s1->mem_index](); + if (s1->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s1->cc_op); + if (c > 0) { + gen_op_incl_T0(); + s1->cc_op = CC_OP_INCB + ot; + } else { + gen_op_decl_T0(); + s1->cc_op = CC_OP_DECB + ot; + } + if (d != OR_TMP0) + gen_op_mov_reg_T0[ot][d](); + else + gen_op_st_T0_A0[ot + s1->mem_index](); + gen_op_update_inc_cc(); +} + +static void gen_shift(DisasContext *s1, int op, int ot, int d, int s) +{ + if (d != OR_TMP0) + gen_op_mov_TN_reg[ot][0][d](); + else + gen_op_ld_T0_A0[ot + s1->mem_index](); + if (s != OR_TMP1) + gen_op_mov_TN_reg[ot][1][s](); + /* for zero counts, flags are not updated, so must do it dynamically */ + if (s1->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s1->cc_op); + + if (d != OR_TMP0) + gen_op_shift_T0_T1_cc[ot][op](); + else + gen_op_shift_mem_T0_T1_cc[ot][op](); + if (d != OR_TMP0) + gen_op_mov_reg_T0[ot][d](); + s1->cc_op = CC_OP_DYNAMIC; /* cannot predict flags after */ +} + +static void gen_shifti(DisasContext *s1, int op, int ot, int d, int c) +{ + /* currently not optimized */ + gen_op_movl_T1_im(c); + gen_shift(s1, op, ot, d, OR_TMP1); +} + +static void gen_lea_modrm(DisasContext *s, int modrm, int *reg_ptr, int *offset_ptr) +{ + int havesib; + int base, disp; + int index; + int scale; + int opreg; + int mod, rm, code, override, must_add_seg; + + override = s->override; + must_add_seg = s->addseg; + if (override >= 0) + must_add_seg = 1; + mod = (modrm >> 6) & 3; + rm = modrm & 7; + + if (s->aflag) { + + havesib = 0; + base = rm; + index = 0; + scale = 0; + + if (base == 4) { + havesib = 1; + code = ldub(s->pc++); + scale = (code >> 6) & 3; + index = (code >> 3) & 7; + base = code & 7; + } + + switch (mod) { + case 0: + if (base == 5) { + base = -1; + disp = ldl(s->pc); + s->pc += 4; + } else { + disp = 0; + } + break; + case 1: + disp = (int8_t)ldub(s->pc++); + break; + default: + case 2: + disp = ldl(s->pc); + s->pc += 4; + break; + } + + if (base >= 0) { + /* for correct popl handling with esp */ + if (base == 4 && s->popl_esp_hack) + disp += s->popl_esp_hack; + gen_op_movl_A0_reg[base](); + if (disp != 0) + gen_op_addl_A0_im(disp); + } else { + gen_op_movl_A0_im(disp); + } + /* XXX: index == 4 is always invalid */ + if (havesib && (index != 4 || scale != 0)) { + gen_op_addl_A0_reg_sN[scale][index](); + } + if (must_add_seg) { + if (override < 0) { + if (base == R_EBP || base == R_ESP) + override = R_SS; + else + override = R_DS; + } + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[override].base)); + } + } else { + switch (mod) { + case 0: + if (rm == 6) { + disp = lduw(s->pc); + s->pc += 2; + gen_op_movl_A0_im(disp); + rm = 0; /* avoid SS override */ + goto no_rm; + } else { + disp = 0; + } + break; + case 1: + disp = (int8_t)ldub(s->pc++); + break; + default: + case 2: + disp = lduw(s->pc); + s->pc += 2; + break; + } + switch(rm) { + case 0: + gen_op_movl_A0_reg[R_EBX](); + gen_op_addl_A0_reg_sN[0][R_ESI](); + break; + case 1: + gen_op_movl_A0_reg[R_EBX](); + gen_op_addl_A0_reg_sN[0][R_EDI](); + break; + case 2: + gen_op_movl_A0_reg[R_EBP](); + gen_op_addl_A0_reg_sN[0][R_ESI](); + break; + case 3: + gen_op_movl_A0_reg[R_EBP](); + gen_op_addl_A0_reg_sN[0][R_EDI](); + break; + case 4: + gen_op_movl_A0_reg[R_ESI](); + break; + case 5: + gen_op_movl_A0_reg[R_EDI](); + break; + case 6: + gen_op_movl_A0_reg[R_EBP](); + break; + default: + case 7: + gen_op_movl_A0_reg[R_EBX](); + break; + } + if (disp != 0) + gen_op_addl_A0_im(disp); + gen_op_andl_A0_ffff(); + no_rm: + if (must_add_seg) { + if (override < 0) { + if (rm == 2 || rm == 3 || rm == 6) + override = R_SS; + else + override = R_DS; + } + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[override].base)); + } + } + + opreg = OR_A0; + disp = 0; + *reg_ptr = opreg; + *offset_ptr = disp; +} + +/* generate modrm memory load or store of 'reg'. TMP0 is used if reg != + OR_TMP0 */ +static void gen_ldst_modrm(DisasContext *s, int modrm, int ot, int reg, int is_store) +{ + int mod, rm, opreg, disp; + + mod = (modrm >> 6) & 3; + rm = modrm & 7; + if (mod == 3) { + if (is_store) { + if (reg != OR_TMP0) + gen_op_mov_TN_reg[ot][0][reg](); + gen_op_mov_reg_T0[ot][rm](); + } else { + gen_op_mov_TN_reg[ot][0][rm](); + if (reg != OR_TMP0) + gen_op_mov_reg_T0[ot][reg](); + } + } else { + gen_lea_modrm(s, modrm, &opreg, &disp); + if (is_store) { + if (reg != OR_TMP0) + gen_op_mov_TN_reg[ot][0][reg](); + gen_op_st_T0_A0[ot + s->mem_index](); + } else { + gen_op_ld_T0_A0[ot + s->mem_index](); + if (reg != OR_TMP0) + gen_op_mov_reg_T0[ot][reg](); + } + } +} + +static inline uint32_t insn_get(DisasContext *s, int ot) +{ + uint32_t ret; + + switch(ot) { + case OT_BYTE: + ret = ldub(s->pc); + s->pc++; + break; + case OT_WORD: + ret = lduw(s->pc); + s->pc += 2; + break; + default: + case OT_LONG: + ret = ldl(s->pc); + s->pc += 4; + break; + } + return ret; +} + +static inline void gen_jcc(DisasContext *s, int b, int val, int next_eip) +{ + TranslationBlock *tb; + int inv, jcc_op; + GenOpFunc3 *func; + + inv = b & 1; + jcc_op = (b >> 1) & 7; + + if (s->jmp_opt) { + switch(s->cc_op) { + /* we optimize the cmp/jcc case */ + case CC_OP_SUBB: + case CC_OP_SUBW: + case CC_OP_SUBL: + func = gen_jcc_sub[s->cc_op - CC_OP_SUBB][jcc_op]; + break; + + /* some jumps are easy to compute */ + case CC_OP_ADDB: + case CC_OP_ADDW: + case CC_OP_ADDL: + case CC_OP_ADCB: + case CC_OP_ADCW: + case CC_OP_ADCL: + case CC_OP_SBBB: + case CC_OP_SBBW: + case CC_OP_SBBL: + case CC_OP_LOGICB: + case CC_OP_LOGICW: + case CC_OP_LOGICL: + case CC_OP_INCB: + case CC_OP_INCW: + case CC_OP_INCL: + case CC_OP_DECB: + case CC_OP_DECW: + case CC_OP_DECL: + case CC_OP_SHLB: + case CC_OP_SHLW: + case CC_OP_SHLL: + case CC_OP_SARB: + case CC_OP_SARW: + case CC_OP_SARL: + switch(jcc_op) { + case JCC_Z: + func = gen_jcc_sub[(s->cc_op - CC_OP_ADDB) % 3][jcc_op]; + break; + case JCC_S: + func = gen_jcc_sub[(s->cc_op - CC_OP_ADDB) % 3][jcc_op]; + break; + default: + func = NULL; + break; + } + break; + default: + func = NULL; + break; + } + + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + + if (!func) { + gen_setcc_slow[jcc_op](); + func = gen_op_jcc; + } + + tb = s->tb; + if (!inv) { + func((long)tb, val, next_eip); + } else { + func((long)tb, next_eip, val); + } + s->is_jmp = 3; + } else { + if (s->cc_op != CC_OP_DYNAMIC) { + gen_op_set_cc_op(s->cc_op); + s->cc_op = CC_OP_DYNAMIC; + } + gen_setcc_slow[jcc_op](); + if (!inv) { + gen_op_jcc_im(val, next_eip); + } else { + gen_op_jcc_im(next_eip, val); + } + gen_eob(s); + } +} + +static void gen_setcc(DisasContext *s, int b) +{ + int inv, jcc_op; + GenOpFunc *func; + + inv = b & 1; + jcc_op = (b >> 1) & 7; + switch(s->cc_op) { + /* we optimize the cmp/jcc case */ + case CC_OP_SUBB: + case CC_OP_SUBW: + case CC_OP_SUBL: + func = gen_setcc_sub[s->cc_op - CC_OP_SUBB][jcc_op]; + if (!func) + goto slow_jcc; + break; + + /* some jumps are easy to compute */ + case CC_OP_ADDB: + case CC_OP_ADDW: + case CC_OP_ADDL: + case CC_OP_LOGICB: + case CC_OP_LOGICW: + case CC_OP_LOGICL: + case CC_OP_INCB: + case CC_OP_INCW: + case CC_OP_INCL: + case CC_OP_DECB: + case CC_OP_DECW: + case CC_OP_DECL: + case CC_OP_SHLB: + case CC_OP_SHLW: + case CC_OP_SHLL: + switch(jcc_op) { + case JCC_Z: + func = gen_setcc_sub[(s->cc_op - CC_OP_ADDB) % 3][jcc_op]; + break; + case JCC_S: + func = gen_setcc_sub[(s->cc_op - CC_OP_ADDB) % 3][jcc_op]; + break; + default: + goto slow_jcc; + } + break; + default: + slow_jcc: + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + func = gen_setcc_slow[jcc_op]; + break; + } + func(); + if (inv) { + gen_op_xor_T0_1(); + } +} + +/* move T0 to seg_reg and compute if the CPU state may change. Never + call this function with seg_reg == R_CS */ +static void gen_movl_seg_T0(DisasContext *s, int seg_reg, unsigned int cur_eip) +{ + if (s->pe && !s->vm86) + gen_op_movl_seg_T0(seg_reg, cur_eip); + else + gen_op_movl_seg_T0_vm(offsetof(CPUX86State,segs[seg_reg])); + /* abort translation because the register may have a non zero base + or because ss32 may change. For R_SS, translation must always + stop as a special handling must be done to disable hardware + interrupts for the next instruction */ + if (seg_reg == R_SS || (!s->addseg && seg_reg < R_FS)) + s->is_jmp = 3; +} + +/* generate a push. It depends on ss32, addseg and dflag */ +static void gen_push_T0(DisasContext *s) +{ + if (s->ss32) { + if (!s->addseg) { + if (s->dflag) + gen_op_pushl_T0(); + else + gen_op_pushw_T0(); + } else { + if (s->dflag) + gen_op_pushl_ss32_T0(); + else + gen_op_pushw_ss32_T0(); + } + } else { + if (s->dflag) + gen_op_pushl_ss16_T0(); + else + gen_op_pushw_ss16_T0(); + } +} + +/* two step pop is necessary for precise exceptions */ +static void gen_pop_T0(DisasContext *s) +{ + if (s->ss32) { + if (!s->addseg) { + if (s->dflag) + gen_op_popl_T0(); + else + gen_op_popw_T0(); + } else { + if (s->dflag) + gen_op_popl_ss32_T0(); + else + gen_op_popw_ss32_T0(); + } + } else { + if (s->dflag) + gen_op_popl_ss16_T0(); + else + gen_op_popw_ss16_T0(); + } +} + +static inline void gen_stack_update(DisasContext *s, int addend) +{ + if (s->ss32) { + if (addend == 2) + gen_op_addl_ESP_2(); + else if (addend == 4) + gen_op_addl_ESP_4(); + else + gen_op_addl_ESP_im(addend); + } else { + if (addend == 2) + gen_op_addw_ESP_2(); + else if (addend == 4) + gen_op_addw_ESP_4(); + else + gen_op_addw_ESP_im(addend); + } +} + +static void gen_pop_update(DisasContext *s) +{ + gen_stack_update(s, 2 << s->dflag); +} + +static void gen_stack_A0(DisasContext *s) +{ + gen_op_movl_A0_ESP(); + if (!s->ss32) + gen_op_andl_A0_ffff(); + gen_op_movl_T1_A0(); + if (s->addseg) + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[R_SS].base)); +} + +/* NOTE: wrap around in 16 bit not fully handled */ +static void gen_pusha(DisasContext *s) +{ + int i; + gen_op_movl_A0_ESP(); + gen_op_addl_A0_im(-16 << s->dflag); + if (!s->ss32) + gen_op_andl_A0_ffff(); + gen_op_movl_T1_A0(); + if (s->addseg) + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[R_SS].base)); + for(i = 0;i < 8; i++) { + gen_op_mov_TN_reg[OT_LONG][0][7 - i](); + gen_op_st_T0_A0[OT_WORD + s->dflag + s->mem_index](); + gen_op_addl_A0_im(2 << s->dflag); + } + gen_op_mov_reg_T1[OT_WORD + s->dflag][R_ESP](); +} + +/* NOTE: wrap around in 16 bit not fully handled */ +static void gen_popa(DisasContext *s) +{ + int i; + gen_op_movl_A0_ESP(); + if (!s->ss32) + gen_op_andl_A0_ffff(); + gen_op_movl_T1_A0(); + gen_op_addl_T1_im(16 << s->dflag); + if (s->addseg) + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[R_SS].base)); + for(i = 0;i < 8; i++) { + /* ESP is not reloaded */ + if (i != 3) { + gen_op_ld_T0_A0[OT_WORD + s->dflag + s->mem_index](); + gen_op_mov_reg_T0[OT_WORD + s->dflag][7 - i](); + } + gen_op_addl_A0_im(2 << s->dflag); + } + gen_op_mov_reg_T1[OT_WORD + s->dflag][R_ESP](); +} + +/* NOTE: wrap around in 16 bit not fully handled */ +/* XXX: check this */ +static void gen_enter(DisasContext *s, int esp_addend, int level) +{ + int ot, level1, addend, opsize; + + ot = s->dflag + OT_WORD; + level &= 0x1f; + level1 = level; + opsize = 2 << s->dflag; + + gen_op_movl_A0_ESP(); + gen_op_addl_A0_im(-opsize); + if (!s->ss32) + gen_op_andl_A0_ffff(); + gen_op_movl_T1_A0(); + if (s->addseg) + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[R_SS].base)); + /* push bp */ + gen_op_mov_TN_reg[OT_LONG][0][R_EBP](); + gen_op_st_T0_A0[ot + s->mem_index](); + if (level) { + while (level--) { + gen_op_addl_A0_im(-opsize); + gen_op_addl_T0_im(-opsize); + gen_op_st_T0_A0[ot + s->mem_index](); + } + gen_op_addl_A0_im(-opsize); + /* XXX: add st_T1_A0 ? */ + gen_op_movl_T0_T1(); + gen_op_st_T0_A0[ot + s->mem_index](); + } + gen_op_mov_reg_T1[ot][R_EBP](); + addend = -esp_addend; + if (level1) + addend -= opsize * (level1 + 1); + gen_op_addl_T1_im(addend); + gen_op_mov_reg_T1[ot][R_ESP](); +} + +static void gen_exception(DisasContext *s, int trapno, unsigned int cur_eip) +{ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(cur_eip); + gen_op_raise_exception(trapno); + s->is_jmp = 3; +} + +/* an interrupt is different from an exception because of the + priviledge checks */ +static void gen_interrupt(DisasContext *s, int intno, + unsigned int cur_eip, unsigned int next_eip) +{ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(cur_eip); + gen_op_raise_interrupt(intno, next_eip); + s->is_jmp = 3; +} + +static void gen_debug(DisasContext *s, unsigned int cur_eip) +{ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(cur_eip); + gen_op_debug(); + s->is_jmp = 3; +} + +/* generate a generic end of block. Trace exception is also generated + if needed */ +static void gen_eob(DisasContext *s) +{ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + if (s->tf) { + gen_op_raise_exception(EXCP01_SSTP); + } else { + gen_op_movl_T0_0(); + gen_op_exit_tb(); + } + s->is_jmp = 3; +} + +/* generate a jump to eip. No segment change must happen before as a + direct call to the next block may occur */ +static void gen_jmp(DisasContext *s, unsigned int eip) +{ + TranslationBlock *tb = s->tb; + + if (s->jmp_opt) { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp((long)tb, eip); + s->is_jmp = 3; + } else { + gen_op_jmp_im(eip); + gen_eob(s); + } +} + +/* convert one instruction. s->is_jmp is set if the translation must + be stopped. Return the next pc value */ +static uint8_t *disas_insn(DisasContext *s, uint8_t *pc_start) +{ + int b, prefixes, aflag, dflag; + int shift, ot; + int modrm, reg, rm, mod, reg_addr, op, opreg, offset_addr, val; + unsigned int next_eip; + + s->pc = pc_start; + prefixes = 0; + aflag = s->code32; + dflag = s->code32; + s->override = -1; + next_byte: + b = ldub(s->pc); + s->pc++; + /* check prefixes */ + switch (b) { + case 0xf3: + prefixes |= PREFIX_REPZ; + goto next_byte; + case 0xf2: + prefixes |= PREFIX_REPNZ; + goto next_byte; + case 0xf0: + prefixes |= PREFIX_LOCK; + goto next_byte; + case 0x2e: + s->override = R_CS; + goto next_byte; + case 0x36: + s->override = R_SS; + goto next_byte; + case 0x3e: + s->override = R_DS; + goto next_byte; + case 0x26: + s->override = R_ES; + goto next_byte; + case 0x64: + s->override = R_FS; + goto next_byte; + case 0x65: + s->override = R_GS; + goto next_byte; + case 0x66: + prefixes |= PREFIX_DATA; + goto next_byte; + case 0x67: + prefixes |= PREFIX_ADR; + goto next_byte; + } + + if (prefixes & PREFIX_DATA) + dflag ^= 1; + if (prefixes & PREFIX_ADR) + aflag ^= 1; + + s->prefix = prefixes; + s->aflag = aflag; + s->dflag = dflag; + + /* lock generation */ + if (prefixes & PREFIX_LOCK) + gen_op_lock(); + + /* now check op code */ + reswitch: + switch(b) { + case 0x0f: + /**************************/ + /* extended op code */ + b = ldub(s->pc++) | 0x100; + goto reswitch; + + /**************************/ + /* arith & logic */ + case 0x00 ... 0x05: + case 0x08 ... 0x0d: + case 0x10 ... 0x15: + case 0x18 ... 0x1d: + case 0x20 ... 0x25: + case 0x28 ... 0x2d: + case 0x30 ... 0x35: + case 0x38 ... 0x3d: + { + int op, f, val; + op = (b >> 3) & 7; + f = (b >> 1) & 3; + + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + switch(f) { + case 0: /* OP Ev, Gv */ + modrm = ldub(s->pc++); + reg = ((modrm >> 3) & 7); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + opreg = OR_TMP0; + } else if (op == OP_XORL && rm == reg) { + xor_zero: + /* xor reg, reg optimisation */ + gen_op_movl_T0_0(); + s->cc_op = CC_OP_LOGICB + ot; + gen_op_mov_reg_T0[ot][reg](); + gen_op_update1_cc(); + break; + } else { + opreg = rm; + } + gen_op_mov_TN_reg[ot][1][reg](); + gen_op(s, op, ot, opreg); + break; + case 1: /* OP Gv, Ev */ + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + reg = ((modrm >> 3) & 7); + rm = modrm & 7; + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T1_A0[ot + s->mem_index](); + } else if (op == OP_XORL && rm == reg) { + goto xor_zero; + } else { + gen_op_mov_TN_reg[ot][1][rm](); + } + gen_op(s, op, ot, reg); + break; + case 2: /* OP A, Iv */ + val = insn_get(s, ot); + gen_op_movl_T1_im(val); + gen_op(s, op, ot, OR_EAX); + break; + } + } + break; + + case 0x80: /* GRP1 */ + case 0x81: + case 0x83: + { + int val; + + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + op = (modrm >> 3) & 7; + + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + opreg = OR_TMP0; + } else { + opreg = rm + OR_EAX; + } + + switch(b) { + default: + case 0x80: + case 0x81: + val = insn_get(s, ot); + break; + case 0x83: + val = (int8_t)insn_get(s, OT_BYTE); + break; + } + gen_op_movl_T1_im(val); + gen_op(s, op, ot, opreg); + } + break; + + /**************************/ + /* inc, dec, and other misc arith */ + case 0x40 ... 0x47: /* inc Gv */ + ot = dflag ? OT_LONG : OT_WORD; + gen_inc(s, ot, OR_EAX + (b & 7), 1); + break; + case 0x48 ... 0x4f: /* dec Gv */ + ot = dflag ? OT_LONG : OT_WORD; + gen_inc(s, ot, OR_EAX + (b & 7), -1); + break; + case 0xf6: /* GRP3 */ + case 0xf7: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + op = (modrm >> 3) & 7; + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_TN_reg[ot][0][rm](); + } + + switch(op) { + case 0: /* test */ + val = insn_get(s, ot); + gen_op_movl_T1_im(val); + gen_op_testl_T0_T1_cc(); + s->cc_op = CC_OP_LOGICB + ot; + break; + case 2: /* not */ + gen_op_notl_T0(); + if (mod != 3) { + gen_op_st_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_reg_T0[ot][rm](); + } + break; + case 3: /* neg */ + gen_op_negl_T0(); + if (mod != 3) { + gen_op_st_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_reg_T0[ot][rm](); + } + gen_op_update_neg_cc(); + s->cc_op = CC_OP_SUBB + ot; + break; + case 4: /* mul */ + switch(ot) { + case OT_BYTE: + gen_op_mulb_AL_T0(); + break; + case OT_WORD: + gen_op_mulw_AX_T0(); + break; + default: + case OT_LONG: + gen_op_mull_EAX_T0(); + break; + } + s->cc_op = CC_OP_MUL; + break; + case 5: /* imul */ + switch(ot) { + case OT_BYTE: + gen_op_imulb_AL_T0(); + break; + case OT_WORD: + gen_op_imulw_AX_T0(); + break; + default: + case OT_LONG: + gen_op_imull_EAX_T0(); + break; + } + s->cc_op = CC_OP_MUL; + break; + case 6: /* div */ + switch(ot) { + case OT_BYTE: + gen_op_divb_AL_T0(pc_start - s->cs_base); + break; + case OT_WORD: + gen_op_divw_AX_T0(pc_start - s->cs_base); + break; + default: + case OT_LONG: + gen_op_divl_EAX_T0(pc_start - s->cs_base); + break; + } + break; + case 7: /* idiv */ + switch(ot) { + case OT_BYTE: + gen_op_idivb_AL_T0(pc_start - s->cs_base); + break; + case OT_WORD: + gen_op_idivw_AX_T0(pc_start - s->cs_base); + break; + default: + case OT_LONG: + gen_op_idivl_EAX_T0(pc_start - s->cs_base); + break; + } + break; + default: + goto illegal_op; + } + break; + + case 0xfe: /* GRP4 */ + case 0xff: /* GRP5 */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + op = (modrm >> 3) & 7; + if (op >= 2 && b == 0xfe) { + goto illegal_op; + } + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + if (op >= 2 && op != 3 && op != 5) + gen_op_ld_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_TN_reg[ot][0][rm](); + } + + switch(op) { + case 0: /* inc Ev */ + if (mod != 3) + opreg = OR_TMP0; + else + opreg = rm; + gen_inc(s, ot, opreg, 1); + break; + case 1: /* dec Ev */ + if (mod != 3) + opreg = OR_TMP0; + else + opreg = rm; + gen_inc(s, ot, opreg, -1); + break; + case 2: /* call Ev */ + /* XXX: optimize if memory (no and is necessary) */ + if (s->dflag == 0) + gen_op_andl_T0_ffff(); + gen_op_jmp_T0(); + next_eip = s->pc - s->cs_base; + gen_op_movl_T0_im(next_eip); + gen_push_T0(s); + gen_eob(s); + break; + case 3: /*< lcall Ev */ + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_addl_A0_im(1 << (ot - OT_WORD + 1)); + gen_op_ld_T0_A0[OT_WORD + s->mem_index](); + do_lcall: + if (s->pe && !s->vm86) { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(pc_start - s->cs_base); + gen_op_lcall_protected_T0_T1(dflag, s->pc - s->cs_base); + } else { + gen_op_lcall_real_T0_T1(dflag, s->pc - s->cs_base); + } + gen_eob(s); + break; + case 4: /* jmp Ev */ + if (s->dflag == 0) + gen_op_andl_T0_ffff(); + gen_op_jmp_T0(); + gen_eob(s); + break; + case 5: /* ljmp Ev */ + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_addl_A0_im(1 << (ot - OT_WORD + 1)); + gen_op_lduw_T0_A0(); + do_ljmp: + if (s->pe && !s->vm86) { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(pc_start - s->cs_base); + gen_op_ljmp_protected_T0_T1(); + } else { + gen_op_movl_seg_T0_vm(offsetof(CPUX86State,segs[R_CS])); + gen_op_movl_T0_T1(); + gen_op_jmp_T0(); + } + gen_eob(s); + break; + case 6: /* push Ev */ + gen_push_T0(s); + break; + default: + goto illegal_op; + } + break; + + case 0x84: /* test Ev, Gv */ + case 0x85: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + reg = (modrm >> 3) & 7; + + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 0); + gen_op_mov_TN_reg[ot][1][reg + OR_EAX](); + gen_op_testl_T0_T1_cc(); + s->cc_op = CC_OP_LOGICB + ot; + break; + + case 0xa8: /* test eAX, Iv */ + case 0xa9: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + val = insn_get(s, ot); + + gen_op_mov_TN_reg[ot][0][OR_EAX](); + gen_op_movl_T1_im(val); + gen_op_testl_T0_T1_cc(); + s->cc_op = CC_OP_LOGICB + ot; + break; + + case 0x98: /* CWDE/CBW */ + if (dflag) + gen_op_movswl_EAX_AX(); + else + gen_op_movsbw_AX_AL(); + break; + case 0x99: /* CDQ/CWD */ + if (dflag) + gen_op_movslq_EDX_EAX(); + else + gen_op_movswl_DX_AX(); + break; + case 0x1af: /* imul Gv, Ev */ + case 0x69: /* imul Gv, Ev, I */ + case 0x6b: + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = ((modrm >> 3) & 7) + OR_EAX; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 0); + if (b == 0x69) { + val = insn_get(s, ot); + gen_op_movl_T1_im(val); + } else if (b == 0x6b) { + val = insn_get(s, OT_BYTE); + gen_op_movl_T1_im(val); + } else { + gen_op_mov_TN_reg[ot][1][reg](); + } + + if (ot == OT_LONG) { + gen_op_imull_T0_T1(); + } else { + gen_op_imulw_T0_T1(); + } + gen_op_mov_reg_T0[ot][reg](); + s->cc_op = CC_OP_MUL; + break; + case 0x1c0: + case 0x1c1: /* xadd Ev, Gv */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + if (mod == 3) { + rm = modrm & 7; + gen_op_mov_TN_reg[ot][0][reg](); + gen_op_mov_TN_reg[ot][1][rm](); + gen_op_addl_T0_T1(); + gen_op_mov_reg_T0[ot][rm](); + gen_op_mov_reg_T1[ot][reg](); + } else { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_mov_TN_reg[ot][0][reg](); + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_addl_T0_T1(); + gen_op_st_T0_A0[ot + s->mem_index](); + gen_op_mov_reg_T1[ot][reg](); + } + gen_op_update2_cc(); + s->cc_op = CC_OP_ADDB + ot; + break; + case 0x1b0: + case 0x1b1: /* cmpxchg Ev, Gv */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + gen_op_mov_TN_reg[ot][1][reg](); + if (mod == 3) { + rm = modrm & 7; + gen_op_mov_TN_reg[ot][0][rm](); + gen_op_cmpxchg_T0_T1_EAX_cc[ot](); + gen_op_mov_reg_T0[ot][rm](); + } else { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T0_A0[ot + s->mem_index](); + gen_op_cmpxchg_mem_T0_T1_EAX_cc[ot](); + } + s->cc_op = CC_OP_SUBB + ot; + break; + case 0x1c7: /* cmpxchg8b */ + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + if (mod == 3) + goto illegal_op; + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_cmpxchg8b(); + s->cc_op = CC_OP_EFLAGS; + break; + + /**************************/ + /* push/pop */ + case 0x50 ... 0x57: /* push */ + gen_op_mov_TN_reg[OT_LONG][0][b & 7](); + gen_push_T0(s); + break; + case 0x58 ... 0x5f: /* pop */ + ot = dflag ? OT_LONG : OT_WORD; + gen_pop_T0(s); + gen_op_mov_reg_T0[ot][b & 7](); + gen_pop_update(s); + break; + case 0x60: /* pusha */ + gen_pusha(s); + break; + case 0x61: /* popa */ + gen_popa(s); + break; + case 0x68: /* push Iv */ + case 0x6a: + ot = dflag ? OT_LONG : OT_WORD; + if (b == 0x68) + val = insn_get(s, ot); + else + val = (int8_t)insn_get(s, OT_BYTE); + gen_op_movl_T0_im(val); + gen_push_T0(s); + break; + case 0x8f: /* pop Ev */ + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + gen_pop_T0(s); + s->popl_esp_hack = 2 << dflag; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 1); + s->popl_esp_hack = 0; + gen_pop_update(s); + break; + case 0xc8: /* enter */ + { + int level; + val = lduw(s->pc); + s->pc += 2; + level = ldub(s->pc++); + gen_enter(s, val, level); + } + break; + case 0xc9: /* leave */ + /* XXX: exception not precise (ESP is updated before potential exception) */ + if (s->ss32) { + gen_op_mov_TN_reg[OT_LONG][0][R_EBP](); + gen_op_mov_reg_T0[OT_LONG][R_ESP](); + } else { + gen_op_mov_TN_reg[OT_WORD][0][R_EBP](); + gen_op_mov_reg_T0[OT_WORD][R_ESP](); + } + gen_pop_T0(s); + ot = dflag ? OT_LONG : OT_WORD; + gen_op_mov_reg_T0[ot][R_EBP](); + gen_pop_update(s); + break; + case 0x06: /* push es */ + case 0x0e: /* push cs */ + case 0x16: /* push ss */ + case 0x1e: /* push ds */ + gen_op_movl_T0_seg(b >> 3); + gen_push_T0(s); + break; + case 0x1a0: /* push fs */ + case 0x1a8: /* push gs */ + gen_op_movl_T0_seg((b >> 3) & 7); + gen_push_T0(s); + break; + case 0x07: /* pop es */ + case 0x17: /* pop ss */ + case 0x1f: /* pop ds */ + reg = b >> 3; + gen_pop_T0(s); + gen_movl_seg_T0(s, reg, pc_start - s->cs_base); + gen_pop_update(s); + if (reg == R_SS) { + /* if reg == SS, inhibit interrupts/trace */ + gen_op_set_inhibit_irq(); + s->tf = 0; + } + if (s->is_jmp) { + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } + break; + case 0x1a1: /* pop fs */ + case 0x1a9: /* pop gs */ + gen_pop_T0(s); + gen_movl_seg_T0(s, (b >> 3) & 7, pc_start - s->cs_base); + gen_pop_update(s); + if (s->is_jmp) { + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } + break; + + /**************************/ + /* mov */ + case 0x88: + case 0x89: /* mov Gv, Ev */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + + /* generate a generic store */ + gen_ldst_modrm(s, modrm, ot, OR_EAX + reg, 1); + break; + case 0xc6: + case 0xc7: /* mov Ev, Iv */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + if (mod != 3) + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + val = insn_get(s, ot); + gen_op_movl_T0_im(val); + if (mod != 3) + gen_op_st_T0_A0[ot + s->mem_index](); + else + gen_op_mov_reg_T0[ot][modrm & 7](); + break; + case 0x8a: + case 0x8b: /* mov Ev, Gv */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 0); + gen_op_mov_reg_T0[ot][reg](); + break; + case 0x8e: /* mov seg, Gv */ + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + if (reg >= 6 || reg == R_CS) + goto illegal_op; + gen_ldst_modrm(s, modrm, OT_WORD, OR_TMP0, 0); + gen_movl_seg_T0(s, reg, pc_start - s->cs_base); + if (reg == R_SS) { + /* if reg == SS, inhibit interrupts/trace */ + gen_op_set_inhibit_irq(); + s->tf = 0; + } + if (s->is_jmp) { + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } + break; + case 0x8c: /* mov Gv, seg */ + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + if (reg >= 6) + goto illegal_op; + gen_op_movl_T0_seg(reg); + ot = OT_WORD; + if (mod == 3 && dflag) + ot = OT_LONG; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 1); + break; + + case 0x1b6: /* movzbS Gv, Eb */ + case 0x1b7: /* movzwS Gv, Eb */ + case 0x1be: /* movsbS Gv, Eb */ + case 0x1bf: /* movswS Gv, Eb */ + { + int d_ot; + /* d_ot is the size of destination */ + d_ot = dflag + OT_WORD; + /* ot is the size of source */ + ot = (b & 1) + OT_BYTE; + modrm = ldub(s->pc++); + reg = ((modrm >> 3) & 7) + OR_EAX; + mod = (modrm >> 6) & 3; + rm = modrm & 7; + + if (mod == 3) { + gen_op_mov_TN_reg[ot][0][rm](); + switch(ot | (b & 8)) { + case OT_BYTE: + gen_op_movzbl_T0_T0(); + break; + case OT_BYTE | 8: + gen_op_movsbl_T0_T0(); + break; + case OT_WORD: + gen_op_movzwl_T0_T0(); + break; + default: + case OT_WORD | 8: + gen_op_movswl_T0_T0(); + break; + } + gen_op_mov_reg_T0[d_ot][reg](); + } else { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + if (b & 8) { + gen_op_lds_T0_A0[ot + s->mem_index](); + } else { + gen_op_ldu_T0_A0[ot + s->mem_index](); + } + gen_op_mov_reg_T0[d_ot][reg](); + } + } + break; + + case 0x8d: /* lea */ + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + /* we must ensure that no segment is added */ + s->override = -1; + val = s->addseg; + s->addseg = 0; + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + s->addseg = val; + gen_op_mov_reg_A0[ot - OT_WORD][reg](); + break; + + case 0xa0: /* mov EAX, Ov */ + case 0xa1: + case 0xa2: /* mov Ov, EAX */ + case 0xa3: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + if (s->aflag) + offset_addr = insn_get(s, OT_LONG); + else + offset_addr = insn_get(s, OT_WORD); + gen_op_movl_A0_im(offset_addr); + /* handle override */ + { + int override, must_add_seg; + must_add_seg = s->addseg; + if (s->override >= 0) { + override = s->override; + must_add_seg = 1; + } else { + override = R_DS; + } + if (must_add_seg) { + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[override].base)); + } + } + if ((b & 2) == 0) { + gen_op_ld_T0_A0[ot + s->mem_index](); + gen_op_mov_reg_T0[ot][R_EAX](); + } else { + gen_op_mov_TN_reg[ot][0][R_EAX](); + gen_op_st_T0_A0[ot + s->mem_index](); + } + break; + case 0xd7: /* xlat */ + gen_op_movl_A0_reg[R_EBX](); + gen_op_addl_A0_AL(); + if (s->aflag == 0) + gen_op_andl_A0_ffff(); + /* handle override */ + { + int override, must_add_seg; + must_add_seg = s->addseg; + override = R_DS; + if (s->override >= 0) { + override = s->override; + must_add_seg = 1; + } else { + override = R_DS; + } + if (must_add_seg) { + gen_op_addl_A0_seg(offsetof(CPUX86State,segs[override].base)); + } + } + gen_op_ldu_T0_A0[OT_BYTE + s->mem_index](); + gen_op_mov_reg_T0[OT_BYTE][R_EAX](); + break; + case 0xb0 ... 0xb7: /* mov R, Ib */ + val = insn_get(s, OT_BYTE); + gen_op_movl_T0_im(val); + gen_op_mov_reg_T0[OT_BYTE][b & 7](); + break; + case 0xb8 ... 0xbf: /* mov R, Iv */ + ot = dflag ? OT_LONG : OT_WORD; + val = insn_get(s, ot); + reg = OR_EAX + (b & 7); + gen_op_movl_T0_im(val); + gen_op_mov_reg_T0[ot][reg](); + break; + + case 0x91 ... 0x97: /* xchg R, EAX */ + ot = dflag ? OT_LONG : OT_WORD; + reg = b & 7; + rm = R_EAX; + goto do_xchg_reg; + case 0x86: + case 0x87: /* xchg Ev, Gv */ + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + if (mod == 3) { + rm = modrm & 7; + do_xchg_reg: + gen_op_mov_TN_reg[ot][0][reg](); + gen_op_mov_TN_reg[ot][1][rm](); + gen_op_mov_reg_T0[ot][rm](); + gen_op_mov_reg_T1[ot][reg](); + } else { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_mov_TN_reg[ot][0][reg](); + /* for xchg, lock is implicit */ + if (!(prefixes & PREFIX_LOCK)) + gen_op_lock(); + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_st_T0_A0[ot + s->mem_index](); + if (!(prefixes & PREFIX_LOCK)) + gen_op_unlock(); + gen_op_mov_reg_T1[ot][reg](); + } + break; + case 0xc4: /* les Gv */ + op = R_ES; + goto do_lxx; + case 0xc5: /* lds Gv */ + op = R_DS; + goto do_lxx; + case 0x1b2: /* lss Gv */ + op = R_SS; + goto do_lxx; + case 0x1b4: /* lfs Gv */ + op = R_FS; + goto do_lxx; + case 0x1b5: /* lgs Gv */ + op = R_GS; + do_lxx: + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + if (mod == 3) + goto illegal_op; + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T1_A0[ot + s->mem_index](); + gen_op_addl_A0_im(1 << (ot - OT_WORD + 1)); + /* load the segment first to handle exceptions properly */ + gen_op_lduw_T0_A0(); + gen_movl_seg_T0(s, op, pc_start - s->cs_base); + /* then put the data */ + gen_op_mov_reg_T1[ot][reg](); + if (s->is_jmp) { + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } + break; + + /************************/ + /* shifts */ + case 0xc0: + case 0xc1: + /* shift Ev,Ib */ + shift = 2; + grp2: + { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + op = (modrm >> 3) & 7; + + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + opreg = OR_TMP0; + } else { + opreg = rm + OR_EAX; + } + + /* simpler op */ + if (shift == 0) { + gen_shift(s, op, ot, opreg, OR_ECX); + } else { + if (shift == 2) { + shift = ldub(s->pc++); + } + gen_shifti(s, op, ot, opreg, shift); + } + } + break; + case 0xd0: + case 0xd1: + /* shift Ev,1 */ + shift = 1; + goto grp2; + case 0xd2: + case 0xd3: + /* shift Ev,cl */ + shift = 0; + goto grp2; + + case 0x1a4: /* shld imm */ + op = 0; + shift = 1; + goto do_shiftd; + case 0x1a5: /* shld cl */ + op = 0; + shift = 0; + goto do_shiftd; + case 0x1ac: /* shrd imm */ + op = 1; + shift = 1; + goto do_shiftd; + case 0x1ad: /* shrd cl */ + op = 1; + shift = 0; + do_shiftd: + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + reg = (modrm >> 3) & 7; + + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_TN_reg[ot][0][rm](); + } + gen_op_mov_TN_reg[ot][1][reg](); + + if (shift) { + val = ldub(s->pc++); + val &= 0x1f; + if (val) { + if (mod == 3) + gen_op_shiftd_T0_T1_im_cc[ot - OT_WORD][op](val); + else + gen_op_shiftd_mem_T0_T1_im_cc[ot - OT_WORD][op](val); + if (op == 0 && ot != OT_WORD) + s->cc_op = CC_OP_SHLB + ot; + else + s->cc_op = CC_OP_SARB + ot; + } + } else { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + if (mod == 3) + gen_op_shiftd_T0_T1_ECX_cc[ot - OT_WORD][op](); + else + gen_op_shiftd_mem_T0_T1_ECX_cc[ot - OT_WORD][op](); + s->cc_op = CC_OP_DYNAMIC; /* cannot predict flags after */ + } + if (mod == 3) { + gen_op_mov_reg_T0[ot][rm](); + } + break; + + /************************/ + /* floats */ + case 0xd8 ... 0xdf: + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + rm = modrm & 7; + op = ((b & 7) << 3) | ((modrm >> 3) & 7); + + if (mod != 3) { + /* memory op */ + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + switch(op) { + case 0x00 ... 0x07: /* fxxxs */ + case 0x10 ... 0x17: /* fixxxl */ + case 0x20 ... 0x27: /* fxxxl */ + case 0x30 ... 0x37: /* fixxx */ + { + int op1; + op1 = op & 7; + + switch(op >> 4) { + case 0: + gen_op_flds_FT0_A0(); + break; + case 1: + gen_op_fildl_FT0_A0(); + break; + case 2: + gen_op_fldl_FT0_A0(); + break; + case 3: + default: + gen_op_fild_FT0_A0(); + break; + } + + gen_op_fp_arith_ST0_FT0[op1](); + if (op1 == 3) { + /* fcomp needs pop */ + gen_op_fpop(); + } + } + break; + case 0x08: /* flds */ + case 0x0a: /* fsts */ + case 0x0b: /* fstps */ + case 0x18: /* fildl */ + case 0x1a: /* fistl */ + case 0x1b: /* fistpl */ + case 0x28: /* fldl */ + case 0x2a: /* fstl */ + case 0x2b: /* fstpl */ + case 0x38: /* filds */ + case 0x3a: /* fists */ + case 0x3b: /* fistps */ + + switch(op & 7) { + case 0: + switch(op >> 4) { + case 0: + gen_op_flds_ST0_A0(); + break; + case 1: + gen_op_fildl_ST0_A0(); + break; + case 2: + gen_op_fldl_ST0_A0(); + break; + case 3: + default: + gen_op_fild_ST0_A0(); + break; + } + break; + default: + switch(op >> 4) { + case 0: + gen_op_fsts_ST0_A0(); + break; + case 1: + gen_op_fistl_ST0_A0(); + break; + case 2: + gen_op_fstl_ST0_A0(); + break; + case 3: + default: + gen_op_fist_ST0_A0(); + break; + } + if ((op & 7) == 3) + gen_op_fpop(); + break; + } + break; + case 0x0c: /* fldenv mem */ + gen_op_fldenv_A0(s->dflag); + break; + case 0x0d: /* fldcw mem */ + gen_op_fldcw_A0(); + break; + case 0x0e: /* fnstenv mem */ + gen_op_fnstenv_A0(s->dflag); + break; + case 0x0f: /* fnstcw mem */ + gen_op_fnstcw_A0(); + break; + case 0x1d: /* fldt mem */ + gen_op_fldt_ST0_A0(); + break; + case 0x1f: /* fstpt mem */ + gen_op_fstt_ST0_A0(); + gen_op_fpop(); + break; + case 0x2c: /* frstor mem */ + gen_op_frstor_A0(s->dflag); + break; + case 0x2e: /* fnsave mem */ + gen_op_fnsave_A0(s->dflag); + break; + case 0x2f: /* fnstsw mem */ + gen_op_fnstsw_A0(); + break; + case 0x3c: /* fbld */ + gen_op_fbld_ST0_A0(); + break; + case 0x3e: /* fbstp */ + gen_op_fbst_ST0_A0(); + gen_op_fpop(); + break; + case 0x3d: /* fildll */ + gen_op_fildll_ST0_A0(); + break; + case 0x3f: /* fistpll */ + gen_op_fistll_ST0_A0(); + gen_op_fpop(); + break; + default: + goto illegal_op; + } + } else { + /* register float ops */ + opreg = rm; + + switch(op) { + case 0x08: /* fld sti */ + gen_op_fpush(); + gen_op_fmov_ST0_STN((opreg + 1) & 7); + break; + case 0x09: /* fxchg sti */ + gen_op_fxchg_ST0_STN(opreg); + break; + case 0x0a: /* grp d9/2 */ + switch(rm) { + case 0: /* fnop */ + break; + default: + goto illegal_op; + } + break; + case 0x0c: /* grp d9/4 */ + switch(rm) { + case 0: /* fchs */ + gen_op_fchs_ST0(); + break; + case 1: /* fabs */ + gen_op_fabs_ST0(); + break; + case 4: /* ftst */ + gen_op_fldz_FT0(); + gen_op_fcom_ST0_FT0(); + break; + case 5: /* fxam */ + gen_op_fxam_ST0(); + break; + default: + goto illegal_op; + } + break; + case 0x0d: /* grp d9/5 */ + { + switch(rm) { + case 0: + gen_op_fpush(); + gen_op_fld1_ST0(); + break; + case 1: + gen_op_fpush(); + gen_op_fldl2t_ST0(); + break; + case 2: + gen_op_fpush(); + gen_op_fldl2e_ST0(); + break; + case 3: + gen_op_fpush(); + gen_op_fldpi_ST0(); + break; + case 4: + gen_op_fpush(); + gen_op_fldlg2_ST0(); + break; + case 5: + gen_op_fpush(); + gen_op_fldln2_ST0(); + break; + case 6: + gen_op_fpush(); + gen_op_fldz_ST0(); + break; + default: + goto illegal_op; + } + } + break; + case 0x0e: /* grp d9/6 */ + switch(rm) { + case 0: /* f2xm1 */ + gen_op_f2xm1(); + break; + case 1: /* fyl2x */ + gen_op_fyl2x(); + break; + case 2: /* fptan */ + gen_op_fptan(); + break; + case 3: /* fpatan */ + gen_op_fpatan(); + break; + case 4: /* fxtract */ + gen_op_fxtract(); + break; + case 5: /* fprem1 */ + gen_op_fprem1(); + break; + case 6: /* fdecstp */ + gen_op_fdecstp(); + break; + default: + case 7: /* fincstp */ + gen_op_fincstp(); + break; + } + break; + case 0x0f: /* grp d9/7 */ + switch(rm) { + case 0: /* fprem */ + gen_op_fprem(); + break; + case 1: /* fyl2xp1 */ + gen_op_fyl2xp1(); + break; + case 2: /* fsqrt */ + gen_op_fsqrt(); + break; + case 3: /* fsincos */ + gen_op_fsincos(); + break; + case 5: /* fscale */ + gen_op_fscale(); + break; + case 4: /* frndint */ + gen_op_frndint(); + break; + case 6: /* fsin */ + gen_op_fsin(); + break; + default: + case 7: /* fcos */ + gen_op_fcos(); + break; + } + break; + case 0x00: case 0x01: case 0x04 ... 0x07: /* fxxx st, sti */ + case 0x20: case 0x21: case 0x24 ... 0x27: /* fxxx sti, st */ + case 0x30: case 0x31: case 0x34 ... 0x37: /* fxxxp sti, st */ + { + int op1; + + op1 = op & 7; + if (op >= 0x20) { + gen_op_fp_arith_STN_ST0[op1](opreg); + if (op >= 0x30) + gen_op_fpop(); + } else { + gen_op_fmov_FT0_STN(opreg); + gen_op_fp_arith_ST0_FT0[op1](); + } + } + break; + case 0x02: /* fcom */ + gen_op_fmov_FT0_STN(opreg); + gen_op_fcom_ST0_FT0(); + break; + case 0x03: /* fcomp */ + gen_op_fmov_FT0_STN(opreg); + gen_op_fcom_ST0_FT0(); + gen_op_fpop(); + break; + case 0x15: /* da/5 */ + switch(rm) { + case 1: /* fucompp */ + gen_op_fmov_FT0_STN(1); + gen_op_fucom_ST0_FT0(); + gen_op_fpop(); + gen_op_fpop(); + break; + default: + goto illegal_op; + } + break; + case 0x1c: + switch(rm) { + case 0: /* feni (287 only, just do nop here) */ + break; + case 1: /* fdisi (287 only, just do nop here) */ + break; + case 2: /* fclex */ + gen_op_fclex(); + break; + case 3: /* fninit */ + gen_op_fninit(); + break; + case 4: /* fsetpm (287 only, just do nop here) */ + break; + default: + goto illegal_op; + } + break; + case 0x1d: /* fucomi */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_fmov_FT0_STN(opreg); + gen_op_fucomi_ST0_FT0(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x1e: /* fcomi */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_fmov_FT0_STN(opreg); + gen_op_fcomi_ST0_FT0(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x2a: /* fst sti */ + gen_op_fmov_STN_ST0(opreg); + break; + case 0x2b: /* fstp sti */ + gen_op_fmov_STN_ST0(opreg); + gen_op_fpop(); + break; + case 0x2c: /* fucom st(i) */ + gen_op_fmov_FT0_STN(opreg); + gen_op_fucom_ST0_FT0(); + break; + case 0x2d: /* fucomp st(i) */ + gen_op_fmov_FT0_STN(opreg); + gen_op_fucom_ST0_FT0(); + gen_op_fpop(); + break; + case 0x33: /* de/3 */ + switch(rm) { + case 1: /* fcompp */ + gen_op_fmov_FT0_STN(1); + gen_op_fcom_ST0_FT0(); + gen_op_fpop(); + gen_op_fpop(); + break; + default: + goto illegal_op; + } + break; + case 0x3c: /* df/4 */ + switch(rm) { + case 0: + gen_op_fnstsw_EAX(); + break; + default: + goto illegal_op; + } + break; + case 0x3d: /* fucomip */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_fmov_FT0_STN(opreg); + gen_op_fucomi_ST0_FT0(); + gen_op_fpop(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x3e: /* fcomip */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_fmov_FT0_STN(opreg); + gen_op_fcomi_ST0_FT0(); + gen_op_fpop(); + s->cc_op = CC_OP_EFLAGS; + break; + default: + goto illegal_op; + } + } + break; + /************************/ + /* string ops */ + + case 0xa4: /* movsS */ + case 0xa5: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) { + gen_repz_movs(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); + } else { + gen_movs(s, ot); + } + break; + + case 0xaa: /* stosS */ + case 0xab: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + + if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) { + gen_repz_stos(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); + } else { + gen_stos(s, ot); + } + break; + case 0xac: /* lodsS */ + case 0xad: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) { + gen_repz_lods(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); + } else { + gen_lods(s, ot); + } + break; + case 0xae: /* scasS */ + case 0xaf: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + if (prefixes & PREFIX_REPNZ) { + gen_repz_scas(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 1); + } else if (prefixes & PREFIX_REPZ) { + gen_repz_scas(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 0); + } else { + gen_scas(s, ot); + s->cc_op = CC_OP_SUBB + ot; + } + break; + + case 0xa6: /* cmpsS */ + case 0xa7: + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + if (prefixes & PREFIX_REPNZ) { + gen_repz_cmps(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 1); + } else if (prefixes & PREFIX_REPZ) { + gen_repz_cmps(s, ot, pc_start - s->cs_base, s->pc - s->cs_base, 0); + } else { + gen_cmps(s, ot); + s->cc_op = CC_OP_SUBB + ot; + } + break; + case 0x6c: /* insS */ + case 0x6d: + if (s->pe && (s->cpl > s->iopl || s->vm86)) { + /* NOTE: even for (E)CX = 0 the exception is raised */ + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) { + gen_repz_ins(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); + } else { + gen_ins(s, ot); + } + } + break; + case 0x6e: /* outsS */ + case 0x6f: + if (s->pe && (s->cpl > s->iopl || s->vm86)) { + /* NOTE: even for (E)CX = 0 the exception is raised */ + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + if (prefixes & (PREFIX_REPZ | PREFIX_REPNZ)) { + gen_repz_outs(s, ot, pc_start - s->cs_base, s->pc - s->cs_base); + } else { + gen_outs(s, ot); + } + } + break; + + /************************/ + /* port I/O */ + case 0xe4: + case 0xe5: + if (s->pe && (s->cpl > s->iopl || s->vm86)) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + val = ldub(s->pc++); + gen_op_movl_T0_im(val); + gen_op_in[ot](); + gen_op_mov_reg_T1[ot][R_EAX](); + } + break; + case 0xe6: + case 0xe7: + if (s->pe && (s->cpl > s->iopl || s->vm86)) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + val = ldub(s->pc++); + gen_op_movl_T0_im(val); + gen_op_mov_TN_reg[ot][1][R_EAX](); + gen_op_out[ot](); + } + break; + case 0xec: + case 0xed: + if (s->pe && (s->cpl > s->iopl || s->vm86)) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + gen_op_mov_TN_reg[OT_WORD][0][R_EDX](); + gen_op_in[ot](); + gen_op_mov_reg_T1[ot][R_EAX](); + } + break; + case 0xee: + case 0xef: + if (s->pe && (s->cpl > s->iopl || s->vm86)) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if ((b & 1) == 0) + ot = OT_BYTE; + else + ot = dflag ? OT_LONG : OT_WORD; + gen_op_mov_TN_reg[OT_WORD][0][R_EDX](); + gen_op_mov_TN_reg[ot][1][R_EAX](); + gen_op_out[ot](); + } + break; + + /************************/ + /* control */ + case 0xc2: /* ret im */ + val = ldsw(s->pc); + s->pc += 2; + gen_pop_T0(s); + gen_stack_update(s, val + (2 << s->dflag)); + if (s->dflag == 0) + gen_op_andl_T0_ffff(); + gen_op_jmp_T0(); + gen_eob(s); + break; + case 0xc3: /* ret */ + gen_pop_T0(s); + gen_pop_update(s); + if (s->dflag == 0) + gen_op_andl_T0_ffff(); + gen_op_jmp_T0(); + gen_eob(s); + break; + case 0xca: /* lret im */ + val = ldsw(s->pc); + s->pc += 2; + do_lret: + if (s->pe && !s->vm86) { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(pc_start - s->cs_base); + gen_op_lret_protected(s->dflag, val); + } else { + gen_stack_A0(s); + /* pop offset */ + gen_op_ld_T0_A0[1 + s->dflag + s->mem_index](); + if (s->dflag == 0) + gen_op_andl_T0_ffff(); + /* NOTE: keeping EIP updated is not a problem in case of + exception */ + gen_op_jmp_T0(); + /* pop selector */ + gen_op_addl_A0_im(2 << s->dflag); + gen_op_ld_T0_A0[1 + s->dflag + s->mem_index](); + gen_op_movl_seg_T0_vm(offsetof(CPUX86State,segs[R_CS])); + /* add stack offset */ + gen_stack_update(s, val + (4 << s->dflag)); + } + gen_eob(s); + break; + case 0xcb: /* lret */ + val = 0; + goto do_lret; + case 0xcf: /* iret */ + if (!s->pe) { + /* real mode */ + gen_op_iret_real(s->dflag); + s->cc_op = CC_OP_EFLAGS; + } else if (s->vm86 && s->iopl != 3) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(pc_start - s->cs_base); + gen_op_iret_protected(s->dflag); + s->cc_op = CC_OP_EFLAGS; + } + gen_eob(s); + break; + case 0xe8: /* call im */ + { + unsigned int next_eip; + ot = dflag ? OT_LONG : OT_WORD; + val = insn_get(s, ot); + next_eip = s->pc - s->cs_base; + val += next_eip; + if (s->dflag == 0) + val &= 0xffff; + gen_op_movl_T0_im(next_eip); + gen_push_T0(s); + gen_jmp(s, val); + } + break; + case 0x9a: /* lcall im */ + { + unsigned int selector, offset; + + ot = dflag ? OT_LONG : OT_WORD; + offset = insn_get(s, ot); + selector = insn_get(s, OT_WORD); + + gen_op_movl_T0_im(selector); + gen_op_movl_T1_im(offset); + } + goto do_lcall; + case 0xe9: /* jmp */ + ot = dflag ? OT_LONG : OT_WORD; + val = insn_get(s, ot); + val += s->pc - s->cs_base; + if (s->dflag == 0) + val = val & 0xffff; + gen_jmp(s, val); + break; + case 0xea: /* ljmp im */ + { + unsigned int selector, offset; + + ot = dflag ? OT_LONG : OT_WORD; + offset = insn_get(s, ot); + selector = insn_get(s, OT_WORD); + + gen_op_movl_T0_im(selector); + gen_op_movl_T1_im(offset); + } + goto do_ljmp; + case 0xeb: /* jmp Jb */ + val = (int8_t)insn_get(s, OT_BYTE); + val += s->pc - s->cs_base; + if (s->dflag == 0) + val = val & 0xffff; + gen_jmp(s, val); + break; + case 0x70 ... 0x7f: /* jcc Jb */ + val = (int8_t)insn_get(s, OT_BYTE); + goto do_jcc; + case 0x180 ... 0x18f: /* jcc Jv */ + if (dflag) { + val = insn_get(s, OT_LONG); + } else { + val = (int16_t)insn_get(s, OT_WORD); + } + do_jcc: + next_eip = s->pc - s->cs_base; + val += next_eip; + if (s->dflag == 0) + val &= 0xffff; + gen_jcc(s, b, val, next_eip); + break; + + case 0x190 ... 0x19f: /* setcc Gv */ + modrm = ldub(s->pc++); + gen_setcc(s, b); + gen_ldst_modrm(s, modrm, OT_BYTE, OR_TMP0, 1); + break; + case 0x140 ... 0x14f: /* cmov Gv, Ev */ + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + gen_setcc(s, b); + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T1_A0[ot + s->mem_index](); + } else { + rm = modrm & 7; + gen_op_mov_TN_reg[ot][1][rm](); + } + gen_op_cmov_reg_T1_T0[ot - OT_WORD][reg](); + break; + + /************************/ + /* flags */ + case 0x9c: /* pushf */ + if (s->vm86 && s->iopl != 3) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_movl_T0_eflags(); + gen_push_T0(s); + } + break; + case 0x9d: /* popf */ + if (s->vm86 && s->iopl != 3) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + gen_pop_T0(s); + if (s->cpl == 0) { + if (s->dflag) { + gen_op_movl_eflags_T0_cpl0(); + } else { + gen_op_movw_eflags_T0_cpl0(); + } + } else { + if (s->dflag) { + gen_op_movl_eflags_T0(); + } else { + gen_op_movw_eflags_T0(); + } + } + gen_pop_update(s); + s->cc_op = CC_OP_EFLAGS; + /* abort translation because TF flag may change */ + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } + break; + case 0x9e: /* sahf */ + gen_op_mov_TN_reg[OT_BYTE][0][R_AH](); + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_movb_eflags_T0(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x9f: /* lahf */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_movl_T0_eflags(); + gen_op_mov_reg_T0[OT_BYTE][R_AH](); + break; + case 0xf5: /* cmc */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_cmc(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0xf8: /* clc */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_clc(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0xf9: /* stc */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_stc(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0xfc: /* cld */ + gen_op_cld(); + break; + case 0xfd: /* std */ + gen_op_std(); + break; + + /************************/ + /* bit operations */ + case 0x1ba: /* bt/bts/btr/btc Gv, im */ + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + op = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + rm = modrm & 7; + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_TN_reg[ot][0][rm](); + } + /* load shift */ + val = ldub(s->pc++); + gen_op_movl_T1_im(val); + if (op < 4) + goto illegal_op; + op -= 4; + gen_op_btx_T0_T1_cc[ot - OT_WORD][op](); + s->cc_op = CC_OP_SARB + ot; + if (op != 0) { + if (mod != 3) + gen_op_st_T0_A0[ot + s->mem_index](); + else + gen_op_mov_reg_T0[ot][rm](); + gen_op_update_bt_cc(); + } + break; + case 0x1a3: /* bt Gv, Ev */ + op = 0; + goto do_btx; + case 0x1ab: /* bts */ + op = 1; + goto do_btx; + case 0x1b3: /* btr */ + op = 2; + goto do_btx; + case 0x1bb: /* btc */ + op = 3; + do_btx: + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + rm = modrm & 7; + gen_op_mov_TN_reg[OT_LONG][1][reg](); + if (mod != 3) { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + /* specific case: we need to add a displacement */ + if (ot == OT_WORD) + gen_op_add_bitw_A0_T1(); + else + gen_op_add_bitl_A0_T1(); + gen_op_ld_T0_A0[ot + s->mem_index](); + } else { + gen_op_mov_TN_reg[ot][0][rm](); + } + gen_op_btx_T0_T1_cc[ot - OT_WORD][op](); + s->cc_op = CC_OP_SARB + ot; + if (op != 0) { + if (mod != 3) + gen_op_st_T0_A0[ot + s->mem_index](); + else + gen_op_mov_reg_T0[ot][rm](); + gen_op_update_bt_cc(); + } + break; + case 0x1bc: /* bsf */ + case 0x1bd: /* bsr */ + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 0); + gen_op_bsx_T0_cc[ot - OT_WORD][b & 1](); + /* NOTE: we always write back the result. Intel doc says it is + undefined if T0 == 0 */ + gen_op_mov_reg_T0[ot][reg](); + s->cc_op = CC_OP_LOGICB + ot; + break; + /************************/ + /* bcd */ + case 0x27: /* daa */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_daa(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x2f: /* das */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_das(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x37: /* aaa */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_aaa(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0x3f: /* aas */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_aas(); + s->cc_op = CC_OP_EFLAGS; + break; + case 0xd4: /* aam */ + val = ldub(s->pc++); + gen_op_aam(val); + s->cc_op = CC_OP_LOGICB; + break; + case 0xd5: /* aad */ + val = ldub(s->pc++); + gen_op_aad(val); + s->cc_op = CC_OP_LOGICB; + break; + /************************/ + /* misc */ + case 0x90: /* nop */ + break; + case 0x9b: /* fwait */ + break; + case 0xcc: /* int3 */ + gen_interrupt(s, EXCP03_INT3, pc_start - s->cs_base, s->pc - s->cs_base); + break; + case 0xcd: /* int N */ + val = ldub(s->pc++); + /* XXX: add error code for vm86 GPF */ + if (!s->vm86) + gen_interrupt(s, val, pc_start - s->cs_base, s->pc - s->cs_base); + else + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + break; + case 0xce: /* into */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_into(s->pc - s->cs_base); + break; + case 0xf1: /* icebp (undocumented, exits to external debugger) */ + gen_debug(s, pc_start - s->cs_base); + break; + case 0xfa: /* cli */ + if (!s->vm86) { + if (s->cpl <= s->iopl) { + gen_op_cli(); + } else { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } + } else { + if (s->iopl == 3) { + gen_op_cli(); + } else { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } + } + break; + case 0xfb: /* sti */ + if (!s->vm86) { + if (s->cpl <= s->iopl) { + gen_sti: + gen_op_sti(); + /* interruptions are enabled only the first insn after sti */ + gen_op_set_inhibit_irq(); + /* give a chance to handle pending irqs */ + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } else { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } + } else { + if (s->iopl == 3) { + goto gen_sti; + } else { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } + } + break; + case 0x62: /* bound */ + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + mod = (modrm >> 6) & 3; + if (mod == 3) + goto illegal_op; + gen_op_mov_reg_T0[ot][reg](); + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + if (ot == OT_WORD) + gen_op_boundw(pc_start - s->cs_base); + else + gen_op_boundl(pc_start - s->cs_base); + break; + case 0x1c8 ... 0x1cf: /* bswap reg */ + reg = b & 7; + gen_op_mov_TN_reg[OT_LONG][0][reg](); + gen_op_bswapl_T0(); + gen_op_mov_reg_T0[OT_LONG][reg](); + break; + case 0xd6: /* salc */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_salc(); + break; + case 0xe0: /* loopnz */ + case 0xe1: /* loopz */ + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + /* FALL THRU */ + case 0xe2: /* loop */ + case 0xe3: /* jecxz */ + val = (int8_t)insn_get(s, OT_BYTE); + next_eip = s->pc - s->cs_base; + val += next_eip; + if (s->dflag == 0) + val &= 0xffff; + gen_op_loop[s->aflag][b & 3](val, next_eip); + gen_eob(s); + break; + case 0x130: /* wrmsr */ + case 0x132: /* rdmsr */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if (b & 2) + gen_op_rdmsr(); + else + gen_op_wrmsr(); + } + break; + case 0x131: /* rdtsc */ + gen_op_rdtsc(); + break; + case 0x1a2: /* cpuid */ + gen_op_cpuid(); + break; + case 0xf4: /* hlt */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + gen_op_jmp_im(s->pc - s->cs_base); + gen_op_hlt(); + s->is_jmp = 3; + } + break; + case 0x100: + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + op = (modrm >> 3) & 7; + switch(op) { + case 0: /* sldt */ + gen_op_movl_T0_env(offsetof(CPUX86State,ldt.selector)); + ot = OT_WORD; + if (mod == 3) + ot += s->dflag; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 1); + break; + case 2: /* lldt */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + gen_ldst_modrm(s, modrm, OT_WORD, OR_TMP0, 0); + gen_op_jmp_im(pc_start - s->cs_base); + gen_op_lldt_T0(); + } + break; + case 1: /* str */ + gen_op_movl_T0_env(offsetof(CPUX86State,tr.selector)); + ot = OT_WORD; + if (mod == 3) + ot += s->dflag; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 1); + break; + case 3: /* ltr */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + gen_ldst_modrm(s, modrm, OT_WORD, OR_TMP0, 0); + gen_op_jmp_im(pc_start - s->cs_base); + gen_op_ltr_T0(); + } + break; + case 4: /* verr */ + case 5: /* verw */ + default: + goto illegal_op; + } + break; + case 0x101: + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + op = (modrm >> 3) & 7; + switch(op) { + case 0: /* sgdt */ + case 1: /* sidt */ + if (mod == 3) + goto illegal_op; + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + if (op == 0) + gen_op_movl_T0_env(offsetof(CPUX86State,gdt.limit)); + else + gen_op_movl_T0_env(offsetof(CPUX86State,idt.limit)); + gen_op_st_T0_A0[OT_WORD + s->mem_index](); + gen_op_addl_A0_im(2); + if (op == 0) + gen_op_movl_T0_env(offsetof(CPUX86State,gdt.base)); + else + gen_op_movl_T0_env(offsetof(CPUX86State,idt.base)); + if (!s->dflag) + gen_op_andl_T0_im(0xffffff); + gen_op_st_T0_A0[OT_LONG + s->mem_index](); + break; + case 2: /* lgdt */ + case 3: /* lidt */ + if (mod == 3) + goto illegal_op; + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_ld_T1_A0[OT_WORD + s->mem_index](); + gen_op_addl_A0_im(2); + gen_op_ld_T0_A0[OT_LONG + s->mem_index](); + if (!s->dflag) + gen_op_andl_T0_im(0xffffff); + if (op == 2) { + gen_op_movl_env_T0(offsetof(CPUX86State,gdt.base)); + gen_op_movl_env_T1(offsetof(CPUX86State,gdt.limit)); + } else { + gen_op_movl_env_T0(offsetof(CPUX86State,idt.base)); + gen_op_movl_env_T1(offsetof(CPUX86State,idt.limit)); + } + } + break; + case 4: /* smsw */ + gen_op_movl_T0_env(offsetof(CPUX86State,cr[0])); + gen_ldst_modrm(s, modrm, OT_WORD, OR_TMP0, 1); + break; + case 6: /* lmsw */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + gen_ldst_modrm(s, modrm, OT_WORD, OR_TMP0, 0); + gen_op_lmsw_T0(); + } + break; + case 7: /* invlpg */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + if (mod == 3) + goto illegal_op; + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + gen_op_invlpg_A0(); + } + break; + default: + goto illegal_op; + } + break; + case 0x102: /* lar */ + case 0x103: /* lsl */ + if (!s->pe || s->vm86) + goto illegal_op; + ot = dflag ? OT_LONG : OT_WORD; + modrm = ldub(s->pc++); + reg = (modrm >> 3) & 7; + gen_ldst_modrm(s, modrm, ot, OR_TMP0, 0); + gen_op_mov_TN_reg[ot][1][reg](); + if (s->cc_op != CC_OP_DYNAMIC) + gen_op_set_cc_op(s->cc_op); + if (b == 0x102) + gen_op_lar(); + else + gen_op_lsl(); + s->cc_op = CC_OP_EFLAGS; + gen_op_mov_reg_T1[ot][reg](); + break; + case 0x118: + modrm = ldub(s->pc++); + mod = (modrm >> 6) & 3; + op = (modrm >> 3) & 7; + switch(op) { + case 0: /* prefetchnta */ + case 1: /* prefetchnt0 */ + case 2: /* prefetchnt0 */ + case 3: /* prefetchnt0 */ + if (mod == 3) + goto illegal_op; + gen_lea_modrm(s, modrm, ®_addr, &offset_addr); + /* nothing more to do */ + break; + default: + goto illegal_op; + } + break; + case 0x120: /* mov reg, crN */ + case 0x122: /* mov crN, reg */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + modrm = ldub(s->pc++); + if ((modrm & 0xc0) != 0xc0) + goto illegal_op; + rm = modrm & 7; + reg = (modrm >> 3) & 7; + switch(reg) { + case 0: + case 2: + case 3: + case 4: + if (b & 2) { + gen_op_mov_TN_reg[OT_LONG][0][rm](); + gen_op_movl_crN_T0(reg); + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } else { + gen_op_movl_T0_env(offsetof(CPUX86State,cr[reg])); + gen_op_mov_reg_T0[OT_LONG][rm](); + } + break; + default: + goto illegal_op; + } + } + break; + case 0x121: /* mov reg, drN */ + case 0x123: /* mov drN, reg */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + modrm = ldub(s->pc++); + if ((modrm & 0xc0) != 0xc0) + goto illegal_op; + rm = modrm & 7; + reg = (modrm >> 3) & 7; + /* XXX: do it dynamically with CR4.DE bit */ + if (reg == 4 || reg == 5) + goto illegal_op; + if (b & 2) { + gen_op_mov_TN_reg[OT_LONG][0][rm](); + gen_op_movl_drN_T0(reg); + gen_op_jmp_im(s->pc - s->cs_base); + gen_eob(s); + } else { + gen_op_movl_T0_env(offsetof(CPUX86State,dr[reg])); + gen_op_mov_reg_T0[OT_LONG][rm](); + } + } + break; + case 0x106: /* clts */ + if (s->cpl != 0) { + gen_exception(s, EXCP0D_GPF, pc_start - s->cs_base); + } else { + gen_op_clts(); + } + break; + default: + goto illegal_op; + } + /* lock generation */ + if (s->prefix & PREFIX_LOCK) + gen_op_unlock(); + return s->pc; + illegal_op: + /* XXX: ensure that no lock was generated */ + gen_exception(s, EXCP06_ILLOP, pc_start - s->cs_base); + return s->pc; +} + +#define CC_OSZAPC (CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C) +#define CC_OSZAP (CC_O | CC_S | CC_Z | CC_A | CC_P) + +/* flags read by an operation */ +static uint16_t opc_read_flags[NB_OPS] = { + [INDEX_op_aas] = CC_A, + [INDEX_op_aaa] = CC_A, + [INDEX_op_das] = CC_A | CC_C, + [INDEX_op_daa] = CC_A | CC_C, + + [INDEX_op_adcb_T0_T1_cc] = CC_C, + [INDEX_op_adcw_T0_T1_cc] = CC_C, + [INDEX_op_adcl_T0_T1_cc] = CC_C, + [INDEX_op_sbbb_T0_T1_cc] = CC_C, + [INDEX_op_sbbw_T0_T1_cc] = CC_C, + [INDEX_op_sbbl_T0_T1_cc] = CC_C, + + [INDEX_op_adcb_mem_T0_T1_cc] = CC_C, + [INDEX_op_adcw_mem_T0_T1_cc] = CC_C, + [INDEX_op_adcl_mem_T0_T1_cc] = CC_C, + [INDEX_op_sbbb_mem_T0_T1_cc] = CC_C, + [INDEX_op_sbbw_mem_T0_T1_cc] = CC_C, + [INDEX_op_sbbl_mem_T0_T1_cc] = CC_C, + + /* subtle: due to the incl/decl implementation, C is used */ + [INDEX_op_update_inc_cc] = CC_C, + + [INDEX_op_into] = CC_O, + + [INDEX_op_jb_subb] = CC_C, + [INDEX_op_jb_subw] = CC_C, + [INDEX_op_jb_subl] = CC_C, + + [INDEX_op_jz_subb] = CC_Z, + [INDEX_op_jz_subw] = CC_Z, + [INDEX_op_jz_subl] = CC_Z, + + [INDEX_op_jbe_subb] = CC_Z | CC_C, + [INDEX_op_jbe_subw] = CC_Z | CC_C, + [INDEX_op_jbe_subl] = CC_Z | CC_C, + + [INDEX_op_js_subb] = CC_S, + [INDEX_op_js_subw] = CC_S, + [INDEX_op_js_subl] = CC_S, + + [INDEX_op_jl_subb] = CC_O | CC_S, + [INDEX_op_jl_subw] = CC_O | CC_S, + [INDEX_op_jl_subl] = CC_O | CC_S, + + [INDEX_op_jle_subb] = CC_O | CC_S | CC_Z, + [INDEX_op_jle_subw] = CC_O | CC_S | CC_Z, + [INDEX_op_jle_subl] = CC_O | CC_S | CC_Z, + + [INDEX_op_loopnzw] = CC_Z, + [INDEX_op_loopnzl] = CC_Z, + [INDEX_op_loopzw] = CC_Z, + [INDEX_op_loopzl] = CC_Z, + + [INDEX_op_seto_T0_cc] = CC_O, + [INDEX_op_setb_T0_cc] = CC_C, + [INDEX_op_setz_T0_cc] = CC_Z, + [INDEX_op_setbe_T0_cc] = CC_Z | CC_C, + [INDEX_op_sets_T0_cc] = CC_S, + [INDEX_op_setp_T0_cc] = CC_P, + [INDEX_op_setl_T0_cc] = CC_O | CC_S, + [INDEX_op_setle_T0_cc] = CC_O | CC_S | CC_Z, + + [INDEX_op_setb_T0_subb] = CC_C, + [INDEX_op_setb_T0_subw] = CC_C, + [INDEX_op_setb_T0_subl] = CC_C, + + [INDEX_op_setz_T0_subb] = CC_Z, + [INDEX_op_setz_T0_subw] = CC_Z, + [INDEX_op_setz_T0_subl] = CC_Z, + + [INDEX_op_setbe_T0_subb] = CC_Z | CC_C, + [INDEX_op_setbe_T0_subw] = CC_Z | CC_C, + [INDEX_op_setbe_T0_subl] = CC_Z | CC_C, + + [INDEX_op_sets_T0_subb] = CC_S, + [INDEX_op_sets_T0_subw] = CC_S, + [INDEX_op_sets_T0_subl] = CC_S, + + [INDEX_op_setl_T0_subb] = CC_O | CC_S, + [INDEX_op_setl_T0_subw] = CC_O | CC_S, + [INDEX_op_setl_T0_subl] = CC_O | CC_S, + + [INDEX_op_setle_T0_subb] = CC_O | CC_S | CC_Z, + [INDEX_op_setle_T0_subw] = CC_O | CC_S | CC_Z, + [INDEX_op_setle_T0_subl] = CC_O | CC_S | CC_Z, + + [INDEX_op_movl_T0_eflags] = CC_OSZAPC, + [INDEX_op_cmc] = CC_C, + [INDEX_op_salc] = CC_C, + + [INDEX_op_rclb_T0_T1_cc] = CC_C, + [INDEX_op_rclw_T0_T1_cc] = CC_C, + [INDEX_op_rcll_T0_T1_cc] = CC_C, + [INDEX_op_rcrb_T0_T1_cc] = CC_C, + [INDEX_op_rcrw_T0_T1_cc] = CC_C, + [INDEX_op_rcrl_T0_T1_cc] = CC_C, + + [INDEX_op_rclb_mem_T0_T1_cc] = CC_C, + [INDEX_op_rclw_mem_T0_T1_cc] = CC_C, + [INDEX_op_rcll_mem_T0_T1_cc] = CC_C, + [INDEX_op_rcrb_mem_T0_T1_cc] = CC_C, + [INDEX_op_rcrw_mem_T0_T1_cc] = CC_C, + [INDEX_op_rcrl_mem_T0_T1_cc] = CC_C, +}; + +/* flags written by an operation */ +static uint16_t opc_write_flags[NB_OPS] = { + [INDEX_op_update2_cc] = CC_OSZAPC, + [INDEX_op_update1_cc] = CC_OSZAPC, + [INDEX_op_cmpl_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_update_neg_cc] = CC_OSZAPC, + /* subtle: due to the incl/decl implementation, C is used */ + [INDEX_op_update_inc_cc] = CC_OSZAPC, + [INDEX_op_testl_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_adcb_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_adcw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_adcl_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sbbb_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sbbw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sbbl_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_adcb_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_adcw_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_adcl_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sbbb_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sbbw_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sbbl_mem_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_mulb_AL_T0] = CC_OSZAPC, + [INDEX_op_imulb_AL_T0] = CC_OSZAPC, + [INDEX_op_mulw_AX_T0] = CC_OSZAPC, + [INDEX_op_imulw_AX_T0] = CC_OSZAPC, + [INDEX_op_mull_EAX_T0] = CC_OSZAPC, + [INDEX_op_imull_EAX_T0] = CC_OSZAPC, + [INDEX_op_imulw_T0_T1] = CC_OSZAPC, + [INDEX_op_imull_T0_T1] = CC_OSZAPC, + + /* bcd */ + [INDEX_op_aam] = CC_OSZAPC, + [INDEX_op_aad] = CC_OSZAPC, + [INDEX_op_aas] = CC_OSZAPC, + [INDEX_op_aaa] = CC_OSZAPC, + [INDEX_op_das] = CC_OSZAPC, + [INDEX_op_daa] = CC_OSZAPC, + + [INDEX_op_movb_eflags_T0] = CC_S | CC_Z | CC_A | CC_P | CC_C, + [INDEX_op_movw_eflags_T0] = CC_OSZAPC, + [INDEX_op_movl_eflags_T0] = CC_OSZAPC, + [INDEX_op_clc] = CC_C, + [INDEX_op_stc] = CC_C, + [INDEX_op_cmc] = CC_C, + + [INDEX_op_rolb_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rolw_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_roll_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rorb_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rorw_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rorl_T0_T1_cc] = CC_O | CC_C, + + [INDEX_op_rclb_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rclw_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcll_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcrb_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcrw_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcrl_T0_T1_cc] = CC_O | CC_C, + + [INDEX_op_shlb_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shlw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shll_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_shrb_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shrw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shrl_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_sarb_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sarw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sarl_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_shldw_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shldl_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shldw_T0_T1_im_cc] = CC_OSZAPC, + [INDEX_op_shldl_T0_T1_im_cc] = CC_OSZAPC, + + [INDEX_op_shrdw_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shrdl_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shrdw_T0_T1_im_cc] = CC_OSZAPC, + [INDEX_op_shrdl_T0_T1_im_cc] = CC_OSZAPC, + + [INDEX_op_rolb_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rolw_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_roll_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rorb_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rorw_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rorl_mem_T0_T1_cc] = CC_O | CC_C, + + [INDEX_op_rclb_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rclw_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcll_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcrb_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcrw_mem_T0_T1_cc] = CC_O | CC_C, + [INDEX_op_rcrl_mem_T0_T1_cc] = CC_O | CC_C, + + [INDEX_op_shlb_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shlw_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shll_mem_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_shrb_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shrw_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_shrl_mem_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_sarb_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sarw_mem_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_sarl_mem_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_shldw_mem_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shldl_mem_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shldw_mem_T0_T1_im_cc] = CC_OSZAPC, + [INDEX_op_shldl_mem_T0_T1_im_cc] = CC_OSZAPC, + + [INDEX_op_shrdw_mem_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shrdl_mem_T0_T1_ECX_cc] = CC_OSZAPC, + [INDEX_op_shrdw_mem_T0_T1_im_cc] = CC_OSZAPC, + [INDEX_op_shrdl_mem_T0_T1_im_cc] = CC_OSZAPC, + + [INDEX_op_btw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btl_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btsw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btsl_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btrw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btrl_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btcw_T0_T1_cc] = CC_OSZAPC, + [INDEX_op_btcl_T0_T1_cc] = CC_OSZAPC, + + [INDEX_op_bsfw_T0_cc] = CC_OSZAPC, + [INDEX_op_bsfl_T0_cc] = CC_OSZAPC, + [INDEX_op_bsrw_T0_cc] = CC_OSZAPC, + [INDEX_op_bsrl_T0_cc] = CC_OSZAPC, + + [INDEX_op_cmpxchgb_T0_T1_EAX_cc] = CC_OSZAPC, + [INDEX_op_cmpxchgw_T0_T1_EAX_cc] = CC_OSZAPC, + [INDEX_op_cmpxchgl_T0_T1_EAX_cc] = CC_OSZAPC, + + [INDEX_op_cmpxchgb_mem_T0_T1_EAX_cc] = CC_OSZAPC, + [INDEX_op_cmpxchgw_mem_T0_T1_EAX_cc] = CC_OSZAPC, + [INDEX_op_cmpxchgl_mem_T0_T1_EAX_cc] = CC_OSZAPC, + + [INDEX_op_cmpxchg8b] = CC_Z, + [INDEX_op_lar] = CC_Z, + [INDEX_op_lsl] = CC_Z, + [INDEX_op_fcomi_ST0_FT0] = CC_Z | CC_P | CC_C, + [INDEX_op_fucomi_ST0_FT0] = CC_Z | CC_P | CC_C, +}; + +/* simpler form of an operation if no flags need to be generated */ +static uint16_t opc_simpler[NB_OPS] = { + [INDEX_op_update2_cc] = INDEX_op_nop, + [INDEX_op_update1_cc] = INDEX_op_nop, + [INDEX_op_update_neg_cc] = INDEX_op_nop, +#if 0 + /* broken: CC_OP logic must be rewritten */ + [INDEX_op_update_inc_cc] = INDEX_op_nop, +#endif + [INDEX_op_rolb_T0_T1_cc] = INDEX_op_rolb_T0_T1, + [INDEX_op_rolw_T0_T1_cc] = INDEX_op_rolw_T0_T1, + [INDEX_op_roll_T0_T1_cc] = INDEX_op_roll_T0_T1, + + [INDEX_op_rorb_T0_T1_cc] = INDEX_op_rorb_T0_T1, + [INDEX_op_rorw_T0_T1_cc] = INDEX_op_rorw_T0_T1, + [INDEX_op_rorl_T0_T1_cc] = INDEX_op_rorl_T0_T1, + + [INDEX_op_rolb_mem_T0_T1_cc] = INDEX_op_rolb_mem_T0_T1, + [INDEX_op_rolw_mem_T0_T1_cc] = INDEX_op_rolw_mem_T0_T1, + [INDEX_op_roll_mem_T0_T1_cc] = INDEX_op_roll_mem_T0_T1, + + [INDEX_op_rorb_mem_T0_T1_cc] = INDEX_op_rorb_mem_T0_T1, + [INDEX_op_rorw_mem_T0_T1_cc] = INDEX_op_rorw_mem_T0_T1, + [INDEX_op_rorl_mem_T0_T1_cc] = INDEX_op_rorl_mem_T0_T1, + + [INDEX_op_shlb_T0_T1_cc] = INDEX_op_shlb_T0_T1, + [INDEX_op_shlw_T0_T1_cc] = INDEX_op_shlw_T0_T1, + [INDEX_op_shll_T0_T1_cc] = INDEX_op_shll_T0_T1, + + [INDEX_op_shrb_T0_T1_cc] = INDEX_op_shrb_T0_T1, + [INDEX_op_shrw_T0_T1_cc] = INDEX_op_shrw_T0_T1, + [INDEX_op_shrl_T0_T1_cc] = INDEX_op_shrl_T0_T1, + + [INDEX_op_sarb_T0_T1_cc] = INDEX_op_sarb_T0_T1, + [INDEX_op_sarw_T0_T1_cc] = INDEX_op_sarw_T0_T1, + [INDEX_op_sarl_T0_T1_cc] = INDEX_op_sarl_T0_T1, +}; + +void optimize_flags_init(void) +{ + int i; + /* put default values in arrays */ + for(i = 0; i < NB_OPS; i++) { + if (opc_simpler[i] == 0) + opc_simpler[i] = i; + } +} + +/* CPU flags computation optimization: we move backward thru the + generated code to see which flags are needed. The operation is + modified if suitable */ +static void optimize_flags(uint16_t *opc_buf, int opc_buf_len) +{ + uint16_t *opc_ptr; + int live_flags, write_flags, op; + + opc_ptr = opc_buf + opc_buf_len; + /* live_flags contains the flags needed by the next instructions + in the code. At the end of the bloc, we consider that all the + flags are live. */ + live_flags = CC_OSZAPC; + while (opc_ptr > opc_buf) { + op = *--opc_ptr; + /* if none of the flags written by the instruction is used, + then we can try to find a simpler instruction */ + write_flags = opc_write_flags[op]; + if ((live_flags & write_flags) == 0) { + *opc_ptr = opc_simpler[op]; + } + /* compute the live flags before the instruction */ + live_flags &= ~write_flags; + live_flags |= opc_read_flags[op]; + } +} + +/* generate intermediate code in gen_opc_buf and gen_opparam_buf for + basic block 'tb'. If search_pc is TRUE, also generate PC + information for each intermediate instruction. */ +static inline int gen_intermediate_code_internal(CPUState *env, + TranslationBlock *tb, + int search_pc) +{ + DisasContext dc1, *dc = &dc1; + uint8_t *pc_ptr; + uint16_t *gen_opc_end; + int flags, j, lj; + uint8_t *pc_start; + uint8_t *cs_base; + + /* generate intermediate code */ + pc_start = (uint8_t *)tb->pc; + cs_base = (uint8_t *)tb->cs_base; + flags = tb->flags; + + dc->pe = env->cr[0] & CR0_PE_MASK; + dc->code32 = (flags >> HF_CS32_SHIFT) & 1; + dc->ss32 = (flags >> HF_SS32_SHIFT) & 1; + dc->addseg = (flags >> HF_ADDSEG_SHIFT) & 1; + dc->f_st = 0; + dc->vm86 = (flags >> VM_SHIFT) & 1; + dc->cpl = (flags >> HF_CPL_SHIFT) & 3; + dc->iopl = (flags >> IOPL_SHIFT) & 3; + dc->tf = (flags >> TF_SHIFT) & 1; + dc->cc_op = CC_OP_DYNAMIC; + dc->cs_base = cs_base; + dc->tb = tb; + dc->popl_esp_hack = 0; + /* select memory access functions */ + dc->mem_index = 0; + if (flags & HF_SOFTMMU_MASK) { + if (dc->cpl == 3) + dc->mem_index = 6; + else + dc->mem_index = 3; + } + dc->jmp_opt = !(dc->tf || env->singlestep_enabled +#ifndef CONFIG_SOFT_MMU + || (flags & HF_SOFTMMU_MASK) +#endif + ); + gen_opc_ptr = gen_opc_buf; + gen_opc_end = gen_opc_buf + OPC_MAX_SIZE; + gen_opparam_ptr = gen_opparam_buf; + + dc->is_jmp = DISAS_NEXT; + pc_ptr = pc_start; + lj = -1; + + /* if irq were inhibited for the next instruction, we can disable + them here as it is simpler (otherwise jumps would have to + handled as special case) */ + if (flags & HF_INHIBIT_IRQ_MASK) { + gen_op_reset_inhibit_irq(); + } + for(;;) { + if (env->nb_breakpoints > 0) { + for(j = 0; j < env->nb_breakpoints; j++) { + if (env->breakpoints[j] == (unsigned long)pc_ptr) { + gen_debug(dc, pc_ptr - dc->cs_base); + break; + } + } + } + if (search_pc) { + j = gen_opc_ptr - gen_opc_buf; + if (lj < j) { + lj++; + while (lj < j) + gen_opc_instr_start[lj++] = 0; + } + gen_opc_pc[lj] = (uint32_t)pc_ptr; + gen_opc_cc_op[lj] = dc->cc_op; + gen_opc_instr_start[lj] = 1; + } + pc_ptr = disas_insn(dc, pc_ptr); + /* stop translation if indicated */ + if (dc->is_jmp) + break; + /* if single step mode, we generate only one instruction and + generate an exception */ + if (dc->tf) { + gen_op_jmp_im(pc_ptr - dc->cs_base); + gen_eob(dc); + break; + } + /* if too long translation, stop generation too */ + if (gen_opc_ptr >= gen_opc_end || + (pc_ptr - pc_start) >= (TARGET_PAGE_SIZE - 32)) { + gen_op_jmp_im(pc_ptr - dc->cs_base); + gen_eob(dc); + break; + } + } + *gen_opc_ptr = INDEX_op_end; + /* we don't forget to fill the last values */ + if (search_pc) { + j = gen_opc_ptr - gen_opc_buf; + lj++; + while (lj <= j) + gen_opc_instr_start[lj++] = 0; + } + +#ifdef DEBUG_DISAS + if (loglevel) { + fprintf(logfile, "----------------\n"); + fprintf(logfile, "IN: %s\n", lookup_symbol(pc_start)); + disas(logfile, pc_start, pc_ptr - pc_start, 0, !dc->code32); + fprintf(logfile, "\n"); + + fprintf(logfile, "OP:\n"); + dump_ops(gen_opc_buf, gen_opparam_buf); + fprintf(logfile, "\n"); + } +#endif + + /* optimize flag computations */ + optimize_flags(gen_opc_buf, gen_opc_ptr - gen_opc_buf); + +#ifdef DEBUG_DISAS + if (loglevel) { + fprintf(logfile, "AFTER FLAGS OPT:\n"); + dump_ops(gen_opc_buf, gen_opparam_buf); + fprintf(logfile, "\n"); + } +#endif + if (!search_pc) + tb->size = pc_ptr - pc_start; + return 0; +} + +int gen_intermediate_code(CPUState *env, TranslationBlock *tb) +{ + return gen_intermediate_code_internal(env, tb, 0); +} + +int gen_intermediate_code_pc(CPUState *env, TranslationBlock *tb) +{ + return gen_intermediate_code_internal(env, tb, 1); +} + |