From 1c152f8e6d987b9e5c57e966407d27107baa6a75 Mon Sep 17 00:00:00 2001 From: Andrzej Zaborowski Date: Fri, 16 Mar 2007 16:32:08 +0100 Subject: [PATCH] iwMMXt instruction set for the PXA270. --- target-arm/cpu.h | 19 + target-arm/exec.h | 2 + target-arm/helper.c | 2 + target-arm/op.c | 690 ++++++++++++++++++++++++++++++++ target-arm/op_mem.h | 20 + target-arm/translate.c | 1007 ++++++++++++++++++++++++++++++++++++++++++++++- tests/Makefile | 3 + tests/test-arm-iwmmxt.s | 49 +++ 8 files changed, 1786 insertions(+), 6 deletions(-) diff --git a/target-arm/cpu.h b/target-arm/cpu.h index 738bb3a..530acb2 100644 --- a/target-arm/cpu.h +++ b/target-arm/cpu.h @@ -128,6 +128,14 @@ typedef struct CPUARMState { float_status fp_status; } vfp; + /* iwMMXt coprocessor state. */ + struct { + uint64_t regs[16]; + uint64_t val; + + uint32_t cregs[16]; + } iwmmxt; + #if defined(CONFIG_USER_ONLY) /* For usermode syscall translation. */ int eabi; @@ -217,10 +225,21 @@ enum arm_cpu_mode { #define ARM_VFP_FPINST 9 #define ARM_VFP_FPINST2 10 +/* iwMMXt coprocessor control registers. */ +#define ARM_IWMMXT_wCID 0 +#define ARM_IWMMXT_wCon 1 +#define ARM_IWMMXT_wCSSF 2 +#define ARM_IWMMXT_wCASF 3 +#define ARM_IWMMXT_wCGR0 8 +#define ARM_IWMMXT_wCGR1 9 +#define ARM_IWMMXT_wCGR2 10 +#define ARM_IWMMXT_wCGR3 11 + enum arm_features { ARM_FEATURE_VFP, ARM_FEATURE_AUXCR, /* ARM1026 Auxiliary control register. */ ARM_FEATURE_XSCALE, /* Intel XScale extensions. */ + ARM_FEATURE_IWMMXT /* Intel iwMMXt extension. */ }; static inline int arm_feature(CPUARMState *env, int feature) diff --git a/target-arm/exec.h b/target-arm/exec.h index af06f48..95b5154 100644 --- a/target-arm/exec.h +++ b/target-arm/exec.h @@ -38,6 +38,8 @@ register uint32_t T2 asm(AREG3); #define FT0d env->vfp.tmp0d #define FT1d env->vfp.tmp1d +#define M0 env->iwmmxt.val + #include "cpu.h" #include "exec-all.h" diff --git a/target-arm/helper.c b/target-arm/helper.c index 50795d6..a05cd21 100644 --- a/target-arm/helper.c +++ b/target-arm/helper.c @@ -42,6 +42,8 @@ static void cpu_reset_model_id(CPUARMState *env, uint32_t id) case ARM_CPUID_PXA270_C5: set_feature(env, ARM_FEATURE_XSCALE); /* JTAG_ID is ((id << 28) | 0x09265013) */ + set_feature(env, ARM_FEATURE_IWMMXT); + env->iwmmxt.cregs[ARM_IWMMXT_wCID] = 0x69051000 | 'Q'; env->cp15.c0_cachetype = 0xd172172; break; default: diff --git a/target-arm/op.c b/target-arm/op.c index 9cfb462..b00ac7d 100644 --- a/target-arm/op.c +++ b/target-arm/op.c @@ -3,6 +3,7 @@ * * Copyright (c) 2003 Fabrice Bellard * Copyright (c) 2005 CodeSourcery, LLC + * Copyright (c) 2007 OpenedHand, Ltd. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public @@ -1213,3 +1214,692 @@ void OPPROTO op_movl_T0_T2(void) { T0 = T2; } + +/* iwMMXt support. */ + +#define M1 env->iwmmxt.regs[PARAM1] + +/* iwMMXt macros extracted from GNU gdb. */ + +/* Set the SIMD wCASF flags for 8, 16, 32 or 64-bit operations. */ +#define SIMD8_SET( v, n, b) ((v != 0) << ((((b) + 1) * 4) + (n))) +#define SIMD16_SET(v, n, h) ((v != 0) << ((((h) + 1) * 8) + (n))) +#define SIMD32_SET(v, n, w) ((v != 0) << ((((w) + 1) * 16) + (n))) +#define SIMD64_SET(v, n) ((v != 0) << (32 + (n))) +/* Flags to pass as "n" above. */ +#define SIMD_NBIT -1 +#define SIMD_ZBIT -2 +#define SIMD_CBIT -3 +#define SIMD_VBIT -4 +/* Various status bit macros. */ +#define NBIT8(x) ((x) & 0x80) +#define NBIT16(x) ((x) & 0x8000) +#define NBIT32(x) ((x) & 0x80000000) +#define NBIT64(x) ((x) & 0x8000000000000000ULL) +#define ZBIT8(x) (((x) & 0xff) == 0) +#define ZBIT16(x) (((x) & 0xffff) == 0) +#define ZBIT32(x) (((x) & 0xffffffff) == 0) +#define ZBIT64(x) (x == 0) +/* Sign extension macros. */ +#define EXTEND8H(a) ((a) & 0x80 ? ((a) | 0xff00) : (a)) +#define EXTEND8(a) ((a) & 0x80 ? ((a) | 0xffffff00) : (a)) +#define EXTEND16(a) ((a) & 0x8000 ? ((a) | 0xffff0000) : (a)) +#define EXTEND32(a) ((a) & 0x80000000ULL ? \ + ((a) | 0xffffffff00000000ULL) : (a)) + +void OPPROTO op_iwmmxt_movl_T0_T1_wRn(void) +{ + T0 = M1 & ~(uint32_t) 0; + T1 = M1 >> 32; +} + +void OPPROTO op_iwmmxt_movl_wRn_T0_T1(void) +{ + M1 = ((uint64_t) T1 << 32) | T0; +} + +void OPPROTO op_iwmmxt_movq_M0_wRn(void) +{ + M0 = M1; +} + +void OPPROTO op_iwmmxt_orq_M0_wRn(void) +{ + M0 |= M1; +} + +void OPPROTO op_iwmmxt_andq_M0_wRn(void) +{ + M0 &= M1; +} + +void OPPROTO op_iwmmxt_xorq_M0_wRn(void) +{ + M0 ^= M1; +} + +void OPPROTO op_iwmmxt_maddsq_M0_wRn(void) +{ + M0 = (( + EXTEND16((M0 >> 0) & 0xffff) * EXTEND16((M1 >> 0) & 0xffff) + + EXTEND16((M0 >> 16) & 0xffff) * EXTEND16((M1 >> 16) & 0xffff) + ) & 0xffffffff) | (( + EXTEND16((M0 >> 32) & 0xffff) * EXTEND16((M1 >> 32) & 0xffff) + + EXTEND16((M0 >> 48) & 0xffff) * EXTEND16((M1 >> 48) & 0xffff) + ) << 32); +} + +void OPPROTO op_iwmmxt_madduq_M0_wRn(void) +{ + M0 = (( + ((M0 >> 0) & 0xffff) * ((M1 >> 0) & 0xffff) + + ((M0 >> 16) & 0xffff) * ((M1 >> 16) & 0xffff) + ) & 0xffffffff) | (( + ((M0 >> 32) & 0xffff) * ((M1 >> 32) & 0xffff) + + ((M0 >> 48) & 0xffff) * ((M1 >> 48) & 0xffff) + ) << 32); +} + +void OPPROTO op_iwmmxt_sadb_M0_wRn(void) +{ +#define abs(x) (((x) >= 0) ? x : -x) +#define SADB(SHR) abs((int) ((M0 >> SHR) & 0xff) - (int) ((M1 >> SHR) & 0xff)) + M0 = + SADB(0) + SADB(8) + SADB(16) + SADB(24) + + SADB(32) + SADB(40) + SADB(48) + SADB(56); +#undef SADB +} + +void OPPROTO op_iwmmxt_sadw_M0_wRn(void) +{ +#define SADW(SHR) \ + abs((int) ((M0 >> SHR) & 0xffff) - (int) ((M1 >> SHR) & 0xffff)) + M0 = SADW(0) + SADW(16) + SADW(32) + SADW(48); +#undef SADW +} + +void OPPROTO op_iwmmxt_addl_M0_wRn(void) +{ + M0 += env->iwmmxt.regs[PARAM1] & 0xffffffff; +} + +void OPPROTO op_iwmmxt_mulsw_M0_wRn(void) +{ +#define MULS(SHR) (((( \ + EXTEND16((M0 >> SHR) & 0xffff) * EXTEND16((M1 >> SHR) & 0xffff) \ + ) >> PARAM2) & 0xffff) << SHR) + M0 = MULS(0) | MULS(16) | MULS(32) | MULS(48); +#undef MULS +} + +void OPPROTO op_iwmmxt_muluw_M0_wRn(void) +{ +#define MULU(SHR) (((( \ + ((M0 >> SHR) & 0xffff) * ((M1 >> SHR) & 0xffff) \ + ) >> PARAM2) & 0xffff) << SHR) + M0 = MULU(0) | MULU(16) | MULU(32) | MULU(48); +#undef MULU +} + +void OPPROTO op_iwmmxt_macsw_M0_wRn(void) +{ +#define MACS(SHR) ( \ + (int32_t) EXTEND16((M0 >> SHR) & 0xffff) * \ + (int32_t) EXTEND16((M1 >> SHR) & 0xffff)) + M0 = (int64_t) (MACS(0) + MACS(16) + MACS(32) + MACS(48)); +#undef MACS +} + +void OPPROTO op_iwmmxt_macuw_M0_wRn(void) +{ +#define MACU(SHR) ( \ + (uint32_t) ((M0 >> SHR) & 0xffff) * \ + (uint32_t) ((M1 >> SHR) & 0xffff)) + M0 = MACU(0) + MACU(16) + MACU(32) + MACU(48); +#undef MACU +} + +void OPPROTO op_iwmmxt_addsq_M0_wRn(void) +{ + M0 = (int64_t) M0 + (int64_t) M1; +} + +void OPPROTO op_iwmmxt_adduq_M0_wRn(void) +{ + M0 += M1; +} + +void OPPROTO op_iwmmxt_movq_wRn_M0(void) +{ + M1 = M0; +} + +void OPPROTO op_iwmmxt_movl_wCx_T0(void) +{ + env->iwmmxt.cregs[PARAM1] = T0; +} + +void OPPROTO op_iwmmxt_movl_T0_wCx(void) +{ + T0 = env->iwmmxt.cregs[PARAM1]; +} + +void OPPROTO op_iwmmxt_movl_T1_wCx(void) +{ + T1 = env->iwmmxt.cregs[PARAM1]; +} + +void OPPROTO op_iwmmxt_set_mup(void) +{ + env->iwmmxt.cregs[ARM_IWMMXT_wCon] |= 2; +} + +void OPPROTO op_iwmmxt_set_cup(void) +{ + env->iwmmxt.cregs[ARM_IWMMXT_wCon] |= 1; +} + +void OPPROTO op_iwmmxt_setpsr_nz(void) +{ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + SIMD64_SET((M0 == 0), SIMD_ZBIT) | + SIMD64_SET((M0 & (1ULL << 63)), SIMD_NBIT); +} + +void OPPROTO op_iwmmxt_negq_M0(void) +{ + M0 = ~M0; +} + +#define NZBIT8(x, i) \ + SIMD8_SET(NBIT8((x) & 0xff), SIMD_NBIT, i) | \ + SIMD8_SET(ZBIT8((x) & 0xff), SIMD_ZBIT, i) +#define NZBIT16(x, i) \ + SIMD16_SET(NBIT16((x) & 0xffff), SIMD_NBIT, i) | \ + SIMD16_SET(ZBIT16((x) & 0xffff), SIMD_ZBIT, i) +#define NZBIT32(x, i) \ + SIMD32_SET(NBIT32((x) & 0xffffffff), SIMD_NBIT, i) | \ + SIMD32_SET(ZBIT32((x) & 0xffffffff), SIMD_ZBIT, i) +#define NZBIT64(x) \ + SIMD64_SET(NBIT64(x), SIMD_NBIT) | \ + SIMD64_SET(ZBIT64(x), SIMD_ZBIT) +#define IWMMXT_OP_UNPACK(S, SH0, SH1, SH2, SH3) \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, b_M0_wRn))(void) \ +{ \ + M0 = \ + (((M0 >> SH0) & 0xff) << 0) | (((M1 >> SH0) & 0xff) << 8) | \ + (((M0 >> SH1) & 0xff) << 16) | (((M1 >> SH1) & 0xff) << 24) | \ + (((M0 >> SH2) & 0xff) << 32) | (((M1 >> SH2) & 0xff) << 40) | \ + (((M0 >> SH3) & 0xff) << 48) | (((M1 >> SH3) & 0xff) << 56); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT8(M0 >> 0, 0) | NZBIT8(M0 >> 8, 1) | \ + NZBIT8(M0 >> 16, 2) | NZBIT8(M0 >> 24, 3) | \ + NZBIT8(M0 >> 32, 4) | NZBIT8(M0 >> 40, 5) | \ + NZBIT8(M0 >> 48, 6) | NZBIT8(M0 >> 56, 7); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, w_M0_wRn))(void) \ +{ \ + M0 = \ + (((M0 >> SH0) & 0xffff) << 0) | \ + (((M1 >> SH0) & 0xffff) << 16) | \ + (((M0 >> SH2) & 0xffff) << 32) | \ + (((M1 >> SH2) & 0xffff) << 48); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT8(M0 >> 0, 0) | NZBIT8(M0 >> 16, 1) | \ + NZBIT8(M0 >> 32, 2) | NZBIT8(M0 >> 48, 3); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, l_M0_wRn))(void) \ +{ \ + M0 = \ + (((M0 >> SH0) & 0xffffffff) << 0) | \ + (((M1 >> SH0) & 0xffffffff) << 32); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, ub_M0))(void) \ +{ \ + M0 = \ + (((M0 >> SH0) & 0xff) << 0) | \ + (((M0 >> SH1) & 0xff) << 16) | \ + (((M0 >> SH2) & 0xff) << 32) | \ + (((M0 >> SH3) & 0xff) << 48); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | \ + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, uw_M0))(void) \ +{ \ + M0 = \ + (((M0 >> SH0) & 0xffff) << 0) | \ + (((M0 >> SH2) & 0xffff) << 32); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, ul_M0))(void) \ +{ \ + M0 = (((M0 >> SH0) & 0xffffffff) << 0); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = NZBIT64(M0 >> 0); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, sb_M0))(void) \ +{ \ + M0 = \ + (EXTEND8H((M0 >> SH0) & 0xff) << 0) | \ + (EXTEND8H((M0 >> SH1) & 0xff) << 16) | \ + (EXTEND8H((M0 >> SH2) & 0xff) << 32) | \ + (EXTEND8H((M0 >> SH3) & 0xff) << 48); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | \ + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, sw_M0))(void) \ +{ \ + M0 = \ + (EXTEND16((M0 >> SH0) & 0xffff) << 0) | \ + (EXTEND16((M0 >> SH2) & 0xffff) << 32); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); \ +} \ +void OPPROTO glue(op_iwmmxt_unpack, glue(S, sl_M0))(void) \ +{ \ + M0 = EXTEND32((M0 >> SH0) & 0xffffffff); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = NZBIT64(M0 >> 0); \ +} +IWMMXT_OP_UNPACK(l, 0, 8, 16, 24) +IWMMXT_OP_UNPACK(h, 32, 40, 48, 56) + +#define IWMMXT_OP_CMP(SUFF, Tb, Tw, Tl, O) \ +void OPPROTO glue(op_iwmmxt_, glue(SUFF, b_M0_wRn))(void) \ +{ \ + M0 = \ + CMP(0, Tb, O, 0xff) | CMP(8, Tb, O, 0xff) | \ + CMP(16, Tb, O, 0xff) | CMP(24, Tb, O, 0xff) | \ + CMP(32, Tb, O, 0xff) | CMP(40, Tb, O, 0xff) | \ + CMP(48, Tb, O, 0xff) | CMP(56, Tb, O, 0xff); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT8(M0 >> 0, 0) | NZBIT8(M0 >> 8, 1) | \ + NZBIT8(M0 >> 16, 2) | NZBIT8(M0 >> 24, 3) | \ + NZBIT8(M0 >> 32, 4) | NZBIT8(M0 >> 40, 5) | \ + NZBIT8(M0 >> 48, 6) | NZBIT8(M0 >> 56, 7); \ +} \ +void OPPROTO glue(op_iwmmxt_, glue(SUFF, w_M0_wRn))(void) \ +{ \ + M0 = CMP(0, Tw, O, 0xffff) | CMP(16, Tw, O, 0xffff) | \ + CMP(32, Tw, O, 0xffff) | CMP(48, Tw, O, 0xffff); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | \ + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); \ +} \ +void OPPROTO glue(op_iwmmxt_, glue(SUFF, l_M0_wRn))(void) \ +{ \ + M0 = CMP(0, Tl, O, 0xffffffff) | \ + CMP(32, Tl, O, 0xffffffff); \ + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = \ + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); \ +} +#define CMP(SHR, TYPE, OPER, MASK) ((((TYPE) ((M0 >> SHR) & MASK) OPER \ + (TYPE) ((M1 >> SHR) & MASK)) ? (uint64_t) MASK : 0) << SHR) +IWMMXT_OP_CMP(cmpeq, uint8_t, uint16_t, uint32_t, ==) +IWMMXT_OP_CMP(cmpgts, int8_t, int16_t, int32_t, >) +IWMMXT_OP_CMP(cmpgtu, uint8_t, uint16_t, uint32_t, >) +#undef CMP +#define CMP(SHR, TYPE, OPER, MASK) ((((TYPE) ((M0 >> SHR) & MASK) OPER \ + (TYPE) ((M1 >> SHR) & MASK)) ? M0 : M1) & ((uint64_t) MASK << SHR)) +IWMMXT_OP_CMP(mins, int8_t, int16_t, int32_t, <) +IWMMXT_OP_CMP(minu, uint8_t, uint16_t, uint32_t, <) +IWMMXT_OP_CMP(maxs, int8_t, int16_t, int32_t, >) +IWMMXT_OP_CMP(maxu, uint8_t, uint16_t, uint32_t, >) +#undef CMP +#define CMP(SHR, TYPE, OPER, MASK) ((uint64_t) (((TYPE) ((M0 >> SHR) & MASK) \ + OPER (TYPE) ((M1 >> SHR) & MASK)) & MASK) << SHR) +IWMMXT_OP_CMP(subn, uint8_t, uint16_t, uint32_t, -) +IWMMXT_OP_CMP(addn, uint8_t, uint16_t, uint32_t, +) +#undef CMP +/* TODO Signed- and Unsigned-Saturation */ +#define CMP(SHR, TYPE, OPER, MASK) ((uint64_t) (((TYPE) ((M0 >> SHR) & MASK) \ + OPER (TYPE) ((M1 >> SHR) & MASK)) & MASK) << SHR) +IWMMXT_OP_CMP(subu, uint8_t, uint16_t, uint32_t, -) +IWMMXT_OP_CMP(addu, uint8_t, uint16_t, uint32_t, +) +IWMMXT_OP_CMP(subs, int8_t, int16_t, int32_t, -) +IWMMXT_OP_CMP(adds, int8_t, int16_t, int32_t, +) +#undef CMP +#undef IWMMXT_OP_CMP + +void OPPROTO op_iwmmxt_avgb_M0_wRn(void) +{ +#define AVGB(SHR) ((( \ + ((M0 >> SHR) & 0xff) + ((M1 >> SHR) & 0xff) + PARAM2) >> 1) << SHR) + M0 = + AVGB(0) | AVGB(8) | AVGB(16) | AVGB(24) | + AVGB(32) | AVGB(40) | AVGB(48) | AVGB(56); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + SIMD8_SET(ZBIT8((M0 >> 0) & 0xff), SIMD_ZBIT, 0) | + SIMD8_SET(ZBIT8((M0 >> 8) & 0xff), SIMD_ZBIT, 1) | + SIMD8_SET(ZBIT8((M0 >> 16) & 0xff), SIMD_ZBIT, 2) | + SIMD8_SET(ZBIT8((M0 >> 24) & 0xff), SIMD_ZBIT, 3) | + SIMD8_SET(ZBIT8((M0 >> 32) & 0xff), SIMD_ZBIT, 4) | + SIMD8_SET(ZBIT8((M0 >> 40) & 0xff), SIMD_ZBIT, 5) | + SIMD8_SET(ZBIT8((M0 >> 48) & 0xff), SIMD_ZBIT, 6) | + SIMD8_SET(ZBIT8((M0 >> 56) & 0xff), SIMD_ZBIT, 7); +#undef AVGB +} + +void OPPROTO op_iwmmxt_avgw_M0_wRn(void) +{ +#define AVGW(SHR) ((( \ + ((M0 >> SHR) & 0xffff) + ((M1 >> SHR) & 0xffff) + PARAM2) >> 1) << SHR) + M0 = AVGW(0) | AVGW(16) | AVGW(32) | AVGW(48); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + SIMD16_SET(ZBIT16((M0 >> 0) & 0xffff), SIMD_ZBIT, 0) | + SIMD16_SET(ZBIT16((M0 >> 16) & 0xffff), SIMD_ZBIT, 1) | + SIMD16_SET(ZBIT16((M0 >> 32) & 0xffff), SIMD_ZBIT, 2) | + SIMD16_SET(ZBIT16((M0 >> 48) & 0xffff), SIMD_ZBIT, 3); +#undef AVGW +} + +void OPPROTO op_iwmmxt_msadb_M0_wRn(void) +{ + M0 = ((((M0 >> 0) & 0xffff) * ((M1 >> 0) & 0xffff) + + ((M0 >> 16) & 0xffff) * ((M1 >> 16) & 0xffff)) & 0xffffffff) | + ((((M0 >> 32) & 0xffff) * ((M1 >> 32) & 0xffff) + + ((M0 >> 48) & 0xffff) * ((M1 >> 48) & 0xffff)) << 32); +} + +void OPPROTO op_iwmmxt_align_M0_T0_wRn(void) +{ + M0 >>= T0 << 3; + M0 |= M1 << (64 - (T0 << 3)); +} + +void OPPROTO op_iwmmxt_insr_M0_T0_T1(void) +{ + M0 &= ~((uint64_t) T1 << PARAM1); + M0 |= (uint64_t) (T0 & T1) << PARAM1; +} + +void OPPROTO op_iwmmxt_extrsb_T0_M0(void) +{ + T0 = EXTEND8((M0 >> PARAM1) & 0xff); +} + +void OPPROTO op_iwmmxt_extrsw_T0_M0(void) +{ + T0 = EXTEND16((M0 >> PARAM1) & 0xffff); +} + +void OPPROTO op_iwmmxt_extru_T0_M0_T1(void) +{ + T0 = (M0 >> PARAM1) & T1; +} + +void OPPROTO op_iwmmxt_bcstb_M0_T0(void) +{ + T0 &= 0xff; + M0 = + ((uint64_t) T0 << 0) | ((uint64_t) T0 << 8) | + ((uint64_t) T0 << 16) | ((uint64_t) T0 << 24) | + ((uint64_t) T0 << 32) | ((uint64_t) T0 << 40) | + ((uint64_t) T0 << 48) | ((uint64_t) T0 << 56); +} + +void OPPROTO op_iwmmxt_bcstw_M0_T0(void) +{ + T0 &= 0xffff; + M0 = + ((uint64_t) T0 << 0) | ((uint64_t) T0 << 16) | + ((uint64_t) T0 << 32) | ((uint64_t) T0 << 48); +} + +void OPPROTO op_iwmmxt_bcstl_M0_T0(void) +{ + M0 = ((uint64_t) T0 << 0) | ((uint64_t) T0 << 32); +} + +void OPPROTO op_iwmmxt_addcb_M0(void) +{ + M0 = + ((M0 >> 0) & 0xff) + ((M0 >> 8) & 0xff) + + ((M0 >> 16) & 0xff) + ((M0 >> 24) & 0xff) + + ((M0 >> 32) & 0xff) + ((M0 >> 40) & 0xff) + + ((M0 >> 48) & 0xff) + ((M0 >> 56) & 0xff); +} + +void OPPROTO op_iwmmxt_addcw_M0(void) +{ + M0 = + ((M0 >> 0) & 0xffff) + ((M0 >> 16) & 0xffff) + + ((M0 >> 32) & 0xffff) + ((M0 >> 48) & 0xffff); +} + +void OPPROTO op_iwmmxt_addcl_M0(void) +{ + M0 = (M0 & 0xffffffff) + (M0 >> 32); +} + +void OPPROTO op_iwmmxt_msbb_T0_M0(void) +{ + T0 = + ((M0 >> 7) & 0x01) | ((M0 >> 14) & 0x02) | + ((M0 >> 21) & 0x04) | ((M0 >> 28) & 0x08) | + ((M0 >> 35) & 0x10) | ((M0 >> 42) & 0x20) | + ((M0 >> 49) & 0x40) | ((M0 >> 56) & 0x80); +} + +void OPPROTO op_iwmmxt_msbw_T0_M0(void) +{ + T0 = + ((M0 >> 15) & 0x01) | ((M0 >> 30) & 0x02) | + ((M0 >> 45) & 0x04) | ((M0 >> 52) & 0x08); +} + +void OPPROTO op_iwmmxt_msbl_T0_M0(void) +{ + T0 = ((M0 >> 31) & 0x01) | ((M0 >> 62) & 0x02); +} + +void OPPROTO op_iwmmxt_srlw_M0_T0(void) +{ + M0 = + (((M0 & (0xffffll << 0)) >> T0) & (0xffffll << 0)) | + (((M0 & (0xffffll << 16)) >> T0) & (0xffffll << 16)) | + (((M0 & (0xffffll << 32)) >> T0) & (0xffffll << 32)) | + (((M0 & (0xffffll << 48)) >> T0) & (0xffffll << 48)); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +void OPPROTO op_iwmmxt_srll_M0_T0(void) +{ + M0 = + ((M0 & (0xffffffffll << 0)) >> T0) | + ((M0 >> T0) & (0xffffffffll << 32)); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); +} + +void OPPROTO op_iwmmxt_srlq_M0_T0(void) +{ + M0 >>= T0; + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = NZBIT64(M0); +} + +void OPPROTO op_iwmmxt_sllw_M0_T0(void) +{ + M0 = + (((M0 & (0xffffll << 0)) << T0) & (0xffffll << 0)) | + (((M0 & (0xffffll << 16)) << T0) & (0xffffll << 16)) | + (((M0 & (0xffffll << 32)) << T0) & (0xffffll << 32)) | + (((M0 & (0xffffll << 48)) << T0) & (0xffffll << 48)); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +void OPPROTO op_iwmmxt_slll_M0_T0(void) +{ + M0 = + ((M0 << T0) & (0xffffffffll << 0)) | + ((M0 & (0xffffffffll << 32)) << T0); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); +} + +void OPPROTO op_iwmmxt_sllq_M0_T0(void) +{ + M0 <<= T0; + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = NZBIT64(M0); +} + +void OPPROTO op_iwmmxt_sraw_M0_T0(void) +{ + M0 = + (((EXTEND16(M0 >> 0) >> T0) & 0xffff) << 0) | + (((EXTEND16(M0 >> 16) >> T0) & 0xffff) << 16) | + (((EXTEND16(M0 >> 32) >> T0) & 0xffff) << 32) | + (((EXTEND16(M0 >> 48) >> T0) & 0xffff) << 48); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +void OPPROTO op_iwmmxt_sral_M0_T0(void) +{ + M0 = + (((EXTEND32(M0 >> 0) >> T0) & 0xffffffff) << 0) | + (((EXTEND32(M0 >> 32) >> T0) & 0xffffffff) << 32); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); +} + +void OPPROTO op_iwmmxt_sraq_M0_T0(void) +{ + M0 = (int64_t) M0 >> T0; + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = NZBIT64(M0); +} + +void OPPROTO op_iwmmxt_rorw_M0_T0(void) +{ + M0 = + ((((M0 & (0xffffll << 0)) >> T0) | + ((M0 & (0xffffll << 0)) << (16 - T0))) & (0xffffll << 0)) | + ((((M0 & (0xffffll << 16)) >> T0) | + ((M0 & (0xffffll << 16)) << (16 - T0))) & (0xffffll << 16)) | + ((((M0 & (0xffffll << 32)) >> T0) | + ((M0 & (0xffffll << 32)) << (16 - T0))) & (0xffffll << 32)) | + ((((M0 & (0xffffll << 48)) >> T0) | + ((M0 & (0xffffll << 48)) << (16 - T0))) & (0xffffll << 48)); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +void OPPROTO op_iwmmxt_rorl_M0_T0(void) +{ + M0 = + ((M0 & (0xffffffffll << 0)) >> T0) | + ((M0 >> T0) & (0xffffffffll << 32)) | + ((M0 << (32 - T0)) & (0xffffffffll << 0)) | + ((M0 & (0xffffffffll << 32)) << (32 - T0)); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); +} + +void OPPROTO op_iwmmxt_rorq_M0_T0(void) +{ + M0 = (M0 >> T0) | (M0 << (64 - T0)); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = NZBIT64(M0); +} + +void OPPROTO op_iwmmxt_shufh_M0_T0(void) +{ + M0 = + (((M0 >> ((T0 << 4) & 0x30)) & 0xffff) << 0) | + (((M0 >> ((T0 << 2) & 0x30)) & 0xffff) << 16) | + (((M0 >> ((T0 << 0) & 0x30)) & 0xffff) << 32) | + (((M0 >> ((T0 >> 2) & 0x30)) & 0xffff) << 48); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +/* TODO: Unsigned-Saturation */ +void OPPROTO op_iwmmxt_packuw_M0_wRn(void) +{ + M0 = + (((M0 >> 0) & 0xff) << 0) | (((M0 >> 16) & 0xff) << 8) | + (((M0 >> 32) & 0xff) << 16) | (((M0 >> 48) & 0xff) << 24) | + (((M1 >> 0) & 0xff) << 32) | (((M1 >> 16) & 0xff) << 40) | + (((M1 >> 32) & 0xff) << 48) | (((M1 >> 48) & 0xff) << 56); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT8(M0 >> 0, 0) | NZBIT8(M0 >> 8, 1) | + NZBIT8(M0 >> 16, 2) | NZBIT8(M0 >> 24, 3) | + NZBIT8(M0 >> 32, 4) | NZBIT8(M0 >> 40, 5) | + NZBIT8(M0 >> 48, 6) | NZBIT8(M0 >> 56, 7); +} + +void OPPROTO op_iwmmxt_packul_M0_wRn(void) +{ + M0 = + (((M0 >> 0) & 0xffff) << 0) | (((M0 >> 32) & 0xffff) << 16) | + (((M1 >> 0) & 0xffff) << 32) | (((M1 >> 32) & 0xffff) << 48); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +void OPPROTO op_iwmmxt_packuq_M0_wRn(void) +{ + M0 = (M0 & 0xffffffff) | ((M1 & 0xffffffff) << 32); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); +} + +/* TODO: Signed-Saturation */ +void OPPROTO op_iwmmxt_packsw_M0_wRn(void) +{ + M0 = + (((M0 >> 0) & 0xff) << 0) | (((M0 >> 16) & 0xff) << 8) | + (((M0 >> 32) & 0xff) << 16) | (((M0 >> 48) & 0xff) << 24) | + (((M1 >> 0) & 0xff) << 32) | (((M1 >> 16) & 0xff) << 40) | + (((M1 >> 32) & 0xff) << 48) | (((M1 >> 48) & 0xff) << 56); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT8(M0 >> 0, 0) | NZBIT8(M0 >> 8, 1) | + NZBIT8(M0 >> 16, 2) | NZBIT8(M0 >> 24, 3) | + NZBIT8(M0 >> 32, 4) | NZBIT8(M0 >> 40, 5) | + NZBIT8(M0 >> 48, 6) | NZBIT8(M0 >> 56, 7); +} + +void OPPROTO op_iwmmxt_packsl_M0_wRn(void) +{ + M0 = + (((M0 >> 0) & 0xffff) << 0) | (((M0 >> 32) & 0xffff) << 16) | + (((M1 >> 0) & 0xffff) << 32) | (((M1 >> 32) & 0xffff) << 48); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT16(M0 >> 0, 0) | NZBIT16(M0 >> 16, 1) | + NZBIT16(M0 >> 32, 2) | NZBIT16(M0 >> 48, 3); +} + +void OPPROTO op_iwmmxt_packsq_M0_wRn(void) +{ + M0 = (M0 & 0xffffffff) | ((M1 & 0xffffffff) << 32); + env->iwmmxt.cregs[ARM_IWMMXT_wCASF] = + NZBIT32(M0 >> 0, 0) | NZBIT32(M0 >> 32, 1); +} + +void OPPROTO op_iwmmxt_muladdsl_M0_T0_T1(void) +{ + M0 += EXTEND32(T0) * EXTEND32(T1); +} + +void OPPROTO op_iwmmxt_muladdsw_M0_T0_T1(void) +{ + M0 += EXTEND32(EXTEND16((T0 >> 0) & 0xffff) * + EXTEND16((T1 >> 0) & 0xffff)); + M0 += EXTEND32(EXTEND16((T0 >> 16) & 0xffff) * + EXTEND16((T1 >> 16) & 0xffff)); +} + +void OPPROTO op_iwmmxt_muladdswl_M0_T0_T1(void) +{ + M0 += EXTEND32(EXTEND16(T0 & 0xffff) * EXTEND16(T1 & 0xffff)); +} diff --git a/target-arm/op_mem.h b/target-arm/op_mem.h index 29fd85b..6bccb06 100644 --- a/target-arm/op_mem.h +++ b/target-arm/op_mem.h @@ -67,4 +67,24 @@ VFP_MEM_OP(d,q) #undef VFP_MEM_OP +/* iwMMXt load/store. Address is in T1 */ +#define MMX_MEM_OP(name, ldname) \ +void OPPROTO glue(op_iwmmxt_ld##name,MEMSUFFIX)(void) \ +{ \ + M0 = glue(ld##ldname,MEMSUFFIX)(T1); \ + FORCE_RET(); \ +} \ +void OPPROTO glue(op_iwmmxt_st##name,MEMSUFFIX)(void) \ +{ \ + glue(st##name,MEMSUFFIX)(T1, M0); \ + FORCE_RET(); \ +} + +MMX_MEM_OP(b, ub) +MMX_MEM_OP(w, uw) +MMX_MEM_OP(l, l) +MMX_MEM_OP(q, q) + +#undef MMX_MEM_OP + #undef MEMSUFFIX diff --git a/target-arm/translate.c b/target-arm/translate.c index 24e8eee..e5032c5 100644 --- a/target-arm/translate.c +++ b/target-arm/translate.c @@ -3,6 +3,7 @@ * * Copyright (c) 2003 Fabrice Bellard * Copyright (c) 2005 CodeSourcery, LLC + * Copyright (c) 2007 OpenedHand, Ltd. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public @@ -492,6 +493,997 @@ static inline void gen_mov_vreg_F0(int dp, int reg) gen_op_vfp_setreg_F0s(vfp_reg_offset(dp, reg)); } +#define ARM_CP_RW_BIT (1 << 20) + +static inline int gen_iwmmxt_address(DisasContext *s, uint32_t insn) +{ + int rd; + uint32_t offset; + + rd = (insn >> 16) & 0xf; + gen_movl_T1_reg(s, rd); + + offset = (insn & 0xff) << ((insn >> 7) & 2); + if (insn & (1 << 24)) { + /* Pre indexed */ + if (insn & (1 << 23)) + gen_op_addl_T1_im(offset); + else + gen_op_addl_T1_im(-offset); + + if (insn & (1 << 21)) + gen_movl_reg_T1(s, rd); + } else if (insn & (1 << 21)) { + /* Post indexed */ + if (insn & (1 << 23)) + gen_op_movl_T0_im(offset); + else + gen_op_movl_T0_im(- offset); + gen_op_addl_T0_T1(); + gen_movl_reg_T0(s, rd); + } else if (!(insn & (1 << 23))) + return 1; + return 0; +} + +static inline int gen_iwmmxt_shift(uint32_t insn, uint32_t mask) +{ + int rd = (insn >> 0) & 0xf; + + if (insn & (1 << 8)) + if (rd < ARM_IWMMXT_wCGR0 || rd > ARM_IWMMXT_wCGR3) + return 1; + else + gen_op_iwmmxt_movl_T0_wCx(rd); + else + gen_op_iwmmxt_movl_T0_T1_wRn(rd); + + gen_op_movl_T1_im(mask); + gen_op_andl_T0_T1(); + return 0; +} + +/* Disassemble an iwMMXt instruction. Returns nonzero if an error occured + (ie. an undefined instruction). */ +static int disas_iwmmxt_insn(CPUState *env, DisasContext *s, uint32_t insn) +{ + int rd, wrd; + int rdhi, rdlo, rd0, rd1, i; + + if (!arm_feature(env, ARM_FEATURE_IWMMXT)) + return 1; + + if ((insn & 0x0e000e00) == 0x0c000000) { + if ((insn & 0x0fe00ff0) == 0x0c400000) { + wrd = insn & 0xf; + rdlo = (insn >> 12) & 0xf; + rdhi = (insn >> 16) & 0xf; + if (insn & ARM_CP_RW_BIT) { /* TMRRC */ + gen_op_iwmmxt_movl_T0_T1_wRn(wrd); + gen_movl_reg_T0(s, rdlo); + gen_movl_reg_T1(s, rdhi); + } else { /* TMCRR */ + gen_movl_T0_reg(s, rdlo); + gen_movl_T1_reg(s, rdhi); + gen_op_iwmmxt_movl_wRn_T0_T1(wrd); + gen_op_iwmmxt_set_mup(); + } + return 0; + } + + wrd = (insn >> 12) & 0xf; + if (gen_iwmmxt_address(s, insn)) + return 1; + if (insn & ARM_CP_RW_BIT) { + if ((insn >> 28) == 0xf) { /* WLDRW wCx */ + gen_ldst(ldl, s); + gen_op_iwmmxt_movl_wCx_T0(wrd); + } else { + if (insn & (1 << 8)) + if (insn & (1 << 22)) /* WLDRD */ + gen_ldst(iwmmxt_ldq, s); + else /* WLDRW wRd */ + gen_ldst(iwmmxt_ldl, s); + else + if (insn & (1 << 22)) /* WLDRH */ + gen_ldst(iwmmxt_ldw, s); + else /* WLDRB */ + gen_ldst(iwmmxt_ldb, s); + gen_op_iwmmxt_movq_wRn_M0(wrd); + } + } else { + if ((insn >> 28) == 0xf) { /* WSTRW wCx */ + gen_op_iwmmxt_movl_T0_wCx(wrd); + gen_ldst(stl, s); + } else { + gen_op_iwmmxt_movq_M0_wRn(wrd); + if (insn & (1 << 8)) + if (insn & (1 << 22)) /* WSTRD */ + gen_ldst(iwmmxt_stq, s); + else /* WSTRW wRd */ + gen_ldst(iwmmxt_stl, s); + else + if (insn & (1 << 22)) /* WSTRH */ + gen_ldst(iwmmxt_ldw, s); + else /* WSTRB */ + gen_ldst(iwmmxt_stb, s); + } + } + return 0; + } + + if ((insn & 0x0f000000) != 0x0e000000) + return 1; + + switch (((insn >> 12) & 0xf00) | ((insn >> 4) & 0xff)) { + case 0x000: /* WOR */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 0) & 0xf; + rd1 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_iwmmxt_orq_M0_wRn(rd1); + gen_op_iwmmxt_setpsr_nz(); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x011: /* TMCR */ + if (insn & 0xf) + return 1; + rd = (insn >> 12) & 0xf; + wrd = (insn >> 16) & 0xf; + switch (wrd) { + case ARM_IWMMXT_wCID: + case ARM_IWMMXT_wCASF: + break; + case ARM_IWMMXT_wCon: + gen_op_iwmmxt_set_cup(); + /* Fall through. */ + case ARM_IWMMXT_wCSSF: + gen_op_iwmmxt_movl_T0_wCx(wrd); + gen_movl_T1_reg(s, rd); + gen_op_bicl_T0_T1(); + gen_op_iwmmxt_movl_wCx_T0(wrd); + break; + case ARM_IWMMXT_wCGR0: + case ARM_IWMMXT_wCGR1: + case ARM_IWMMXT_wCGR2: + case ARM_IWMMXT_wCGR3: + gen_op_iwmmxt_set_cup(); + gen_movl_reg_T0(s, rd); + gen_op_iwmmxt_movl_wCx_T0(wrd); + break; + default: + return 1; + } + break; + case 0x100: /* WXOR */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 0) & 0xf; + rd1 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_iwmmxt_xorq_M0_wRn(rd1); + gen_op_iwmmxt_setpsr_nz(); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x111: /* TMRC */ + if (insn & 0xf) + return 1; + rd = (insn >> 12) & 0xf; + wrd = (insn >> 16) & 0xf; + gen_op_iwmmxt_movl_T0_wCx(wrd); + gen_movl_reg_T0(s, rd); + break; + case 0x300: /* WANDN */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 0) & 0xf; + rd1 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_iwmmxt_negq_M0(); + gen_op_iwmmxt_andq_M0_wRn(rd1); + gen_op_iwmmxt_setpsr_nz(); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x200: /* WAND */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 0) & 0xf; + rd1 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_iwmmxt_andq_M0_wRn(rd1); + gen_op_iwmmxt_setpsr_nz(); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x810: case 0xa10: /* WMADD */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 0) & 0xf; + rd1 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (insn & (1 << 21)) + gen_op_iwmmxt_maddsq_M0_wRn(rd1); + else + gen_op_iwmmxt_madduq_M0_wRn(rd1); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x10e: case 0x50e: case 0x90e: case 0xd0e: /* WUNPCKIL */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + gen_op_iwmmxt_unpacklb_M0_wRn(rd1); + break; + case 1: + gen_op_iwmmxt_unpacklw_M0_wRn(rd1); + break; + case 2: + gen_op_iwmmxt_unpackll_M0_wRn(rd1); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x10c: case 0x50c: case 0x90c: case 0xd0c: /* WUNPCKIH */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + gen_op_iwmmxt_unpackhb_M0_wRn(rd1); + break; + case 1: + gen_op_iwmmxt_unpackhw_M0_wRn(rd1); + break; + case 2: + gen_op_iwmmxt_unpackhl_M0_wRn(rd1); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x012: case 0x112: case 0x412: case 0x512: /* WSAD */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (insn & (1 << 22)) + gen_op_iwmmxt_sadw_M0_wRn(rd1); + else + gen_op_iwmmxt_sadb_M0_wRn(rd1); + if (!(insn & (1 << 20))) + gen_op_iwmmxt_addl_M0_wRn(wrd); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x010: case 0x110: case 0x210: case 0x310: /* WMUL */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (insn & (1 << 21)) + gen_op_iwmmxt_mulsw_M0_wRn(rd1, (insn & (1 << 20)) ? 16 : 0); + else + gen_op_iwmmxt_muluw_M0_wRn(rd1, (insn & (1 << 20)) ? 16 : 0); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x410: case 0x510: case 0x610: case 0x710: /* WMAC */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (insn & (1 << 21)) + gen_op_iwmmxt_macsw_M0_wRn(rd1); + else + gen_op_iwmmxt_macuw_M0_wRn(rd1); + if (!(insn & (1 << 20))) { + if (insn & (1 << 21)) + gen_op_iwmmxt_addsq_M0_wRn(wrd); + else + gen_op_iwmmxt_adduq_M0_wRn(wrd); + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x006: case 0x406: case 0x806: case 0xc06: /* WCMPEQ */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + gen_op_iwmmxt_cmpeqb_M0_wRn(rd1); + break; + case 1: + gen_op_iwmmxt_cmpeqw_M0_wRn(rd1); + break; + case 2: + gen_op_iwmmxt_cmpeql_M0_wRn(rd1); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x800: case 0x900: case 0xc00: case 0xd00: /* WAVG2 */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (insn & (1 << 22)) + gen_op_iwmmxt_avgw_M0_wRn(rd1, (insn >> 20) & 1); + else + gen_op_iwmmxt_avgb_M0_wRn(rd1, (insn >> 20) & 1); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x802: case 0x902: case 0xa02: case 0xb02: /* WALIGNR */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_iwmmxt_movl_T0_wCx(ARM_IWMMXT_wCGR0 + ((insn >> 20) & 3)); + gen_op_movl_T1_im(7); + gen_op_andl_T0_T1(); + gen_op_iwmmxt_align_M0_T0_wRn(rd1); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x601: case 0x605: case 0x609: case 0x60d: /* TINSR */ + rd = (insn >> 12) & 0xf; + wrd = (insn >> 16) & 0xf; + gen_movl_T0_reg(s, rd); + gen_op_iwmmxt_movq_M0_wRn(wrd); + switch ((insn >> 6) & 3) { + case 0: + gen_op_movl_T1_im(0xff); + gen_op_iwmmxt_insr_M0_T0_T1((insn & 7) << 3); + break; + case 1: + gen_op_movl_T1_im(0xffff); + gen_op_iwmmxt_insr_M0_T0_T1((insn & 3) << 4); + break; + case 2: + gen_op_movl_T1_im(0xffffffff); + gen_op_iwmmxt_insr_M0_T0_T1((insn & 1) << 5); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x107: case 0x507: case 0x907: case 0xd07: /* TEXTRM */ + rd = (insn >> 12) & 0xf; + wrd = (insn >> 16) & 0xf; + if (rd == 15) + return 1; + gen_op_iwmmxt_movq_M0_wRn(wrd); + switch ((insn >> 22) & 3) { + case 0: + if (insn & 8) + gen_op_iwmmxt_extrsb_T0_M0((insn & 7) << 3); + else { + gen_op_movl_T1_im(0xff); + gen_op_iwmmxt_extru_T0_M0_T1((insn & 7) << 3); + } + break; + case 1: + if (insn & 8) + gen_op_iwmmxt_extrsw_T0_M0((insn & 3) << 4); + else { + gen_op_movl_T1_im(0xffff); + gen_op_iwmmxt_extru_T0_M0_T1((insn & 3) << 4); + } + break; + case 2: + gen_op_movl_T1_im(0xffffffff); + gen_op_iwmmxt_extru_T0_M0_T1((insn & 1) << 5); + break; + case 3: + return 1; + } + gen_op_movl_reg_TN[0][rd](); + break; + case 0x117: case 0x517: case 0x917: case 0xd17: /* TEXTRC */ + if ((insn & 0x000ff008) != 0x0003f000) + return 1; + gen_op_iwmmxt_movl_T1_wCx(ARM_IWMMXT_wCASF); + switch ((insn >> 22) & 3) { + case 0: + gen_op_shrl_T1_im(((insn & 7) << 2) + 0); + break; + case 1: + gen_op_shrl_T1_im(((insn & 3) << 3) + 4); + break; + case 2: + gen_op_shrl_T1_im(((insn & 1) << 4) + 12); + break; + case 3: + return 1; + } + gen_op_shll_T1_im(28); + gen_op_movl_T0_T1(); + gen_op_movl_cpsr_T0(0xf0000000); + break; + case 0x401: case 0x405: case 0x409: case 0x40d: /* TBCST */ + rd = (insn >> 12) & 0xf; + wrd = (insn >> 16) & 0xf; + gen_movl_T0_reg(s, rd); + switch ((insn >> 6) & 3) { + case 0: + gen_op_iwmmxt_bcstb_M0_T0(); + break; + case 1: + gen_op_iwmmxt_bcstw_M0_T0(); + break; + case 2: + gen_op_iwmmxt_bcstl_M0_T0(); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x113: case 0x513: case 0x913: case 0xd13: /* TANDC */ + if ((insn & 0x000ff00f) != 0x0003f000) + return 1; + gen_op_iwmmxt_movl_T1_wCx(ARM_IWMMXT_wCASF); + switch ((insn >> 22) & 3) { + case 0: + for (i = 0; i < 7; i ++) { + gen_op_shll_T1_im(4); + gen_op_andl_T0_T1(); + } + break; + case 1: + for (i = 0; i < 3; i ++) { + gen_op_shll_T1_im(8); + gen_op_andl_T0_T1(); + } + break; + case 2: + gen_op_shll_T1_im(16); + gen_op_andl_T0_T1(); + break; + case 3: + return 1; + } + gen_op_movl_cpsr_T0(0xf0000000); + break; + case 0x01c: case 0x41c: case 0x81c: case 0xc1c: /* WACC */ + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + gen_op_iwmmxt_addcb_M0(); + break; + case 1: + gen_op_iwmmxt_addcw_M0(); + break; + case 2: + gen_op_iwmmxt_addcl_M0(); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x115: case 0x515: case 0x915: case 0xd15: /* TORC */ + if ((insn & 0x000ff00f) != 0x0003f000) + return 1; + gen_op_iwmmxt_movl_T1_wCx(ARM_IWMMXT_wCASF); + switch ((insn >> 22) & 3) { + case 0: + for (i = 0; i < 7; i ++) { + gen_op_shll_T1_im(4); + gen_op_orl_T0_T1(); + } + break; + case 1: + for (i = 0; i < 3; i ++) { + gen_op_shll_T1_im(8); + gen_op_orl_T0_T1(); + } + break; + case 2: + gen_op_shll_T1_im(16); + gen_op_orl_T0_T1(); + break; + case 3: + return 1; + } + gen_op_movl_T1_im(0xf0000000); + gen_op_andl_T0_T1(); + gen_op_movl_cpsr_T0(0xf0000000); + break; + case 0x103: case 0x503: case 0x903: case 0xd03: /* TMOVMSK */ + rd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + if ((insn & 0xf) != 0) + return 1; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + gen_op_iwmmxt_msbb_T0_M0(); + break; + case 1: + gen_op_iwmmxt_msbw_T0_M0(); + break; + case 2: + gen_op_iwmmxt_msbl_T0_M0(); + break; + case 3: + return 1; + } + gen_movl_reg_T0(s, rd); + break; + case 0x106: case 0x306: case 0x506: case 0x706: /* WCMPGT */ + case 0x906: case 0xb06: case 0xd06: case 0xf06: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + if (insn & (1 << 21)) + gen_op_iwmmxt_cmpgtsb_M0_wRn(rd1); + else + gen_op_iwmmxt_cmpgtub_M0_wRn(rd1); + break; + case 1: + if (insn & (1 << 21)) + gen_op_iwmmxt_cmpgtsw_M0_wRn(rd1); + else + gen_op_iwmmxt_cmpgtuw_M0_wRn(rd1); + break; + case 2: + if (insn & (1 << 21)) + gen_op_iwmmxt_cmpgtsl_M0_wRn(rd1); + else + gen_op_iwmmxt_cmpgtul_M0_wRn(rd1); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x00e: case 0x20e: case 0x40e: case 0x60e: /* WUNPCKEL */ + case 0x80e: case 0xa0e: case 0xc0e: case 0xe0e: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + if (insn & (1 << 21)) + gen_op_iwmmxt_unpacklsb_M0(); + else + gen_op_iwmmxt_unpacklub_M0(); + break; + case 1: + if (insn & (1 << 21)) + gen_op_iwmmxt_unpacklsw_M0(); + else + gen_op_iwmmxt_unpackluw_M0(); + break; + case 2: + if (insn & (1 << 21)) + gen_op_iwmmxt_unpacklsl_M0(); + else + gen_op_iwmmxt_unpacklul_M0(); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x00c: case 0x20c: case 0x40c: case 0x60c: /* WUNPCKEH */ + case 0x80c: case 0xa0c: case 0xc0c: case 0xe0c: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + if (insn & (1 << 21)) + gen_op_iwmmxt_unpackhsb_M0(); + else + gen_op_iwmmxt_unpackhub_M0(); + break; + case 1: + if (insn & (1 << 21)) + gen_op_iwmmxt_unpackhsw_M0(); + else + gen_op_iwmmxt_unpackhuw_M0(); + break; + case 2: + if (insn & (1 << 21)) + gen_op_iwmmxt_unpackhsl_M0(); + else + gen_op_iwmmxt_unpackhul_M0(); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x204: case 0x604: case 0xa04: case 0xe04: /* WSRL */ + case 0x214: case 0x614: case 0xa14: case 0xe14: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (gen_iwmmxt_shift(insn, 0xff)) + return 1; + switch ((insn >> 22) & 3) { + case 0: + return 1; + case 1: + gen_op_iwmmxt_srlw_M0_T0(); + break; + case 2: + gen_op_iwmmxt_srll_M0_T0(); + break; + case 3: + gen_op_iwmmxt_srlq_M0_T0(); + break; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x004: case 0x404: case 0x804: case 0xc04: /* WSRA */ + case 0x014: case 0x414: case 0x814: case 0xc14: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (gen_iwmmxt_shift(insn, 0xff)) + return 1; + switch ((insn >> 22) & 3) { + case 0: + return 1; + case 1: + gen_op_iwmmxt_sraw_M0_T0(); + break; + case 2: + gen_op_iwmmxt_sral_M0_T0(); + break; + case 3: + gen_op_iwmmxt_sraq_M0_T0(); + break; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x104: case 0x504: case 0x904: case 0xd04: /* WSLL */ + case 0x114: case 0x514: case 0x914: case 0xd14: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (gen_iwmmxt_shift(insn, 0xff)) + return 1; + switch ((insn >> 22) & 3) { + case 0: + return 1; + case 1: + gen_op_iwmmxt_sllw_M0_T0(); + break; + case 2: + gen_op_iwmmxt_slll_M0_T0(); + break; + case 3: + gen_op_iwmmxt_sllq_M0_T0(); + break; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x304: case 0x704: case 0xb04: case 0xf04: /* WROR */ + case 0x314: case 0x714: case 0xb14: case 0xf14: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + return 1; + case 1: + if (gen_iwmmxt_shift(insn, 0xf)) + return 1; + gen_op_iwmmxt_rorw_M0_T0(); + break; + case 2: + if (gen_iwmmxt_shift(insn, 0x1f)) + return 1; + gen_op_iwmmxt_rorl_M0_T0(); + break; + case 3: + if (gen_iwmmxt_shift(insn, 0x3f)) + return 1; + gen_op_iwmmxt_rorq_M0_T0(); + break; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x116: case 0x316: case 0x516: case 0x716: /* WMIN */ + case 0x916: case 0xb16: case 0xd16: case 0xf16: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + if (insn & (1 << 21)) + gen_op_iwmmxt_minsb_M0_wRn(rd1); + else + gen_op_iwmmxt_minub_M0_wRn(rd1); + break; + case 1: + if (insn & (1 << 21)) + gen_op_iwmmxt_minsw_M0_wRn(rd1); + else + gen_op_iwmmxt_minuw_M0_wRn(rd1); + break; + case 2: + if (insn & (1 << 21)) + gen_op_iwmmxt_minsl_M0_wRn(rd1); + else + gen_op_iwmmxt_minul_M0_wRn(rd1); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x016: case 0x216: case 0x416: case 0x616: /* WMAX */ + case 0x816: case 0xa16: case 0xc16: case 0xe16: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 22) & 3) { + case 0: + if (insn & (1 << 21)) + gen_op_iwmmxt_maxsb_M0_wRn(rd1); + else + gen_op_iwmmxt_maxub_M0_wRn(rd1); + break; + case 1: + if (insn & (1 << 21)) + gen_op_iwmmxt_maxsw_M0_wRn(rd1); + else + gen_op_iwmmxt_maxuw_M0_wRn(rd1); + break; + case 2: + if (insn & (1 << 21)) + gen_op_iwmmxt_maxsl_M0_wRn(rd1); + else + gen_op_iwmmxt_maxul_M0_wRn(rd1); + break; + case 3: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x002: case 0x102: case 0x202: case 0x302: /* WALIGNI */ + case 0x402: case 0x502: case 0x602: case 0x702: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_movl_T0_im((insn >> 20) & 3); + gen_op_iwmmxt_align_M0_T0_wRn(rd1); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + case 0x01a: case 0x11a: case 0x21a: case 0x31a: /* WSUB */ + case 0x41a: case 0x51a: case 0x61a: case 0x71a: + case 0x81a: case 0x91a: case 0xa1a: case 0xb1a: + case 0xc1a: case 0xd1a: case 0xe1a: case 0xf1a: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 20) & 0xf) { + case 0x0: + gen_op_iwmmxt_subnb_M0_wRn(rd1); + break; + case 0x1: + gen_op_iwmmxt_subub_M0_wRn(rd1); + break; + case 0x3: + gen_op_iwmmxt_subsb_M0_wRn(rd1); + break; + case 0x4: + gen_op_iwmmxt_subnw_M0_wRn(rd1); + break; + case 0x5: + gen_op_iwmmxt_subuw_M0_wRn(rd1); + break; + case 0x7: + gen_op_iwmmxt_subsw_M0_wRn(rd1); + break; + case 0x8: + gen_op_iwmmxt_subnl_M0_wRn(rd1); + break; + case 0x9: + gen_op_iwmmxt_subul_M0_wRn(rd1); + break; + case 0xb: + gen_op_iwmmxt_subsl_M0_wRn(rd1); + break; + default: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x01e: case 0x11e: case 0x21e: case 0x31e: /* WSHUFH */ + case 0x41e: case 0x51e: case 0x61e: case 0x71e: + case 0x81e: case 0x91e: case 0xa1e: case 0xb1e: + case 0xc1e: case 0xd1e: case 0xe1e: case 0xf1e: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + gen_op_movl_T0_im(((insn >> 16) & 0xf0) | (insn & 0x0f)); + gen_op_iwmmxt_shufh_M0_T0(); + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x018: case 0x118: case 0x218: case 0x318: /* WADD */ + case 0x418: case 0x518: case 0x618: case 0x718: + case 0x818: case 0x918: case 0xa18: case 0xb18: + case 0xc18: case 0xd18: case 0xe18: case 0xf18: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + switch ((insn >> 20) & 0xf) { + case 0x0: + gen_op_iwmmxt_addnb_M0_wRn(rd1); + break; + case 0x1: + gen_op_iwmmxt_addub_M0_wRn(rd1); + break; + case 0x3: + gen_op_iwmmxt_addsb_M0_wRn(rd1); + break; + case 0x4: + gen_op_iwmmxt_addnw_M0_wRn(rd1); + break; + case 0x5: + gen_op_iwmmxt_adduw_M0_wRn(rd1); + break; + case 0x7: + gen_op_iwmmxt_addsw_M0_wRn(rd1); + break; + case 0x8: + gen_op_iwmmxt_addnl_M0_wRn(rd1); + break; + case 0x9: + gen_op_iwmmxt_addul_M0_wRn(rd1); + break; + case 0xb: + gen_op_iwmmxt_addsl_M0_wRn(rd1); + break; + default: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x008: case 0x108: case 0x208: case 0x308: /* WPACK */ + case 0x408: case 0x508: case 0x608: case 0x708: + case 0x808: case 0x908: case 0xa08: case 0xb08: + case 0xc08: case 0xd08: case 0xe08: case 0xf08: + wrd = (insn >> 12) & 0xf; + rd0 = (insn >> 16) & 0xf; + rd1 = (insn >> 0) & 0xf; + gen_op_iwmmxt_movq_M0_wRn(rd0); + if (!(insn & (1 << 20))) + return 1; + switch ((insn >> 22) & 3) { + case 0: + return 1; + case 1: + if (insn & (1 << 21)) + gen_op_iwmmxt_packsw_M0_wRn(rd1); + else + gen_op_iwmmxt_packuw_M0_wRn(rd1); + break; + case 2: + if (insn & (1 << 21)) + gen_op_iwmmxt_packsl_M0_wRn(rd1); + else + gen_op_iwmmxt_packul_M0_wRn(rd1); + break; + case 3: + if (insn & (1 << 21)) + gen_op_iwmmxt_packsq_M0_wRn(rd1); + else + gen_op_iwmmxt_packuq_M0_wRn(rd1); + break; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + gen_op_iwmmxt_set_cup(); + break; + case 0x201: case 0x203: case 0x205: case 0x207: + case 0x209: case 0x20b: case 0x20d: case 0x20f: + case 0x211: case 0x213: case 0x215: case 0x217: + case 0x219: case 0x21b: case 0x21d: case 0x21f: + wrd = (insn >> 5) & 0xf; + rd0 = (insn >> 12) & 0xf; + rd1 = (insn >> 0) & 0xf; + if (rd0 == 0xf || rd1 == 0xf) + return 1; + gen_op_iwmmxt_movq_M0_wRn(wrd); + switch ((insn >> 16) & 0xf) { + case 0x0: /* TMIA */ + gen_op_movl_TN_reg[0][rd0](); + gen_op_movl_TN_reg[1][rd1](); + gen_op_iwmmxt_muladdsl_M0_T0_T1(); + break; + case 0x8: /* TMIAPH */ + gen_op_movl_TN_reg[0][rd0](); + gen_op_movl_TN_reg[1][rd1](); + gen_op_iwmmxt_muladdsw_M0_T0_T1(); + break; + case 0xc: case 0xd: case 0xe: case 0xf: /* TMIAxy */ + gen_op_movl_TN_reg[1][rd0](); + if (insn & (1 << 16)) + gen_op_shrl_T1_im(16); + gen_op_movl_T0_T1(); + gen_op_movl_TN_reg[1][rd1](); + if (insn & (1 << 17)) + gen_op_shrl_T1_im(16); + gen_op_iwmmxt_muladdswl_M0_T0_T1(); + break; + default: + return 1; + } + gen_op_iwmmxt_movq_wRn_M0(wrd); + gen_op_iwmmxt_set_mup(); + break; + default: + return 1; + } + + return 0; +} + /* Disassemble system coprocessor instruction. Return nonzero if instruction is not defined. */ static int disas_cp_insn(DisasContext *s, uint32_t insn) @@ -501,7 +1493,7 @@ static int disas_cp_insn(DisasContext *s, uint32_t insn) return 1; } - if (insn & (1 << 20)) { + if (insn & ARM_CP_RW_BIT) { gen_op_movl_T0_cp(insn); gen_movl_reg_T0(s, rd); } else { @@ -535,7 +1527,7 @@ static int disas_cp15_insn(DisasContext *s, uint32_t insn) return 0; } rd = (insn >> 12) & 0xf; - if (insn & (1 << 20)) { + if (insn & ARM_CP_RW_BIT) { gen_op_movl_T0_cp15(insn); /* If the destination register is r15 then sets condition codes. */ if (rd != 15) @@ -582,7 +1574,7 @@ static int disas_vfp_insn(CPUState * env, DisasContext *s, uint32_t insn) we only set half the register. */ gen_mov_F0_vreg(1, rn); gen_op_vfp_mrrd(); - if (insn & (1 << 20)) { + if (insn & ARM_CP_RW_BIT) { /* vfp->arm */ if (insn & (1 << 21)) gen_movl_reg_T1(s, rd); @@ -599,7 +1591,7 @@ static int disas_vfp_insn(CPUState * env, DisasContext *s, uint32_t insn) } } else { rn = ((insn >> 15) & 0x1e) | ((insn >> 7) & 1); - if (insn & (1 << 20)) { + if (insn & ARM_CP_RW_BIT) { /* vfp->arm */ if (insn & (1 << 21)) { /* system register */ @@ -933,7 +1925,7 @@ static int disas_vfp_insn(CPUState * env, DisasContext *s, uint32_t insn) } else rm = ((insn << 1) & 0x1e) | ((insn >> 5) & 1); - if (insn & (1 << 20)) { + if (insn & ARM_CP_RW_BIT) { /* vfp->arm */ if (dp) { gen_mov_F0_vreg(1, rm); @@ -1000,7 +1992,7 @@ static int disas_vfp_insn(CPUState * env, DisasContext *s, uint32_t insn) else offset = 4; for (i = 0; i < n; i++) { - if (insn & (1 << 20)) { + if (insn & ARM_CP_RW_BIT) { /* load */ gen_vfp_ld(s, dp); gen_mov_vreg_F0(dp, rd + i); @@ -1825,6 +2817,9 @@ static void disas_arm_insn(CPUState * env, DisasContext *s) goto illegal_op; switch (op1) { case 0 ... 1: + if (disas_iwmmxt_insn(env, s, insn)) + goto illegal_op; + break; case 2 ... 9: case 12 ... 14: if (disas_cp_insn(s, insn)) diff --git a/tests/Makefile b/tests/Makefile index 79a3d55..49f1c3f 100644 --- a/tests/Makefile +++ b/tests/Makefile @@ -82,6 +82,9 @@ hello-arm: hello-arm.o hello-arm.o: hello-arm.c arm-linux-gcc -Wall -g -O2 -c -o $@ $< +test-arm-iwmmxt: test-arm-iwmmxt.s + cpp < $< | arm-palmte-linux-gnu-gcc -Wall -static -march=iwmmxt -mabi=aapcs -x assembler - -o $@ + # MIPS test hello-mips: hello-mips.c mips-linux-gnu-gcc -nostdlib -static -mno-abicalls -fno-PIC -mabi=32 -Wall -Wextra -g -O2 -o $@ $< diff --git a/tests/test-arm-iwmmxt.s b/tests/test-arm-iwmmxt.s new file mode 100644 index 0000000..d647f94 --- /dev/null +++ b/tests/test-arm-iwmmxt.s @@ -0,0 +1,49 @@ +@ Checks whether iwMMXt is functional. +.code 32 +.globl main + +main: +ldr r0, =data0 +ldr r1, =data1 +ldr r2, =data2 +#ifndef FPA +wldrd wr0, [r0, #0] +wldrd wr1, [r0, #8] +wldrd wr2, [r1, #0] +wldrd wr3, [r1, #8] +wsubb wr2, wr2, wr0 +wsubb wr3, wr3, wr1 +wldrd wr0, [r2, #0] +wldrd wr1, [r2, #8] +waddb wr0, wr0, wr2 +waddb wr1, wr1, wr3 +wstrd wr0, [r2, #0] +wstrd wr1, [r2, #8] +#else +ldfe f0, [r0, #0] +ldfe f1, [r0, #8] +ldfe f2, [r1, #0] +ldfe f3, [r1, #8] +adfdp f2, f2, f0 +adfdp f3, f3, f1 +ldfe f0, [r2, #0] +ldfe f1, [r2, #8] +adfd f0, f0, f2 +adfd f1, f1, f3 +stfe f0, [r2, #0] +stfe f1, [r2, #8] +#endif +mov r0, #1 +mov r1, r2 +mov r2, #0x11 +swi #0x900004 +mov r0, #0 +swi #0x900001 + +.data +data0: +.string "aaaabbbbccccdddd" +data1: +.string "bbbbccccddddeeee" +data2: +.string "hvLLWs\x1fsdrs9\x1fNJ-\n" -- 1.4.4.3