From ee17db83d2dce35792e9bf03366af193e5e0e5c9 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 29 Mar 2020 10:11:56 -0700 Subject: tcg: Consolidate 3 bits into enum TCGTempKind The temp_fixed, temp_global, temp_local bits are all related. Combine them into a single enumeration. Reviewed-by: Alex Bennée Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- include/tcg/tcg.h | 20 ++++++++++++-------- 1 file changed, 12 insertions(+), 8 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index 95fe5604eb..571d4e0fa3 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -483,23 +483,27 @@ typedef enum TCGTempVal { TEMP_VAL_CONST, } TCGTempVal; +typedef enum TCGTempKind { + /* Temp is dead at the end of all basic blocks. */ + TEMP_NORMAL, + /* Temp is saved across basic blocks but dead at the end of TBs. */ + TEMP_LOCAL, + /* Temp is saved across both basic blocks and translation blocks. */ + TEMP_GLOBAL, + /* Temp is in a fixed register. */ + TEMP_FIXED, +} TCGTempKind; + typedef struct TCGTemp { TCGReg reg:8; TCGTempVal val_type:8; TCGType base_type:8; TCGType type:8; - unsigned int fixed_reg:1; + TCGTempKind kind:3; unsigned int indirect_reg:1; unsigned int indirect_base:1; unsigned int mem_coherent:1; unsigned int mem_allocated:1; - /* If true, the temp is saved across both basic blocks and - translation blocks. */ - unsigned int temp_global:1; - /* If true, the temp is saved across basic blocks but dead - at the end of translation blocks. If false, the temp is - dead at the end of basic blocks. */ - unsigned int temp_local:1; unsigned int temp_allocated:1; tcg_target_long val; -- cgit v1.2.3-55-g7522 From e01fa97dea857a35be5bb8cce0d632a62e72c689 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 29 Mar 2020 10:40:49 -0700 Subject: tcg: Add temp_readonly In most, but not all, places that we check for TEMP_FIXED, we are really testing that we do not modify the temporary. Reviewed-by: Alex Bennée Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Richard Henderson --- include/tcg/tcg.h | 5 +++++ tcg/tcg.c | 21 ++++++++++----------- 2 files changed, 15 insertions(+), 11 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index 571d4e0fa3..2bdaeaa69c 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -679,6 +679,11 @@ struct TCGContext { target_ulong gen_insn_data[TCG_MAX_INSNS][TARGET_INSN_START_WORDS]; }; +static inline bool temp_readonly(TCGTemp *ts) +{ + return ts->kind == TEMP_FIXED; +} + extern TCGContext tcg_init_ctx; extern __thread TCGContext *tcg_ctx; extern const void *tcg_code_gen_epilogue; diff --git a/tcg/tcg.c b/tcg/tcg.c index 143794a585..e02bb71953 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -3254,7 +3254,7 @@ static void temp_load(TCGContext *, TCGTemp *, TCGRegSet, TCGRegSet, TCGRegSet); mark it free; otherwise mark it dead. */ static void temp_free_or_dead(TCGContext *s, TCGTemp *ts, int free_or_dead) { - if (ts->kind == TEMP_FIXED) { + if (temp_readonly(ts)) { return; } if (ts->val_type == TEMP_VAL_REG) { @@ -3278,7 +3278,7 @@ static inline void temp_dead(TCGContext *s, TCGTemp *ts) static void temp_sync(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs, TCGRegSet preferred_regs, int free_or_dead) { - if (ts->kind == TEMP_FIXED) { + if (temp_readonly(ts)) { return; } if (!ts->mem_coherent) { @@ -3461,8 +3461,7 @@ static void temp_save(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs) { /* The liveness analysis already ensures that globals are back in memory. Keep an tcg_debug_assert for safety. */ - tcg_debug_assert(ts->val_type == TEMP_VAL_MEM - || ts->kind == TEMP_FIXED); + tcg_debug_assert(ts->val_type == TEMP_VAL_MEM || temp_readonly(ts)); } /* save globals to their canonical location and assume they can be @@ -3542,7 +3541,7 @@ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, TCGRegSet preferred_regs) { /* ENV should not be modified. */ - tcg_debug_assert(ots->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ots)); /* The movi is not explicitly generated here. */ if (ots->val_type == TEMP_VAL_REG) { @@ -3582,7 +3581,7 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) ts = arg_temp(op->args[1]); /* ENV should not be modified. */ - tcg_debug_assert(ots->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ots)); /* Note that otype != itype for no-op truncation. */ otype = ots->type; @@ -3643,7 +3642,7 @@ static void tcg_reg_alloc_mov(TCGContext *s, const TCGOp *op) * Store the source register into the destination slot * and leave the destination temp as TEMP_VAL_MEM. */ - assert(ots->kind != TEMP_FIXED); + assert(!temp_readonly(ots)); if (!ts->mem_allocated) { temp_allocate_frame(s, ots); } @@ -3680,7 +3679,7 @@ static void tcg_reg_alloc_dup(TCGContext *s, const TCGOp *op) its = arg_temp(op->args[1]); /* ENV should not be modified. */ - tcg_debug_assert(ots->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ots)); itype = its->type; vece = TCGOP_VECE(op); @@ -3912,7 +3911,7 @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) ts = arg_temp(arg); /* ENV should not be modified. */ - tcg_debug_assert(ts->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ts)); if (arg_ct->oalias && !const_args[arg_ct->alias_index]) { reg = new_args[arg_ct->alias_index]; @@ -3953,7 +3952,7 @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) ts = arg_temp(op->args[i]); /* ENV should not be modified. */ - tcg_debug_assert(ts->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ts)); if (NEED_SYNC_ARG(i)) { temp_sync(s, ts, o_allocated_regs, 0, IS_DEAD_ARG(i)); @@ -4085,7 +4084,7 @@ static void tcg_reg_alloc_call(TCGContext *s, TCGOp *op) ts = arg_temp(arg); /* ENV should not be modified. */ - tcg_debug_assert(ts->kind != TEMP_FIXED); + tcg_debug_assert(!temp_readonly(ts)); reg = tcg_target_call_oarg_regs[i]; tcg_debug_assert(s->reg_to_temp[reg] == NULL); -- cgit v1.2.3-55-g7522 From bdb38b95f72ebbef2d24e057828dd18ba9c81f63 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 6 Sep 2020 11:31:44 -0700 Subject: tcg: Expand TCGTemp.val to 64-bits This will reduce the differences between 32-bit and 64-bit hosts, allowing full 64-bit constants to be created with the same interface. Signed-off-by: Richard Henderson --- include/tcg/tcg.h | 2 +- tcg/tcg.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index 2bdaeaa69c..e7adc7e265 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -506,7 +506,7 @@ typedef struct TCGTemp { unsigned int mem_allocated:1; unsigned int temp_allocated:1; - tcg_target_long val; + int64_t val; struct TCGTemp *mem_base; intptr_t mem_offset; const char *name; diff --git a/tcg/tcg.c b/tcg/tcg.c index e02bb71953..545dd2b0b2 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -3176,7 +3176,7 @@ static void dump_regs(TCGContext *s) tcg_target_reg_names[ts->mem_base->reg]); break; case TEMP_VAL_CONST: - printf("$0x%" TCG_PRIlx, ts->val); + printf("$0x%" PRIx64, ts->val); break; case TEMP_VAL_DEAD: printf("D"); -- cgit v1.2.3-55-g7522 From c0522136adf550c7a0ef7c0755c1f9d1560d2757 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 29 Mar 2020 18:55:52 -0700 Subject: tcg: Introduce TYPE_CONST temporaries These will hold a single constant for the duration of the TB. They are hashed, so that each value has one temp across the TB. Not used yet, this is all infrastructure. Signed-off-by: Richard Henderson --- include/tcg/tcg.h | 24 +++++- tcg/optimize.c | 13 +++- tcg/tcg.c | 224 ++++++++++++++++++++++++++++++++++++++++++------------ 3 files changed, 211 insertions(+), 50 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index e7adc7e265..eeeb70ad43 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -492,6 +492,8 @@ typedef enum TCGTempKind { TEMP_GLOBAL, /* Temp is in a fixed register. */ TEMP_FIXED, + /* Temp is a fixed constant. */ + TEMP_CONST, } TCGTempKind; typedef struct TCGTemp { @@ -665,6 +667,7 @@ struct TCGContext { QSIMPLEQ_HEAD(, TCGOp) plugin_ops; #endif + GHashTable *const_table[TCG_TYPE_COUNT]; TCGTempSet free_temps[TCG_TYPE_COUNT * 2]; TCGTemp temps[TCG_MAX_TEMPS]; /* globals first, temps after */ @@ -681,7 +684,7 @@ struct TCGContext { static inline bool temp_readonly(TCGTemp *ts) { - return ts->kind == TEMP_FIXED; + return ts->kind >= TEMP_FIXED; } extern TCGContext tcg_init_ctx; @@ -1079,6 +1082,7 @@ TCGOp *tcg_op_insert_after(TCGContext *s, TCGOp *op, TCGOpcode opc); void tcg_optimize(TCGContext *s); +/* Allocate a new temporary and initialize it with a constant. */ TCGv_i32 tcg_const_i32(int32_t val); TCGv_i64 tcg_const_i64(int64_t val); TCGv_i32 tcg_const_local_i32(int32_t val); @@ -1088,6 +1092,24 @@ TCGv_vec tcg_const_ones_vec(TCGType); TCGv_vec tcg_const_zeros_vec_matching(TCGv_vec); TCGv_vec tcg_const_ones_vec_matching(TCGv_vec); +/* + * Locate or create a read-only temporary that is a constant. + * This kind of temporary need not and should not be freed. + */ +TCGTemp *tcg_constant_internal(TCGType type, int64_t val); + +static inline TCGv_i32 tcg_constant_i32(int32_t val) +{ + return temp_tcgv_i32(tcg_constant_internal(TCG_TYPE_I32, val)); +} + +static inline TCGv_i64 tcg_constant_i64(int64_t val) +{ + return temp_tcgv_i64(tcg_constant_internal(TCG_TYPE_I64, val)); +} + +TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val); + #if UINTPTR_MAX == UINT32_MAX # define tcg_const_ptr(x) ((TCGv_ptr)tcg_const_i32((intptr_t)(x))) # define tcg_const_local_ptr(x) ((TCGv_ptr)tcg_const_local_i32((intptr_t)(x))) diff --git a/tcg/optimize.c b/tcg/optimize.c index 433d2540f4..16b0aa7229 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -99,8 +99,17 @@ static void init_ts_info(TempOptInfo *infos, ts->state_ptr = ti; ti->next_copy = ts; ti->prev_copy = ts; - ti->is_const = false; - ti->mask = -1; + if (ts->kind == TEMP_CONST) { + ti->is_const = true; + ti->val = ti->mask = ts->val; + if (TCG_TARGET_REG_BITS > 32 && ts->type == TCG_TYPE_I32) { + /* High bits of a 32-bit quantity are garbage. */ + ti->mask |= ~0xffffffffull; + } + } else { + ti->is_const = false; + ti->mask = -1; + } set_bit(idx, temps_used->l); } } diff --git a/tcg/tcg.c b/tcg/tcg.c index 545dd2b0b2..802f0b8a32 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -1184,6 +1184,13 @@ void tcg_func_start(TCGContext *s) /* No temps have been previously allocated for size or locality. */ memset(s->free_temps, 0, sizeof(s->free_temps)); + /* No constant temps have been previously allocated. */ + for (int i = 0; i < TCG_TYPE_COUNT; ++i) { + if (s->const_table[i]) { + g_hash_table_remove_all(s->const_table[i]); + } + } + s->nb_ops = 0; s->nb_labels = 0; s->current_frame_offset = s->frame_start; @@ -1255,13 +1262,19 @@ TCGTemp *tcg_global_mem_new_internal(TCGType type, TCGv_ptr base, bigendian = 1; #endif - if (base_ts->kind != TEMP_FIXED) { + switch (base_ts->kind) { + case TEMP_FIXED: + break; + case TEMP_GLOBAL: /* We do not support double-indirect registers. */ tcg_debug_assert(!base_ts->indirect_reg); base_ts->indirect_base = 1; s->nb_indirects += (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64 ? 2 : 1); indirect_reg = 1; + break; + default: + g_assert_not_reached(); } if (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64) { @@ -1386,6 +1399,11 @@ void tcg_temp_free_internal(TCGTemp *ts) TCGContext *s = tcg_ctx; int k, idx; + /* In order to simplify users of tcg_constant_*, silently ignore free. */ + if (ts->kind == TEMP_CONST) { + return; + } + #if defined(CONFIG_DEBUG_TCG) s->temps_in_use--; if (s->temps_in_use < 0) { @@ -1402,6 +1420,60 @@ void tcg_temp_free_internal(TCGTemp *ts) set_bit(idx, s->free_temps[k].l); } +TCGTemp *tcg_constant_internal(TCGType type, int64_t val) +{ + TCGContext *s = tcg_ctx; + GHashTable *h = s->const_table[type]; + TCGTemp *ts; + + if (h == NULL) { + h = g_hash_table_new(g_int64_hash, g_int64_equal); + s->const_table[type] = h; + } + + ts = g_hash_table_lookup(h, &val); + if (ts == NULL) { + ts = tcg_temp_alloc(s); + + if (TCG_TARGET_REG_BITS == 32 && type == TCG_TYPE_I64) { + TCGTemp *ts2 = tcg_temp_alloc(s); + + ts->base_type = TCG_TYPE_I64; + ts->type = TCG_TYPE_I32; + ts->kind = TEMP_CONST; + ts->temp_allocated = 1; + /* + * Retain the full value of the 64-bit constant in the low + * part, so that the hash table works. Actual uses will + * truncate the value to the low part. + */ + ts->val = val; + + tcg_debug_assert(ts2 == ts + 1); + ts2->base_type = TCG_TYPE_I64; + ts2->type = TCG_TYPE_I32; + ts2->kind = TEMP_CONST; + ts2->temp_allocated = 1; + ts2->val = val >> 32; + } else { + ts->base_type = type; + ts->type = type; + ts->kind = TEMP_CONST; + ts->temp_allocated = 1; + ts->val = val; + } + g_hash_table_insert(h, &ts->val, ts); + } + + return ts; +} + +TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val) +{ + val = dup_const(vece, val); + return temp_tcgv_vec(tcg_constant_internal(type, val)); +} + TCGv_i32 tcg_const_i32(int32_t val) { TCGv_i32 t0; @@ -1937,6 +2009,9 @@ static void tcg_reg_alloc_start(TCGContext *s) TCGTempVal val = TEMP_VAL_MEM; switch (ts->kind) { + case TEMP_CONST: + val = TEMP_VAL_CONST; + break; case TEMP_FIXED: val = TEMP_VAL_REG; break; @@ -1973,6 +2048,26 @@ static char *tcg_get_arg_str_ptr(TCGContext *s, char *buf, int buf_size, case TEMP_NORMAL: snprintf(buf, buf_size, "tmp%d", idx - s->nb_globals); break; + case TEMP_CONST: + switch (ts->type) { + case TCG_TYPE_I32: + snprintf(buf, buf_size, "$0x%x", (int32_t)ts->val); + break; +#if TCG_TARGET_REG_BITS > 32 + case TCG_TYPE_I64: + snprintf(buf, buf_size, "$0x%" PRIx64, ts->val); + break; +#endif + case TCG_TYPE_V64: + case TCG_TYPE_V128: + case TCG_TYPE_V256: + snprintf(buf, buf_size, "v%d$0x%" PRIx64, + 64 << (ts->type - TCG_TYPE_V64), ts->val); + break; + default: + g_assert_not_reached(); + } + break; } return buf; } @@ -2574,6 +2669,7 @@ static void la_bb_end(TCGContext *s, int ng, int nt) state = TS_DEAD | TS_MEM; break; case TEMP_NORMAL: + case TEMP_CONST: state = TS_DEAD; break; default: @@ -2608,14 +2704,24 @@ static void la_bb_sync(TCGContext *s, int ng, int nt) la_global_sync(s, ng); for (int i = ng; i < nt; ++i) { - if (s->temps[i].kind == TEMP_LOCAL) { - int state = s->temps[i].state; - s->temps[i].state = state | TS_MEM; + TCGTemp *ts = &s->temps[i]; + int state; + + switch (ts->kind) { + case TEMP_LOCAL: + state = ts->state; + ts->state = state | TS_MEM; if (state != TS_DEAD) { continue; } - } else { + break; + case TEMP_NORMAL: s->temps[i].state = TS_DEAD; + break; + case TEMP_CONST: + continue; + default: + g_assert_not_reached(); } la_reset_pref(&s->temps[i]); } @@ -3254,15 +3360,28 @@ static void temp_load(TCGContext *, TCGTemp *, TCGRegSet, TCGRegSet, TCGRegSet); mark it free; otherwise mark it dead. */ static void temp_free_or_dead(TCGContext *s, TCGTemp *ts, int free_or_dead) { - if (temp_readonly(ts)) { + TCGTempVal new_type; + + switch (ts->kind) { + case TEMP_FIXED: return; + case TEMP_GLOBAL: + case TEMP_LOCAL: + new_type = TEMP_VAL_MEM; + break; + case TEMP_NORMAL: + new_type = free_or_dead < 0 ? TEMP_VAL_MEM : TEMP_VAL_DEAD; + break; + case TEMP_CONST: + new_type = TEMP_VAL_CONST; + break; + default: + g_assert_not_reached(); } if (ts->val_type == TEMP_VAL_REG) { s->reg_to_temp[ts->reg] = NULL; } - ts->val_type = (free_or_dead < 0 - || ts->kind != TEMP_NORMAL - ? TEMP_VAL_MEM : TEMP_VAL_DEAD); + ts->val_type = new_type; } /* Mark a temporary as dead. */ @@ -3278,10 +3397,7 @@ static inline void temp_dead(TCGContext *s, TCGTemp *ts) static void temp_sync(TCGContext *s, TCGTemp *ts, TCGRegSet allocated_regs, TCGRegSet preferred_regs, int free_or_dead) { - if (temp_readonly(ts)) { - return; - } - if (!ts->mem_coherent) { + if (!temp_readonly(ts) && !ts->mem_coherent) { if (!ts->mem_allocated) { temp_allocate_frame(s, ts); } @@ -3499,12 +3615,22 @@ static void tcg_reg_alloc_bb_end(TCGContext *s, TCGRegSet allocated_regs) for (i = s->nb_globals; i < s->nb_temps; i++) { TCGTemp *ts = &s->temps[i]; - if (ts->kind == TEMP_LOCAL) { + + switch (ts->kind) { + case TEMP_LOCAL: temp_save(s, ts, allocated_regs); - } else { + break; + case TEMP_NORMAL: /* The liveness analysis already ensures that temps are dead. Keep an tcg_debug_assert for safety. */ tcg_debug_assert(ts->val_type == TEMP_VAL_DEAD); + break; + case TEMP_CONST: + /* Similarly, we should have freed any allocated register. */ + tcg_debug_assert(ts->val_type == TEMP_VAL_CONST); + break; + default: + g_assert_not_reached(); } } @@ -3525,10 +3651,17 @@ static void tcg_reg_alloc_cbranch(TCGContext *s, TCGRegSet allocated_regs) * The liveness analysis already ensures that temps are dead. * Keep tcg_debug_asserts for safety. */ - if (ts->kind == TEMP_LOCAL) { + switch (ts->kind) { + case TEMP_LOCAL: tcg_debug_assert(ts->val_type != TEMP_VAL_REG || ts->mem_coherent); - } else { + break; + case TEMP_NORMAL: tcg_debug_assert(ts->val_type == TEMP_VAL_DEAD); + break; + case TEMP_CONST: + break; + default: + g_assert_not_reached(); } } } @@ -3819,45 +3952,42 @@ static void tcg_reg_alloc_op(TCGContext *s, const TCGOp *op) i_preferred_regs = o_preferred_regs = 0; if (arg_ct->ialias) { o_preferred_regs = op->output_pref[arg_ct->alias_index]; - if (ts->kind == TEMP_FIXED) { - /* if fixed register, we must allocate a new register - if the alias is not the same register */ - if (arg != op->args[arg_ct->alias_index]) { - goto allocate_in_reg; - } - } else { - /* if the input is aliased to an output and if it is - not dead after the instruction, we must allocate - a new register and move it */ - if (!IS_DEAD_ARG(i)) { - goto allocate_in_reg; - } - /* check if the current register has already been allocated - for another input aliased to an output */ - if (ts->val_type == TEMP_VAL_REG) { - int k2, i2; - reg = ts->reg; - for (k2 = 0 ; k2 < k ; k2++) { - i2 = def->args_ct[nb_oargs + k2].sort_index; - if (def->args_ct[i2].ialias && reg == new_args[i2]) { - goto allocate_in_reg; - } + /* + * If the input is readonly, then it cannot also be an + * output and aliased to itself. If the input is not + * dead after the instruction, we must allocate a new + * register and move it. + */ + if (temp_readonly(ts) || !IS_DEAD_ARG(i)) { + goto allocate_in_reg; + } + + /* + * Check if the current register has already been allocated + * for another input aliased to an output. + */ + if (ts->val_type == TEMP_VAL_REG) { + reg = ts->reg; + for (int k2 = 0; k2 < k; k2++) { + int i2 = def->args_ct[nb_oargs + k2].sort_index; + if (def->args_ct[i2].ialias && reg == new_args[i2]) { + goto allocate_in_reg; } } - i_preferred_regs = o_preferred_regs; } + i_preferred_regs = o_preferred_regs; } temp_load(s, ts, arg_ct->regs, i_allocated_regs, i_preferred_regs); reg = ts->reg; - if (tcg_regset_test_reg(arg_ct->regs, reg)) { - /* nothing to do : the constraint is satisfied */ - } else { - allocate_in_reg: - /* allocate a new register matching the constraint - and move the temporary register into it */ + if (!tcg_regset_test_reg(arg_ct->regs, reg)) { + allocate_in_reg: + /* + * Allocate a new register matching the constraint + * and move the temporary register into it. + */ temp_load(s, ts, tcg_target_available_regs[ts->type], i_allocated_regs, 0); reg = tcg_reg_alloc(s, arg_ct->regs, i_allocated_regs, -- cgit v1.2.3-55-g7522 From 0e1ea43a9dc296c3ab2eab998e3e9c6c7ca488c5 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Fri, 17 Apr 2020 09:31:55 -0700 Subject: tcg: Use tcg_constant_i32 with icount expander We must do this before we adjust tcg_out_movi_i32, lest the under-the-hood poking that we do for icount be broken. Reviewed-by: Alex Bennée Signed-off-by: Richard Henderson --- include/exec/gen-icount.h | 25 +++++++++++++------------ 1 file changed, 13 insertions(+), 12 deletions(-) (limited to 'include') diff --git a/include/exec/gen-icount.h b/include/exec/gen-icount.h index aa4b44354a..298e01eef4 100644 --- a/include/exec/gen-icount.h +++ b/include/exec/gen-icount.h @@ -34,7 +34,7 @@ static inline void gen_io_end(void) static inline void gen_tb_start(const TranslationBlock *tb) { - TCGv_i32 count, imm; + TCGv_i32 count; tcg_ctx->exitreq_label = gen_new_label(); if (tb_cflags(tb) & CF_USE_ICOUNT) { @@ -48,15 +48,13 @@ static inline void gen_tb_start(const TranslationBlock *tb) offsetof(ArchCPU, env)); if (tb_cflags(tb) & CF_USE_ICOUNT) { - imm = tcg_temp_new_i32(); - /* We emit a movi with a dummy immediate argument. Keep the insn index - * of the movi so that we later (when we know the actual insn count) - * can update the immediate argument with the actual insn count. */ - tcg_gen_movi_i32(imm, 0xdeadbeef); + /* + * We emit a sub with a dummy immediate argument. Keep the insn index + * of the sub so that we later (when we know the actual insn count) + * can update the argument with the actual insn count. + */ + tcg_gen_sub_i32(count, count, tcg_constant_i32(0)); icount_start_insn = tcg_last_op(); - - tcg_gen_sub_i32(count, count, imm); - tcg_temp_free_i32(imm); } tcg_gen_brcondi_i32(TCG_COND_LT, count, 0, tcg_ctx->exitreq_label); @@ -74,9 +72,12 @@ static inline void gen_tb_start(const TranslationBlock *tb) static inline void gen_tb_end(const TranslationBlock *tb, int num_insns) { if (tb_cflags(tb) & CF_USE_ICOUNT) { - /* Update the num_insn immediate parameter now that we know - * the actual insn count. */ - tcg_set_insn_param(icount_start_insn, 1, num_insns); + /* + * Update the num_insn immediate parameter now that we know + * the actual insn count. + */ + tcg_set_insn_param(icount_start_insn, 2, + tcgv_i32_arg(tcg_constant_i32(num_insns))); } gen_set_label(tcg_ctx->exitreq_label); -- cgit v1.2.3-55-g7522 From 11d11d61bd9e82ac917c8159f6a2b736829231ae Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Sun, 29 Mar 2020 20:07:08 -0700 Subject: tcg: Use tcg_constant_{i32,i64} with tcg int expanders Signed-off-by: Richard Henderson --- include/tcg/tcg-op.h | 13 +-- tcg/tcg-op.c | 227 ++++++++++++++++++++++++--------------------------- 2 files changed, 109 insertions(+), 131 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg-op.h b/include/tcg/tcg-op.h index 901b19f32a..ed8de045e2 100644 --- a/include/tcg/tcg-op.h +++ b/include/tcg/tcg-op.h @@ -271,6 +271,7 @@ void tcg_gen_mb(TCGBar); /* 32 bit ops */ +void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg); void tcg_gen_addi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2); void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2); void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2); @@ -349,11 +350,6 @@ static inline void tcg_gen_mov_i32(TCGv_i32 ret, TCGv_i32 arg) } } -static inline void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg) -{ - tcg_gen_op2i_i32(INDEX_op_movi_i32, ret, arg); -} - static inline void tcg_gen_ld8u_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset) { @@ -467,6 +463,7 @@ static inline void tcg_gen_not_i32(TCGv_i32 ret, TCGv_i32 arg) /* 64 bit ops */ +void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg); void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2); void tcg_gen_subfi_i64(TCGv_i64 ret, int64_t arg1, TCGv_i64 arg2); void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2); @@ -550,11 +547,6 @@ static inline void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg) } } -static inline void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg) -{ - tcg_gen_op2i_i64(INDEX_op_movi_i64, ret, arg); -} - static inline void tcg_gen_ld8u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset) { @@ -698,7 +690,6 @@ static inline void tcg_gen_sub_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2) void tcg_gen_discard_i64(TCGv_i64 arg); void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg); -void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg); void tcg_gen_ld8u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset); void tcg_gen_ld8s_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset); void tcg_gen_ld16u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset); diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c index 0374b5d56d..70475773f4 100644 --- a/tcg/tcg-op.c +++ b/tcg/tcg-op.c @@ -104,15 +104,18 @@ void tcg_gen_mb(TCGBar mb_type) /* 32 bit ops */ +void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg) +{ + tcg_gen_mov_i32(ret, tcg_constant_i32(arg)); +} + void tcg_gen_addi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) { /* some cases can be optimized here */ if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_add_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_add_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -122,9 +125,7 @@ void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2) /* Don't recurse with tcg_gen_neg_i32. */ tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg2); } else { - TCGv_i32 t0 = tcg_const_i32(arg1); - tcg_gen_sub_i32(ret, t0, arg2); - tcg_temp_free_i32(t0); + tcg_gen_sub_i32(ret, tcg_constant_i32(arg1), arg2); } } @@ -134,15 +135,12 @@ void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_sub_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_sub_i32(ret, arg1, tcg_constant_i32(arg2)); } } void tcg_gen_andi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) { - TCGv_i32 t0; /* Some cases can be optimized here. */ switch (arg2) { case 0: @@ -165,9 +163,8 @@ void tcg_gen_andi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) } break; } - t0 = tcg_const_i32(arg2); - tcg_gen_and_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + + tcg_gen_and_i32(ret, arg1, tcg_constant_i32(arg2)); } void tcg_gen_ori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) @@ -178,9 +175,7 @@ void tcg_gen_ori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_or_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_or_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -193,9 +188,7 @@ void tcg_gen_xori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) /* Don't recurse with tcg_gen_not_i32. */ tcg_gen_op2_i32(INDEX_op_not_i32, ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_xor_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_xor_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -205,9 +198,7 @@ void tcg_gen_shli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_shl_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_shl_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -217,9 +208,7 @@ void tcg_gen_shri_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_shr_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_shr_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -229,9 +218,7 @@ void tcg_gen_sari_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_sar_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_sar_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -250,9 +237,7 @@ void tcg_gen_brcondi_i32(TCGCond cond, TCGv_i32 arg1, int32_t arg2, TCGLabel *l) if (cond == TCG_COND_ALWAYS) { tcg_gen_br(l); } else if (cond != TCG_COND_NEVER) { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_brcond_i32(cond, arg1, t0, l); - tcg_temp_free_i32(t0); + tcg_gen_brcond_i32(cond, arg1, tcg_constant_i32(arg2), l); } } @@ -271,9 +256,7 @@ void tcg_gen_setcond_i32(TCGCond cond, TCGv_i32 ret, void tcg_gen_setcondi_i32(TCGCond cond, TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_setcond_i32(cond, ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_setcond_i32(cond, ret, arg1, tcg_constant_i32(arg2)); } void tcg_gen_muli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) @@ -283,9 +266,7 @@ void tcg_gen_muli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) } else if (is_power_of_2(arg2)) { tcg_gen_shli_i32(ret, arg1, ctz32(arg2)); } else { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_mul_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_mul_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -433,9 +414,7 @@ void tcg_gen_clz_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2) void tcg_gen_clzi_i32(TCGv_i32 ret, TCGv_i32 arg1, uint32_t arg2) { - TCGv_i32 t = tcg_const_i32(arg2); - tcg_gen_clz_i32(ret, arg1, t); - tcg_temp_free_i32(t); + tcg_gen_clz_i32(ret, arg1, tcg_constant_i32(arg2)); } void tcg_gen_ctz_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2) @@ -468,10 +447,9 @@ void tcg_gen_ctz_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2) tcg_gen_clzi_i32(t, t, 32); tcg_gen_xori_i32(t, t, 31); } - z = tcg_const_i32(0); + z = tcg_constant_i32(0); tcg_gen_movcond_i32(TCG_COND_EQ, ret, arg1, z, arg2, t); tcg_temp_free_i32(t); - tcg_temp_free_i32(z); } else { gen_helper_ctz_i32(ret, arg1, arg2); } @@ -487,9 +465,7 @@ void tcg_gen_ctzi_i32(TCGv_i32 ret, TCGv_i32 arg1, uint32_t arg2) tcg_gen_ctpop_i32(ret, t); tcg_temp_free_i32(t); } else { - TCGv_i32 t = tcg_const_i32(arg2); - tcg_gen_ctz_i32(ret, arg1, t); - tcg_temp_free_i32(t); + tcg_gen_ctz_i32(ret, arg1, tcg_constant_i32(arg2)); } } @@ -547,9 +523,7 @@ void tcg_gen_rotli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2) if (arg2 == 0) { tcg_gen_mov_i32(ret, arg1); } else if (TCG_TARGET_HAS_rot_i32) { - TCGv_i32 t0 = tcg_const_i32(arg2); - tcg_gen_rotl_i32(ret, arg1, t0); - tcg_temp_free_i32(t0); + tcg_gen_rotl_i32(ret, arg1, tcg_constant_i32(arg2)); } else { TCGv_i32 t0, t1; t0 = tcg_temp_new_i32(); @@ -653,9 +627,8 @@ void tcg_gen_deposit_z_i32(TCGv_i32 ret, TCGv_i32 arg, tcg_gen_andi_i32(ret, arg, (1u << len) - 1); } else if (TCG_TARGET_HAS_deposit_i32 && TCG_TARGET_deposit_i32_valid(ofs, len)) { - TCGv_i32 zero = tcg_const_i32(0); + TCGv_i32 zero = tcg_constant_i32(0); tcg_gen_op5ii_i32(INDEX_op_deposit_i32, ret, zero, arg, ofs, len); - tcg_temp_free_i32(zero); } else { /* To help two-operand hosts we prefer to zero-extend first, which allows ARG to stay live. */ @@ -1052,7 +1025,7 @@ void tcg_gen_bswap32_i32(TCGv_i32 ret, TCGv_i32 arg) } else { TCGv_i32 t0 = tcg_temp_new_i32(); TCGv_i32 t1 = tcg_temp_new_i32(); - TCGv_i32 t2 = tcg_const_i32(0x00ff00ff); + TCGv_i32 t2 = tcg_constant_i32(0x00ff00ff); /* arg = abcd */ tcg_gen_shri_i32(t0, arg, 8); /* t0 = .abc */ @@ -1067,7 +1040,6 @@ void tcg_gen_bswap32_i32(TCGv_i32 ret, TCGv_i32 arg) tcg_temp_free_i32(t0); tcg_temp_free_i32(t1); - tcg_temp_free_i32(t2); } } @@ -1114,8 +1086,15 @@ void tcg_gen_discard_i64(TCGv_i64 arg) void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg) { - tcg_gen_mov_i32(TCGV_LOW(ret), TCGV_LOW(arg)); - tcg_gen_mov_i32(TCGV_HIGH(ret), TCGV_HIGH(arg)); + TCGTemp *ts = tcgv_i64_temp(arg); + + /* Canonicalize TCGv_i64 TEMP_CONST into TCGv_i32 TEMP_CONST. */ + if (ts->kind == TEMP_CONST) { + tcg_gen_movi_i64(ret, ts->val); + } else { + tcg_gen_mov_i32(TCGV_LOW(ret), TCGV_LOW(arg)); + tcg_gen_mov_i32(TCGV_HIGH(ret), TCGV_HIGH(arg)); + } } void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg) @@ -1237,6 +1216,14 @@ void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2) tcg_temp_free_i64(t0); tcg_temp_free_i32(t1); } + +#else + +void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg) +{ + tcg_gen_mov_i64(ret, tcg_constant_i64(arg)); +} + #endif /* TCG_TARGET_REG_SIZE == 32 */ void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) @@ -1244,10 +1231,12 @@ void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) /* some cases can be optimized here */ if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); + } else if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_add_i64(ret, arg1, tcg_constant_i64(arg2)); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_add_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_add2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), tcg_constant_i32(arg2 >> 32)); } } @@ -1256,10 +1245,12 @@ void tcg_gen_subfi_i64(TCGv_i64 ret, int64_t arg1, TCGv_i64 arg2) if (arg1 == 0 && TCG_TARGET_HAS_neg_i64) { /* Don't recurse with tcg_gen_neg_i64. */ tcg_gen_op2_i64(INDEX_op_neg_i64, ret, arg2); + } else if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_sub_i64(ret, tcg_constant_i64(arg1), arg2); } else { - TCGv_i64 t0 = tcg_const_i64(arg1); - tcg_gen_sub_i64(ret, t0, arg2); - tcg_temp_free_i64(t0); + tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), + tcg_constant_i32(arg1), tcg_constant_i32(arg1 >> 32), + TCGV_LOW(arg2), TCGV_HIGH(arg2)); } } @@ -1268,17 +1259,17 @@ void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) /* some cases can be optimized here */ if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); + } else if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_sub_i64(ret, arg1, tcg_constant_i64(arg2)); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_sub_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), tcg_constant_i32(arg2 >> 32)); } } void tcg_gen_andi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) { - TCGv_i64 t0; - if (TCG_TARGET_REG_BITS == 32) { tcg_gen_andi_i32(TCGV_LOW(ret), TCGV_LOW(arg1), arg2); tcg_gen_andi_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), arg2 >> 32); @@ -1313,9 +1304,8 @@ void tcg_gen_andi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } break; } - t0 = tcg_const_i64(arg2); - tcg_gen_and_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + + tcg_gen_and_i64(ret, arg1, tcg_constant_i64(arg2)); } void tcg_gen_ori_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) @@ -1331,9 +1321,7 @@ void tcg_gen_ori_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_or_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_or_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -1351,9 +1339,7 @@ void tcg_gen_xori_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) /* Don't recurse with tcg_gen_not_i64. */ tcg_gen_op2_i64(INDEX_op_not_i64, ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_xor_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_xor_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -1415,9 +1401,7 @@ void tcg_gen_shli_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_shl_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_shl_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -1429,9 +1413,7 @@ void tcg_gen_shri_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_shr_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_shr_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -1443,9 +1425,7 @@ void tcg_gen_sari_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) } else if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_sar_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_sar_i64(ret, arg1, tcg_constant_i64(arg2)); } } @@ -1468,12 +1448,17 @@ void tcg_gen_brcond_i64(TCGCond cond, TCGv_i64 arg1, TCGv_i64 arg2, TCGLabel *l) void tcg_gen_brcondi_i64(TCGCond cond, TCGv_i64 arg1, int64_t arg2, TCGLabel *l) { - if (cond == TCG_COND_ALWAYS) { + if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_brcond_i64(cond, arg1, tcg_constant_i64(arg2), l); + } else if (cond == TCG_COND_ALWAYS) { tcg_gen_br(l); } else if (cond != TCG_COND_NEVER) { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_brcond_i64(cond, arg1, t0, l); - tcg_temp_free_i64(t0); + l->refs++; + tcg_gen_op6ii_i32(INDEX_op_brcond2_i32, + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), + tcg_constant_i32(arg2 >> 32), + cond, label_arg(l)); } } @@ -1499,9 +1484,19 @@ void tcg_gen_setcond_i64(TCGCond cond, TCGv_i64 ret, void tcg_gen_setcondi_i64(TCGCond cond, TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_setcond_i64(cond, ret, arg1, t0); - tcg_temp_free_i64(t0); + if (TCG_TARGET_REG_BITS == 64) { + tcg_gen_setcond_i64(cond, ret, arg1, tcg_constant_i64(arg2)); + } else if (cond == TCG_COND_ALWAYS) { + tcg_gen_movi_i64(ret, 1); + } else if (cond == TCG_COND_NEVER) { + tcg_gen_movi_i64(ret, 0); + } else { + tcg_gen_op6i_i32(INDEX_op_setcond2_i32, TCGV_LOW(ret), + TCGV_LOW(arg1), TCGV_HIGH(arg1), + tcg_constant_i32(arg2), + tcg_constant_i32(arg2 >> 32), cond); + tcg_gen_movi_i32(TCGV_HIGH(ret), 0); + } } void tcg_gen_muli_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) @@ -1690,7 +1685,7 @@ void tcg_gen_bswap32_i64(TCGv_i64 ret, TCGv_i64 arg) } else { TCGv_i64 t0 = tcg_temp_new_i64(); TCGv_i64 t1 = tcg_temp_new_i64(); - TCGv_i64 t2 = tcg_const_i64(0x00ff00ff); + TCGv_i64 t2 = tcg_constant_i64(0x00ff00ff); /* arg = ....abcd */ tcg_gen_shri_i64(t0, arg, 8); /* t0 = .....abc */ @@ -1706,7 +1701,6 @@ void tcg_gen_bswap32_i64(TCGv_i64 ret, TCGv_i64 arg) tcg_temp_free_i64(t0); tcg_temp_free_i64(t1); - tcg_temp_free_i64(t2); } } @@ -1850,16 +1844,16 @@ void tcg_gen_clzi_i64(TCGv_i64 ret, TCGv_i64 arg1, uint64_t arg2) if (TCG_TARGET_REG_BITS == 32 && TCG_TARGET_HAS_clz_i32 && arg2 <= 0xffffffffu) { - TCGv_i32 t = tcg_const_i32((uint32_t)arg2 - 32); - tcg_gen_clz_i32(t, TCGV_LOW(arg1), t); + TCGv_i32 t = tcg_temp_new_i32(); + tcg_gen_clzi_i32(t, TCGV_LOW(arg1), arg2 - 32); tcg_gen_addi_i32(t, t, 32); tcg_gen_clz_i32(TCGV_LOW(ret), TCGV_HIGH(arg1), t); tcg_gen_movi_i32(TCGV_HIGH(ret), 0); tcg_temp_free_i32(t); } else { - TCGv_i64 t = tcg_const_i64(arg2); - tcg_gen_clz_i64(ret, arg1, t); - tcg_temp_free_i64(t); + TCGv_i64 t0 = tcg_const_i64(arg2); + tcg_gen_clz_i64(ret, arg1, t0); + tcg_temp_free_i64(t0); } } @@ -1881,7 +1875,7 @@ void tcg_gen_ctz_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2) tcg_gen_clzi_i64(t, t, 64); tcg_gen_xori_i64(t, t, 63); } - z = tcg_const_i64(0); + z = tcg_constant_i64(0); tcg_gen_movcond_i64(TCG_COND_EQ, ret, arg1, z, arg2, t); tcg_temp_free_i64(t); tcg_temp_free_i64(z); @@ -1895,8 +1889,8 @@ void tcg_gen_ctzi_i64(TCGv_i64 ret, TCGv_i64 arg1, uint64_t arg2) if (TCG_TARGET_REG_BITS == 32 && TCG_TARGET_HAS_ctz_i32 && arg2 <= 0xffffffffu) { - TCGv_i32 t32 = tcg_const_i32((uint32_t)arg2 - 32); - tcg_gen_ctz_i32(t32, TCGV_HIGH(arg1), t32); + TCGv_i32 t32 = tcg_temp_new_i32(); + tcg_gen_ctzi_i32(t32, TCGV_HIGH(arg1), arg2 - 32); tcg_gen_addi_i32(t32, t32, 32); tcg_gen_ctz_i32(TCGV_LOW(ret), TCGV_LOW(arg1), t32); tcg_gen_movi_i32(TCGV_HIGH(ret), 0); @@ -1911,9 +1905,9 @@ void tcg_gen_ctzi_i64(TCGv_i64 ret, TCGv_i64 arg1, uint64_t arg2) tcg_gen_ctpop_i64(ret, t); tcg_temp_free_i64(t); } else { - TCGv_i64 t64 = tcg_const_i64(arg2); - tcg_gen_ctz_i64(ret, arg1, t64); - tcg_temp_free_i64(t64); + TCGv_i64 t0 = tcg_const_i64(arg2); + tcg_gen_ctz_i64(ret, arg1, t0); + tcg_temp_free_i64(t0); } } @@ -1969,9 +1963,7 @@ void tcg_gen_rotli_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2) if (arg2 == 0) { tcg_gen_mov_i64(ret, arg1); } else if (TCG_TARGET_HAS_rot_i64) { - TCGv_i64 t0 = tcg_const_i64(arg2); - tcg_gen_rotl_i64(ret, arg1, t0); - tcg_temp_free_i64(t0); + tcg_gen_rotl_i64(ret, arg1, tcg_constant_i64(arg2)); } else { TCGv_i64 t0, t1; t0 = tcg_temp_new_i64(); @@ -2089,9 +2081,8 @@ void tcg_gen_deposit_z_i64(TCGv_i64 ret, TCGv_i64 arg, tcg_gen_andi_i64(ret, arg, (1ull << len) - 1); } else if (TCG_TARGET_HAS_deposit_i64 && TCG_TARGET_deposit_i64_valid(ofs, len)) { - TCGv_i64 zero = tcg_const_i64(0); + TCGv_i64 zero = tcg_constant_i64(0); tcg_gen_op5ii_i64(INDEX_op_deposit_i64, ret, zero, arg, ofs, len); - tcg_temp_free_i64(zero); } else { if (TCG_TARGET_REG_BITS == 32) { if (ofs >= 32) { @@ -3117,9 +3108,8 @@ void tcg_gen_atomic_cmpxchg_i32(TCGv_i32 retv, TCGv addr, TCGv_i32 cmpv, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop & ~MO_SIGN, idx)); - gen(retv, cpu_env, addr, cmpv, newv, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); } #else gen(retv, cpu_env, addr, cmpv, newv); @@ -3162,9 +3152,8 @@ void tcg_gen_atomic_cmpxchg_i64(TCGv_i64 retv, TCGv addr, TCGv_i64 cmpv, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop, idx)); - gen(retv, cpu_env, addr, cmpv, newv, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop, idx); + gen(retv, cpu_env, addr, cmpv, newv, tcg_constant_i32(oi)); } #else gen(retv, cpu_env, addr, cmpv, newv); @@ -3226,9 +3215,8 @@ static void do_atomic_op_i32(TCGv_i32 ret, TCGv addr, TCGv_i32 val, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop & ~MO_SIGN, idx)); - gen(ret, cpu_env, addr, val, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); } #else gen(ret, cpu_env, addr, val); @@ -3272,9 +3260,8 @@ static void do_atomic_op_i64(TCGv_i64 ret, TCGv addr, TCGv_i64 val, #ifdef CONFIG_SOFTMMU { - TCGv_i32 oi = tcg_const_i32(make_memop_idx(memop & ~MO_SIGN, idx)); - gen(ret, cpu_env, addr, val, oi); - tcg_temp_free_i32(oi); + TCGMemOpIdx oi = make_memop_idx(memop & ~MO_SIGN, idx); + gen(ret, cpu_env, addr, val, tcg_constant_i32(oi)); } #else gen(ret, cpu_env, addr, val); -- cgit v1.2.3-55-g7522 From 88d4005b098427638d7551aa04ebde4fdd06835b Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Thu, 3 Sep 2020 18:18:08 -0700 Subject: tcg: Use tcg_constant_{i32,i64,vec} with gvec expanders Signed-off-by: Richard Henderson --- include/tcg/tcg.h | 1 + tcg/tcg-op-gvec.c | 127 +++++++++++++++++++++--------------------------------- tcg/tcg.c | 8 ++++ 3 files changed, 59 insertions(+), 77 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h index eeeb70ad43..504c5e9bb0 100644 --- a/include/tcg/tcg.h +++ b/include/tcg/tcg.h @@ -1109,6 +1109,7 @@ static inline TCGv_i64 tcg_constant_i64(int64_t val) } TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val); +TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val); #if UINTPTR_MAX == UINT32_MAX # define tcg_const_ptr(x) ((TCGv_ptr)tcg_const_i32((intptr_t)(x))) diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c index 1a41dfa908..498a959839 100644 --- a/tcg/tcg-op-gvec.c +++ b/tcg/tcg-op-gvec.c @@ -115,7 +115,7 @@ void tcg_gen_gvec_2_ool(uint32_t dofs, uint32_t aofs, gen_helper_gvec_2 *fn) { TCGv_ptr a0, a1; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -127,7 +127,6 @@ void tcg_gen_gvec_2_ool(uint32_t dofs, uint32_t aofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with two vector operands @@ -137,7 +136,7 @@ void tcg_gen_gvec_2i_ool(uint32_t dofs, uint32_t aofs, TCGv_i64 c, gen_helper_gvec_2i *fn) { TCGv_ptr a0, a1; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -149,7 +148,6 @@ void tcg_gen_gvec_2i_ool(uint32_t dofs, uint32_t aofs, TCGv_i64 c, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with three vector operands. */ @@ -158,7 +156,7 @@ void tcg_gen_gvec_3_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, gen_helper_gvec_3 *fn) { TCGv_ptr a0, a1, a2; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -173,7 +171,6 @@ void tcg_gen_gvec_3_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with four vector operands. */ @@ -182,7 +179,7 @@ void tcg_gen_gvec_4_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, int32_t data, gen_helper_gvec_4 *fn) { TCGv_ptr a0, a1, a2, a3; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -200,7 +197,6 @@ void tcg_gen_gvec_4_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with five vector operands. */ @@ -209,7 +205,7 @@ void tcg_gen_gvec_5_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, uint32_t maxsz, int32_t data, gen_helper_gvec_5 *fn) { TCGv_ptr a0, a1, a2, a3, a4; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -230,7 +226,6 @@ void tcg_gen_gvec_5_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); tcg_temp_free_ptr(a4); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with three vector operands @@ -240,7 +235,7 @@ void tcg_gen_gvec_2_ptr(uint32_t dofs, uint32_t aofs, int32_t data, gen_helper_gvec_2_ptr *fn) { TCGv_ptr a0, a1; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -252,7 +247,6 @@ void tcg_gen_gvec_2_ptr(uint32_t dofs, uint32_t aofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with three vector operands @@ -262,7 +256,7 @@ void tcg_gen_gvec_3_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, int32_t data, gen_helper_gvec_3_ptr *fn) { TCGv_ptr a0, a1, a2; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -277,7 +271,6 @@ void tcg_gen_gvec_3_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a0); tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with four vector operands @@ -288,7 +281,7 @@ void tcg_gen_gvec_4_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, gen_helper_gvec_4_ptr *fn) { TCGv_ptr a0, a1, a2, a3; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -306,7 +299,6 @@ void tcg_gen_gvec_4_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a1); tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); - tcg_temp_free_i32(desc); } /* Generate a call to a gvec-style helper with five vector operands @@ -317,7 +309,7 @@ void tcg_gen_gvec_5_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, gen_helper_gvec_5_ptr *fn) { TCGv_ptr a0, a1, a2, a3, a4; - TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data)); + TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data)); a0 = tcg_temp_new_ptr(); a1 = tcg_temp_new_ptr(); @@ -338,7 +330,6 @@ void tcg_gen_gvec_5_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs, tcg_temp_free_ptr(a2); tcg_temp_free_ptr(a3); tcg_temp_free_ptr(a4); - tcg_temp_free_i32(desc); } /* Return true if we want to implement something of OPRSZ bytes @@ -605,9 +596,9 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, || (TCG_TARGET_REG_BITS == 64 && (in_c == 0 || in_c == -1 || !check_size_impl(oprsz, 4)))) { - t_64 = tcg_const_i64(in_c); + t_64 = tcg_constant_i64(in_c); } else { - t_32 = tcg_const_i32(in_c); + t_32 = tcg_constant_i32(in_c); } } @@ -648,11 +639,11 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, t_val = tcg_temp_new_i32(); tcg_gen_extrl_i64_i32(t_val, in_64); } else { - t_val = tcg_const_i32(in_c); + t_val = tcg_constant_i32(in_c); } gen_helper_memset(t_ptr, t_ptr, t_val, t_size); - if (!in_32) { + if (in_64) { tcg_temp_free_i32(t_val); } tcg_temp_free_ptr(t_size); @@ -660,15 +651,14 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, return; } - t_desc = tcg_const_i32(simd_desc(oprsz, maxsz, 0)); + t_desc = tcg_constant_i32(simd_desc(oprsz, maxsz, 0)); if (vece == MO_64) { if (in_64) { gen_helper_gvec_dup64(t_ptr, t_desc, in_64); } else { - t_64 = tcg_const_i64(in_c); + t_64 = tcg_constant_i64(in_c); gen_helper_gvec_dup64(t_ptr, t_desc, t_64); - tcg_temp_free_i64(t_64); } } else { typedef void dup_fn(TCGv_ptr, TCGv_i32, TCGv_i32); @@ -680,24 +670,23 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz, if (in_32) { fns[vece](t_ptr, t_desc, in_32); - } else { + } else if (in_64) { t_32 = tcg_temp_new_i32(); - if (in_64) { - tcg_gen_extrl_i64_i32(t_32, in_64); - } else if (vece == MO_8) { - tcg_gen_movi_i32(t_32, in_c & 0xff); + tcg_gen_extrl_i64_i32(t_32, in_64); + fns[vece](t_ptr, t_desc, t_32); + tcg_temp_free_i32(t_32); + } else { + if (vece == MO_8) { + in_c &= 0xff; } else if (vece == MO_16) { - tcg_gen_movi_i32(t_32, in_c & 0xffff); - } else { - tcg_gen_movi_i32(t_32, in_c); + in_c &= 0xffff; } + t_32 = tcg_constant_i32(in_c); fns[vece](t_ptr, t_desc, t_32); - tcg_temp_free_i32(t_32); } } tcg_temp_free_ptr(t_ptr); - tcg_temp_free_i32(t_desc); return; done: @@ -1247,10 +1236,9 @@ void tcg_gen_gvec_2i(uint32_t dofs, uint32_t aofs, uint32_t oprsz, if (g->fno) { tcg_gen_gvec_2_ool(dofs, aofs, oprsz, maxsz, c, g->fno); } else { - TCGv_i64 tcg_c = tcg_const_i64(c); + TCGv_i64 tcg_c = tcg_constant_i64(c); tcg_gen_gvec_2i_ool(dofs, aofs, tcg_c, oprsz, maxsz, c, g->fnoi); - tcg_temp_free_i64(tcg_c); } oprsz = maxsz; } @@ -1744,16 +1732,14 @@ static void gen_addv_mask(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 m) void tcg_gen_vec_add8_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80)); gen_addv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_add16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000)); gen_addv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_add32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) @@ -1837,9 +1823,8 @@ void tcg_gen_gvec_adds(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_addi(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(c); + TCGv_i64 tmp = tcg_constant_i64(c); tcg_gen_gvec_adds(vece, dofs, aofs, tmp, oprsz, maxsz); - tcg_temp_free_i64(tmp); } static const TCGOpcode vecop_list_sub[] = { INDEX_op_sub_vec, 0 }; @@ -1897,16 +1882,14 @@ static void gen_subv_mask(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 m) void tcg_gen_vec_sub8_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80)); gen_subv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_sub16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000)); gen_subv_mask(d, a, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_sub32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) @@ -2017,9 +2000,8 @@ void tcg_gen_gvec_muls(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_muli(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(c); + TCGv_i64 tmp = tcg_constant_i64(c); tcg_gen_gvec_muls(vece, dofs, aofs, tmp, oprsz, maxsz); - tcg_temp_free_i64(tmp); } void tcg_gen_gvec_ssadd(unsigned vece, uint32_t dofs, uint32_t aofs, @@ -2076,18 +2058,16 @@ void tcg_gen_gvec_sssub(unsigned vece, uint32_t dofs, uint32_t aofs, static void tcg_gen_usadd_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b) { - TCGv_i32 max = tcg_const_i32(-1); + TCGv_i32 max = tcg_constant_i32(-1); tcg_gen_add_i32(d, a, b); tcg_gen_movcond_i32(TCG_COND_LTU, d, d, a, max, d); - tcg_temp_free_i32(max); } static void tcg_gen_usadd_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 max = tcg_const_i64(-1); + TCGv_i64 max = tcg_constant_i64(-1); tcg_gen_add_i64(d, a, b); tcg_gen_movcond_i64(TCG_COND_LTU, d, d, a, max, d); - tcg_temp_free_i64(max); } void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs, @@ -2120,18 +2100,16 @@ void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs, static void tcg_gen_ussub_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b) { - TCGv_i32 min = tcg_const_i32(0); + TCGv_i32 min = tcg_constant_i32(0); tcg_gen_sub_i32(d, a, b); tcg_gen_movcond_i32(TCG_COND_LTU, d, a, b, min, d); - tcg_temp_free_i32(min); } static void tcg_gen_ussub_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b) { - TCGv_i64 min = tcg_const_i64(0); + TCGv_i64 min = tcg_constant_i64(0); tcg_gen_sub_i64(d, a, b); tcg_gen_movcond_i64(TCG_COND_LTU, d, a, b, min, d); - tcg_temp_free_i64(min); } void tcg_gen_gvec_ussub(unsigned vece, uint32_t dofs, uint32_t aofs, @@ -2292,16 +2270,14 @@ static void gen_negv_mask(TCGv_i64 d, TCGv_i64 b, TCGv_i64 m) void tcg_gen_vec_neg8_i64(TCGv_i64 d, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80)); gen_negv_mask(d, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_neg16_i64(TCGv_i64 d, TCGv_i64 b) { - TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000)); + TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000)); gen_negv_mask(d, b, m); - tcg_temp_free_i64(m); } void tcg_gen_vec_neg32_i64(TCGv_i64 d, TCGv_i64 b) @@ -2570,9 +2546,8 @@ void tcg_gen_gvec_ands(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_andi(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c)); + TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c)); tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_ands); - tcg_temp_free_i64(tmp); } static const GVecGen2s gop_xors = { @@ -2595,9 +2570,8 @@ void tcg_gen_gvec_xors(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_xori(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c)); + TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c)); tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_xors); - tcg_temp_free_i64(tmp); } static const GVecGen2s gop_ors = { @@ -2620,9 +2594,8 @@ void tcg_gen_gvec_ors(unsigned vece, uint32_t dofs, uint32_t aofs, void tcg_gen_gvec_ori(unsigned vece, uint32_t dofs, uint32_t aofs, int64_t c, uint32_t oprsz, uint32_t maxsz) { - TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c)); + TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c)); tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_ors); - tcg_temp_free_i64(tmp); } void tcg_gen_vec_shl8i_i64(TCGv_i64 d, TCGv_i64 a, int64_t c) @@ -3110,9 +3083,9 @@ static void tcg_gen_shlv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_shlv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -3173,9 +3146,9 @@ static void tcg_gen_shrv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_shrv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -3236,9 +3209,9 @@ static void tcg_gen_sarv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_sarv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -3299,9 +3272,9 @@ static void tcg_gen_rotlv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_rotlv_vec(vece, d, a, t); tcg_temp_free_vec(t); } @@ -3358,9 +3331,9 @@ static void tcg_gen_rotrv_mod_vec(unsigned vece, TCGv_vec d, TCGv_vec a, TCGv_vec b) { TCGv_vec t = tcg_temp_new_vec_matching(d); + TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1); - tcg_gen_dupi_vec(vece, t, (8 << vece) - 1); - tcg_gen_and_vec(vece, t, t, b); + tcg_gen_and_vec(vece, t, b, m); tcg_gen_rotrv_vec(vece, d, a, t); tcg_temp_free_vec(t); } diff --git a/tcg/tcg.c b/tcg/tcg.c index ad1348d811..1cadfe070c 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -1474,6 +1474,14 @@ TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val) return temp_tcgv_vec(tcg_constant_internal(type, val)); } +TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val) +{ + TCGTemp *t = tcgv_vec_temp(match); + + tcg_debug_assert(t->temp_allocated != 0); + return tcg_constant_vec(t->base_type, vece, val); +} + TCGv_i32 tcg_const_i32(int32_t val) { TCGv_i32 t0; -- cgit v1.2.3-55-g7522 From 1bd1af98d7b166ced72e2fb8126b484c86d5357b Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Fri, 17 Apr 2020 13:19:47 -0700 Subject: tcg/tci: Add special tci_movi_{i32,i64} opcodes The normal movi opcodes are going away. We need something for TCI to use internally. Reviewed-by: Alex Bennée Signed-off-by: Richard Henderson --- include/tcg/tcg-opc.h | 8 ++++++++ tcg/tci.c | 4 ++-- tcg/tci/tcg-target.c.inc | 4 ++-- 3 files changed, 12 insertions(+), 4 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h index 70a76646c4..43711634f6 100644 --- a/include/tcg/tcg-opc.h +++ b/include/tcg/tcg-opc.h @@ -278,6 +278,14 @@ DEF(last_generic, 0, 0, 0, TCG_OPF_NOT_PRESENT) #include "tcg-target.opc.h" #endif +#ifdef TCG_TARGET_INTERPRETER +/* These opcodes are only for use between the tci generator and interpreter. */ +DEF(tci_movi_i32, 1, 0, 1, TCG_OPF_NOT_PRESENT) +#if TCG_TARGET_REG_BITS == 64 +DEF(tci_movi_i64, 1, 0, 1, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) +#endif +#endif + #undef TLADDR_ARGS #undef DATA64_ARGS #undef IMPL diff --git a/tcg/tci.c b/tcg/tci.c index 2bcc4409be..2311aa7d3a 100644 --- a/tcg/tci.c +++ b/tcg/tci.c @@ -593,7 +593,7 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, t1 = tci_read_r32(regs, &tb_ptr); tci_write_reg32(regs, t0, t1); break; - case INDEX_op_movi_i32: + case INDEX_op_tci_movi_i32: t0 = *tb_ptr++; t1 = tci_read_i32(&tb_ptr); tci_write_reg32(regs, t0, t1); @@ -864,7 +864,7 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env, t1 = tci_read_r64(regs, &tb_ptr); tci_write_reg64(regs, t0, t1); break; - case INDEX_op_movi_i64: + case INDEX_op_tci_movi_i64: t0 = *tb_ptr++; t1 = tci_read_i64(&tb_ptr); tci_write_reg64(regs, t0, t1); diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc index d5a4d9d37c..e9e5f94c04 100644 --- a/tcg/tci/tcg-target.c.inc +++ b/tcg/tci/tcg-target.c.inc @@ -529,13 +529,13 @@ static void tcg_out_movi(TCGContext *s, TCGType type, uint8_t *old_code_ptr = s->code_ptr; uint32_t arg32 = arg; if (type == TCG_TYPE_I32 || arg == arg32) { - tcg_out_op_t(s, INDEX_op_movi_i32); + tcg_out_op_t(s, INDEX_op_tci_movi_i32); tcg_out_r(s, t0); tcg_out32(s, arg32); } else { tcg_debug_assert(type == TCG_TYPE_I64); #if TCG_TARGET_REG_BITS == 64 - tcg_out_op_t(s, INDEX_op_movi_i64); + tcg_out_op_t(s, INDEX_op_tci_movi_i64); tcg_out_r(s, t0); tcg_out64(s, arg); #else -- cgit v1.2.3-55-g7522 From c58f4c97b2ad9247c5ee85d625a934370862fba1 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Fri, 17 Apr 2020 13:22:43 -0700 Subject: tcg: Remove movi and dupi opcodes These are now completely covered by mov from a TYPE_CONST temporary. Reviewed-by: Alex Bennée Reviewed-by: Aleksandar Markovic Signed-off-by: Richard Henderson --- include/tcg/tcg-opc.h | 3 --- tcg/aarch64/tcg-target.c.inc | 3 --- tcg/arm/tcg-target.c.inc | 1 - tcg/i386/tcg-target.c.inc | 3 --- tcg/mips/tcg-target.c.inc | 2 -- tcg/optimize.c | 4 ---- tcg/ppc/tcg-target.c.inc | 3 --- tcg/riscv/tcg-target.c.inc | 2 -- tcg/s390/tcg-target.c.inc | 2 -- tcg/sparc/tcg-target.c.inc | 2 -- tcg/tcg-op-vec.c | 1 - tcg/tcg.c | 18 +----------------- tcg/tci/tcg-target.c.inc | 2 -- 13 files changed, 1 insertion(+), 45 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h index 43711634f6..900984c005 100644 --- a/include/tcg/tcg-opc.h +++ b/include/tcg/tcg-opc.h @@ -45,7 +45,6 @@ DEF(br, 0, 0, 1, TCG_OPF_BB_END) DEF(mb, 0, 0, 1, 0) DEF(mov_i32, 1, 1, 0, TCG_OPF_NOT_PRESENT) -DEF(movi_i32, 1, 0, 1, TCG_OPF_NOT_PRESENT) DEF(setcond_i32, 1, 2, 1, 0) DEF(movcond_i32, 1, 4, 1, IMPL(TCG_TARGET_HAS_movcond_i32)) /* load/store */ @@ -111,7 +110,6 @@ DEF(ctz_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_ctz_i32)) DEF(ctpop_i32, 1, 1, 0, IMPL(TCG_TARGET_HAS_ctpop_i32)) DEF(mov_i64, 1, 1, 0, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) -DEF(movi_i64, 1, 0, 1, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT) DEF(setcond_i64, 1, 2, 1, IMPL64) DEF(movcond_i64, 1, 4, 1, IMPL64 | IMPL(TCG_TARGET_HAS_movcond_i64)) /* load/store */ @@ -221,7 +219,6 @@ DEF(qemu_st8_i32, 0, TLADDR_ARGS + 1, 1, #define IMPLVEC TCG_OPF_VECTOR | IMPL(TCG_TARGET_MAYBE_vec) DEF(mov_vec, 1, 1, 0, TCG_OPF_VECTOR | TCG_OPF_NOT_PRESENT) -DEF(dupi_vec, 1, 0, 1, TCG_OPF_VECTOR | TCG_OPF_NOT_PRESENT) DEF(dup_vec, 1, 1, 0, IMPLVEC) DEF(dup2_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_REG_BITS == 32)) diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc index be6d3ea2a8..e370b7e61c 100644 --- a/tcg/aarch64/tcg-target.c.inc +++ b/tcg/aarch64/tcg-target.c.inc @@ -2257,8 +2257,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: g_assert_not_reached(); @@ -2466,7 +2464,6 @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc index 0fd1126454..c2b26b3c45 100644 --- a/tcg/arm/tcg-target.c.inc +++ b/tcg/arm/tcg-target.c.inc @@ -2068,7 +2068,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc index 9f81e11773..1706b7c776 100644 --- a/tcg/i386/tcg-target.c.inc +++ b/tcg/i386/tcg-target.c.inc @@ -2666,8 +2666,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); @@ -2953,7 +2951,6 @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc index add157f6c3..7293169ab2 100644 --- a/tcg/mips/tcg-target.c.inc +++ b/tcg/mips/tcg-target.c.inc @@ -2141,8 +2141,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); diff --git a/tcg/optimize.c b/tcg/optimize.c index dbb03ef96b..37c902283e 100644 --- a/tcg/optimize.c +++ b/tcg/optimize.c @@ -1103,10 +1103,6 @@ void tcg_optimize(TCGContext *s) CASE_OP_32_64_VEC(mov): tcg_opt_gen_mov(s, op, op->args[0], op->args[1]); break; - CASE_OP_32_64(movi): - case INDEX_op_dupi_vec: - tcg_opt_gen_movi(s, &temps_used, op, op->args[0], op->args[1]); - break; case INDEX_op_dup_vec: if (arg_is_const(op->args[1])) { diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc index d00ec20203..1fbb1b6db0 100644 --- a/tcg/ppc/tcg-target.c.inc +++ b/tcg/ppc/tcg-target.c.inc @@ -2977,8 +2977,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); @@ -3326,7 +3324,6 @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, return; case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ - case INDEX_op_dupi_vec: /* Always emitted via tcg_out_movi. */ case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ default: g_assert_not_reached(); diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc index c60b91ba58..71c0badc02 100644 --- a/tcg/riscv/tcg-target.c.inc +++ b/tcg/riscv/tcg-target.c.inc @@ -1563,8 +1563,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: g_assert_not_reached(); diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc index d7ef079055..8517e55232 100644 --- a/tcg/s390/tcg-target.c.inc +++ b/tcg/s390/tcg-target.c.inc @@ -2295,8 +2295,6 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc index 922ae96481..28b5b6559a 100644 --- a/tcg/sparc/tcg-target.c.inc +++ b/tcg/sparc/tcg-target.c.inc @@ -1586,8 +1586,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c index 9fbed1366c..ce0d2f6e0e 100644 --- a/tcg/tcg-op-vec.c +++ b/tcg/tcg-op-vec.c @@ -83,7 +83,6 @@ bool tcg_can_emit_vecop_list(const TCGOpcode *list, case INDEX_op_xor_vec: case INDEX_op_mov_vec: case INDEX_op_dup_vec: - case INDEX_op_dupi_vec: case INDEX_op_dup2_vec: case INDEX_op_ld_vec: case INDEX_op_st_vec: diff --git a/tcg/tcg.c b/tcg/tcg.c index 1cadfe070c..5b0e42be91 100644 --- a/tcg/tcg.c +++ b/tcg/tcg.c @@ -1564,7 +1564,6 @@ bool tcg_op_supported(TCGOpcode op) return TCG_TARGET_HAS_goto_ptr; case INDEX_op_mov_i32: - case INDEX_op_movi_i32: case INDEX_op_setcond_i32: case INDEX_op_brcond_i32: case INDEX_op_ld8u_i32: @@ -1658,7 +1657,6 @@ bool tcg_op_supported(TCGOpcode op) return TCG_TARGET_REG_BITS == 32; case INDEX_op_mov_i64: - case INDEX_op_movi_i64: case INDEX_op_setcond_i64: case INDEX_op_brcond_i64: case INDEX_op_ld8u_i64: @@ -1764,7 +1762,6 @@ bool tcg_op_supported(TCGOpcode op) case INDEX_op_mov_vec: case INDEX_op_dup_vec: - case INDEX_op_dupi_vec: case INDEX_op_dupm_vec: case INDEX_op_ld_vec: case INDEX_op_st_vec: @@ -3670,7 +3667,7 @@ static void tcg_reg_alloc_cbranch(TCGContext *s, TCGRegSet allocated_regs) } /* - * Specialized code generation for INDEX_op_movi_*. + * Specialized code generation for INDEX_op_mov_* with a constant. */ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, tcg_target_ulong val, TCGLifeData arg_life, @@ -3693,14 +3690,6 @@ static void tcg_reg_alloc_do_movi(TCGContext *s, TCGTemp *ots, } } -static void tcg_reg_alloc_movi(TCGContext *s, const TCGOp *op) -{ - TCGTemp *ots = arg_temp(op->args[0]); - tcg_target_ulong val = op->args[1]; - - tcg_reg_alloc_do_movi(s, ots, val, op->life, op->output_pref[0]); -} - /* * Specialized code generation for INDEX_op_mov_*. */ @@ -4481,11 +4470,6 @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb) case INDEX_op_mov_vec: tcg_reg_alloc_mov(s, op); break; - case INDEX_op_movi_i32: - case INDEX_op_movi_i64: - case INDEX_op_dupi_vec: - tcg_reg_alloc_movi(s, op); - break; case INDEX_op_dup_vec: tcg_reg_alloc_dup(s, op); break; diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc index e9e5f94c04..15981265db 100644 --- a/tcg/tci/tcg-target.c.inc +++ b/tcg/tci/tcg-target.c.inc @@ -814,8 +814,6 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args, break; case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ case INDEX_op_mov_i64: - case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi. */ - case INDEX_op_movi_i64: case INDEX_op_call: /* Always emitted via tcg_out_call. */ default: tcg_abort(); -- cgit v1.2.3-55-g7522 From be986adb35e3594b02ee0d7f1cbec96b08bb29b7 Mon Sep 17 00:00:00 2001 From: Richard Henderson Date: Tue, 31 Mar 2020 20:10:20 -0700 Subject: tcg: Remove tcg_gen_dup{8,16,32,64}i_vec These interfaces have been replaced by tcg_gen_dupi_vec and tcg_constant_vec. Reviewed-by: Alex Bennée Signed-off-by: Richard Henderson --- include/tcg/tcg-op.h | 4 ---- tcg/tcg-op-vec.c | 20 -------------------- 2 files changed, 24 deletions(-) (limited to 'include') diff --git a/include/tcg/tcg-op.h b/include/tcg/tcg-op.h index ed8de045e2..2cd1faf9c4 100644 --- a/include/tcg/tcg-op.h +++ b/include/tcg/tcg-op.h @@ -959,10 +959,6 @@ void tcg_gen_mov_vec(TCGv_vec, TCGv_vec); void tcg_gen_dup_i32_vec(unsigned vece, TCGv_vec, TCGv_i32); void tcg_gen_dup_i64_vec(unsigned vece, TCGv_vec, TCGv_i64); void tcg_gen_dup_mem_vec(unsigned vece, TCGv_vec, TCGv_ptr, tcg_target_long); -void tcg_gen_dup8i_vec(TCGv_vec, uint32_t); -void tcg_gen_dup16i_vec(TCGv_vec, uint32_t); -void tcg_gen_dup32i_vec(TCGv_vec, uint32_t); -void tcg_gen_dup64i_vec(TCGv_vec, uint64_t); void tcg_gen_dupi_vec(unsigned vece, TCGv_vec, uint64_t); void tcg_gen_add_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b); void tcg_gen_sub_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b); diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c index ce0d2f6e0e..d19aa7373e 100644 --- a/tcg/tcg-op-vec.c +++ b/tcg/tcg-op-vec.c @@ -241,26 +241,6 @@ TCGv_vec tcg_const_ones_vec_matching(TCGv_vec m) return tcg_const_ones_vec(t->base_type); } -void tcg_gen_dup64i_vec(TCGv_vec r, uint64_t a) -{ - tcg_gen_dupi_vec(MO_64, r, a); -} - -void tcg_gen_dup32i_vec(TCGv_vec r, uint32_t a) -{ - tcg_gen_dupi_vec(MO_32, r, a); -} - -void tcg_gen_dup16i_vec(TCGv_vec r, uint32_t a) -{ - tcg_gen_dupi_vec(MO_16, r, a); -} - -void tcg_gen_dup8i_vec(TCGv_vec r, uint32_t a) -{ - tcg_gen_dupi_vec(MO_8, r, a); -} - void tcg_gen_dupi_vec(unsigned vece, TCGv_vec r, uint64_t a) { TCGTemp *rt = tcgv_vec_temp(r); -- cgit v1.2.3-55-g7522