@@ -6432,17 +6432,6 @@ static void handle_shri_with_rndacc(TCGv_i64 tcg_res, TCGv_i64 tcg_src,
}
}
-/* Common SHL/SLI - Shift left with an optional insert */
-static void handle_shli_with_ins(TCGv_i64 tcg_res, TCGv_i64 tcg_src,
- bool insert, int shift)
-{
- if (insert) { /* SLI */
- tcg_gen_deposit_i64(tcg_res, tcg_res, tcg_src, shift, 64 - shift);
- } else { /* SHL */
- tcg_gen_shli_i64(tcg_res, tcg_src, shift);
- }
-}
-
/* SRI: shift right with insert */
static void handle_shri_with_ins(TCGv_i64 tcg_res, TCGv_i64 tcg_src,
int size, int shift)
@@ -6546,7 +6535,11 @@ static void handle_scalar_simd_shli(DisasContext *s, bool insert,
tcg_rn = read_fp_dreg(s, rn);
tcg_rd = insert ? read_fp_dreg(s, rd) : tcg_temp_new_i64();
- handle_shli_with_ins(tcg_rd, tcg_rn, insert, shift);
+ if (insert) {
+ tcg_gen_deposit_i64(tcg_rd, tcg_rd, tcg_rn, shift, 64 - shift);
+ } else {
+ tcg_gen_shli_i64(tcg_rd, tcg_rn, shift);
+ }
write_fp_dreg(s, rd, tcg_rd);
@@ -8283,16 +8276,195 @@ static void disas_simd_scalar_two_reg_misc(DisasContext *s, uint32_t insn)
}
}
+static void gen_ssra8_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_vec_sar8i_i64(a, a, shift);
+ tcg_gen_vec_add8_i64(d, d, a);
+}
+
+static void gen_ssra16_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_vec_sar16i_i64(a, a, shift);
+ tcg_gen_vec_add16_i64(d, d, a);
+}
+
+static void gen_ssra32_i32(TCGv_i32 d, TCGv_i32 a, unsigned shift)
+{
+ tcg_gen_sari_i32(a, a, shift);
+ tcg_gen_add_i32(d, d, a);
+}
+
+static void gen_ssra64_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_sari_i64(a, a, shift);
+ tcg_gen_add_i64(d, d, a);
+}
+
+static void gen_ssra_vec(unsigned vece, TCGv_vec d, TCGv_vec a, unsigned sh)
+{
+ tcg_gen_sari_vec(vece, a, a, sh);
+ tcg_gen_add_vec(vece, d, d, a);
+}
+
+static void gen_usra8_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_vec_shr8i_i64(a, a, shift);
+ tcg_gen_vec_add8_i64(d, d, a);
+}
+
+static void gen_usra16_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_vec_shr16i_i64(a, a, shift);
+ tcg_gen_vec_add16_i64(d, d, a);
+}
+
+static void gen_usra32_i32(TCGv_i32 d, TCGv_i32 a, unsigned shift)
+{
+ tcg_gen_shri_i32(a, a, shift);
+ tcg_gen_add_i32(d, d, a);
+}
+
+static void gen_usra64_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_shri_i64(a, a, shift);
+ tcg_gen_add_i64(d, d, a);
+}
+
+static void gen_usra_vec(unsigned vece, TCGv_vec d, TCGv_vec a, unsigned sh)
+{
+ tcg_gen_shri_vec(vece, a, a, sh);
+ tcg_gen_add_vec(vece, d, d, a);
+}
+
+static void gen_shr8_ins_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ uint64_t mask = (0xff >> shift) * (-1ull / 0xff);
+ TCGv_i64 t = tcg_temp_new_i64();
+
+ tcg_gen_shri_i64(t, a, shift);
+ tcg_gen_andi_i64(t, t, mask);
+ tcg_gen_andi_i64(d, d, ~mask);
+ tcg_gen_or_i64(d, d, t);
+ tcg_temp_free_i64(t);
+}
+
+static void gen_shr16_ins_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ uint64_t mask = (0xffff >> shift) * (-1ull / 0xffff);
+ TCGv_i64 t = tcg_temp_new_i64();
+
+ tcg_gen_shri_i64(t, a, shift);
+ tcg_gen_andi_i64(t, t, mask);
+ tcg_gen_andi_i64(d, d, ~mask);
+ tcg_gen_or_i64(d, d, t);
+ tcg_temp_free_i64(t);
+}
+
+static void gen_shr32_ins_i32(TCGv_i32 d, TCGv_i32 a, unsigned shift)
+{
+ tcg_gen_shri_i32(a, a, shift);
+ tcg_gen_deposit_i32(d, d, a, 0, 32 - shift);
+}
+
+static void gen_shr64_ins_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_shri_i64(a, a, shift);
+ tcg_gen_deposit_i64(d, d, a, 0, 64 - shift);
+}
+
+static void gen_shr_ins_vec(unsigned vece, TCGv_vec d, TCGv_vec a, unsigned sh)
+{
+ uint64_t mask = (2ull << ((8 << vece) - 1)) - 1;
+ TCGv_vec t = tcg_temp_new_vec_matching(d);
+ TCGv_vec m = tcg_temp_new_vec_matching(d);
+
+ tcg_gen_dupi_vec(vece, m, mask ^ (mask >> sh));
+ tcg_gen_shri_vec(vece, t, a, sh);
+ tcg_gen_and_vec(vece, d, d, m);
+ tcg_gen_or_vec(vece, d, d, t);
+
+ tcg_temp_free_vec(t);
+ tcg_temp_free_vec(m);
+}
+
/* SSHR[RA]/USHR[RA] - Vector shift right (optional rounding/accumulate) */
static void handle_vec_simd_shri(DisasContext *s, bool is_q, bool is_u,
int immh, int immb, int opcode, int rn, int rd)
{
+ static const GVecGen2i ssra_op[4] = {
+ { .fni8 = gen_ssra8_i64,
+ .fniv = gen_ssra_vec,
+ .load_dest = true,
+ .opc = INDEX_op_sari_vec,
+ .vece = MO_8 },
+ { .fni8 = gen_ssra16_i64,
+ .fniv = gen_ssra_vec,
+ .load_dest = true,
+ .opc = INDEX_op_sari_vec,
+ .vece = MO_16 },
+ { .fni4 = gen_ssra32_i32,
+ .fniv = gen_ssra_vec,
+ .load_dest = true,
+ .opc = INDEX_op_sari_vec,
+ .vece = MO_32 },
+ { .fni8 = gen_ssra64_i64,
+ .fniv = gen_ssra_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .opc = INDEX_op_sari_vec,
+ .vece = MO_64 },
+ };
+ static const GVecGen2i usra_op[4] = {
+ { .fni8 = gen_usra8_i64,
+ .fniv = gen_usra_vec,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_8, },
+ { .fni8 = gen_usra16_i64,
+ .fniv = gen_usra_vec,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_16, },
+ { .fni4 = gen_usra32_i32,
+ .fniv = gen_usra_vec,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_32, },
+ { .fni8 = gen_usra64_i64,
+ .fniv = gen_usra_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_64, },
+ };
+ static const GVecGen2i sri_op[4] = {
+ { .fni8 = gen_shr8_ins_i64,
+ .fniv = gen_shr_ins_vec,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_8 },
+ { .fni8 = gen_shr16_ins_i64,
+ .fniv = gen_shr_ins_vec,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_16 },
+ { .fni4 = gen_shr32_ins_i32,
+ .fniv = gen_shr_ins_vec,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_32 },
+ { .fni8 = gen_shr64_ins_i64,
+ .fniv = gen_shr_ins_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .opc = INDEX_op_shri_vec,
+ .vece = MO_64 },
+ };
+
int size = 32 - clz32(immh) - 1;
int immhb = immh << 3 | immb;
int shift = 2 * (8 << size) - immhb;
bool accumulate = false;
- bool round = false;
- bool insert = false;
int dsize = is_q ? 128 : 64;
int esize = 8 << size;
int elements = dsize/esize;
@@ -8300,6 +8472,8 @@ static void handle_vec_simd_shri(DisasContext *s, bool is_q, bool is_u,
TCGv_i64 tcg_rn = new_tmp_a64(s);
TCGv_i64 tcg_rd = new_tmp_a64(s);
TCGv_i64 tcg_round;
+ uint64_t round_const;
+ const GVecGen2i *gvec_op;
int i;
if (extract32(immh, 3, 1) && !is_q) {
@@ -8318,64 +8492,141 @@ static void handle_vec_simd_shri(DisasContext *s, bool is_q, bool is_u,
switch (opcode) {
case 0x02: /* SSRA / USRA (accumulate) */
- accumulate = true;
- break;
+ if (is_u) {
+ /* Shift count same as element size produces zero to add. */
+ if (shift == 8 << size) {
+ goto done;
+ }
+ gvec_op = &usra_op[size];
+ } else {
+ /* Shift count same as element size produces all sign to add. */
+ if (shift == 8 << size) {
+ shift -= 1;
+ }
+ gvec_op = &ssra_op[size];
+ }
+ goto do_gvec;
+ case 0x08: /* SRI */
+ /* Shift count same as element size is valid but does nothing. */
+ if (shift == 8 << size) {
+ goto done;
+ }
+ gvec_op = &sri_op[size];
+ do_gvec:
+ tcg_gen_gvec_2i(vec_full_reg_offset(s, rd),
+ vec_full_reg_offset(s, rn), is_q ? 16 : 8,
+ vec_full_reg_size(s), shift, gvec_op);
+ return;
+
+ case 0x00: /* SSHR / USHR */
+ if (is_u) {
+ if (shift == 8 << size) {
+ /* Shift count the same size as element size produces zero. */
+ tcg_gen_gvec_dup8i(vec_full_reg_offset(s, rd),
+ is_q ? 16 : 8, vec_full_reg_size(s), 0);
+ } else {
+ tcg_gen_gvec_shri(size, vec_full_reg_offset(s, rd),
+ vec_full_reg_offset(s, rn), is_q ? 16 : 8,
+ vec_full_reg_size(s), shift);
+ }
+ } else {
+ /* Shift count the same size as element size produces all sign. */
+ if (shift == 8 << size) {
+ shift -= 1;
+ }
+ tcg_gen_gvec_sari(size, vec_full_reg_offset(s, rd),
+ vec_full_reg_offset(s, rn), is_q ? 16 : 8,
+ vec_full_reg_size(s), shift);
+ }
+ return;
+
case 0x04: /* SRSHR / URSHR (rounding) */
- round = true;
break;
case 0x06: /* SRSRA / URSRA (accum + rounding) */
- accumulate = round = true;
- break;
- case 0x08: /* SRI */
- insert = true;
+ accumulate = true;
break;
+ default:
+ g_assert_not_reached();
}
- if (round) {
- uint64_t round_const = 1ULL << (shift - 1);
- tcg_round = tcg_const_i64(round_const);
- } else {
- tcg_round = NULL;
- }
+ round_const = 1ULL << (shift - 1);
+ tcg_round = tcg_const_i64(round_const);
for (i = 0; i < elements; i++) {
read_vec_element(s, tcg_rn, rn, i, memop);
- if (accumulate || insert) {
+ if (accumulate) {
read_vec_element(s, tcg_rd, rd, i, memop);
}
- if (insert) {
- handle_shri_with_ins(tcg_rd, tcg_rn, size, shift);
- } else {
- handle_shri_with_rndacc(tcg_rd, tcg_rn, tcg_round,
- accumulate, is_u, size, shift);
- }
+ handle_shri_with_rndacc(tcg_rd, tcg_rn, tcg_round,
+ accumulate, is_u, size, shift);
write_vec_element(s, tcg_rd, rd, i, size);
}
+ tcg_temp_free_i64(tcg_round);
+ done:
if (!is_q) {
clear_vec_high(s, rd);
}
+}
- if (round) {
- tcg_temp_free_i64(tcg_round);
- }
+static void gen_shl8_ins_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ uint64_t mask = ((0xff << shift) & 0xff) * (-1ull / 0xff);
+ TCGv_i64 t = tcg_temp_new_i64();
+
+ tcg_gen_shli_i64(t, a, shift);
+ tcg_gen_andi_i64(t, t, mask);
+ tcg_gen_andi_i64(d, d, ~mask);
+ tcg_gen_or_i64(d, d, t);
+ tcg_temp_free_i64(t);
+}
+
+static void gen_shl16_ins_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ uint64_t mask = ((0xffff << shift) & 0xffff) * (-1ull / 0xffff);
+ TCGv_i64 t = tcg_temp_new_i64();
+
+ tcg_gen_shli_i64(t, a, shift);
+ tcg_gen_andi_i64(t, t, mask);
+ tcg_gen_andi_i64(d, d, ~mask);
+ tcg_gen_or_i64(d, d, t);
+ tcg_temp_free_i64(t);
+}
+
+static void gen_shl32_ins_i32(TCGv_i32 d, TCGv_i32 a, unsigned shift)
+{
+ tcg_gen_deposit_i32(d, d, a, shift, 32 - shift);
+}
+
+static void gen_shl64_ins_i64(TCGv_i64 d, TCGv_i64 a, unsigned shift)
+{
+ tcg_gen_deposit_i64(d, d, a, shift, 64 - shift);
+}
+
+static void gen_shl_ins_vec(unsigned vece, TCGv_vec d, TCGv_vec a, unsigned sh)
+{
+ uint64_t mask = (1ull << sh) - 1;
+ TCGv_vec t = tcg_temp_new_vec_matching(d);
+ TCGv_vec m = tcg_temp_new_vec_matching(d);
+
+ tcg_gen_dupi_vec(vece, m, mask);
+ tcg_gen_shli_vec(vece, t, a, sh);
+ tcg_gen_and_vec(vece, d, d, m);
+ tcg_gen_or_vec(vece, d, d, t);
+
+ tcg_temp_free_vec(t);
+ tcg_temp_free_vec(m);
}
/* SHL/SLI - Vector shift left */
static void handle_vec_simd_shli(DisasContext *s, bool is_q, bool insert,
- int immh, int immb, int opcode, int rn, int rd)
+ int immh, int immb, int opcode, int rn, int rd)
{
int size = 32 - clz32(immh) - 1;
int immhb = immh << 3 | immb;
int shift = immhb - (8 << size);
- int dsize = is_q ? 128 : 64;
- int esize = 8 << size;
- int elements = dsize/esize;
- TCGv_i64 tcg_rn = new_tmp_a64(s);
- TCGv_i64 tcg_rd = new_tmp_a64(s);
- int i;
if (extract32(immh, 3, 1) && !is_q) {
unallocated_encoding(s);
@@ -8391,19 +8642,40 @@ static void handle_vec_simd_shli(DisasContext *s, bool is_q, bool insert,
return;
}
- for (i = 0; i < elements; i++) {
- read_vec_element(s, tcg_rn, rn, i, size);
- if (insert) {
- read_vec_element(s, tcg_rd, rd, i, size);
- }
-
- handle_shli_with_ins(tcg_rd, tcg_rn, insert, shift);
-
- write_vec_element(s, tcg_rd, rd, i, size);
- }
-
- if (!is_q) {
- clear_vec_high(s, rd);
+ if (insert) {
+ static const GVecGen2i shi_op[4] = {
+ { .fni8 = gen_shl8_ins_i64,
+ .fniv = gen_shl_ins_vec,
+ .opc = INDEX_op_shli_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .vece = MO_8 },
+ { .fni8 = gen_shl16_ins_i64,
+ .fniv = gen_shl_ins_vec,
+ .opc = INDEX_op_shli_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .vece = MO_16 },
+ { .fni4 = gen_shl32_ins_i32,
+ .fniv = gen_shl_ins_vec,
+ .opc = INDEX_op_shli_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .vece = MO_32 },
+ { .fni8 = gen_shl64_ins_i64,
+ .fniv = gen_shl_ins_vec,
+ .opc = INDEX_op_shli_vec,
+ .prefer_i64 = TCG_TARGET_REG_BITS == 64,
+ .load_dest = true,
+ .vece = MO_64 },
+ };
+ tcg_gen_gvec_2i(vec_full_reg_offset(s, rd),
+ vec_full_reg_offset(s, rn), is_q ? 16 : 8,
+ vec_full_reg_size(s), shift, &shi_op[size]);
+ } else {
+ tcg_gen_gvec_shli(size, vec_full_reg_offset(s, rd),
+ vec_full_reg_offset(s, rn), is_q ? 16 : 8,
+ vec_full_reg_size(s), shift);
}
}
@@ -1290,7 +1290,11 @@ void tcg_gen_gvec_shli(unsigned vece, uint32_t dofs, uint32_t aofs,
};
tcg_debug_assert(vece <= MO_64);
- tcg_gen_gvec_2i(dofs, aofs, opsz, clsz, shift, &g[vece]);
+ if (shift == 0) {
+ tcg_gen_gvec_mov(vece, dofs, aofs, opsz, clsz);
+ } else {
+ tcg_gen_gvec_2i(dofs, aofs, opsz, clsz, shift, &g[vece]);
+ }
}
void tcg_gen_vec_shr8i_i64(TCGv_i64 d, TCGv_i64 a, unsigned c)
@@ -1335,7 +1339,11 @@ void tcg_gen_gvec_shri(unsigned vece, uint32_t dofs, uint32_t aofs,
};
tcg_debug_assert(vece <= MO_64);
- tcg_gen_gvec_2i(dofs, aofs, opsz, clsz, shift, &g[vece]);
+ if (shift == 0) {
+ tcg_gen_gvec_mov(vece, dofs, aofs, opsz, clsz);
+ } else {
+ tcg_gen_gvec_2i(dofs, aofs, opsz, clsz, shift, &g[vece]);
+ }
}
void tcg_gen_vec_sar8i_i64(TCGv_i64 d, TCGv_i64 a, unsigned c)
@@ -1394,7 +1402,11 @@ void tcg_gen_gvec_sari(unsigned vece, uint32_t dofs, uint32_t aofs,
};
tcg_debug_assert(vece <= MO_64);
- tcg_gen_gvec_2i(dofs, aofs, opsz, clsz, shift, &g[vece]);
+ if (shift == 0) {
+ tcg_gen_gvec_mov(vece, dofs, aofs, opsz, clsz);
+ } else {
+ tcg_gen_gvec_2i(dofs, aofs, opsz, clsz, shift, &g[vece]);
+ }
}
static void do_zip(unsigned vece, uint32_t dofs, uint32_t aofs,
@@ -389,11 +389,16 @@ static void do_shifti(TCGOpcode opc, unsigned vece,
TCGArg ri = temp_arg(rt);
TCGArg ai = temp_arg(at);
TCGType type = rt->base_type;
- unsigned vecl = type - TCG_TYPE_V64;
int can;
tcg_debug_assert(at->base_type == type);
tcg_debug_assert(i < (8 << vece));
+
+ if (i == 0) {
+ tcg_gen_mov_vec(r, a);
+ return;
+ }
+
can = tcg_can_emit_vec_op(opc, type, vece);
if (can > 0) {
vec_gen_3(opc, type, vece, ri, ai, i);
@@ -402,7 +407,7 @@ static void do_shifti(TCGOpcode opc, unsigned vece,
to the target. Often, but not always, dupi can feed a vector
shift easier than a scalar. */
tcg_debug_assert(can < 0);
- tcg_expand_vec_op(opc, vecl, vece, ri, ai, i);
+ tcg_expand_vec_op(opc, type, vece, ri, ai, i);
}
}
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- target/arm/translate-a64.c | 386 ++++++++++++++++++++++++++++++++++++++------- tcg/tcg-op-gvec.c | 18 ++- tcg/tcg-op-vec.c | 9 +- 3 files changed, 351 insertions(+), 62 deletions(-) -- 2.14.3