diff mbox series

[RFC,v2,20/44] target/loongarch: Implement vsigncov

Message ID 20230328030631.3117129-21-gaosong@loongson.cn
State New
Headers show
Series Add LoongArch LSX instructions | expand

Commit Message

gaosong March 28, 2023, 3:06 a.m. UTC
This patch includes:
- VSIGNCOV.{B/H/W/D}.

Signed-off-by: Song Gao <gaosong@loongson.cn>
---
 target/loongarch/disas.c                    |  5 ++
 target/loongarch/helper.h                   |  5 ++
 target/loongarch/insn_trans/trans_lsx.c.inc | 54 +++++++++++++++++++++
 target/loongarch/insns.decode               |  5 ++
 target/loongarch/lsx_helper.c               | 19 ++++++++
 5 files changed, 88 insertions(+)

Comments

Richard Henderson April 1, 2023, 5:11 a.m. UTC | #1
On 3/27/23 20:06, Song Gao wrote:
> +static void gen_vsigncov(unsigned vece, TCGv_vec t, TCGv_vec a, TCGv_vec b)
> +{
> +    TCGv_vec t1, t2;
> +
> +    t1 = tcg_temp_new_vec_matching(t);
> +    t2 = tcg_temp_new_vec_matching(t);
> +
> +    tcg_gen_neg_vec(vece, t1, b);
> +    tcg_gen_dupi_vec(vece, t2, 0);

tcg_constant_vec_matching.

Otherwise,
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~
diff mbox series

Patch

diff --git a/target/loongarch/disas.c b/target/loongarch/disas.c
index 412c1cedcb..46e808c321 100644
--- a/target/loongarch/disas.c
+++ b/target/loongarch/disas.c
@@ -1079,3 +1079,8 @@  INSN_LSX(vexth_hu_bu,      vv)
 INSN_LSX(vexth_wu_hu,      vv)
 INSN_LSX(vexth_du_wu,      vv)
 INSN_LSX(vexth_qu_du,      vv)
+
+INSN_LSX(vsigncov_b,       vvv)
+INSN_LSX(vsigncov_h,       vvv)
+INSN_LSX(vsigncov_w,       vvv)
+INSN_LSX(vsigncov_d,       vvv)
diff --git a/target/loongarch/helper.h b/target/loongarch/helper.h
index 0876aa3331..a7394b2eb7 100644
--- a/target/loongarch/helper.h
+++ b/target/loongarch/helper.h
@@ -353,3 +353,8 @@  DEF_HELPER_3(vexth_hu_bu, void, env, i32, i32)
 DEF_HELPER_3(vexth_wu_hu, void, env, i32, i32)
 DEF_HELPER_3(vexth_du_wu, void, env, i32, i32)
 DEF_HELPER_3(vexth_qu_du, void, env, i32, i32)
+
+DEF_HELPER_FLAGS_4(vsigncov_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(vsigncov_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(vsigncov_w, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(vsigncov_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
diff --git a/target/loongarch/insn_trans/trans_lsx.c.inc b/target/loongarch/insn_trans/trans_lsx.c.inc
index f6058c1360..865485ea10 100644
--- a/target/loongarch/insn_trans/trans_lsx.c.inc
+++ b/target/loongarch/insn_trans/trans_lsx.c.inc
@@ -2507,3 +2507,57 @@  TRANS(vexth_hu_bu, gen_vv, gen_helper_vexth_hu_bu)
 TRANS(vexth_wu_hu, gen_vv, gen_helper_vexth_wu_hu)
 TRANS(vexth_du_wu, gen_vv, gen_helper_vexth_du_wu)
 TRANS(vexth_qu_du, gen_vv, gen_helper_vexth_qu_du)
+
+static void gen_vsigncov(unsigned vece, TCGv_vec t, TCGv_vec a, TCGv_vec b)
+{
+    TCGv_vec t1, t2;
+
+    t1 = tcg_temp_new_vec_matching(t);
+    t2 = tcg_temp_new_vec_matching(t);
+
+    tcg_gen_neg_vec(vece, t1, b);
+    tcg_gen_dupi_vec(vece, t2, 0);
+    tcg_gen_cmpsel_vec(TCG_COND_LT, vece, t, a, t2, t1, b);
+    tcg_gen_cmpsel_vec(TCG_COND_EQ, vece, t, a, t2, t2, t);
+}
+
+static void do_vsigncov(unsigned vece, uint32_t vd_ofs, uint32_t vj_ofs,
+                        uint32_t vk_ofs, uint32_t oprsz, uint32_t maxsz)
+{
+    static const TCGOpcode vecop_list[] = {
+        INDEX_op_neg_vec, INDEX_op_cmpsel_vec, 0
+        };
+    static const GVecGen3 op[4] = {
+        {
+            .fniv = gen_vsigncov,
+            .fno = gen_helper_vsigncov_b,
+            .opt_opc = vecop_list,
+            .vece = MO_8
+        },
+        {
+            .fniv = gen_vsigncov,
+            .fno = gen_helper_vsigncov_h,
+            .opt_opc = vecop_list,
+            .vece = MO_16
+        },
+        {
+            .fniv = gen_vsigncov,
+            .fno = gen_helper_vsigncov_w,
+            .opt_opc = vecop_list,
+            .vece = MO_32
+        },
+        {
+            .fniv = gen_vsigncov,
+            .fno = gen_helper_vsigncov_d,
+            .opt_opc = vecop_list,
+            .vece = MO_64
+        },
+    };
+
+    tcg_gen_gvec_3(vd_ofs, vj_ofs, vk_ofs, oprsz, maxsz, &op[vece]);
+}
+
+TRANS(vsigncov_b, gvec_vvv, MO_8, do_vsigncov)
+TRANS(vsigncov_h, gvec_vvv, MO_16, do_vsigncov)
+TRANS(vsigncov_w, gvec_vvv, MO_32, do_vsigncov)
+TRANS(vsigncov_d, gvec_vvv, MO_64, do_vsigncov)
diff --git a/target/loongarch/insns.decode b/target/loongarch/insns.decode
index 39c582d098..4233dd7404 100644
--- a/target/loongarch/insns.decode
+++ b/target/loongarch/insns.decode
@@ -778,3 +778,8 @@  vexth_hu_bu      0111 00101001 11101 11100 ..... .....    @vv
 vexth_wu_hu      0111 00101001 11101 11101 ..... .....    @vv
 vexth_du_wu      0111 00101001 11101 11110 ..... .....    @vv
 vexth_qu_du      0111 00101001 11101 11111 ..... .....    @vv
+
+vsigncov_b       0111 00010010 11100 ..... ..... .....    @vvv
+vsigncov_h       0111 00010010 11101 ..... ..... .....    @vvv
+vsigncov_w       0111 00010010 11110 ..... ..... .....    @vvv
+vsigncov_d       0111 00010010 11111 ..... ..... .....    @vvv
diff --git a/target/loongarch/lsx_helper.c b/target/loongarch/lsx_helper.c
index 9a0b358576..b3a9b8cb66 100644
--- a/target/loongarch/lsx_helper.c
+++ b/target/loongarch/lsx_helper.c
@@ -871,3 +871,22 @@  VEXTH(vexth_d_w, 64, int64_t, int32_t, D, W)
 VEXTH(vexth_hu_bu, 16, uint16_t, uint8_t, H, B)
 VEXTH(vexth_wu_hu, 32, uint32_t, uint16_t, W, H)
 VEXTH(vexth_du_wu, 64, uint64_t, uint32_t, D, W)
+
+#define DO_SIGNCOV(a, b)  (a == 0 ? 0 : a < 0 ? -b : b)
+
+#define VSIGNCOV(NAME, BIT, E, DO_OP)                       \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t v) \
+{                                                           \
+    int i;                                                  \
+    VReg *Vd = (VReg *)vd;                                  \
+    VReg *Vj = (VReg *)vj;                                  \
+    VReg *Vk = (VReg *)vk;                                  \
+    for (i = 0; i < LSX_LEN/BIT; i++) {                     \
+        Vd->E(i) = DO_OP(Vj->E(i),  Vk->E(i));              \
+    }                                                       \
+}
+
+VSIGNCOV(vsigncov_b, 8, B, DO_SIGNCOV)
+VSIGNCOV(vsigncov_h, 16, H, DO_SIGNCOV)
+VSIGNCOV(vsigncov_w, 32, W, DO_SIGNCOV)
+VSIGNCOV(vsigncov_d, 64, D, DO_SIGNCOV)