target/arm: Convert the VCVT-to-f16 insns to decodetree

Convert the VCVTT and VCVTB instructions which convert from
f32 and f64 to f16 to decodetree.

Since we're no longer constrained to the old decoder's style
using cpu_F0s and cpu_F0d we can perform a direct 16 bit
store of the right half of the input single-precision register
rather than doing a load/modify/store sequence on the full
32 bits.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

3 files changed, 69 insertions, 78 deletions

diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c
index 732bf6020a..a19ede8671 100644
--- a/target/arm/translate-vfp.inc.c
+++ b/target/arm/translate-vfp.inc.c
@@ -2095,3 +2095,65 @@ static bool trans_VCVT_f64_f16(DisasContext *s, arg_VCVT_f64_f16 *a)
     tcg_temp_free_i64(vd);
     return true;
 }
+
+static bool trans_VCVT_f16_f32(DisasContext *s, arg_VCVT_f16_f32 *a)
+{
+    TCGv_ptr fpst;
+    TCGv_i32 ahp_mode;
+    TCGv_i32 tmp;
+
+    if (!dc_isar_feature(aa32_fp16_spconv, s)) {
+        return false;
+    }
+
+    if (!vfp_access_check(s)) {
+        return true;
+    }
+
+    fpst = get_fpstatus_ptr(false);
+    ahp_mode = get_ahp_flag();
+    tmp = tcg_temp_new_i32();
+
+    neon_load_reg32(tmp, a->vm);
+    gen_helper_vfp_fcvt_f32_to_f16(tmp, tmp, fpst, ahp_mode);
+    tcg_gen_st16_i32(tmp, cpu_env, vfp_f16_offset(a->vd, a->t));
+    tcg_temp_free_i32(ahp_mode);
+    tcg_temp_free_ptr(fpst);
+    tcg_temp_free_i32(tmp);
+    return true;
+}
+
+static bool trans_VCVT_f16_f64(DisasContext *s, arg_VCVT_f16_f64 *a)
+{
+    TCGv_ptr fpst;
+    TCGv_i32 ahp_mode;
+    TCGv_i32 tmp;
+    TCGv_i64 vm;
+
+    if (!dc_isar_feature(aa32_fp16_dpconv, s)) {
+        return false;
+    }
+
+    /* UNDEF accesses to D16-D31 if they don't exist. */
+    if (!dc_isar_feature(aa32_fp_d32, s) && (a->vm  & 0x10)) {
+        return false;
+    }
+
+    if (!vfp_access_check(s)) {
+        return true;
+    }
+
+    fpst = get_fpstatus_ptr(false);
+    ahp_mode = get_ahp_flag();
+    tmp = tcg_temp_new_i32();
+    vm = tcg_temp_new_i64();
+
+    neon_load_reg64(vm, a->vm);
+    gen_helper_vfp_fcvt_f64_to_f16(tmp, vm, fpst, ahp_mode);
+    tcg_temp_free_i64(vm);
+    tcg_gen_st16_i32(tmp, cpu_env, vfp_f16_offset(a->vd, a->t));
+    tcg_temp_free_i32(ahp_mode);
+    tcg_temp_free_ptr(fpst);
+    tcg_temp_free_i32(tmp);
+    return true;
+}
diff --git a/target/arm/translate.c b/target/arm/translate.c
index f22b8ddf68..b53bf93382 100644
--- a/target/arm/translate.c
+++ b/target/arm/translate.c
@@ -2963,20 +2963,6 @@ static int disas_dsp_insn(DisasContext *s, uint32_t insn)
 #define VFP_SREG_M(insn) VFP_SREG(insn,  0,  5)
 #define VFP_DREG_M(reg, insn) VFP_DREG(reg, insn,  0,  5)
 
-/* Move between integer and VFP cores.  */
-static TCGv_i32 gen_vfp_mrs(void)
-{
-    TCGv_i32 tmp = tcg_temp_new_i32();
-    tcg_gen_mov_i32(tmp, cpu_F0s);
-    return tmp;
-}
-
-static void gen_vfp_msr(TCGv_i32 tmp)
-{
-    tcg_gen_mov_i32(cpu_F0s, tmp);
-    tcg_temp_free_i32(tmp);
-}
-
 static void gen_neon_dup_low16(TCGv_i32 var)
 {
     TCGv_i32 tmp = tcg_temp_new_i32();
@@ -3003,8 +2989,6 @@ static int disas_vfp_insn(DisasContext *s, uint32_t insn)
 {
     uint32_t rd, rn, rm, op, delta_d, delta_m, bank_mask;
     int dp, veclen;
-    TCGv_i32 tmp;
-    TCGv_i32 tmp2;
 
     if (!arm_dc_feature(s, ARM_FEATURE_VFP)) {
         return 1;
@@ -3066,8 +3050,7 @@ static int disas_vfp_insn(DisasContext *s, uint32_t insn)
                 return 1;
             case 15:
                 switch (rn) {
-                case 0 ... 5:
-                case 8 ... 11:
+                case 0 ... 11:
                     /* Already handled by decodetree */
                     return 1;
                 default:
@@ -3080,20 +3063,6 @@ static int disas_vfp_insn(DisasContext *s, uint32_t insn)
             if (op == 15) {
                 /* rn is opcode, encoded as per VFP_SREG_N. */
                 switch (rn) {
-                case 0x06: /* vcvtb.f16.f32, vcvtb.f16.f64 */
-                case 0x07: /* vcvtt.f16.f32, vcvtt.f16.f64 */
-                    if (dp) {
-                        if (!dc_isar_feature(aa32_fp16_dpconv, s)) {
-                            return 1;
-                        }
-                    } else {
-                        if (!dc_isar_feature(aa32_fp16_spconv, s)) {
-                            return 1;
-                        }
-                    }
-                    rd_is_dp = false;
-                    break;
-
                 case 0x0c: /* vrintr */
                 case 0x0d: /* vrintz */
                 case 0x0e: /* vrintx */
@@ -3221,52 +3190,6 @@ static int disas_vfp_insn(DisasContext *s, uint32_t insn)
                 switch (op) {
                 case 15: /* extension space */
                     switch (rn) {
-                    case 6: /* vcvtb.f16.f32, vcvtb.f16.f64 */
-                    {
-                        TCGv_ptr fpst = get_fpstatus_ptr(false);
-                        TCGv_i32 ahp = get_ahp_flag();
-                        tmp = tcg_temp_new_i32();
-
-                        if (dp) {
-                            gen_helper_vfp_fcvt_f64_to_f16(tmp, cpu_F0d,
-                                                           fpst, ahp);
-                        } else {
-                            gen_helper_vfp_fcvt_f32_to_f16(tmp, cpu_F0s,
-                                                           fpst, ahp);
-                        }
-                        tcg_temp_free_i32(ahp);
-                        tcg_temp_free_ptr(fpst);
-                        gen_mov_F0_vreg(0, rd);
-                        tmp2 = gen_vfp_mrs();
-                        tcg_gen_andi_i32(tmp2, tmp2, 0xffff0000);
-                        tcg_gen_or_i32(tmp, tmp, tmp2);
-                        tcg_temp_free_i32(tmp2);
-                        gen_vfp_msr(tmp);
-                        break;
-                    }
-                    case 7: /* vcvtt.f16.f32, vcvtt.f16.f64 */
-                    {
-                        TCGv_ptr fpst = get_fpstatus_ptr(false);
-                        TCGv_i32 ahp = get_ahp_flag();
-                        tmp = tcg_temp_new_i32();
-                        if (dp) {
-                            gen_helper_vfp_fcvt_f64_to_f16(tmp, cpu_F0d,
-                                                           fpst, ahp);
-                        } else {
-                            gen_helper_vfp_fcvt_f32_to_f16(tmp, cpu_F0s,
-                                                           fpst, ahp);
-                        }
-                        tcg_temp_free_i32(ahp);
-                        tcg_temp_free_ptr(fpst);
-                        tcg_gen_shli_i32(tmp, tmp, 16);
-                        gen_mov_F0_vreg(0, rd);
-                        tmp2 = gen_vfp_mrs();
-                        tcg_gen_ext16u_i32(tmp2, tmp2);
-                        tcg_gen_or_i32(tmp, tmp, tmp2);
-                        tcg_temp_free_i32(tmp2);
-                        gen_vfp_msr(tmp);
-                        break;
-                    }
                     case 12: /* vrintr */
                     {
                         TCGv_ptr fpst = get_fpstatus_ptr(0);
diff --git a/target/arm/vfp.decode b/target/arm/vfp.decode
index 53d9544f7c..b88d1d06f0 100644
--- a/target/arm/vfp.decode
+++ b/target/arm/vfp.decode
@@ -187,3 +187,9 @@ VCVT_f32_f16 ---- 1110 1.11 0010 .... 1010 t:1 1.0 .... \
              vd=%vd_sp vm=%vm_sp
 VCVT_f64_f16 ---- 1110 1.11 0010 .... 1011 t:1 1.0 .... \
              vd=%vd_dp vm=%vm_sp
+
+# VCVTB and VCVTT to f16: Vd format is always vd_sp; Vm format depends on size bit
+VCVT_f16_f32 ---- 1110 1.11 0011 .... 1010 t:1 1.0 .... \
+             vd=%vd_sp vm=%vm_sp
+VCVT_f16_f64 ---- 1110 1.11 0011 .... 1011 t:1 1.0 .... \
+             vd=%vd_sp vm=%vm_dp