target/arm: Implement SME MOVA

We can reuse the SVE functions for implementing moves to/from
horizontal tile slices, but we need new ones for moves to/from
vertical tile slices.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-20-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

1 files changed, 127 insertions, 0 deletions

diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c
index 9715045..79c33d3 100644
--- a/target/arm/translate-sme.c
+++ b/target/arm/translate-sme.c
@@ -35,6 +35,74 @@
 #include "decode-sme.c.inc"
 
 
+/*
+ * Resolve tile.size[index] to a host pointer, where tile and index
+ * are always decoded together, dependent on the element size.
+ */
+static TCGv_ptr get_tile_rowcol(DisasContext *s, int esz, int rs,
+                                int tile_index, bool vertical)
+{
+    int tile = tile_index >> (4 - esz);
+    int index = esz == MO_128 ? 0 : extract32(tile_index, 0, 4 - esz);
+    int pos, len, offset;
+    TCGv_i32 tmp;
+    TCGv_ptr addr;
+
+    /* Compute the final index, which is Rs+imm. */
+    tmp = tcg_temp_new_i32();
+    tcg_gen_trunc_tl_i32(tmp, cpu_reg(s, rs));
+    tcg_gen_addi_i32(tmp, tmp, index);
+
+    /* Prepare a power-of-two modulo via extraction of @len bits. */
+    len = ctz32(streaming_vec_reg_size(s)) - esz;
+
+    if (vertical) {
+        /*
+         * Compute the byte offset of the index within the tile:
+         *     (index % (svl / size)) * size
+         *   = (index % (svl >> esz)) << esz
+         * Perform the power-of-two modulo via extraction of the low @len bits.
+         * Perform the multiply by shifting left by @pos bits.
+         * Perform these operations simultaneously via deposit into zero.
+         */
+        pos = esz;
+        tcg_gen_deposit_z_i32(tmp, tmp, pos, len);
+
+        /*
+         * For big-endian, adjust the indexed column byte offset within
+         * the uint64_t host words that make up env->zarray[].
+         */
+        if (HOST_BIG_ENDIAN && esz < MO_64) {
+            tcg_gen_xori_i32(tmp, tmp, 8 - (1 << esz));
+        }
+    } else {
+        /*
+         * Compute the byte offset of the index within the tile:
+         *     (index % (svl / size)) * (size * sizeof(row))
+         *   = (index % (svl >> esz)) << (esz + log2(sizeof(row)))
+         */
+        pos = esz + ctz32(sizeof(ARMVectorReg));
+        tcg_gen_deposit_z_i32(tmp, tmp, pos, len);
+
+        /* Row slices are always aligned and need no endian adjustment. */
+    }
+
+    /* The tile byte offset within env->zarray is the row. */
+    offset = tile * sizeof(ARMVectorReg);
+
+    /* Include the byte offset of zarray to make this relative to env. */
+    offset += offsetof(CPUARMState, zarray);
+    tcg_gen_addi_i32(tmp, tmp, offset);
+
+    /* Add the byte offset to env to produce the final pointer. */
+    addr = tcg_temp_new_ptr();
+    tcg_gen_ext_i32_ptr(addr, tmp);
+    tcg_temp_free_i32(tmp);
+    tcg_gen_add_ptr(addr, addr, cpu_env);
+
+    return addr;
+}
+
 static bool trans_ZERO(DisasContext *s, arg_ZERO *a)
 {
     if (!dc_isar_feature(aa64_sme, s)) {
@@ -46,3 +114,62 @@ static bool trans_ZERO(DisasContext *s, arg_ZERO *a)
     }
     return true;
 }
+
+static bool trans_MOVA(DisasContext *s, arg_MOVA *a)
+{
+    static gen_helper_gvec_4 * const h_fns[5] = {
+        gen_helper_sve_sel_zpzz_b, gen_helper_sve_sel_zpzz_h,
+        gen_helper_sve_sel_zpzz_s, gen_helper_sve_sel_zpzz_d,
+        gen_helper_sve_sel_zpzz_q
+    };
+    static gen_helper_gvec_3 * const cz_fns[5] = {
+        gen_helper_sme_mova_cz_b, gen_helper_sme_mova_cz_h,
+        gen_helper_sme_mova_cz_s, gen_helper_sme_mova_cz_d,
+        gen_helper_sme_mova_cz_q,
+    };
+    static gen_helper_gvec_3 * const zc_fns[5] = {
+        gen_helper_sme_mova_zc_b, gen_helper_sme_mova_zc_h,
+        gen_helper_sme_mova_zc_s, gen_helper_sme_mova_zc_d,
+        gen_helper_sme_mova_zc_q,
+    };
+
+    TCGv_ptr t_za, t_zr, t_pg;
+    TCGv_i32 t_desc;
+    int svl;
+
+    if (!dc_isar_feature(aa64_sme, s)) {
+        return false;
+    }
+    if (!sme_smza_enabled_check(s)) {
+        return true;
+    }
+
+    t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v);
+    t_zr = vec_full_reg_ptr(s, a->zr);
+    t_pg = pred_full_reg_ptr(s, a->pg);
+
+    svl = streaming_vec_reg_size(s);
+    t_desc = tcg_constant_i32(simd_desc(svl, svl, 0));
+
+    if (a->v) {
+        /* Vertical slice -- use sme mova helpers. */
+        if (a->to_vec) {
+            zc_fns[a->esz](t_zr, t_za, t_pg, t_desc);
+        } else {
+            cz_fns[a->esz](t_za, t_zr, t_pg, t_desc);
+        }
+    } else {
+        /* Horizontal slice -- reuse sve sel helpers. */
+        if (a->to_vec) {
+            h_fns[a->esz](t_zr, t_za, t_zr, t_pg, t_desc);
+        } else {
+            h_fns[a->esz](t_za, t_zr, t_za, t_pg, t_desc);
+        }
+    }
+
+    tcg_temp_free_ptr(t_za);
+    tcg_temp_free_ptr(t_zr);
+    tcg_temp_free_ptr(t_pg);
+
+    return true;
+}