target/arm: Implement BFMOPA, BFMOPS

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-26-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

target/arm/helper-sme.h

@@ -125,3 +125,5 @@ DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, ptr, ptr, ptr, i32)
 DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, ptr, i32)

target/arm/sme.decode

@@ -73,3 +73,5 @@ ADDVA_d         11000000 11 01000 1 ... ... ..... 00 ...        @adda_64
 
 FMOPA_s         10000000 100 ..... ... ... ..... . 00 ..        @op_32
 FMOPA_d         10000000 110 ..... ... ... ..... . 0 ...        @op_64
+
+BFMOPA          10000001 100 ..... ... ... ..... . 00 ..        @op_32

target/arm/sme_helper.c

@@ -987,3 +987,59 @@ void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn,
         }
     }
 }
+
+/*
+ * Alter PAIR as needed for controlling predicates being false,
+ * and for NEG on an enabled row element.
+ */
+static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg)
+{
+    /*
+     * The pseudocode uses a conditional negate after the conditional zero.
+     * It is simpler here to unconditionally negate before conditional zero.
+     */
+    pair ^= neg;
+    if (!(pg & 1)) {
+        pair &= 0xffff0000u;
+    }
+    if (!(pg & 4)) {
+        pair &= 0x0000ffffu;
+    }
+    return pair;
+}
+
+void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn,
+                        void *vpm, uint32_t desc)
+{
+    intptr_t row, col, oprsz = simd_maxsz(desc);
+    uint32_t neg = simd_data(desc) * 0x80008000u;
+    uint16_t *pn = vpn, *pm = vpm;
+
+    for (row = 0; row < oprsz; ) {
+        uint16_t prow = pn[H2(row >> 4)];
+        do {
+            void *vza_row = vza + tile_vslice_offset(row);
+            uint32_t n = *(uint32_t *)(vzn + H1_4(row));
+
+            n = f16mop_adj_pair(n, prow, neg);
+
+            for (col = 0; col < oprsz; ) {
+                uint16_t pcol = pm[H2(col >> 4)];
+                do {
+                    if (prow & pcol & 0b0101) {
+                        uint32_t *a = vza_row + H1_4(col);
+                        uint32_t m = *(uint32_t *)(vzm + H1_4(col));
+
+                        m = f16mop_adj_pair(m, pcol, 0);
+                        *a = bfdotadd(*a, n, m);
+
+                        col += 4;
+                        pcol >>= 4;
+                    }
+                } while (col & 15);
+            }
+            row += 4;
+            prow >>= 4;
+        } while (row & 15);
+    }
+}

target/arm/translate-sme.c

@@ -299,6 +299,33 @@ TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s)
 TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d)
 TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d)
 
+static bool do_outprod(DisasContext *s, arg_op *a, MemOp esz,
+                       gen_helper_gvec_5 *fn)
+{
+    int svl = streaming_vec_reg_size(s);
+    uint32_t desc = simd_desc(svl, svl, a->sub);
+    TCGv_ptr za, zn, zm, pn, pm;
+
+    if (!sme_smza_enabled_check(s)) {
+        return true;
+    }
+
+    /* Sum XZR+zad to find ZAd. */
+    za = get_tile_rowcol(s, esz, 31, a->zad, false);
+    zn = vec_full_reg_ptr(s, a->zn);
+    zm = vec_full_reg_ptr(s, a->zm);
+    pn = pred_full_reg_ptr(s, a->pn);
+    pm = pred_full_reg_ptr(s, a->pm);
+
+    fn(za, zn, zm, pn, pm, tcg_constant_i32(desc));
+
+    tcg_temp_free_ptr(za);
+    tcg_temp_free_ptr(zn);
+    tcg_temp_free_ptr(pn);
+    tcg_temp_free_ptr(pm);
+    return true;
+}
+
 static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz,
                             gen_helper_gvec_5_ptr *fn)
 {
@@ -330,3 +357,6 @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz,
 
 TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s)
 TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d)
+
+/* TODO: FEAT_EBF16 */
+TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa)