target/arm: Load/store integer pair with one tcg operation

This is required for LSE2, where the pair must be treated atomically if
it does not cross a 16-byte boundary.  But it simplifies the code to do
this always.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-10-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm/tcg/translate-a64.c

@@ -2942,26 +2942,66 @@ static void disas_ldst_pair(DisasContext *s, uint32_t insn)
     } else {
         TCGv_i64 tcg_rt = cpu_reg(s, rt);
         TCGv_i64 tcg_rt2 = cpu_reg(s, rt2);
+        MemOp mop = size + 1;
+
+        /*
+         * With LSE2, non-sign-extending pairs are treated atomically if
+         * aligned, and if unaligned one of the pair will be completely
+         * within a 16-byte block and that element will be atomic.
+         * Otherwise each element is separately atomic.
+         * In all cases, issue one operation with the correct atomicity.
+         *
+         * This treats sign-extending loads like zero-extending loads,
+         * since that reuses the most code below.
+         */
+        if (s->align_mem) {
+            mop |= (size == 2 ? MO_ALIGN_4 : MO_ALIGN_8);
+        }
+        mop = finalize_memop_pair(s, mop);
 
         if (is_load) {
-            TCGv_i64 tmp = tcg_temp_new_i64();
+            if (size == 2) {
+                int o2 = s->be_data == MO_LE ? 32 : 0;
+                int o1 = o2 ^ 32;
 
-            /* Do not modify tcg_rt before recognizing any exception
-             * from the second load.
-             */
-            do_gpr_ld(s, tmp, clean_addr, size + is_signed * MO_SIGN,
-                      false, false, 0, false, false);
-            tcg_gen_addi_i64(clean_addr, clean_addr, 1 << size);
-            do_gpr_ld(s, tcg_rt2, clean_addr, size + is_signed * MO_SIGN,
-                      false, false, 0, false, false);
+                tcg_gen_qemu_ld_i64(tcg_rt, clean_addr, get_mem_index(s), mop);
+                if (is_signed) {
+                    tcg_gen_sextract_i64(tcg_rt2, tcg_rt, o2, 32);
+                    tcg_gen_sextract_i64(tcg_rt, tcg_rt, o1, 32);
+                } else {
+                    tcg_gen_extract_i64(tcg_rt2, tcg_rt, o2, 32);
+                    tcg_gen_extract_i64(tcg_rt, tcg_rt, o1, 32);
+                }
+            } else {
+                TCGv_i128 tmp = tcg_temp_new_i128();
 
-            tcg_gen_mov_i64(tcg_rt, tmp);
+                tcg_gen_qemu_ld_i128(tmp, clean_addr, get_mem_index(s), mop);
+                if (s->be_data == MO_LE) {
+                    tcg_gen_extr_i128_i64(tcg_rt, tcg_rt2, tmp);
+                } else {
+                    tcg_gen_extr_i128_i64(tcg_rt2, tcg_rt, tmp);
+                }
+            }
         } else {
-            do_gpr_st(s, tcg_rt, clean_addr, size,
-                      false, 0, false, false);
-            tcg_gen_addi_i64(clean_addr, clean_addr, 1 << size);
-            do_gpr_st(s, tcg_rt2, clean_addr, size,
-                      false, 0, false, false);
+            if (size == 2) {
+                TCGv_i64 tmp = tcg_temp_new_i64();
+
+                if (s->be_data == MO_LE) {
+                    tcg_gen_concat32_i64(tmp, tcg_rt, tcg_rt2);
+                } else {
+                    tcg_gen_concat32_i64(tmp, tcg_rt2, tcg_rt);
+                }
+                tcg_gen_qemu_st_i64(tmp, clean_addr, get_mem_index(s), mop);
+            } else {
+                TCGv_i128 tmp = tcg_temp_new_i128();
+
+                if (s->be_data == MO_LE) {
+                    tcg_gen_concat_i64_i128(tmp, tcg_rt, tcg_rt2);
+                } else {
+                    tcg_gen_concat_i64_i128(tmp, tcg_rt2, tcg_rt);
+                }
+                tcg_gen_qemu_st_i128(tmp, clean_addr, get_mem_index(s), mop);
+            }
         }
     }