target/arm: Convert sq{, r}dmulh to gvec for aa64 advsimd

qemu/aarch64.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_aarch64
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_aarch64
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_aarch64
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_aarch64
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_aarch64
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_aarch64
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_aarch64
 #define helper_neon_sub_u16 helper_neon_sub_u16_aarch64
 #define helper_neon_sub_u8 helper_neon_sub_u8_aarch64
 #define helper_neon_subl_u16 helper_neon_subl_u16_aarch64
@@ -3661,6 +3665,10 @@
 #define helper_neon_ceq_f64 helper_neon_ceq_f64_aarch64
 #define helper_neon_cge_f64 helper_neon_cge_f64_aarch64
 #define helper_neon_cgt_f64 helper_neon_cgt_f64_aarch64
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_aarch64
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_aarch64
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_aarch64
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_aarch64
 #define helper_pacda helper_pacda_aarch64
 #define helper_pacdb helper_pacdb_aarch64
 #define helper_pacga helper_pacga_aarch64

qemu/aarch64eb.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_aarch64eb
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_aarch64eb
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_aarch64eb
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_aarch64eb
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_aarch64eb
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_aarch64eb
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_aarch64eb
 #define helper_neon_sub_u16 helper_neon_sub_u16_aarch64eb
 #define helper_neon_sub_u8 helper_neon_sub_u8_aarch64eb
 #define helper_neon_subl_u16 helper_neon_subl_u16_aarch64eb
@@ -3661,6 +3665,10 @@
 #define helper_neon_ceq_f64 helper_neon_ceq_f64_aarch64eb
 #define helper_neon_cge_f64 helper_neon_cge_f64_aarch64eb
 #define helper_neon_cgt_f64 helper_neon_cgt_f64_aarch64eb
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_aarch64eb
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_aarch64eb
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_aarch64eb
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_aarch64eb
 #define helper_pacda helper_pacda_aarch64eb
 #define helper_pacdb helper_pacdb_aarch64eb
 #define helper_pacga helper_pacga_aarch64eb

qemu/arm.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_arm
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_arm
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_arm
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_arm
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_arm
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_arm
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_arm
 #define helper_neon_sub_u16 helper_neon_sub_u16_arm
 #define helper_neon_sub_u8 helper_neon_sub_u8_arm
 #define helper_neon_subl_u16 helper_neon_subl_u16_arm
@@ -3523,6 +3527,15 @@
 #define gen_ushl_i32 gen_ushl_i32_arm
 #define gen_ushl_i64 gen_ushl_i64_arm
 #define helper_fjcvtzs helper_fjcvtzs_arm
+#define helper_gvec_mla_idx_d helper_gvec_mla_idx_d_arm
+#define helper_gvec_mla_idx_h helper_gvec_mla_idx_h_arm
+#define helper_gvec_mla_idx_s helper_gvec_mla_idx_s_arm
+#define helper_gvec_mls_idx_d helper_gvec_mls_idx_d_arm
+#define helper_gvec_mls_idx_h helper_gvec_mls_idx_h_arm
+#define helper_gvec_mls_idx_s helper_gvec_mls_idx_s_arm
+#define helper_gvec_mul_idx_d helper_gvec_mul_idx_d_arm
+#define helper_gvec_mul_idx_h helper_gvec_mul_idx_h_arm
+#define helper_gvec_mul_idx_s helper_gvec_mul_idx_s_arm
 #define helper_gvec_saba_b helper_gvec_saba_b_arm
 #define helper_gvec_saba_d helper_gvec_saba_d_arm
 #define helper_gvec_saba_h helper_gvec_saba_h_arm

qemu/armeb.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_armeb
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_armeb
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_armeb
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_armeb
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_armeb
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_armeb
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_armeb
 #define helper_neon_sub_u16 helper_neon_sub_u16_armeb
 #define helper_neon_sub_u8 helper_neon_sub_u8_armeb
 #define helper_neon_subl_u16 helper_neon_subl_u16_armeb
@@ -3523,6 +3527,15 @@
 #define gen_ushl_i32 gen_ushl_i32_armeb
 #define gen_ushl_i64 gen_ushl_i64_armeb
 #define helper_fjcvtzs helper_fjcvtzs_armeb
+#define helper_gvec_mla_idx_d helper_gvec_mla_idx_d_armeb
+#define helper_gvec_mla_idx_h helper_gvec_mla_idx_h_armeb
+#define helper_gvec_mla_idx_s helper_gvec_mla_idx_s_armeb
+#define helper_gvec_mls_idx_d helper_gvec_mls_idx_d_armeb
+#define helper_gvec_mls_idx_h helper_gvec_mls_idx_h_armeb
+#define helper_gvec_mls_idx_s helper_gvec_mls_idx_s_armeb
+#define helper_gvec_mul_idx_d helper_gvec_mul_idx_d_armeb
+#define helper_gvec_mul_idx_h helper_gvec_mul_idx_h_armeb
+#define helper_gvec_mul_idx_s helper_gvec_mul_idx_s_armeb
 #define helper_gvec_saba_b helper_gvec_saba_b_armeb
 #define helper_gvec_saba_d helper_gvec_saba_d_armeb
 #define helper_gvec_saba_h helper_gvec_saba_h_armeb

qemu/header_gen.py

@@ -1713,6 +1713,10 @@ symbols = (
     'helper_neon_sqadd_u32',
     'helper_neon_sqadd_u64',
     'helper_neon_sqadd_u8',
+    'helper_neon_sqdmulh_h',
+    'helper_neon_sqdmulh_s',
+    'helper_neon_sqrdmulh_h',
+    'helper_neon_sqrdmulh_s',
     'helper_neon_sub_u16',
     'helper_neon_sub_u8',
     'helper_neon_subl_u16',
@@ -3532,6 +3536,15 @@ arm_symbols = (
     'gen_ushl_i32',
     'gen_ushl_i64',
     'helper_fjcvtzs',
+    'helper_gvec_mla_idx_d',
+    'helper_gvec_mla_idx_h',
+    'helper_gvec_mla_idx_s',
+    'helper_gvec_mls_idx_d',
+    'helper_gvec_mls_idx_h',
+    'helper_gvec_mls_idx_s',
+    'helper_gvec_mul_idx_d',
+    'helper_gvec_mul_idx_h',
+    'helper_gvec_mul_idx_s',
     'helper_gvec_saba_b',
     'helper_gvec_saba_d',
     'helper_gvec_saba_h',
@@ -3801,6 +3814,10 @@ aarch64_symbols = (
     'helper_neon_ceq_f64',
     'helper_neon_cge_f64',
     'helper_neon_cgt_f64',
+    'helper_neon_sqdmulh_h',
+    'helper_neon_sqdmulh_s',
+    'helper_neon_sqrdmulh_h',
+    'helper_neon_sqrdmulh_s',
     'helper_pacda',
     'helper_pacdb',
     'helper_pacga',

qemu/m68k.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_m68k
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_m68k
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_m68k
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_m68k
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_m68k
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_m68k
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_m68k
 #define helper_neon_sub_u16 helper_neon_sub_u16_m68k
 #define helper_neon_sub_u8 helper_neon_sub_u8_m68k
 #define helper_neon_subl_u16 helper_neon_subl_u16_m68k

qemu/mips.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_mips
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_mips
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_mips
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_mips
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_mips
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_mips
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_mips
 #define helper_neon_sub_u16 helper_neon_sub_u16_mips
 #define helper_neon_sub_u8 helper_neon_sub_u8_mips
 #define helper_neon_subl_u16 helper_neon_subl_u16_mips

qemu/mips64.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_mips64
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_mips64
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_mips64
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_mips64
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_mips64
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_mips64
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_mips64
 #define helper_neon_sub_u16 helper_neon_sub_u16_mips64
 #define helper_neon_sub_u8 helper_neon_sub_u8_mips64
 #define helper_neon_subl_u16 helper_neon_subl_u16_mips64

qemu/mips64el.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_mips64el
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_mips64el
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_mips64el
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_mips64el
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_mips64el
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_mips64el
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_mips64el
 #define helper_neon_sub_u16 helper_neon_sub_u16_mips64el
 #define helper_neon_sub_u8 helper_neon_sub_u8_mips64el
 #define helper_neon_subl_u16 helper_neon_subl_u16_mips64el

qemu/mipsel.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_mipsel
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_mipsel
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_mipsel
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_mipsel
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_mipsel
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_mipsel
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_mipsel
 #define helper_neon_sub_u16 helper_neon_sub_u16_mipsel
 #define helper_neon_sub_u8 helper_neon_sub_u8_mipsel
 #define helper_neon_subl_u16 helper_neon_subl_u16_mipsel

qemu/powerpc.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_powerpc
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_powerpc
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_powerpc
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_powerpc
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_powerpc
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_powerpc
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_powerpc
 #define helper_neon_sub_u16 helper_neon_sub_u16_powerpc
 #define helper_neon_sub_u8 helper_neon_sub_u8_powerpc
 #define helper_neon_subl_u16 helper_neon_subl_u16_powerpc

qemu/riscv32.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_riscv32
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_riscv32
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_riscv32
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_riscv32
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_riscv32
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_riscv32
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_riscv32
 #define helper_neon_sub_u16 helper_neon_sub_u16_riscv32
 #define helper_neon_sub_u8 helper_neon_sub_u8_riscv32
 #define helper_neon_subl_u16 helper_neon_subl_u16_riscv32

qemu/riscv64.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_riscv64
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_riscv64
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_riscv64
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_riscv64
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_riscv64
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_riscv64
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_riscv64
 #define helper_neon_sub_u16 helper_neon_sub_u16_riscv64
 #define helper_neon_sub_u8 helper_neon_sub_u8_riscv64
 #define helper_neon_subl_u16 helper_neon_subl_u16_riscv64

qemu/sparc.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_sparc
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_sparc
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_sparc
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_sparc
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_sparc
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_sparc
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_sparc
 #define helper_neon_sub_u16 helper_neon_sub_u16_sparc
 #define helper_neon_sub_u8 helper_neon_sub_u8_sparc
 #define helper_neon_subl_u16 helper_neon_subl_u16_sparc

qemu/sparc64.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_sparc64
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_sparc64
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_sparc64
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_sparc64
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_sparc64
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_sparc64
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_sparc64
 #define helper_neon_sub_u16 helper_neon_sub_u16_sparc64
 #define helper_neon_sub_u8 helper_neon_sub_u8_sparc64
 #define helper_neon_subl_u16 helper_neon_subl_u16_sparc64

qemu/target/arm/helper.h

@@ -774,6 +774,16 @@ DEF_HELPER_FLAGS_5(gvec_mls_idx_s, TCG_CALL_NO_RWG,
 DEF_HELPER_FLAGS_5(gvec_mls_idx_d, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, ptr, i32)
 
+DEF_HELPER_FLAGS_5(neon_sqdmulh_h, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_5(neon_sqdmulh_s, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, i32)
+
+DEF_HELPER_FLAGS_5(neon_sqrdmulh_h, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_5(neon_sqrdmulh_s, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, i32)
+
 #ifdef TARGET_ARM
 #define helper_clz helper_clz_arm
 #define gen_helper_clz gen_helper_clz_arm

qemu/target/arm/translate-a64.c

@@ -851,6 +851,21 @@ static void gen_gvec_op3_fpst(DisasContext *s, bool is_q, int rd, int rn,
     tcg_temp_free_ptr(tcg_ctx, fpst);
 }
 
+/* Expand a 3-operand + qc + operation using an out-of-line helper.  */
+static void gen_gvec_op3_qc(DisasContext *s, bool is_q, int rd, int rn,
+                            int rm, gen_helper_gvec_3_ptr *fn)
+{
+    TCGContext *tcg_ctx = s->uc->tcg_ctx;
+    TCGv_ptr qc_ptr = tcg_temp_new_ptr(tcg_ctx);
+
+    tcg_gen_addi_ptr(tcg_ctx, qc_ptr, tcg_ctx->cpu_env, offsetof(CPUARMState, vfp.qc));
+    tcg_gen_gvec_3_ptr(tcg_ctx, vec_full_reg_offset(s, rd),
+                       vec_full_reg_offset(s, rn),
+                       vec_full_reg_offset(s, rm), qc_ptr,
+                       is_q ? 16 : 8, vec_full_reg_size(s), 0, fn);
+    tcg_temp_free_ptr(tcg_ctx, qc_ptr);
+}
+
 /* Set ZF and NF based on a 64 bit result. This is alas fiddlier
  * than the 32 bit equivalent.
  */
@@ -12033,6 +12048,15 @@ static void disas_simd_3same_int(DisasContext *s, uint32_t insn)
             gen_gvec_fn3(s, is_q, rd, rn, rm, gen_gvec_mla, size);
         }
         return;
+    case 0x16: /* SQDMULH, SQRDMULH */
+        {
+            static gen_helper_gvec_3_ptr * const fns[2][2] = {
+                { gen_helper_neon_sqdmulh_h, gen_helper_neon_sqrdmulh_h },
+                { gen_helper_neon_sqdmulh_s, gen_helper_neon_sqrdmulh_s },
+            };
+            gen_gvec_op3_qc(s, is_q, rd, rn, rm, fns[size - 1][u]);
+        }
+        return;
     case 0x11:
         if (!u) { /* CMTST */
             gen_gvec_fn3(s, is_q, rd, rn, rm, gen_gvec_cmtst, size);
@@ -12144,16 +12168,6 @@ static void disas_simd_3same_int(DisasContext *s, uint32_t insn)
                 genenvfn = fns[size][u];
                 break;
             }
-            case 0x16: /* SQDMULH, SQRDMULH */
-            {
-                static NeonGenTwoOpEnvFn * const fns[2][2] = {
-                    { gen_helper_neon_qdmulh_s16, gen_helper_neon_qrdmulh_s16 },
-                    { gen_helper_neon_qdmulh_s32, gen_helper_neon_qrdmulh_s32 },
-                };
-                assert(size == 1 || size == 2);
-                genenvfn = fns[size - 1][u];
-                break;
-            }
             default:
                 g_assert_not_reached();
             }

qemu/target/arm/vec_helper.c

@@ -110,6 +110,30 @@ void HELPER(gvec_qrdmlsh_s16)(void *vd, void *vn, void *vm,
     clear_tail(d, opr_sz, simd_maxsz(desc));
 }
 
+void HELPER(neon_sqdmulh_h)(void *vd, void *vn, void *vm,
+                            void *vq, uint32_t desc)
+{
+    intptr_t i, opr_sz = simd_oprsz(desc);
+    int16_t *d = vd, *n = vn, *m = vm;
+
+    for (i = 0; i < opr_sz / 2; ++i) {
+        d[i] = do_sqrdmlah_h(n[i], m[i], 0, false, false, vq);
+    }
+    clear_tail(d, opr_sz, simd_maxsz(desc));
+}
+
+void HELPER(neon_sqrdmulh_h)(void *vd, void *vn, void *vm,
+                             void *vq, uint32_t desc)
+{
+    intptr_t i, opr_sz = simd_oprsz(desc);
+    int16_t *d = vd, *n = vn, *m = vm;
+
+    for (i = 0; i < opr_sz / 2; ++i) {
+        d[i] = do_sqrdmlah_h(n[i], m[i], 0, false, true, vq);
+    }
+    clear_tail(d, opr_sz, simd_maxsz(desc));
+}
+
 /* Signed saturating rounding doubling multiply-accumulate high half, 32-bit */
 static int32_t do_sqrdmlah_s(int32_t src1, int32_t src2, int32_t src3,
                              bool neg, bool round, uint32_t *sat)
@@ -173,6 +197,30 @@ void HELPER(gvec_qrdmlsh_s32)(void *vd, void *vn, void *vm,
     clear_tail(d, opr_sz, simd_maxsz(desc));
 }
 
+void HELPER(neon_sqdmulh_s)(void *vd, void *vn, void *vm,
+                            void *vq, uint32_t desc)
+{
+    intptr_t i, opr_sz = simd_oprsz(desc);
+    int32_t *d = vd, *n = vn, *m = vm;
+
+    for (i = 0; i < opr_sz / 4; ++i) {
+        d[i] = do_sqrdmlah_s(n[i], m[i], 0, false, false, vq);
+    }
+    clear_tail(d, opr_sz, simd_maxsz(desc));
+}
+
+void HELPER(neon_sqrdmulh_s)(void *vd, void *vn, void *vm,
+                             void *vq, uint32_t desc)
+{
+    intptr_t i, opr_sz = simd_oprsz(desc);
+    int32_t *d = vd, *n = vn, *m = vm;
+
+    for (i = 0; i < opr_sz / 4; ++i) {
+        d[i] = do_sqrdmlah_s(n[i], m[i], 0, false, true, vq);
+    }
+    clear_tail(d, opr_sz, simd_maxsz(desc));
+}
+
 /* Integer 8 and 16-bit dot-product.
  *
  * Note that for the loops herein, host endianness does not matter

qemu/x86_64.h

@@ -1707,6 +1707,10 @@
 #define helper_neon_sqadd_u32 helper_neon_sqadd_u32_x86_64
 #define helper_neon_sqadd_u64 helper_neon_sqadd_u64_x86_64
 #define helper_neon_sqadd_u8 helper_neon_sqadd_u8_x86_64
+#define helper_neon_sqdmulh_h helper_neon_sqdmulh_h_x86_64
+#define helper_neon_sqdmulh_s helper_neon_sqdmulh_s_x86_64
+#define helper_neon_sqrdmulh_h helper_neon_sqrdmulh_h_x86_64
+#define helper_neon_sqrdmulh_s helper_neon_sqrdmulh_s_x86_64
 #define helper_neon_sub_u16 helper_neon_sub_u16_x86_64
 #define helper_neon_sub_u8 helper_neon_sub_u8_x86_64
 #define helper_neon_subl_u16 helper_neon_subl_u16_x86_64