[PATCH 6/6] target/arm: Split helper_crypto_sm3tt

[Richard Henderson <richard.henderson@linaro.org>]

Rather than passing an opcode to a helper, fully decode the
operation at translate time.  Use clear_tail_16 to zap the
balance of the SVE register with the AdvSIMD write.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper.h        |  5 ++++-
 target/arm/crypto_helper.c | 24 ++++++++++++++++++------
 target/arm/translate-a64.c | 21 +++++----------------
 3 files changed, 27 insertions(+), 23 deletions(-)

diff --git a/target/arm/helper.h b/target/arm/helper.h
index 13475ecf81..2a20c8174c 100644
--- a/target/arm/helper.h
+++ b/target/arm/helper.h
@@ -531,7 +531,10 @@ DEF_HELPER_FLAGS_3(crypto_sha512su0, TCG_CALL_NO_RWG, void, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(crypto_sha512su1, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, i32)
 
-DEF_HELPER_FLAGS_5(crypto_sm3tt, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32, i32)
+DEF_HELPER_FLAGS_4(crypto_sm3tt1a, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(crypto_sm3tt1b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(crypto_sm3tt2a, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(crypto_sm3tt2b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(crypto_sm3partw1, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(crypto_sm3partw2, TCG_CALL_NO_RWG,
diff --git a/target/arm/crypto_helper.c b/target/arm/crypto_helper.c
index 636683d0f1..c76806dc8d 100644
--- a/target/arm/crypto_helper.c
+++ b/target/arm/crypto_helper.c
@@ -632,15 +632,14 @@ void HELPER(crypto_sm3partw2)(void *vd, void *vn, void *vm, uint32_t desc)
     clear_tail_16(vd, desc);
 }
 
-void HELPER(crypto_sm3tt)(void *vd, void *vn, void *vm, uint32_t imm2,
-                          uint32_t opcode)
+static inline void QEMU_ALWAYS_INLINE
+crypto_sm3tt(uint64_t *rd, uint64_t *rn, uint64_t *rm,
+             uint32_t desc, uint32_t opcode)
 {
-    uint64_t *rd = vd;
-    uint64_t *rn = vn;
-    uint64_t *rm = vm;
     union CRYPTO_STATE d = { .l = { rd[0], rd[1] } };
     union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
     union CRYPTO_STATE m = { .l = { rm[0], rm[1] } };
+    uint32_t imm2 = simd_data(desc);
     uint32_t t;
 
     assert(imm2 < 4);
@@ -655,7 +654,7 @@ void HELPER(crypto_sm3tt)(void *vd, void *vn, void *vm, uint32_t imm2,
         /* SM3TT2B */
         t = cho(CR_ST_WORD(d, 3), CR_ST_WORD(d, 2), CR_ST_WORD(d, 1));
     } else {
-        g_assert_not_reached();
+        qemu_build_not_reached();
     }
 
     t += CR_ST_WORD(d, 0) + CR_ST_WORD(m, imm2);
@@ -680,8 +679,21 @@ void HELPER(crypto_sm3tt)(void *vd, void *vn, void *vm, uint32_t imm2,
 
     rd[0] = d.l[0];
     rd[1] = d.l[1];
+
+    clear_tail_16(rd, desc);
 }
 
+#define DO_SM3TT(NAME, OPCODE) \
+    void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \
+    { crypto_sm3tt(vd, vn, vm, desc, OPCODE); }
+
+DO_SM3TT(crypto_sm3tt1a, 0)
+DO_SM3TT(crypto_sm3tt1b, 1)
+DO_SM3TT(crypto_sm3tt2a, 2)
+DO_SM3TT(crypto_sm3tt2b, 3)
+
+#undef DO_SM3TT
+
 static uint8_t const sm4_sbox[] = {
     0xd6, 0x90, 0xe9, 0xfe, 0xcc, 0xe1, 0x3d, 0xb7,
     0x16, 0xb6, 0x14, 0xc2, 0x28, 0xfb, 0x2c, 0x05,
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 49ca7ac76e..9c1ebcc8e3 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -13861,13 +13861,15 @@ static void disas_crypto_xar(DisasContext *s, uint32_t insn)
  */
 static void disas_crypto_three_reg_imm2(DisasContext *s, uint32_t insn)
 {
+    static gen_helper_gvec_3 * const fns[4] = {
+        gen_helper_crypto_sm3tt1a, gen_helper_crypto_sm3tt1b,
+        gen_helper_crypto_sm3tt2a, gen_helper_crypto_sm3tt2b,
+    };
     int opcode = extract32(insn, 10, 2);
     int imm2 = extract32(insn, 12, 2);
     int rm = extract32(insn, 16, 5);
     int rn = extract32(insn, 5, 5);
     int rd = extract32(insn, 0, 5);
-    TCGv_ptr tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr;
-    TCGv_i32 tcg_imm2, tcg_opcode;
 
     if (!dc_isar_feature(aa64_sm3, s)) {
         unallocated_encoding(s);
@@ -13878,20 +13880,7 @@ static void disas_crypto_three_reg_imm2(DisasContext *s, uint32_t insn)
         return;
     }
 
-    tcg_rd_ptr = vec_full_reg_ptr(s, rd);
-    tcg_rn_ptr = vec_full_reg_ptr(s, rn);
-    tcg_rm_ptr = vec_full_reg_ptr(s, rm);
-    tcg_imm2   = tcg_const_i32(imm2);
-    tcg_opcode = tcg_const_i32(opcode);
-
-    gen_helper_crypto_sm3tt(tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr, tcg_imm2,
-                            tcg_opcode);
-
-    tcg_temp_free_ptr(tcg_rd_ptr);
-    tcg_temp_free_ptr(tcg_rn_ptr);
-    tcg_temp_free_ptr(tcg_rm_ptr);
-    tcg_temp_free_i32(tcg_imm2);
-    tcg_temp_free_i32(tcg_opcode);
+    gen_gvec_op3_ool(s, true, rd, rn, rm, imm2, fns[opcode]);
 }
 
 /* C3.6 Data processing - SIMD, inc Crypto
-- 
2.20.1