[PATCH v3 28/44] target/arm: Implement MVE VQDMULH and VQRDMULH (scalar)

[Peter Maydell <peter.maydell@linaro.org>]

Implement the MVE VQDMULH and VQRDMULH scalar insns, which multiply
elements by the scalar, double, possibly round, take the high half
and saturate.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper-mve.h    |  8 ++++++++
 target/arm/mve.decode      |  3 +++
 target/arm/mve_helper.c    | 25 +++++++++++++++++++++++++
 target/arm/translate-mve.c |  2 ++
 4 files changed, 38 insertions(+)

diff --git a/target/arm/helper-mve.h b/target/arm/helper-mve.h
index 092efdab475..a0a01d0cc3b 100644
--- a/target/arm/helper-mve.h
+++ b/target/arm/helper-mve.h
@@ -189,6 +189,14 @@ DEF_HELPER_FLAGS_4(mve_vqsubu_scalarb, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(mve_vqsubu_scalarh, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(mve_vqsubu_scalarw, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
 
+DEF_HELPER_FLAGS_4(mve_vqdmulh_scalarb, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(mve_vqdmulh_scalarh, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(mve_vqdmulh_scalarw, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
+
+DEF_HELPER_FLAGS_4(mve_vqrdmulh_scalarb, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(mve_vqrdmulh_scalarh, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
+DEF_HELPER_FLAGS_4(mve_vqrdmulh_scalarw, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
+
 DEF_HELPER_FLAGS_4(mve_vbrsrb, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(mve_vbrsrh, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
 DEF_HELPER_FLAGS_4(mve_vbrsrw, TCG_CALL_NO_WG, void, env, ptr, ptr, i32)
diff --git a/target/arm/mve.decode b/target/arm/mve.decode
index c85227c675a..47ce6ebb83b 100644
--- a/target/arm/mve.decode
+++ b/target/arm/mve.decode
@@ -174,6 +174,9 @@ VQSUB_S_scalar   1110 1110 0 . .. ... 0 ... 1 1111 . 110 .... @2scalar
 VQSUB_U_scalar   1111 1110 0 . .. ... 0 ... 1 1111 . 110 .... @2scalar
 VBRSR            1111 1110 0 . .. ... 1 ... 1 1110 . 110 .... @2scalar
 
+VQDMULH_scalar   1110 1110 0 . .. ... 1 ... 0 1110 . 110 .... @2scalar
+VQRDMULH_scalar  1111 1110 0 . .. ... 1 ... 0 1110 . 110 .... @2scalar
+
 # Predicate operations
 %mask_22_13      22:1 13:3
 VPST             1111 1110 0 . 11 000 1 ... 0 1111 0100 1101 mask=%mask_22_13
diff --git a/target/arm/mve_helper.c b/target/arm/mve_helper.c
index 0ae2aeadbb2..0e168ab47f2 100644
--- a/target/arm/mve_helper.c
+++ b/target/arm/mve_helper.c
@@ -533,6 +533,24 @@ static inline int32_t do_sat_bhw(int64_t val, int64_t min, int64_t max, bool *s)
 #define DO_UQSUB_H(n, m, s) do_sat_bhw((int64_t)n - m, 0, UINT16_MAX, s)
 #define DO_UQSUB_W(n, m, s) do_sat_bhw((int64_t)n - m, 0, UINT32_MAX, s)
 
+/*
+ * For QDMULH and QRDMULH we simplify "double and shift by esize" into
+ * "shift by esize-1", adjusting the QRDMULH rounding constant to match.
+ */
+#define DO_QDMULH_B(n, m, s) do_sat_bhw(((int64_t)n * m) >> 7, \
+                                        INT8_MIN, INT8_MAX, s)
+#define DO_QDMULH_H(n, m, s) do_sat_bhw(((int64_t)n * m) >> 15, \
+                                        INT16_MIN, INT16_MAX, s)
+#define DO_QDMULH_W(n, m, s) do_sat_bhw(((int64_t)n * m) >> 31, \
+                                        INT32_MIN, INT32_MAX, s)
+
+#define DO_QRDMULH_B(n, m, s) do_sat_bhw(((int64_t)n * m + (1 << 6)) >> 7, \
+                                         INT8_MIN, INT8_MAX, s)
+#define DO_QRDMULH_H(n, m, s) do_sat_bhw(((int64_t)n * m + (1 << 14)) >> 15, \
+                                         INT16_MIN, INT16_MAX, s)
+#define DO_QRDMULH_W(n, m, s) do_sat_bhw(((int64_t)n * m + (1 << 30)) >> 31, \
+                                         INT32_MIN, INT32_MAX, s)
+
 #define DO_2OP_SCALAR(OP, ESIZE, TYPE, FN)                              \
     void HELPER(glue(mve_, OP))(CPUARMState *env, void *vd, void *vn,   \
                                 uint32_t rm)                            \
@@ -600,6 +618,13 @@ DO_2OP_SAT_SCALAR(vqsubs_scalarb, 1, int8_t, DO_SQSUB_B)
 DO_2OP_SAT_SCALAR(vqsubs_scalarh, 2, int16_t, DO_SQSUB_H)
 DO_2OP_SAT_SCALAR(vqsubs_scalarw, 4, int32_t, DO_SQSUB_W)
 
+DO_2OP_SAT_SCALAR(vqdmulh_scalarb, 1, int8_t, DO_QDMULH_B)
+DO_2OP_SAT_SCALAR(vqdmulh_scalarh, 2, int16_t, DO_QDMULH_H)
+DO_2OP_SAT_SCALAR(vqdmulh_scalarw, 4, int32_t, DO_QDMULH_W)
+DO_2OP_SAT_SCALAR(vqrdmulh_scalarb, 1, int8_t, DO_QRDMULH_B)
+DO_2OP_SAT_SCALAR(vqrdmulh_scalarh, 2, int16_t, DO_QRDMULH_H)
+DO_2OP_SAT_SCALAR(vqrdmulh_scalarw, 4, int32_t, DO_QRDMULH_W)
+
 static inline uint32_t do_vbrsrb(uint32_t n, uint32_t m)
 {
     m &= 0xff;
diff --git a/target/arm/translate-mve.c b/target/arm/translate-mve.c
index 27c69d9c7dd..84a7320cf80 100644
--- a/target/arm/translate-mve.c
+++ b/target/arm/translate-mve.c
@@ -450,6 +450,8 @@ DO_2OP_SCALAR(VQADD_S_scalar, vqadds_scalar)
 DO_2OP_SCALAR(VQADD_U_scalar, vqaddu_scalar)
 DO_2OP_SCALAR(VQSUB_S_scalar, vqsubs_scalar)
 DO_2OP_SCALAR(VQSUB_U_scalar, vqsubu_scalar)
+DO_2OP_SCALAR(VQDMULH_scalar, vqdmulh_scalar)
+DO_2OP_SCALAR(VQRDMULH_scalar, vqrdmulh_scalar)
 DO_2OP_SCALAR(VBRSR, vbrsr)
 
 static bool do_long_dual_acc(DisasContext *s, arg_vmlaldav *a,
-- 
2.20.1