target/arm: Implement MVE FP max/min across vector

Implement the MVE VMAXNMV, VMINNMV, VMAXNMAV, VMINNMAV insns.  These
calculate the maximum or minimum of floating point elements across a
vector, starting with a value in a general purpose register and
returning the result there.

The pseudocode silences a possible SNaN in the accumulating result
on every iteration (by calling FPConvertNaN), but we do it only
on the input ra, because if none of the inputs to float*_maxnum
or float*_minnum are SNaNs then the result can't be an SNaN.

Note that we can't use the float*_maxnuma() etc functions we defined
earlier for VMAXNMA and VMINNMA, because we mustn't take the absolute
value of the starting general-purpose register value, which could be
negative.

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

4 files changed, 102 insertions, 6 deletions

diff --git a/target/arm/helper-mve.h b/target/arm/helper-mve.h
index cb7b642..47fd18d 100644
--- a/target/arm/helper-mve.h
+++ b/target/arm/helper-mve.h
@@ -614,6 +614,18 @@ DEF_HELPER_FLAGS_3(mve_vminavb, TCG_CALL_NO_WG, i32, env, ptr, i32)
 DEF_HELPER_FLAGS_3(mve_vminavh, TCG_CALL_NO_WG, i32, env, ptr, i32)
 DEF_HELPER_FLAGS_3(mve_vminavw, TCG_CALL_NO_WG, i32, env, ptr, i32)
 
+DEF_HELPER_FLAGS_3(mve_vmaxnmvh, TCG_CALL_NO_WG, i32, env, ptr, i32)
+DEF_HELPER_FLAGS_3(mve_vmaxnmvs, TCG_CALL_NO_WG, i32, env, ptr, i32)
+
+DEF_HELPER_FLAGS_3(mve_vminnmvh, TCG_CALL_NO_WG, i32, env, ptr, i32)
+DEF_HELPER_FLAGS_3(mve_vminnmvs, TCG_CALL_NO_WG, i32, env, ptr, i32)
+
+DEF_HELPER_FLAGS_3(mve_vmaxnmavh, TCG_CALL_NO_WG, i32, env, ptr, i32)
+DEF_HELPER_FLAGS_3(mve_vmaxnmavs, TCG_CALL_NO_WG, i32, env, ptr, i32)
+
+DEF_HELPER_FLAGS_3(mve_vminnmavh, TCG_CALL_NO_WG, i32, env, ptr, i32)
+DEF_HELPER_FLAGS_3(mve_vminnmavs, TCG_CALL_NO_WG, i32, env, ptr, i32)
+
 DEF_HELPER_FLAGS_3(mve_vaddlv_s, TCG_CALL_NO_WG, i64, env, ptr, i64)
 DEF_HELPER_FLAGS_3(mve_vaddlv_u, TCG_CALL_NO_WG, i64, env, ptr, i64)
 
diff --git a/target/arm/mve.decode b/target/arm/mve.decode
index d2bd681..a46372f 100644
--- a/target/arm/mve.decode
+++ b/target/arm/mve.decode
@@ -137,6 +137,10 @@
 @vmaxnma  .... .... .... .... .... .... .... .... &2op \
           qd=%qd qn=%qd qm=%qm
 
+# Here also we don't decode the bit 28 size in the format to avoid
+# awkward nested overlap groups
+@vmaxnmv          .... .... .... .... rda:4 .... .... .... &vmaxv qm=%qm
+
 @2op_fp_scalar .... .... .... .... .... .... .... rm:4 &2scalar \
                qd=%qd qn=%qn size=%2op_fp_scalar_size
 
@@ -440,17 +444,33 @@ VMLADAV_S        1110 1110 1111  ... 0 ... . 1111 . 0 . 0 ... 1 @vmladav_nosz
 VMLADAV_U        1111 1110 1111  ... 0 ... . 1111 . 0 . 0 ... 1 @vmladav_nosz
 
 {
-  VMAXV_S        1110 1110 1110  .. 10 ....  1111 0 0 . 0 ... 0 @vmaxv
-  VMINV_S        1110 1110 1110  .. 10 ....  1111 1 0 . 0 ... 0 @vmaxv
-  VMAXAV         1110 1110 1110  .. 00 ....  1111 0 0 . 0 ... 0 @vmaxv
-  VMINAV         1110 1110 1110  .. 00 ....  1111 1 0 . 0 ... 0 @vmaxv
+  [
+    VMAXNMAV     1110 1110 1110  11 00 ....  1111 0 0 . 0 ... 0 @vmaxnmv size=2
+    VMINNMAV     1110 1110 1110  11 00 ....  1111 1 0 . 0 ... 0 @vmaxnmv size=2
+    VMAXNMV      1110 1110 1110  11 10 ....  1111 0 0 . 0 ... 0 @vmaxnmv size=2
+    VMINNMV      1110 1110 1110  11 10 ....  1111 1 0 . 0 ... 0 @vmaxnmv size=2
+  ]
+  [
+    VMAXV_S      1110 1110 1110  .. 10 ....  1111 0 0 . 0 ... 0 @vmaxv
+    VMINV_S      1110 1110 1110  .. 10 ....  1111 1 0 . 0 ... 0 @vmaxv
+    VMAXAV       1110 1110 1110  .. 00 ....  1111 0 0 . 0 ... 0 @vmaxv
+    VMINAV       1110 1110 1110  .. 00 ....  1111 1 0 . 0 ... 0 @vmaxv
+  ]
   VMLADAV_S      1110 1110 1111  ... 0 ... . 1111 . 0 . 0 ... 0 @vmladav_nosz
   VRMLALDAVH_S   1110 1110 1 ... ... 0 ... . 1111 . 0 . 0 ... 0 @vmlaldav_nosz
 }
 
 {
-  VMAXV_U        1111 1110 1110  .. 10 ....  1111 0 0 . 0 ... 0 @vmaxv
-  VMINV_U        1111 1110 1110  .. 10 ....  1111 1 0 . 0 ... 0 @vmaxv
+  [
+    VMAXNMAV     1111 1110 1110  11 00 ....  1111 0 0 . 0 ... 0 @vmaxnmv size=1
+    VMINNMAV     1111 1110 1110  11 00 ....  1111 1 0 . 0 ... 0 @vmaxnmv size=1
+    VMAXNMV      1111 1110 1110  11 10 ....  1111 0 0 . 0 ... 0 @vmaxnmv size=1
+    VMINNMV      1111 1110 1110  11 10 ....  1111 1 0 . 0 ... 0 @vmaxnmv size=1
+  ]
+  [
+    VMAXV_U      1111 1110 1110  .. 10 ....  1111 0 0 . 0 ... 0 @vmaxv
+    VMINV_U      1111 1110 1110  .. 10 ....  1111 1 0 . 0 ... 0 @vmaxv
+  ]
   VMLADAV_U      1111 1110 1111  ... 0 ... . 1111 . 0 . 0 ... 0 @vmladav_nosz
   VRMLALDAVH_U   1111 1110 1 ... ... 0 ... . 1111 . 0 . 0 ... 0 @vmlaldav_nosz
 }
diff --git a/target/arm/mve_helper.c b/target/arm/mve_helper.c
index 36f0910..52e5a8f 100644
--- a/target/arm/mve_helper.c
+++ b/target/arm/mve_helper.c
@@ -3112,3 +3112,47 @@ DO_2OP_FP_ACC_SCALAR(vfma_scalarh, 2, float16, float16_muladd)
 DO_2OP_FP_ACC_SCALAR(vfma_scalars, 4, float32, float32_muladd)
 DO_2OP_FP_ACC_SCALAR(vfmas_scalarh, 2, float16, DO_VFMAS_SCALARH)
 DO_2OP_FP_ACC_SCALAR(vfmas_scalars, 4, float32, DO_VFMAS_SCALARS)
+
+/* Floating point max/min across vector. */
+#define DO_FP_VMAXMINV(OP, ESIZE, TYPE, ABS, FN)                \
+    uint32_t HELPER(glue(mve_, OP))(CPUARMState *env, void *vm, \
+                                    uint32_t ra_in)             \
+    {                                                           \
+        uint16_t mask = mve_element_mask(env);                  \
+        unsigned e;                                             \
+        TYPE *m = vm;                                           \
+        TYPE ra = (TYPE)ra_in;                                  \
+        float_status *fpst = (ESIZE == 2) ?                     \
+            &env->vfp.standard_fp_status_f16 :                  \
+            &env->vfp.standard_fp_status;                       \
+        for (e = 0; e < 16 / ESIZE; e++, mask >>= ESIZE) {      \
+            if (mask & 1) {                                     \
+                TYPE v = m[H##ESIZE(e)];                        \
+                if (TYPE##_is_signaling_nan(ra, fpst)) {        \
+                    ra = TYPE##_silence_nan(ra, fpst);          \
+                    float_raise(float_flag_invalid, fpst);      \
+                }                                               \
+                if (TYPE##_is_signaling_nan(v, fpst)) {         \
+                    v = TYPE##_silence_nan(v, fpst);            \
+                    float_raise(float_flag_invalid, fpst);      \
+                }                                               \
+                if (ABS) {                                      \
+                    v = TYPE##_abs(v);                          \
+                }                                               \
+                ra = FN(ra, v, fpst);                           \
+            }                                                   \
+        }                                                       \
+        mve_advance_vpt(env);                                   \
+        return ra;                                              \
+    }                                                           \
+
+#define NOP(X) (X)
+
+DO_FP_VMAXMINV(vmaxnmvh, 2, float16, false, float16_maxnum)
+DO_FP_VMAXMINV(vmaxnmvs, 4, float32, false, float32_maxnum)
+DO_FP_VMAXMINV(vminnmvh, 2, float16, false, float16_minnum)
+DO_FP_VMAXMINV(vminnmvs, 4, float32, false, float32_minnum)
+DO_FP_VMAXMINV(vmaxnmavh, 2, float16, true, float16_maxnum)
+DO_FP_VMAXMINV(vmaxnmavs, 4, float32, true, float32_maxnum)
+DO_FP_VMAXMINV(vminnmavh, 2, float16, true, float16_minnum)
+DO_FP_VMAXMINV(vminnmavs, 4, float32, true, float32_minnum)
diff --git a/target/arm/translate-mve.c b/target/arm/translate-mve.c
index 3627ba2..4e2aa2c 100644
--- a/target/arm/translate-mve.c
+++ b/target/arm/translate-mve.c
@@ -1806,6 +1806,26 @@ DO_VMAXV(VMINV_S, vminvs)
 DO_VMAXV(VMINV_U, vminvu)
 DO_VMAXV(VMINAV, vminav)
 
+#define DO_VMAXV_FP(INSN, FN)                                   \
+    static bool trans_##INSN(DisasContext *s, arg_vmaxv *a)     \
+    {                                                           \
+        static MVEGenVADDVFn * const fns[] = {                  \
+            NULL,                                               \
+            gen_helper_mve_##FN##h,                             \
+            gen_helper_mve_##FN##s,                             \
+            NULL,                                               \
+        };                                                      \
+        if (!dc_isar_feature(aa32_mve_fp, s)) {                 \
+            return false;                                       \
+        }                                                       \
+        return do_vmaxv(s, a, fns[a->size]);                    \
+    }
+
+DO_VMAXV_FP(VMAXNMV, vmaxnmv)
+DO_VMAXV_FP(VMINNMV, vminnmv)
+DO_VMAXV_FP(VMAXNMAV, vmaxnmav)
+DO_VMAXV_FP(VMINNMAV, vminnmav)
+
 static bool do_vabav(DisasContext *s, arg_vabav *a, MVEGenVABAVFn *fn)
 {
     /* Absolute difference accumulated across vector */