risc-v: fix RVV backend on clang with undefined CV_RVV_SCALABLE

- v_interleave_quads
- v_pack_triplets
- v_signmask

diff --git a/modules/core/include/opencv2/core/hal/intrin_rvv.hpp b/modules/core/include/opencv2/core/hal/intrin_rvv.hpp
index a592976827c7beafc3b42a8b6581ea8896ef5230..dca54a27d1e557471df083dd45f41faf60f2ce10 100644 (file)
--- a/modules/core/include/opencv2/core/hal/intrin_rvv.hpp
+++ b/modules/core/include/opencv2/core/hal/intrin_rvv.hpp
@@ -2646,14 +2646,14 @@ inline v_##_Tpvec v_interleave_quads(const v_##_Tpvec& vec) \
     v_store(ptrvec, vec); \
     for (int i = 0; i < v_##_Tpvec::nlanes/8; i++) \
     { \
-        ptr[8*i  ] = ptrvec[4*i  ]; \
-        ptr[8*i+1] = ptrvec[4*i+4]; \
-        ptr[8*i+2] = ptrvec[4*i+1]; \
-        ptr[8*i+3] = ptrvec[4*i+5]; \
-        ptr[8*i+4] = ptrvec[4*i+2]; \
-        ptr[8*i+5] = ptrvec[4*i+6]; \
-        ptr[8*i+6] = ptrvec[4*i+3]; \
-        ptr[8*i+7] = ptrvec[4*i+7]; \
+        ptr[8*i  ] = ptrvec[8*i  ]; \
+        ptr[8*i+1] = ptrvec[8*i+4]; \
+        ptr[8*i+2] = ptrvec[8*i+1]; \
+        ptr[8*i+3] = ptrvec[8*i+5]; \
+        ptr[8*i+4] = ptrvec[8*i+2]; \
+        ptr[8*i+5] = ptrvec[8*i+6]; \
+        ptr[8*i+6] = ptrvec[8*i+3]; \
+        ptr[8*i+7] = ptrvec[8*i+7]; \
     } \
     return v_load(ptr); \
 }
@@ -2753,7 +2753,7 @@ inline int v_signmask(const _Tpvec& a) \
 { \
     uint8_t ans[16] = {0};\
     vsm(ans, vmslt(a, 0, vl), vl);\
-    return reinterpret_cast<int*>(ans)[0];\
+    return reinterpret_cast<int*>(ans)[0] & ((1 << (vl)) - 1);\
 }
 
 OPENCV_HAL_IMPL_RVV_SIGNMASK_OP(v_int8x16, 8, 16)
@@ -2810,7 +2810,7 @@ OPENCV_HAL_IMPL_RVV_SCAN_FORWOARD_OP(v_float64x2, double, f64)
 #ifndef __clang__
 inline v_int8x16 v_pack_triplets(const v_int8x16& vec)
 {
-    uint64 ptr[2] = {0x0908060504020100, 0xFFFFFFFF0E0D0C0A};
+    uint64 ptr[2] = {0x0908060504020100, 0xFFFFFF0F0E0D0C0A};
     return v_int8x16((vint8m1_t)vrgather_vv_u8m1((vuint8m1_t)vint8m1_t(vec), (vuint8m1_t)vle64_v_u64m1(ptr, 2), 16));
 }
 inline v_uint8x16 v_pack_triplets(const v_uint8x16& vec)
@@ -2820,7 +2820,7 @@ inline v_uint8x16 v_pack_triplets(const v_uint8x16& vec)
 
 inline v_int16x8 v_pack_triplets(const v_int16x8& vec)
 {
-    uint64 ptr[2] = {0x0908060504020100, 0xFFFFFFFF0E0D0C0A};
+    uint64 ptr[2] = {0x0908050403020100, 0xFFFF0F0E0D0C0B0A};
     return v_int16x8((vint16m1_t)vrgather_vv_u8m1((vuint8m1_t)vint16m1_t(vec), (vuint8m1_t)vle64_v_u64m1(ptr, 2), 16));
 }
 inline v_uint16x8 v_pack_triplets(const v_uint16x8& vec)
@@ -2836,7 +2836,7 @@ inline v_float32x4 v_pack_triplets(const v_float32x4& vec) { return vec; }
 
 inline v_int8x16 v_pack_triplets(const v_int8x16& vec)
 {
-    uint64 ptr[2] = {0x0908060504020100, 0xFFFFFFFF0E0D0C0A};
+    uint64 ptr[2] = {0x0908060504020100, 0xFFFFFF0F0E0D0C0A};
     return v_int8x16(vreinterpret_i8m1(vrgather_vv_u8m1(v_reinterpret_as_u8(vec), vreinterpret_u8m1(vle64_v_u64m1(ptr, 2)), 16)));
 }
 inline v_uint8x16 v_pack_triplets(const v_uint8x16& vec)
@@ -2846,7 +2846,7 @@ inline v_uint8x16 v_pack_triplets(const v_uint8x16& vec)
 
 inline v_int16x8 v_pack_triplets(const v_int16x8& vec)
 {
-    uint64 ptr[2] = {0x0908060504020100, 0xFFFFFFFF0E0D0C0A};
+    uint64 ptr[2] = {0x0908050403020100, 0xFFFF0F0E0D0C0B0A};
     return v_int16x8(v_reinterpret_as_s16(v_uint8x16(vrgather_vv_u8m1(v_reinterpret_as_u8(vec), vreinterpret_u8m1(vle64_v_u64m1(ptr, 2)), 16))));
 }
 inline v_uint16x8 v_pack_triplets(const v_uint16x8& vec)