vppinfra: more vector inlines

Type: improvement
Change-Id: Ie0de374b89ec3a17befecf3f08e94951597609ec
Signed-off-by: Damjan Marion <damarion@cisco.com>
diff --git a/src/vppinfra/vector_sse42.h b/src/vppinfra/vector_sse42.h
index 918ded3..8c28dd7 100644
--- a/src/vppinfra/vector_sse42.h
+++ b/src/vppinfra/vector_sse42.h
@@ -650,6 +650,14 @@
   return (u32x4) _mm_hadd_epi32 ((__m128i) v1, (__m128i) v2);
 }
 
+static_always_inline u32 __clib_unused
+u32x4_sum_elts (u32x4 sum4)
+{
+  sum4 += (u32x4) u8x16_align_right (sum4, sum4, 8);
+  sum4 += (u32x4) u8x16_align_right (sum4, sum4, 4);
+  return sum4[0];
+}
+
 static_always_inline u8x16
 u8x16_shuffle (u8x16 v, u8x16 m)
 {
@@ -756,6 +764,14 @@
   return a ^ b ^ c;
 }
 
+#ifdef __AVX512F__
+static_always_inline u8x16
+u8x16_mask_load (u8x16 a, void *p, u16 mask)
+{
+  return (u8x16) _mm_mask_loadu_epi8 ((__m128i) a, mask, p);
+}
+#endif
+
 #endif /* included_vector_sse2_h */
 
 /*