diff --git a/.travis.yml b/.travis.yml
index c2c0c84..9f2a546 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -57,3 +57,6 @@ after_success:
         --gcov-options '\-lp'
         --verbose;
     fi
+
+after_failure:
+  - cat ./test-suite.log
diff --git a/include/cglm/quat.h b/include/cglm/quat.h
index f5f29af..2ebc8b3 100644
--- a/include/cglm/quat.h
+++ b/include/cglm/quat.h
@@ -233,7 +233,7 @@ glm_quat_normalize_to(versor q, versor dest) {
   dot = glm_vec4_norm2(q);
 
   if (dot <= 0.0f) {
-    glm_quat_identity(q);
+    glm_quat_identity(dest);
     return;
   }
 
diff --git a/include/cglm/vec4.h b/include/cglm/vec4.h
index b94cf61..d053ed5 100644
--- a/include/cglm/vec4.h
+++ b/include/cglm/vec4.h
@@ -690,18 +690,9 @@ CGLM_INLINE
 float
 glm_vec4_distance(vec4 a, vec4 b) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  __m128 x0;
-  x0 = _mm_sub_ps(glmm_load(b), glmm_load(a));
-  x0 = _mm_mul_ps(x0, x0);
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 3, 2));
-  return _mm_cvtss_f32(_mm_sqrt_ss(_mm_add_ss(x0,
-                                              glmm_shuff1(x0, 0, 1, 0, 1))));
+  return glmm_norm(_mm_sub_ps(glmm_load(b), glmm_load(a)));
 #elif defined(CGLM_NEON_FP)
-  float32x4_t v0;
-  float32_t   r;
-  v0 = vsubq_f32(vld1q_f32(a), vld1q_f32(b));
-  r  = vaddvq_f32(vmulq_f32(v0, v0));
-  return sqrtf(r);
+  return glmm_norm(vsubq_f32(glmm_load(a), glmm_load(b)));
 #else
   return sqrtf(glm_pow2(b[0] - a[0])
              + glm_pow2(b[1] - a[1])
diff --git a/post-build.sh b/post-build.sh
index 9861ac4..6fa6e7a 100644
--- a/post-build.sh
+++ b/post-build.sh
@@ -8,12 +8,17 @@
 
 cd $(dirname "$0")
 
-mkdir -p .libs
+mkdir -p "$(pwd)/.libs"
+
+libmocka_folder=$(pwd)/test/lib/cmocka/build/src/
 
 if [ "$(uname)" = "Darwin" ]; then
-  ln -sf $(pwd)/test/lib/cmocka/build/src/libcmocka.0.dylib \
-      .libs/libcmocka.0.dylib;
+  libcmocka=libcmocka.0.dylib
 else
-  ln -sf $(pwd)/test/lib/cmocka/build/src/libcmocka.so.0 \
-      .libs/libcmocka.so.0;
+  libcmocka=libcmocka.so.0
+fi
+
+libcmocka_path="$libmocka_folder$libcmocka"
+if [ -f "$libcmocka_path" ]; then
+  ln -sf "$libcmocka_path" "$(pwd)/.libs/$libcmocka";
 fi
diff --git a/test/src/test_bezier.c b/test/src/test_bezier.c
index bd7c796..1c7bb11 100644
--- a/test/src/test_bezier.c
+++ b/test/src/test_bezier.c
@@ -51,8 +51,8 @@ test_bezier(void **state) {
   Bs_plain = test_bezier_plain(s, p0, c0, c1, p1);
 
   assert_true(glm_eq(Bs,  Bs_plain));
-  assert_true(glm_eq(smc, Bs_plain));
-  assert_true(glm_eq(Bs,  smc));
+  test_assert_eqf(smc, Bs_plain);
+  test_assert_eqf(Bs,  smc);
 
   /* test cubic hermite */
   smc      = glm_smc(s, GLM_HERMITE_MAT, (vec4){p0, p1, c0, c1});
diff --git a/test/src/test_common.c b/test/src/test_common.c
index f835235..02ccaec 100644
--- a/test/src/test_common.c
+++ b/test/src/test_common.c
@@ -5,6 +5,7 @@
 
 #include "test_common.h"
 #include <stdlib.h>
+#include <math.h>
 
 #define m 4
 #define n 4
diff --git a/test/src/test_vec4.c b/test/src/test_vec4.c
index 5fe3be8..02137b0 100644
--- a/test/src/test_vec4.c
+++ b/test/src/test_vec4.c
@@ -93,6 +93,13 @@ test_vec4(void **state) {
     /* 3. test SIMD norm2 */
     test_rand_vec4(v);
     test_assert_eqf(test_vec4_norm2(v), glm_vec4_norm2(v));
+
+    /* 4. test SSE/SIMD distance */
+    test_rand_vec4(v1);
+    test_rand_vec4(v2);
+    d1 = glm_vec4_distance(v1, v2);
+    d2 = sqrtf(powf(v1[0]-v2[0], 2.0f) + pow(v1[1]-v2[1], 2.0f) + pow(v1[2]-v2[2], 2.0f) + pow(v1[3]-v2[3], 2.0f));
+    assert_true(fabsf(d1 - d2) <= 0.000009);
   }
 
   /* test zero */