diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index cd6f20b..5bee326 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -14,7 +14,7 @@ jobs:
   ruby-versions:
     uses: ruby/actions/.github/workflows/ruby_versions.yml@master
     with:
-      engine: cruby
+      engine: cruby-truffleruby
       min_version: 2.5
       versions: '["debug"]'
 
@@ -35,6 +35,7 @@ jobs:
         - { os: windows-latest , ruby: mingw }
         - { os: windows-latest , ruby: mswin }
         exclude:
+        - { os: macos-latest   , ruby: "2.5" }
         - { os: macos-arm-oss  , ruby: "2.5" }
         - { os: windows-latest , ruby: debug }
         - { os: windows-latest , ruby: truffleruby }
diff --git a/CHANGES.md b/CHANGES.md
index 2123976..299819c 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -1,5 +1,20 @@
 # CHANGES
 
+## 3.1.8
+
+* Add missing documents [GH-277]
+
+  **@nobu**
+
+* Fix memory leak in VpAlloc [GH-294] [GH-290]
+
+  Reoprted by **@MaxLap**
+
+## 3.1.7
+
+* Only consists of CI settings and test changes.
+  This release is needed for developing Ruby to run `make test-bundled-gems` with the released version of bigdecimal.
+
 ## 3.1.6
 
 * Add LICENSE file to gem files [GH-282]
diff --git a/bigdecimal.gemspec b/bigdecimal.gemspec
index 6942383..b6ef8fd 100644
--- a/bigdecimal.gemspec
+++ b/bigdecimal.gemspec
@@ -52,4 +52,6 @@ Gem::Specification.new do |s|
   end
 
   s.required_ruby_version = Gem::Requirement.new(">= 2.5.0")
+
+  s.metadata["changelog_uri"] = s.homepage + "/blob/master/CHANGES.md"
 end
diff --git a/ext/bigdecimal/bigdecimal.c b/ext/bigdecimal/bigdecimal.c
index 3e14364..aa2bf21 100644
--- a/ext/bigdecimal/bigdecimal.c
+++ b/ext/bigdecimal/bigdecimal.c
@@ -31,7 +31,7 @@
 #include "bits.h"
 #include "static_assert.h"
 
-#define BIGDECIMAL_VERSION "3.1.7"
+#define BIGDECIMAL_VERSION "3.1.8"
 
 /* #define ENABLE_NUMERIC_STRING */
 
@@ -1780,6 +1780,17 @@ BigDecimal_neg(VALUE self)
     return VpCheckGetValue(c);
 }
 
+/*
+ * call-seq:
+ *   a * b   -> bigdecimal
+ *
+ * Multiply by the specified value.
+ *
+ * The result precision will be the precision of the sum of each precision.
+ *
+ * See BigDecimal#mult.
+ */
+
 static VALUE
 BigDecimal_mult(VALUE self, VALUE r)
 {
@@ -3257,10 +3268,11 @@ BigDecimal_initialize_copy(VALUE self, VALUE other)
     return self;
 }
 
+/* :nodoc: */
 static VALUE
 BigDecimal_clone(VALUE self)
 {
-  return self;
+    return self;
 }
 
 #ifdef HAVE_RB_OPTS_EXCEPTION_P
@@ -3758,6 +3770,12 @@ f_BigDecimal(int argc, VALUE *argv, VALUE self)
     return rb_convert_to_BigDecimal(val, digs, exception);
 }
 
+/*  call-seq:
+ *    BigDecimal.interpret_loosely(string) -> bigdecimal
+ *
+ *  Returns the +BigDecimal+ converted loosely from +string+.
+ */
+
 static VALUE
 BigDecimal_s_interpret_loosely(VALUE klass, VALUE str)
 {
@@ -4238,6 +4256,17 @@ BigDecimal_negative_zero(void)
     return BIGDECIMAL_NEGATIVE_ZERO;
 }
 
+static inline VALUE
+BigDecimal_literal(const char *str)
+{
+    VALUE arg = rb_str_new_cstr(str);
+    VALUE val = f_BigDecimal(1, &arg, rb_cBigDecimal);
+    rb_gc_register_mark_object(val);
+    return val;
+}
+
+#define BIGDECIMAL_LITERAL(var, val) (BIGDECIMAL_ ## var = BigDecimal_literal(#val))
+
 /* Document-class: BigDecimal
  * BigDecimal provides arbitrary-precision floating point decimal arithmetic.
  *
@@ -4394,7 +4423,6 @@ Init_bigdecimal(void)
 #ifdef HAVE_RB_EXT_RACTOR_SAFE
     rb_ext_ractor_safe(true);
 #endif
-    VALUE arg;
 
     id_BigDecimal_exception_mode = rb_intern_const("BigDecimal.exception_mode");
     id_BigDecimal_rounding_mode = rb_intern_const("BigDecimal.rounding_mode");
@@ -4532,33 +4560,19 @@ Init_bigdecimal(void)
     rb_define_const(rb_cBigDecimal, "SIGN_NEGATIVE_INFINITE", INT2FIX(VP_SIGN_NEGATIVE_INFINITE));
 
     /* Positive zero value. */
-    arg = rb_str_new2("+0");
-    BIGDECIMAL_POSITIVE_ZERO = f_BigDecimal(1, &arg, rb_cBigDecimal);
-    rb_gc_register_mark_object(BIGDECIMAL_POSITIVE_ZERO);
+    BIGDECIMAL_LITERAL(POSITIVE_ZERO, +0);
 
     /* Negative zero value. */
-    arg = rb_str_new2("-0");
-    BIGDECIMAL_NEGATIVE_ZERO = f_BigDecimal(1, &arg, rb_cBigDecimal);
-    rb_gc_register_mark_object(BIGDECIMAL_NEGATIVE_ZERO);
+    BIGDECIMAL_LITERAL(NEGATIVE_ZERO, -0);
 
-    /* Positive infinity value. */
-    arg = rb_str_new2("+Infinity");
-    BIGDECIMAL_POSITIVE_INFINITY = f_BigDecimal(1, &arg, rb_cBigDecimal);
-    rb_gc_register_mark_object(BIGDECIMAL_POSITIVE_INFINITY);
+    /* Positive infinity[rdoc-ref:BigDecimal@Infinity] value. */
+    rb_define_const(rb_cBigDecimal, "INFINITY", BIGDECIMAL_LITERAL(POSITIVE_INFINITY, +Infinity));
 
     /* Negative infinity value. */
-    arg = rb_str_new2("-Infinity");
-    BIGDECIMAL_NEGATIVE_INFINITY = f_BigDecimal(1, &arg, rb_cBigDecimal);
-    rb_gc_register_mark_object(BIGDECIMAL_NEGATIVE_INFINITY);
+    BIGDECIMAL_LITERAL(NEGATIVE_INFINITY, -Infinity);
 
-    /* 'Not a Number' value. */
-    arg = rb_str_new2("NaN");
-    BIGDECIMAL_NAN = f_BigDecimal(1, &arg, rb_cBigDecimal);
-    rb_gc_register_mark_object(BIGDECIMAL_NAN);
-
-    /* Special value constants */
-    rb_define_const(rb_cBigDecimal, "INFINITY", BIGDECIMAL_POSITIVE_INFINITY);
-    rb_define_const(rb_cBigDecimal, "NAN", BIGDECIMAL_NAN);
+    /* '{Not a Number}[rdoc-ref:BigDecimal@Not+a+Number]' value. */
+    rb_define_const(rb_cBigDecimal, "NAN", BIGDECIMAL_LITERAL(NAN, NaN));
 
     /* instance methods */
     rb_define_method(rb_cBigDecimal, "precs", BigDecimal_prec, 0);
@@ -5203,6 +5217,48 @@ bigdecimal_parse_special_string(const char *str)
     return NULL;
 }
 
+struct VpCtoV_args {
+  Real *a;
+  const char *int_chr;
+  size_t ni;
+  const char *frac;
+  size_t nf;
+  const char *exp_chr;
+  size_t ne;
+};
+
+static VALUE
+call_VpCtoV(VALUE arg)
+{
+  struct VpCtoV_args *x = (struct VpCtoV_args *)arg;
+  return (VALUE)VpCtoV(x->a, x->int_chr, x->ni, x->frac, x->nf, x->exp_chr, x->ne);
+}
+
+static int
+protected_VpCtoV(Real *a, const char *int_chr, size_t ni, const char *frac, size_t nf, const char *exp_chr, size_t ne, int free_on_error)
+{
+  struct VpCtoV_args args;
+  int state = 0;
+
+  args.a = a;
+  args.int_chr = int_chr;
+  args.ni = ni;
+  args.frac = frac;
+  args.nf = nf;
+  args.exp_chr = exp_chr;
+  args.ne = ne;
+
+  VALUE result = rb_protect(call_VpCtoV, (VALUE)&args, &state);
+  if (state) {
+    if (free_on_error) {
+      rbd_free_struct(a);
+    }
+    rb_jump_tag(state);
+  }
+
+  return (int)result;
+}
+
 /*
  * Allocates variable.
  * [Input]
@@ -5422,7 +5478,7 @@ VpAlloc(size_t mx, const char *szVal, int strict_p, int exc)
     vp = rbd_allocate_struct(len);
     vp->MaxPrec = len;        /* set max precision */
     VpSetZero(vp, sign);
-    VpCtoV(vp, psz, ni, psz + ipf, nf, psz + ipe, ne);
+    protected_VpCtoV(vp, psz, ni, psz + ipf, nf, psz + ipe, ne, true);
     rb_str_resize(buf, 0);
     return vp;
 }