Move String RVALUES between pools

And re-embed any strings that can now fit inside the slot they've been
moved to

1 files changed, 73 insertions, 4 deletions

diff --git a/string.c b/string.c
index 84af905deb..4dc462f5c4 100644
--- a/string.c
+++ b/string.c
@@ -221,17 +221,51 @@ str_embed_capa(VALUE str)
 #endif
 }
 
+bool
+rb_str_reembeddable_p(VALUE str)
+{
+    return !FL_TEST(str, STR_NOFREE|STR_SHARED_ROOT|STR_SHARED);
+}
+
 static inline size_t
-str_embed_size(long capa)
+rb_str_embed_size(long capa)
 {
     return offsetof(struct RString, as.embed.ary) + capa;
 }
 
+bool
+rb_str_shared_root_p(VALUE str)
+{
+    return FL_TEST_RAW(str, STR_SHARED_ROOT);
+}
+
+size_t
+rb_str_size_as_embedded(VALUE str)
+{
+    size_t real_size;
+#if USE_RVARGC
+    if (STR_EMBED_P(str)) {
+        real_size = rb_str_embed_size(RSTRING(str)->as.embed.len) + TERM_LEN(str);
+    }
+    /* if the string is not currently embedded, but it can be embedded, how
+     * much space would it require */
+    else if (rb_str_reembeddable_p(str)) {
+        real_size = rb_str_embed_size(RSTRING(str)->as.heap.len) + TERM_LEN(str);
+    }
+    else {
+#endif
+        real_size = sizeof(struct RString);
+#if USE_RVARGC
+    }
+#endif
+    return real_size;
+}
+
 static inline bool
 STR_EMBEDDABLE_P(long len, long termlen)
 {
 #if USE_RVARGC
-    return rb_gc_size_allocatable_p(str_embed_size(len + termlen));
+    return rb_gc_size_allocatable_p(rb_str_embed_size(len + termlen));
 #else
     return len <= RSTRING_EMBED_LEN_MAX + 1 - termlen;
 #endif
@@ -265,6 +299,41 @@ rb_str_make_independent(VALUE str)
 }
 
 void
+rb_str_make_embedded(VALUE str) {
+    RUBY_ASSERT(rb_str_reembeddable_p(str));
+    RUBY_ASSERT(!STR_EMBED_P(str));
+
+    char *buf = RSTRING_PTR(str);
+    long len = RSTRING_LEN(str);
+
+    STR_SET_EMBED(str);
+    STR_SET_EMBED_LEN(str, len);
+
+    memmove(RSTRING_PTR(str), buf, len);
+    ruby_xfree(buf);
+}
+
+void
+rb_str_update_shared_ary(VALUE str, VALUE old_root, VALUE new_root)
+{
+    // if the root location hasn't changed, we don't need to update
+    if (new_root == old_root) {
+        return;
+    }
+
+    // if the root string isn't embedded, we don't need to touch the ponter.
+    // it already points to the shame shared buffer
+    if (!STR_EMBED_P(new_root)) {
+        return;
+    }
+
+    size_t offset = (size_t)((uintptr_t)RSTRING(str)->as.heap.ptr - (uintptr_t)RSTRING(old_root)->as.embed.ary);
+
+    RUBY_ASSERT(RSTRING(str)->as.heap.ptr >= RSTRING(old_root)->as.embed.ary);
+    RSTRING(str)->as.heap.ptr = RSTRING(new_root)->as.embed.ary + offset;
+}
+
+void
 rb_debug_rstring_null_ptr(const char *func)
 {
     fprintf(stderr, "%s is returning NULL!! "
@@ -849,7 +918,7 @@ str_alloc(VALUE klass, size_t size)
 static inline VALUE
 str_alloc_embed(VALUE klass, size_t capa)
 {
-    size_t size = str_embed_size(capa);
+    size_t size = rb_str_embed_size(capa);
     assert(rb_gc_size_allocatable_p(size));
 #if !USE_RVARGC
     assert(size <= sizeof(struct RString));
@@ -1693,7 +1762,7 @@ ec_str_alloc(struct rb_execution_context_struct *ec, VALUE klass, size_t size)
 static inline VALUE
 ec_str_alloc_embed(struct rb_execution_context_struct *ec, VALUE klass, size_t capa)
 {
-    size_t size = str_embed_size(capa);
+    size_t size = rb_str_embed_size(capa);
     assert(rb_gc_size_allocatable_p(size));
 #if !USE_RVARGC
     assert(size <= sizeof(struct RString));