[Feature] Add mmap support to lua_text

7 files changed, 44 insertions, 27 deletions

diff --git a/src/lua/lua_common.h b/src/lua/lua_common.h
index 6fdff5990..3f2fa6ba5 100644
--- a/src/lua/lua_common.h
+++ b/src/lua/lua_common.h
@@ -64,10 +64,12 @@ struct rspamd_lua_ip {
 	rspamd_inet_addr_t *addr;
 };
 
+#define RSPAMD_TEXT_FLAG_OWN (1 << 0)
+#define RSPAMD_TEXT_FLAG_MMAPED (1 << 1)
 struct rspamd_lua_text {
 	const gchar *start;
 	guint len;
-	gboolean own;
+	guint flags;
 };
 
 struct rspamd_lua_url {
diff --git a/src/lua/lua_html.c b/src/lua/lua_html.c
index bad085450..1e1ab10ae 100644
--- a/src/lua/lua_html.c
+++ b/src/lua/lua_html.c
@@ -355,7 +355,7 @@ lua_html_push_block (lua_State *L, struct html_block *bl)
 		t = lua_newuserdata (L, sizeof (*t));
 		t->start = bl->style.start;
 		t->len = bl->style.len;
-		t->own = FALSE;
+		t->flags = 0;
 		lua_settable (L, -3);
 	}
 }
@@ -562,7 +562,7 @@ lua_html_tag_get_content (lua_State *L)
 			rspamd_lua_setclass (L, "rspamd{text}", -1);
 			t->start = tag->content;
 			t->len = tag->content_length;
-			t->own = FALSE;
+			t->flags = 0;
 		}
 		else {
 			lua_pushnil (L);
diff --git a/src/lua/lua_mimepart.c b/src/lua/lua_mimepart.c
index a2b6cc49c..e8c7bd9c8 100644
--- a/src/lua/lua_mimepart.c
+++ b/src/lua/lua_mimepart.c
@@ -357,7 +357,7 @@ lua_textpart_get_content (lua_State * L)
 	rspamd_lua_setclass (L, "rspamd{text}", -1);
 	t->start = part->content->data;
 	t->len = part->content->len;
-	t->own = FALSE;
+	t->flags = 0;
 
 	return 1;
 }
@@ -377,7 +377,7 @@ lua_textpart_get_raw_content (lua_State * L)
 	rspamd_lua_setclass (L, "rspamd{text}", -1);
 	t->start = part->orig->data;
 	t->len = part->orig->len;
-	t->own = FALSE;
+	t->flags = 0;
 
 	return 1;
 }
@@ -397,7 +397,7 @@ lua_textpart_get_content_oneline (lua_State * L)
 	rspamd_lua_setclass (L, "rspamd{text}", -1);
 	t->start = part->stripped_content->data;
 	t->len = part->stripped_content->len;
-	t->own = FALSE;
+	t->flags = 0;
 
 	return 1;
 }
@@ -609,7 +609,7 @@ lua_mimepart_get_content (lua_State * L)
 	rspamd_lua_setclass (L, "rspamd{text}", -1);
 	t->start = part->content->data;
 	t->len = part->content->len;
-	t->own = FALSE;
+	t->flags = 0;
 
 	return 1;
 }
diff --git a/src/lua/lua_regexp.c b/src/lua/lua_regexp.c
index 80b752172..1e6122b2e 100644
--- a/src/lua/lua_regexp.c
+++ b/src/lua/lua_regexp.c
@@ -640,7 +640,7 @@ lua_regexp_split (lua_State *L)
 						rspamd_lua_setclass (L, "rspamd{text}", -1);
 						t->start = old_start;
 						t->len = start - old_start;
-						t->own = FALSE;
+						t->flags = 0;
 					}
 
 					lua_rawseti (L, -2, ++i);
@@ -665,7 +665,7 @@ lua_regexp_split (lua_State *L)
 					rspamd_lua_setclass (L, "rspamd{text}", -1);
 					t->start = end;
 					t->len = (data + len) - end;
-					t->own = FALSE;
+					t->flags = 0;
 				}
 
 				lua_rawseti (L, -2, ++i);
diff --git a/src/lua/lua_task.c b/src/lua/lua_task.c
index 4706d69a0..a3354acb6 100644
--- a/src/lua/lua_task.c
+++ b/src/lua/lua_task.c
@@ -25,6 +25,7 @@
 #include "email_addr.h"
 #include "utlist.h"
 #include "cryptobox.h"
+#include "unix-std.h"
 
 /***
  * @module rspamd_task
@@ -1184,7 +1185,7 @@ lua_task_get_content (lua_State * L)
 		rspamd_lua_setclass (L, "rspamd{text}", -1);
 		t->len = task->msg.len;
 		t->start = task->msg.begin;
-		t->own = FALSE;
+		t->flags = 0;
 	}
 	else {
 		return luaL_error (L, "invalid arguments");
@@ -1213,7 +1214,7 @@ lua_task_get_rawbody (lua_State * L)
 			t->start = task->msg.begin;
 		}
 
-		t->own = FALSE;
+		t->flags = 0;
 	}
 	else {
 		return luaL_error (L, "invalid arguments");
@@ -1311,7 +1312,7 @@ lua_task_get_request_header (lua_State *L)
 			rspamd_lua_setclass (L, "rspamd{text}", -1);
 			t->start = hdr->begin;
 			t->len = hdr->len;
-			t->own = FALSE;
+			t->flags = 0;
 
 			return 1;
 		}
@@ -1490,7 +1491,7 @@ lua_task_get_raw_headers (lua_State *L)
 		rspamd_lua_setclass (L, "rspamd{text}", -1);
 		t->start = task->raw_headers_content.begin;
 		t->len = task->raw_headers_content.len;
-		t->own = FALSE;
+		t->flags = 0;
 	}
 	else {
 		return luaL_error (L, "invalid arguments");
@@ -3473,8 +3474,16 @@ lua_text_gc (lua_State *L)
 {
 	struct rspamd_lua_text *t = lua_check_text (L, 1);
 
-	if (t != NULL && t->own) {
-		g_free ((gpointer)t->start);
+	if (t != NULL) {
+		if (t->flags & RSPAMD_TEXT_FLAG_OWN) {
+			if (t->flags & RSPAMD_TEXT_FLAG_MMAPED) {
+				munmap ((gpointer)t->start, t->len);
+			}
+			else {
+				g_free ((gpointer)t->start);
+			}
+		}
+
 	}
 
 	return 0;
diff --git a/src/lua/lua_tcp.c b/src/lua/lua_tcp.c
index 23140b79a..c80c311f8 100644
--- a/src/lua/lua_tcp.c
+++ b/src/lua/lua_tcp.c
@@ -199,7 +199,7 @@ lua_tcp_push_data (struct lua_tcp_cbdata *cbd, const guint8 *str, gsize len)
 	rspamd_lua_setclass (cbd->L, "rspamd{text}", -1);
 	t->start = (const gchar *)str;
 	t->len = len;
-	t->own = FALSE;
+	t->flags = 0;
 	/* Connection */
 	pcbd = lua_newuserdata (cbd->L, sizeof (*pcbd));
 	*pcbd = cbd;
@@ -467,9 +467,9 @@ lua_tcp_arg_toiovec (lua_State *L, gint pos, rspamd_mempool_t *pool,
 			vec->iov_base = (void *)t->start;
 			vec->iov_len = t->len;
 
-			if (t->own) {
+			if (t->flags & RSPAMD_TEXT_FLAG_OWN) {
 				/* Steal ownership */
-				t->own = FALSE;
+				t->flags = 0;
 				rspamd_mempool_add_destructor (pool, g_free, (void *)t->start);
 			}
 		}
diff --git a/src/lua/lua_util.c b/src/lua/lua_util.c
index 9c37bc861..4ce7cbe64 100644
--- a/src/lua/lua_util.c
+++ b/src/lua/lua_util.c
@@ -627,7 +627,7 @@ lua_util_encode_base64 (lua_State *L)
 			t->start = out;
 			t->len = outlen;
 			/* Need destruction */
-			t->own = TRUE;
+			t->flags = RSPAMD_TEXT_FLAG_OWN;
 		}
 		else {
 			lua_pushnil (L);
@@ -657,8 +657,8 @@ lua_util_decode_base64 (lua_State *L)
 			s = t->start;
 			inlen = t->len;
 			zero_copy = TRUE;
-			if (t->own) {
-				t->own = FALSE;
+			if (t->flags & RSPAMD_TEXT_FLAG_OWN) {
+				t->flags = 0;
 				grab_own = TRUE;
 			}
 		}
@@ -672,7 +672,13 @@ lua_util_decode_base64 (lua_State *L)
 			rspamd_lua_setclass (L, "rspamd{text}", -1);
 			t->start = s;
 			t->len = outlen;
-			t->own = grab_own;
+
+			if (grab_own) {
+				t->flags |= RSPAMD_TEXT_FLAG_OWN;
+			}
+			else {
+				t->flags = 0;
+			}
 		}
 		else {
 			t = lua_newuserdata (L, sizeof (*t));
@@ -682,7 +688,7 @@ lua_util_decode_base64 (lua_State *L)
 			outlen = g_base64_decode_step (s, inlen, (guchar *)t->start,
 					&state, &save);
 			t->len = outlen;
-			t->own = TRUE;
+			t->flags = RSPAMD_TEXT_FLAG_OWN;
 		}
 	}
 	else {
@@ -725,7 +731,7 @@ lua_util_encode_base32 (lua_State *L)
 			t->start = out;
 			t->len = outlen;
 			/* Need destruction */
-			t->own = TRUE;
+			t->flags = RSPAMD_TEXT_FLAG_OWN;
 		}
 		else {
 			lua_pushnil (L);
@@ -759,7 +765,7 @@ lua_util_decode_base32 (lua_State *L)
 		rspamd_lua_setclass (L, "rspamd{text}", -1);
 		t->start = rspamd_decode_base32 (s, inlen, &outlen);
 		t->len = outlen;
-		t->own = TRUE;
+		t->flags = RSPAMD_TEXT_FLAG_OWN;
 	}
 	else {
 		lua_pushnil (L);
@@ -793,7 +799,7 @@ lua_util_decode_url (lua_State *L)
 		t->start = g_malloc (inlen);
 		memcpy ((char *)t->start, s, inlen);
 		t->len = rspamd_decode_url ((char *)t->start, s, inlen);
-		t->own = TRUE;
+		t->flags = RSPAMD_TEXT_FLAG_OWN;
 	}
 	else {
 		lua_pushnil (L);
@@ -938,7 +944,7 @@ lua_util_parse_html (lua_State *L)
 		rspamd_lua_setclass (L, "rspamd{text}", -1);
 		t->start = res->data;
 		t->len = res->len;
-		t->own = TRUE;
+		t->flags = RSPAMD_TEXT_FLAG_OWN;
 
 		g_byte_array_free (res, FALSE);
 		g_byte_array_free (in, TRUE);