Merge pull request #1138 from tempesta-tech/ik-move-parser-to-connection

Move http parser to connection

tempesta_fw/addr.c

@@ -53,7 +53,7 @@ tfw_addr_pton_v4(const TfwStr *s, TfwAddr *addr)
 
 	TFW_STR_FOR_EACH_CHUNK(c, s, end) {
 		for (k = 0; k != c->len; ++k) {
-			p = c->ptr + k;
+			p = (char *)c->ptr + k;
 			if (isdigit(*p)) {
 				octet = (octet == -1)
 					? *p - '0'
@@ -101,7 +101,7 @@ tfw_addr_pton_v6(const TfwStr *s, TfwAddr *addr)
 
 	TFW_STR_FOR_EACH_CHUNK(c, s, end) {
 		for (k = 0; k != c->len; ++k) {
-			p = c->ptr + k;
+			p = (char *)c->ptr + k;
 			if (i > 7 && !(i == 8 && port == 1))
 				return -EINVAL;
 
@@ -249,7 +249,7 @@ tfw_addr_pton(const TfwStr *str, TfwAddr *addr)
 		TFW_STR_FOR_EACH_CHUNK(c, str, end) {
 			int i;
 			for (i = 0; i != c->len; ++i) {
-				pos = c->ptr + i;
+				pos = (char *)c->ptr + i;
 				if (!isdigit(*pos))
 					goto delim;
 			}

tempesta_fw/connection.h

@@ -29,6 +29,8 @@
 #include "msg.h"
 #include "peer.h"
 
+#include "http_parser.h"
+
 #include "sync_socket.h"
 #include "tls.h"
 
@@ -93,6 +95,7 @@ enum {
 #define TFW_CONN_COMMON					\
 	SsProto			proto;			\
 	TfwGState		state;			\
+	TfwHttpParser		parser;			\
 	struct list_head	list;			\
 	atomic_t		refcnt;			\
 	struct timer_list	timer;			\

tempesta_fw/http.c

@@ -28,6 +28,7 @@
 #include "cache.h"
 #include "http_limits.h"
 #include "http_tbl.h"
+#include "http_parser.h"
 #include "client.h"
 #include "http_msg.h"
 #include "http_sess.h"
@@ -1685,14 +1686,20 @@ tfw_http_resp_pair(TfwHttpMsg *hmresp)
 static TfwMsg *
 tfw_http_conn_msg_alloc(TfwConn *conn)
 {
-	TfwHttpMsg *hm = __tfw_http_msg_alloc(TFW_CONN_TYPE(conn), true);
+	int type = TFW_CONN_TYPE(conn);
+	TfwHttpMsg *hm = __tfw_http_msg_alloc(type, true);
 	if (unlikely(!hm))
 		return NULL;
 
 	hm->conn = conn;
 	tfw_connection_get(conn);
 
-	if (TFW_CONN_TYPE(conn) & Conn_Clnt) {
+	if (type & Conn_Clnt)
+		tfw_http_init_parser_req((TfwHttpReq *)hm);
+	else
+		tfw_http_init_parser_resp((TfwHttpResp *)hm);
+
+	if (type & Conn_Clnt) {
 		TFW_INC_STAT_BH(clnt.rx_messages);
 	} else {
 		if (unlikely(tfw_http_resp_pair(hm))) {
@@ -2699,7 +2706,7 @@ tfw_http_req_process(TfwConn *conn, const TfwFsmData *data)
 		bool block = false;
 		TfwHttpMsg *hmsib = NULL;
 		TfwHttpReq *req = (TfwHttpReq *)conn->msg;
-		TfwHttpParser *parser = &req->parser;
+		TfwHttpParser *parser = &conn->parser;
 
 		/*
 		 * Process/parse data in the SKB.
@@ -3183,7 +3190,7 @@ tfw_http_resp_process(TfwConn *conn, const TfwFsmData *data)
 		TfwHttpParser *parser;
 
 		hmresp = (TfwHttpMsg *)conn->msg;
-		parser = &hmresp->parser;
+		parser = &conn->parser;
 
 		/*
 		 * Process/parse data in the SKB.
@@ -4196,10 +4203,6 @@ tfw_http_init(void)
 {
 	int r;
 
-	/* Make sure @req->httperr doesn't take too much space. */
-	BUILD_BUG_ON(FIELD_SIZEOF(TfwHttpMsg, httperr)
-		     > FIELD_SIZEOF(TfwHttpMsg, parser));
-
 	r = tfw_gfsm_register_fsm(TFW_FSM_HTTP, tfw_http_msg_process);
 	if (r)
 		return r;

tempesta_fw/http.h

@@ -23,15 +23,14 @@
 
 #include <crypto/sha.h>
 
+#include "http_types.h"
 #include "connection.h"
 #include "gfsm.h"
 #include "msg.h"
 #include "server.h"
 #include "str.h"
 #include "vhost.h"
 
-typedef struct tfw_http_sess_t TfwHttpSess;
-
 /**
  * HTTP Generic FSM states.
  *
@@ -204,82 +203,6 @@ typedef enum {
 	TFW_HTTP_HDR_NUM	= 16,
 } tfw_http_hdr_t;
 
-typedef struct {
-	unsigned int	size;	/* number of elements in the table */
-	unsigned int	off;
-	TfwStr		tbl[0];
-} TfwHttpHdrTbl;
-
-#define __HHTBL_SZ(o)			(TFW_HTTP_HDR_NUM * (o))
-#define TFW_HHTBL_EXACTSZ(s)		(sizeof(TfwHttpHdrTbl)		\
-					 + sizeof(TfwStr) * (s))
-#define TFW_HHTBL_SZ(o)			TFW_HHTBL_EXACTSZ(__HHTBL_SZ(o))
-
-/** Maximum of hop-by-hop tokens listed in Connection header. */
-#define TFW_HBH_TOKENS_MAX		16
-
-/**
- * Non-cacheable hop-by-hop headers in terms of RFC 7230.
- *
- * We don't store the headers in cache and create them from scratch if needed.
- * Adding a header is faster then modify it, so this speeds up headers
- * adjusting as well as saves cache storage.
- *
- * Headers unconditionally treated as hop-by-hop must be listed in
- * tfw_http_init_parser_req()/tfw_http_init_parser_resp() functions and must be
- * members of Special headers.
- * group.
- *
- * @spec	- bit array for special headers. Hop-by-hop special header is
- *		  stored as (0x1 << tfw_http_hdr_t[hid]);
- * @raw		- table of raw headers names, parsed form connection field;
- * @off		- offset of last added raw header name;
- */
-typedef struct {
-	unsigned int	spec;
-	unsigned int	off;
-	TfwStr		raw[TFW_HBH_TOKENS_MAX];
-} TfwHttpHbhHdrs;
-
-/**
- * We use goto/switch-driven automaton, so compiler typically generates binary
- * search code over jump labels, so it gives log(N) lookup complexity where
- * N is number of states. However, DFA for full HTTP processing can be quite
- * large and log(N) becomes expensive and hard to code.
- *
- * So we use states space splitting to avoid states explosion.
- * @_i_st is used to save current state and go to interior sub-automaton
- * (e.g. process OWS using @state while current state is saved in @_i_st
- * or using @_i_st parse value of a header described.
- *
- * @_cnt	- currently the count of hex digits in a body chunk size;
- * @to_go	- remaining number of bytes to process in the data chunk;
- *		  (limited by single packet size and never exceeds 64KB)
- * @state	- current parser state;
- * @_i_st	- helping (interior) state;
- * @to_read	- remaining number of bytes to read;
- * @_hdr_tag	- stores header id which must be closed on generic EoL handling
- *		  (see RGEN_EOL());
- * @_acc	- integer accumulator for parsing chunked integers;
- * @_tmp_chunk	- currently parsed (sub)string, possibly chunked;
- * @hdr		- currently parsed header.
- * @hbh_parser	- list of special and raw headers names to be treated as
- *		  hop-by-hop
- * @_date	- currently parsed http date value;
- */
-typedef struct {
-	unsigned short	to_go;
-	unsigned short	_cnt;
-	unsigned int	_hdr_tag;
-	int		state;
-	int		_i_st;
-	long		to_read;
-	unsigned long	_acc;
-	time_t		_date;
-	TfwStr		_tmp_chunk;
-	TfwStr		hdr;
-	TfwHttpHbhHdrs	hbh_parser;
-} TfwHttpParser;
 
 enum {
 	/* Common flags for requests and responses. */
@@ -353,7 +276,7 @@ enum {
 #define TFW_HTTP_F_RESP_STALE		(1U << TFW_HTTP_B_RESP_STALE)
 #define TFW_HTTP_F_RESP_READY		(1U << TFW_HTTP_B_RESP_READY)
 
-/*
+/**
  * The structure to hold data for an HTTP error response.
  * An error response is sent later in an unlocked queue context.
  *
@@ -365,17 +288,12 @@ typedef struct {
 	unsigned short	status;
 }TfwHttpError;
 
-typedef struct tfw_http_msg_t	TfwHttpMsg;
-typedef struct tfw_http_req_t	TfwHttpReq;
-typedef struct tfw_http_resp_t	TfwHttpResp;
-
 /**
  * Common HTTP message members.
  *
  * @msg			- the base data of an HTTP message;
  * @pool		- message's memory allocation pool;
  * @h_tbl		- table of message's HTTP headers in internal form;
- * @parser		- parser state data while a message is parsed;
  * @httperr		- HTTP error data used to form an error response;
  * @pair		- the message paired with this one;
  * @req			- the request paired with this response;
@@ -395,14 +313,6 @@ typedef struct tfw_http_resp_t	TfwHttpResp;
  *
  * TfwStr members must be the last for efficient scanning.
  *
- * NOTE: The space taken by @parser member is shared with @httperr member.
- * When a message results in an error, the parser state data is not used
- * anymore, so this is safe. The reason for reuse is that it's imperative
- * that saving the error data succeeds. If it fails, that will lead to
- * a request without a response - a hole in @seq_queue. Alternatively,
- * memory allocation from pool may fail, even if that's a rare case.
- * BUILD_BUG_ON() is used in tfw_http_init() to ensure that @httperr
- * doesn't make the whole structure bigger.
  */
 #define TFW_HTTP_MSG_COMMON						\
 	TfwMsg		msg;						\
@@ -413,10 +323,7 @@ typedef struct tfw_http_resp_t	TfwHttpResp;
 		TfwHttpReq	*req;					\
 		TfwHttpResp	*resp;					\
 	};								\
-	union {								\
-		TfwHttpParser	parser;					\
-		TfwHttpError	httperr;				\
-	};								\
+	TfwHttpError	httperr;					\
 	TfwCacheControl	cache_ctl;					\
 	unsigned char	version;					\
 	unsigned int	flags;						\
@@ -594,13 +501,6 @@ tfw_http_resp_code_range(const int n)
 
 typedef void (*tfw_http_cache_cb_t)(TfwHttpMsg *);
 
-/* Internal (parser) HTTP functions. */
-void tfw_http_init_parser_req(TfwHttpReq *req);
-void tfw_http_init_parser_resp(TfwHttpResp *resp);
-int tfw_http_parse_req(void *req_data, unsigned char *data, size_t len);
-int tfw_http_parse_resp(void *resp_data, unsigned char *data, size_t len);
-bool tfw_http_parse_terminate(TfwHttpMsg *hm);
-
 /* External HTTP functions. */
 int tfw_http_msg_process(void *conn, const TfwFsmData *data);
 unsigned long tfw_http_req_key_calc(TfwHttpReq *req);

tempesta_fw/http_limits.c

@@ -481,9 +481,11 @@ __frang_http_field_len(const TfwHttpReq *req, FrangAcc *ra,
 static int
 frang_http_field_len(const TfwHttpReq *req, FrangAcc *ra, unsigned int field_len)
 {
-	if (req->parser.hdr.len > field_len) {
+	TfwHttpParser *parser = &req->conn->parser;
+
+	if (parser->hdr.len > field_len) {
 		frang_limmsg("HTTP in-progress field length",
-			     req->parser.hdr.len, field_len,
+			     parser->hdr.len, field_len,
 			     &FRANG_ACC2CLI(ra)->addr);
 		return TFW_BLOCK;
 	}

tempesta_fw/http_msg.c

@@ -25,6 +25,7 @@
 #include "lib/str.h"
 #include "gfsm.h"
 #include "http_msg.h"
+#include "http_parser.h"
 #include "ss_skb.h"
 
 /**
@@ -315,7 +316,7 @@ __hdr_lookup(TfwHttpMsg *hm, const TfwStr *hdr)
 void
 tfw_http_msg_hdr_open(TfwHttpMsg *hm, unsigned char *hdr_start)
 {
-	TfwStr *hdr = &hm->parser.hdr;
+	TfwStr *hdr = &hm->conn->parser.hdr;
 
 	BUG_ON(!TFW_STR_EMPTY(hdr));
 
@@ -337,12 +338,13 @@ tfw_http_msg_hdr_close(TfwHttpMsg *hm, unsigned int id)
 {
 	TfwStr *h;
 	TfwHttpHdrTbl *ht = hm->h_tbl;
+	TfwHttpParser *parser = &hm->conn->parser;
 
-	BUG_ON(hm->parser.hdr.flags & TFW_STR_DUPLICATE);
+	BUG_ON(parser->hdr.flags & TFW_STR_DUPLICATE);
 	BUG_ON(id > TFW_HTTP_HDR_RAW);
 
 	/* Close just parsed header. */
-	hm->parser.hdr.flags |= TFW_STR_COMPLETE;
+	parser->hdr.flags |= TFW_STR_COMPLETE;
 
 	/* Quick path for special headers. */
 	if (likely(id < TFW_HTTP_HDR_RAW)) {
@@ -373,7 +375,7 @@ tfw_http_msg_hdr_close(TfwHttpMsg *hm, unsigned int id)
 	 * Both the headers, the new one and existing one, can already be
 	 * compound.
 	 */
-	id = __hdr_lookup(hm, &hm->parser.hdr);
+	id = __hdr_lookup(hm, &parser->hdr);
 
 	/* Allocate some more room if not enough to store the header. */
 	if (unlikely(id == ht->size)) {
@@ -392,14 +394,14 @@ tfw_http_msg_hdr_close(TfwHttpMsg *hm, unsigned int id)
 duplicate:
 	h = tfw_str_add_duplicate(hm->pool, h);
 	if (unlikely(!h)) {
-		TFW_WARN("Cannot close header %p id=%d\n", &hm->parser.hdr, id);
+		TFW_WARN("Cannot close header %p id=%d\n", &parser->hdr, id);
 		return TFW_BLOCK;
 	}
 
 done:
-	*h = hm->parser.hdr;
+	*h = parser->hdr;
 
-	TFW_STR_INIT(&hm->parser.hdr);
+	TFW_STR_INIT(&parser->hdr);
 	TFW_DBG3("store header w/ ptr=%p len=%lu eolen=%u flags=%x id=%d\n",
 		 h->ptr, h->len, h->eolen, h->flags, id);
 
@@ -1012,11 +1014,6 @@ __tfw_http_msg_alloc(int type, bool full)
 		hm->h_tbl->size = __HHTBL_SZ(1);
 		hm->h_tbl->off = TFW_HTTP_HDR_RAW;
 		bzero_fast(hm->h_tbl->tbl, __HHTBL_SZ(1) * sizeof(TfwStr));
-
-		if (type & Conn_Clnt)
-			tfw_http_init_parser_req((TfwHttpReq *)hm);
-		else
-			tfw_http_init_parser_resp((TfwHttpResp *)hm);
 	}
 
 	hm->msg.skb_head = NULL;