/*
 * HTTP/1 protocol analyzer
 *
 * Copyright 2000-2017 Willy Tarreau <w@1wt.eu>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 *
 */

#include <common/config.h>

#include <proto/h1.h>
#include <proto/hdr_idx.h>

/* It is about twice as fast on recent architectures to lookup a byte in a
 * table than to perform a boolean AND or OR between two tests. Refer to
 * RFC2616/RFC5234/RFC7230 for those chars. A token is any ASCII char that is
 * neither a separator nor a CTL char. An http ver_token is any ASCII which can
 * be found in an HTTP version, which includes 'H', 'T', 'P', '/', '.' and any
 * digit. Note: please do not overwrite values in assignment since gcc-2.95
 * will not handle them correctly. It's worth noting that chars 128..255 are
 * nothing, not even control chars.
 */
const unsigned char h1_char_classes[256] = {
	[  0] = H1_FLG_CTL,
	[  1] = H1_FLG_CTL,
	[  2] = H1_FLG_CTL,
	[  3] = H1_FLG_CTL,
	[  4] = H1_FLG_CTL,
	[  5] = H1_FLG_CTL,
	[  6] = H1_FLG_CTL,
	[  7] = H1_FLG_CTL,
	[  8] = H1_FLG_CTL,
	[  9] = H1_FLG_SPHT | H1_FLG_LWS | H1_FLG_SEP | H1_FLG_CTL,
	[ 10] = H1_FLG_CRLF | H1_FLG_LWS | H1_FLG_CTL,
	[ 11] = H1_FLG_CTL,
	[ 12] = H1_FLG_CTL,
	[ 13] = H1_FLG_CRLF | H1_FLG_LWS | H1_FLG_CTL,
	[ 14] = H1_FLG_CTL,
	[ 15] = H1_FLG_CTL,
	[ 16] = H1_FLG_CTL,
	[ 17] = H1_FLG_CTL,
	[ 18] = H1_FLG_CTL,
	[ 19] = H1_FLG_CTL,
	[ 20] = H1_FLG_CTL,
	[ 21] = H1_FLG_CTL,
	[ 22] = H1_FLG_CTL,
	[ 23] = H1_FLG_CTL,
	[ 24] = H1_FLG_CTL,
	[ 25] = H1_FLG_CTL,
	[ 26] = H1_FLG_CTL,
	[ 27] = H1_FLG_CTL,
	[ 28] = H1_FLG_CTL,
	[ 29] = H1_FLG_CTL,
	[ 30] = H1_FLG_CTL,
	[ 31] = H1_FLG_CTL,
	[' '] = H1_FLG_SPHT | H1_FLG_LWS | H1_FLG_SEP,
	['!'] = H1_FLG_TOK,
	['"'] = H1_FLG_SEP,
	['#'] = H1_FLG_TOK,
	['$'] = H1_FLG_TOK,
	['%'] = H1_FLG_TOK,
	['&'] = H1_FLG_TOK,
	[ 39] = H1_FLG_TOK,
	['('] = H1_FLG_SEP,
	[')'] = H1_FLG_SEP,
	['*'] = H1_FLG_TOK,
	['+'] = H1_FLG_TOK,
	[','] = H1_FLG_SEP,
	['-'] = H1_FLG_TOK,
	['.'] = H1_FLG_TOK | H1_FLG_VER,
	['/'] = H1_FLG_SEP | H1_FLG_VER,
	['0'] = H1_FLG_TOK | H1_FLG_VER,
	['1'] = H1_FLG_TOK | H1_FLG_VER,
	['2'] = H1_FLG_TOK | H1_FLG_VER,
	['3'] = H1_FLG_TOK | H1_FLG_VER,
	['4'] = H1_FLG_TOK | H1_FLG_VER,
	['5'] = H1_FLG_TOK | H1_FLG_VER,
	['6'] = H1_FLG_TOK | H1_FLG_VER,
	['7'] = H1_FLG_TOK | H1_FLG_VER,
	['8'] = H1_FLG_TOK | H1_FLG_VER,
	['9'] = H1_FLG_TOK | H1_FLG_VER,
	[':'] = H1_FLG_SEP,
	[';'] = H1_FLG_SEP,
	['<'] = H1_FLG_SEP,
	['='] = H1_FLG_SEP,
	['>'] = H1_FLG_SEP,
	['?'] = H1_FLG_SEP,
	['@'] = H1_FLG_SEP,
	['A'] = H1_FLG_TOK,
	['B'] = H1_FLG_TOK,
	['C'] = H1_FLG_TOK,
	['D'] = H1_FLG_TOK,
	['E'] = H1_FLG_TOK,
	['F'] = H1_FLG_TOK,
	['G'] = H1_FLG_TOK,
	['H'] = H1_FLG_TOK | H1_FLG_VER,
	['I'] = H1_FLG_TOK,
	['J'] = H1_FLG_TOK,
	['K'] = H1_FLG_TOK,
	['L'] = H1_FLG_TOK,
	['M'] = H1_FLG_TOK,
	['N'] = H1_FLG_TOK,
	['O'] = H1_FLG_TOK,
	['P'] = H1_FLG_TOK | H1_FLG_VER,
	['Q'] = H1_FLG_TOK,
	['R'] = H1_FLG_TOK | H1_FLG_VER,
	['S'] = H1_FLG_TOK | H1_FLG_VER,
	['T'] = H1_FLG_TOK | H1_FLG_VER,
	['U'] = H1_FLG_TOK,
	['V'] = H1_FLG_TOK,
	['W'] = H1_FLG_TOK,
	['X'] = H1_FLG_TOK,
	['Y'] = H1_FLG_TOK,
	['Z'] = H1_FLG_TOK,
	['['] = H1_FLG_SEP,
	[ 92] = H1_FLG_SEP,
	[']'] = H1_FLG_SEP,
	['^'] = H1_FLG_TOK,
	['_'] = H1_FLG_TOK,
	['`'] = H1_FLG_TOK,
	['a'] = H1_FLG_TOK,
	['b'] = H1_FLG_TOK,
	['c'] = H1_FLG_TOK,
	['d'] = H1_FLG_TOK,
	['e'] = H1_FLG_TOK,
	['f'] = H1_FLG_TOK,
	['g'] = H1_FLG_TOK,
	['h'] = H1_FLG_TOK,
	['i'] = H1_FLG_TOK,
	['j'] = H1_FLG_TOK,
	['k'] = H1_FLG_TOK,
	['l'] = H1_FLG_TOK,
	['m'] = H1_FLG_TOK,
	['n'] = H1_FLG_TOK,
	['o'] = H1_FLG_TOK,
	['p'] = H1_FLG_TOK,
	['q'] = H1_FLG_TOK,
	['r'] = H1_FLG_TOK,
	['s'] = H1_FLG_TOK,
	['t'] = H1_FLG_TOK,
	['u'] = H1_FLG_TOK,
	['v'] = H1_FLG_TOK,
	['w'] = H1_FLG_TOK,
	['x'] = H1_FLG_TOK,
	['y'] = H1_FLG_TOK,
	['z'] = H1_FLG_TOK,
	['{'] = H1_FLG_SEP,
	['|'] = H1_FLG_TOK,
	['}'] = H1_FLG_SEP,
	['~'] = H1_FLG_TOK,
	[127] = H1_FLG_CTL,
};


/*
 * This function parses a status line between <ptr> and <end>, starting with
 * parser state <state>. Only states HTTP_MSG_RPVER, HTTP_MSG_RPVER_SP,
 * HTTP_MSG_RPCODE, HTTP_MSG_RPCODE_SP and HTTP_MSG_RPREASON are handled. Others
 * will give undefined results.
 * Note that it is upon the caller's responsibility to ensure that ptr < end,
 * and that msg->sol points to the beginning of the response.
 * If a complete line is found (which implies that at least one CR or LF is
 * found before <end>, the updated <ptr> is returned, otherwise NULL is
 * returned indicating an incomplete line (which does not mean that parts have
 * not been updated). In the incomplete case, if <ret_ptr> or <ret_state> are
 * non-NULL, they are fed with the new <ptr> and <state> values to be passed
 * upon next call.
 *
 * This function was intentionally designed to be called from
 * http_msg_analyzer() with the lowest overhead. It should integrate perfectly
 * within its state machine and use the same macros, hence the need for same
 * labels and variable names. Note that msg->sol is left unchanged.
 */
const char *http_parse_stsline(struct http_msg *msg,
                               enum h1_state state, const char *ptr, const char *end,
                               unsigned int *ret_ptr, enum h1_state *ret_state)
{
	const char *msg_start = msg->chn->buf->p;

	switch (state)	{
	case HTTP_MSG_RPVER:
	http_msg_rpver:
		if (likely(HTTP_IS_VER_TOKEN(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpver, http_msg_ood, state, HTTP_MSG_RPVER);

		if (likely(HTTP_IS_SPHT(*ptr))) {
			msg->sl.st.v_l = ptr - msg_start;
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpver_sp, http_msg_ood, state, HTTP_MSG_RPVER_SP);
		}
		msg->err_state = HTTP_MSG_RPVER;
		state = HTTP_MSG_ERROR;
		break;

	case HTTP_MSG_RPVER_SP:
	http_msg_rpver_sp:
		if (likely(!HTTP_IS_LWS(*ptr))) {
			msg->sl.st.c = ptr - msg_start;
			goto http_msg_rpcode;
		}
		if (likely(HTTP_IS_SPHT(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpver_sp, http_msg_ood, state, HTTP_MSG_RPVER_SP);
		/* so it's a CR/LF, this is invalid */
		msg->err_state = HTTP_MSG_RPVER_SP;
		state = HTTP_MSG_ERROR;
		break;

	case HTTP_MSG_RPCODE:
	http_msg_rpcode:
		if (likely(!HTTP_IS_LWS(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpcode, http_msg_ood, state, HTTP_MSG_RPCODE);

		if (likely(HTTP_IS_SPHT(*ptr))) {
			msg->sl.st.c_l = ptr - msg_start - msg->sl.st.c;
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpcode_sp, http_msg_ood, state, HTTP_MSG_RPCODE_SP);
		}

		/* so it's a CR/LF, so there is no reason phrase */
		msg->sl.st.c_l = ptr - msg_start - msg->sl.st.c;
	http_msg_rsp_reason:
		/* FIXME: should we support HTTP responses without any reason phrase ? */
		msg->sl.st.r = ptr - msg_start;
		msg->sl.st.r_l = 0;
		goto http_msg_rpline_eol;

	case HTTP_MSG_RPCODE_SP:
	http_msg_rpcode_sp:
		if (likely(!HTTP_IS_LWS(*ptr))) {
			msg->sl.st.r = ptr - msg_start;
			goto http_msg_rpreason;
		}
		if (likely(HTTP_IS_SPHT(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpcode_sp, http_msg_ood, state, HTTP_MSG_RPCODE_SP);
		/* so it's a CR/LF, so there is no reason phrase */
		goto http_msg_rsp_reason;

	case HTTP_MSG_RPREASON:
	http_msg_rpreason:
		if (likely(!HTTP_IS_CRLF(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpreason, http_msg_ood, state, HTTP_MSG_RPREASON);
		msg->sl.st.r_l = ptr - msg_start - msg->sl.st.r;
	http_msg_rpline_eol:
		/* We have seen the end of line. Note that we do not
		 * necessarily have the \n yet, but at least we know that we
		 * have EITHER \r OR \n, otherwise the response would not be
		 * complete. We can then record the response length and return
		 * to the caller which will be able to register it.
		 */
		msg->sl.st.l = ptr - msg_start - msg->sol;
		return ptr;

	default:
#ifdef DEBUG_FULL
		fprintf(stderr, "FIXME !!!! impossible state at %s:%d = %d\n", __FILE__, __LINE__, state);
		exit(1);
#endif
		;
	}

 http_msg_ood:
	/* out of valid data */
	if (ret_state)
		*ret_state = state;
	if (ret_ptr)
		*ret_ptr = ptr - msg_start;
	return NULL;
}

/*
 * This function parses a request line between <ptr> and <end>, starting with
 * parser state <state>. Only states HTTP_MSG_RQMETH, HTTP_MSG_RQMETH_SP,
 * HTTP_MSG_RQURI, HTTP_MSG_RQURI_SP and HTTP_MSG_RQVER are handled. Others
 * will give undefined results.
 * Note that it is upon the caller's responsibility to ensure that ptr < end,
 * and that msg->sol points to the beginning of the request.
 * If a complete line is found (which implies that at least one CR or LF is
 * found before <end>, the updated <ptr> is returned, otherwise NULL is
 * returned indicating an incomplete line (which does not mean that parts have
 * not been updated). In the incomplete case, if <ret_ptr> or <ret_state> are
 * non-NULL, they are fed with the new <ptr> and <state> values to be passed
 * upon next call.
 *
 * This function was intentionally designed to be called from
 * http_msg_analyzer() with the lowest overhead. It should integrate perfectly
 * within its state machine and use the same macros, hence the need for same
 * labels and variable names. Note that msg->sol is left unchanged.
 */
const char *http_parse_reqline(struct http_msg *msg,
			       enum h1_state state, const char *ptr, const char *end,
			       unsigned int *ret_ptr, enum h1_state *ret_state)
{
	const char *msg_start = msg->chn->buf->p;

	switch (state)	{
	case HTTP_MSG_RQMETH:
	http_msg_rqmeth:
		if (likely(HTTP_IS_TOKEN(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqmeth, http_msg_ood, state, HTTP_MSG_RQMETH);

		if (likely(HTTP_IS_SPHT(*ptr))) {
			msg->sl.rq.m_l = ptr - msg_start;
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqmeth_sp, http_msg_ood, state, HTTP_MSG_RQMETH_SP);
		}

		if (likely(HTTP_IS_CRLF(*ptr))) {
			/* HTTP 0.9 request */
			msg->sl.rq.m_l = ptr - msg_start;
		http_msg_req09_uri:
			msg->sl.rq.u = ptr - msg_start;
		http_msg_req09_uri_e:
			msg->sl.rq.u_l = ptr - msg_start - msg->sl.rq.u;
		http_msg_req09_ver:
			msg->sl.rq.v = ptr - msg_start;
			msg->sl.rq.v_l = 0;
			goto http_msg_rqline_eol;
		}
		msg->err_state = HTTP_MSG_RQMETH;
		state = HTTP_MSG_ERROR;
		break;

	case HTTP_MSG_RQMETH_SP:
	http_msg_rqmeth_sp:
		if (likely(!HTTP_IS_LWS(*ptr))) {
			msg->sl.rq.u = ptr - msg_start;
			goto http_msg_rquri;
		}
		if (likely(HTTP_IS_SPHT(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqmeth_sp, http_msg_ood, state, HTTP_MSG_RQMETH_SP);
		/* so it's a CR/LF, meaning an HTTP 0.9 request */
		goto http_msg_req09_uri;

	case HTTP_MSG_RQURI:
	http_msg_rquri:
#if defined(__x86_64__) ||						\
    defined(__i386__) || defined(__i486__) || defined(__i586__) || defined(__i686__) || \
    defined(__ARM_ARCH_7A__)
		/* speedup: skip bytes not between 0x21 and 0x7e inclusive */
		while (ptr <= end - sizeof(int)) {
			int x = *(int *)ptr - 0x21212121;
			if (x & 0x80808080)
				break;

			x -= 0x5e5e5e5e;
			if (!(x & 0x80808080))
				break;

			ptr += sizeof(int);
		}
#endif
		if (ptr >= end) {
			state = HTTP_MSG_RQURI;
			goto http_msg_ood;
		}
	http_msg_rquri2:
		if (likely((unsigned char)(*ptr - 33) <= 93)) /* 33 to 126 included */
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rquri2, http_msg_ood, state, HTTP_MSG_RQURI);

		if (likely(HTTP_IS_SPHT(*ptr))) {
			msg->sl.rq.u_l = ptr - msg_start - msg->sl.rq.u;
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rquri_sp, http_msg_ood, state, HTTP_MSG_RQURI_SP);
		}

		if (likely((unsigned char)*ptr >= 128)) {
			/* non-ASCII chars are forbidden unless option
			 * accept-invalid-http-request is enabled in the frontend.
			 * In any case, we capture the faulty char.
			 */
			if (msg->err_pos < -1)
				goto invalid_char;
			if (msg->err_pos == -1)
				msg->err_pos = ptr - msg_start;
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rquri, http_msg_ood, state, HTTP_MSG_RQURI);
		}

		if (likely(HTTP_IS_CRLF(*ptr))) {
			/* so it's a CR/LF, meaning an HTTP 0.9 request */
			goto http_msg_req09_uri_e;
		}

		/* OK forbidden chars, 0..31 or 127 */
	invalid_char:
		msg->err_pos = ptr - msg_start;
		msg->err_state = HTTP_MSG_RQURI;
		state = HTTP_MSG_ERROR;
		break;

	case HTTP_MSG_RQURI_SP:
	http_msg_rquri_sp:
		if (likely(!HTTP_IS_LWS(*ptr))) {
			msg->sl.rq.v = ptr - msg_start;
			goto http_msg_rqver;
		}
		if (likely(HTTP_IS_SPHT(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rquri_sp, http_msg_ood, state, HTTP_MSG_RQURI_SP);
		/* so it's a CR/LF, meaning an HTTP 0.9 request */
		goto http_msg_req09_ver;

	case HTTP_MSG_RQVER:
	http_msg_rqver:
		if (likely(HTTP_IS_VER_TOKEN(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqver, http_msg_ood, state, HTTP_MSG_RQVER);

		if (likely(HTTP_IS_CRLF(*ptr))) {
			msg->sl.rq.v_l = ptr - msg_start - msg->sl.rq.v;
		http_msg_rqline_eol:
			/* We have seen the end of line. Note that we do not
			 * necessarily have the \n yet, but at least we know that we
			 * have EITHER \r OR \n, otherwise the request would not be
			 * complete. We can then record the request length and return
			 * to the caller which will be able to register it.
			 */
			msg->sl.rq.l = ptr - msg_start - msg->sol;
			return ptr;
		}

		/* neither an HTTP_VER token nor a CRLF */
		msg->err_state = HTTP_MSG_RQVER;
		state = HTTP_MSG_ERROR;
		break;

	default:
#ifdef DEBUG_FULL
		fprintf(stderr, "FIXME !!!! impossible state at %s:%d = %d\n", __FILE__, __LINE__, state);
		exit(1);
#endif
		;
	}

 http_msg_ood:
	/* out of valid data */
	if (ret_state)
		*ret_state = state;
	if (ret_ptr)
		*ret_ptr = ptr - msg_start;
	return NULL;
}

/*
 * This function parses an HTTP message, either a request or a response,
 * depending on the initial msg->msg_state. The caller is responsible for
 * ensuring that the message does not wrap. The function can be preempted
 * everywhere when data are missing and recalled at the exact same location
 * with no information loss. The message may even be realigned between two
 * calls. The header index is re-initialized when switching from
 * MSG_R[PQ]BEFORE to MSG_RPVER|MSG_RQMETH. It modifies msg->sol among other
 * fields. Note that msg->sol will be initialized after completing the first
 * state, so that none of the msg pointers has to be initialized prior to the
 * first call.
 */
void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
{
	enum h1_state state;       /* updated only when leaving the FSM */
	register char *ptr, *end; /* request pointers, to avoid dereferences */
	struct buffer *buf;

	state = msg->msg_state;
	buf = msg->chn->buf;
	ptr = buf->p + msg->next;
	end = buf->p + buf->i;

	if (unlikely(ptr >= end))
		goto http_msg_ood;

	switch (state)	{
	/*
	 * First, states that are specific to the response only.
	 * We check them first so that request and headers are
	 * closer to each other (accessed more often).
	 */
	case HTTP_MSG_RPBEFORE:
	http_msg_rpbefore:
		if (likely(HTTP_IS_TOKEN(*ptr))) {
			/* we have a start of message, but we have to check
			 * first if we need to remove some CRLF. We can only
			 * do this when o=0.
			 */
			if (unlikely(ptr != buf->p)) {
				if (buf->o)
					goto http_msg_ood;
				/* Remove empty leading lines, as recommended by RFC2616. */
				bi_fast_delete(buf, ptr - buf->p);
			}
			msg->sol = 0;
			msg->sl.st.l = 0; /* used in debug mode */
			hdr_idx_init(idx);
			state = HTTP_MSG_RPVER;
			goto http_msg_rpver;
		}

		if (unlikely(!HTTP_IS_CRLF(*ptr))) {
			state = HTTP_MSG_RPBEFORE;
			goto http_msg_invalid;
		}

		if (unlikely(*ptr == '\n'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpbefore, http_msg_ood, state, HTTP_MSG_RPBEFORE);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpbefore_cr, http_msg_ood, state, HTTP_MSG_RPBEFORE_CR);
		/* stop here */

	case HTTP_MSG_RPBEFORE_CR:
	http_msg_rpbefore_cr:
		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_RPBEFORE_CR);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpbefore, http_msg_ood, state, HTTP_MSG_RPBEFORE);
		/* stop here */

	case HTTP_MSG_RPVER:
	http_msg_rpver:
	case HTTP_MSG_RPVER_SP:
	case HTTP_MSG_RPCODE:
	case HTTP_MSG_RPCODE_SP:
	case HTTP_MSG_RPREASON:
		ptr = (char *)http_parse_stsline(msg,
						 state, ptr, end,
						 &msg->next, &msg->msg_state);
		if (unlikely(!ptr))
			return;

		/* we have a full response and we know that we have either a CR
		 * or an LF at <ptr>.
		 */
		hdr_idx_set_start(idx, msg->sl.st.l, *ptr == '\r');

		msg->sol = ptr - buf->p;
		if (likely(*ptr == '\r'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpline_end, http_msg_ood, state, HTTP_MSG_RPLINE_END);
		goto http_msg_rpline_end;

	case HTTP_MSG_RPLINE_END:
	http_msg_rpline_end:
		/* msg->sol must point to the first of CR or LF. */
		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_RPLINE_END);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_first, http_msg_ood, state, HTTP_MSG_HDR_FIRST);
		/* stop here */

	/*
	 * Second, states that are specific to the request only
	 */
	case HTTP_MSG_RQBEFORE:
	http_msg_rqbefore:
		if (likely(HTTP_IS_TOKEN(*ptr))) {
			/* we have a start of message, but we have to check
			 * first if we need to remove some CRLF. We can only
			 * do this when o=0.
			 */
			if (likely(ptr != buf->p)) {
				if (buf->o)
					goto http_msg_ood;
				/* Remove empty leading lines, as recommended by RFC2616. */
				bi_fast_delete(buf, ptr - buf->p);
			}
			msg->sol = 0;
			msg->sl.rq.l = 0; /* used in debug mode */
			state = HTTP_MSG_RQMETH;
			goto http_msg_rqmeth;
		}

		if (unlikely(!HTTP_IS_CRLF(*ptr))) {
			state = HTTP_MSG_RQBEFORE;
			goto http_msg_invalid;
		}

		if (unlikely(*ptr == '\n'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqbefore, http_msg_ood, state, HTTP_MSG_RQBEFORE);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqbefore_cr, http_msg_ood, state, HTTP_MSG_RQBEFORE_CR);
		/* stop here */

	case HTTP_MSG_RQBEFORE_CR:
	http_msg_rqbefore_cr:
		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_RQBEFORE_CR);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqbefore, http_msg_ood, state, HTTP_MSG_RQBEFORE);
		/* stop here */

	case HTTP_MSG_RQMETH:
	http_msg_rqmeth:
	case HTTP_MSG_RQMETH_SP:
	case HTTP_MSG_RQURI:
	case HTTP_MSG_RQURI_SP:
	case HTTP_MSG_RQVER:
		ptr = (char *)http_parse_reqline(msg,
						 state, ptr, end,
						 &msg->next, &msg->msg_state);
		if (unlikely(!ptr))
			return;

		/* we have a full request and we know that we have either a CR
		 * or an LF at <ptr>.
		 */
		hdr_idx_set_start(idx, msg->sl.rq.l, *ptr == '\r');

		msg->sol = ptr - buf->p;
		if (likely(*ptr == '\r'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqline_end, http_msg_ood, state, HTTP_MSG_RQLINE_END);
		goto http_msg_rqline_end;

	case HTTP_MSG_RQLINE_END:
	http_msg_rqline_end:
		/* check for HTTP/0.9 request : no version information available.
		 * msg->sol must point to the first of CR or LF.
		 */
		if (unlikely(msg->sl.rq.v_l == 0))
			goto http_msg_last_lf;

		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_RQLINE_END);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_first, http_msg_ood, state, HTTP_MSG_HDR_FIRST);
		/* stop here */

	/*
	 * Common states below
	 */
	case HTTP_MSG_HDR_FIRST:
	http_msg_hdr_first:
		msg->sol = ptr - buf->p;
		if (likely(!HTTP_IS_CRLF(*ptr))) {
			goto http_msg_hdr_name;
		}

		if (likely(*ptr == '\r'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_last_lf, http_msg_ood, state, HTTP_MSG_LAST_LF);
		goto http_msg_last_lf;

	case HTTP_MSG_HDR_NAME:
	http_msg_hdr_name:
		/* assumes msg->sol points to the first char */
		if (likely(HTTP_IS_TOKEN(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_name, http_msg_ood, state, HTTP_MSG_HDR_NAME);

		if (likely(*ptr == ':'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l1_sp, http_msg_ood, state, HTTP_MSG_HDR_L1_SP);

		if (likely(msg->err_pos < -1) || *ptr == '\n') {
			state = HTTP_MSG_HDR_NAME;
			goto http_msg_invalid;
		}

		if (msg->err_pos == -1) /* capture error pointer */
			msg->err_pos = ptr - buf->p; /* >= 0 now */

		/* and we still accept this non-token character */
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_name, http_msg_ood, state, HTTP_MSG_HDR_NAME);

	case HTTP_MSG_HDR_L1_SP:
	http_msg_hdr_l1_sp:
		/* assumes msg->sol points to the first char */
		if (likely(HTTP_IS_SPHT(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l1_sp, http_msg_ood, state, HTTP_MSG_HDR_L1_SP);

		/* header value can be basically anything except CR/LF */
		msg->sov = ptr - buf->p;

		if (likely(!HTTP_IS_CRLF(*ptr))) {
			goto http_msg_hdr_val;
		}

		if (likely(*ptr == '\r'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l1_lf, http_msg_ood, state, HTTP_MSG_HDR_L1_LF);
		goto http_msg_hdr_l1_lf;

	case HTTP_MSG_HDR_L1_LF:
	http_msg_hdr_l1_lf:
		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_HDR_L1_LF);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l1_lws, http_msg_ood, state, HTTP_MSG_HDR_L1_LWS);

	case HTTP_MSG_HDR_L1_LWS:
	http_msg_hdr_l1_lws:
		if (likely(HTTP_IS_SPHT(*ptr))) {
			/* replace HT,CR,LF with spaces */
			for (; buf->p + msg->sov < ptr; msg->sov++)
				buf->p[msg->sov] = ' ';
			goto http_msg_hdr_l1_sp;
		}
		/* we had a header consisting only in spaces ! */
		msg->eol = msg->sov;
		goto http_msg_complete_header;

	case HTTP_MSG_HDR_VAL:
	http_msg_hdr_val:
		/* assumes msg->sol points to the first char, and msg->sov
		 * points to the first character of the value.
		 */

		/* speedup: we'll skip packs of 4 or 8 bytes not containing bytes 0x0D
		 * and lower. In fact since most of the time is spent in the loop, we
		 * also remove the sign bit test so that bytes 0x8e..0x0d break the
		 * loop, but we don't care since they're very rare in header values.
		 */
#if defined(__x86_64__)
		while (ptr <= end - sizeof(long)) {
			if ((*(long *)ptr - 0x0e0e0e0e0e0e0e0eULL) & 0x8080808080808080ULL)
				goto http_msg_hdr_val2;
			ptr += sizeof(long);
		}
#endif
#if defined(__x86_64__) || \
    defined(__i386__) || defined(__i486__) || defined(__i586__) || defined(__i686__) || \
    defined(__ARM_ARCH_7A__)
		while (ptr <= end - sizeof(int)) {
			if ((*(int*)ptr - 0x0e0e0e0e) & 0x80808080)
				goto http_msg_hdr_val2;
			ptr += sizeof(int);
		}
#endif
		if (ptr >= end) {
			state = HTTP_MSG_HDR_VAL;
			goto http_msg_ood;
		}
	http_msg_hdr_val2:
		if (likely(!HTTP_IS_CRLF(*ptr)))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_val2, http_msg_ood, state, HTTP_MSG_HDR_VAL);

		msg->eol = ptr - buf->p;
		/* Note: we could also copy eol into ->eoh so that we have the
		 * real header end in case it ends with lots of LWS, but is this
		 * really needed ?
		 */
		if (likely(*ptr == '\r'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l2_lf, http_msg_ood, state, HTTP_MSG_HDR_L2_LF);
		goto http_msg_hdr_l2_lf;

	case HTTP_MSG_HDR_L2_LF:
	http_msg_hdr_l2_lf:
		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_HDR_L2_LF);
		EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l2_lws, http_msg_ood, state, HTTP_MSG_HDR_L2_LWS);

	case HTTP_MSG_HDR_L2_LWS:
	http_msg_hdr_l2_lws:
		if (unlikely(HTTP_IS_SPHT(*ptr))) {
			/* LWS: replace HT,CR,LF with spaces */
			for (; buf->p + msg->eol < ptr; msg->eol++)
				buf->p[msg->eol] = ' ';
			goto http_msg_hdr_val;
		}
	http_msg_complete_header:
		/*
		 * It was a new header, so the last one is finished.
		 * Assumes msg->sol points to the first char, msg->sov points
		 * to the first character of the value and msg->eol to the
		 * first CR or LF so we know how the line ends. We insert last
		 * header into the index.
		 */
		if (unlikely(hdr_idx_add(msg->eol - msg->sol, buf->p[msg->eol] == '\r',
					 idx, idx->tail) < 0)) {
			state = HTTP_MSG_HDR_L2_LWS;
			goto http_msg_invalid;
		}

		msg->sol = ptr - buf->p;
		if (likely(!HTTP_IS_CRLF(*ptr))) {
			goto http_msg_hdr_name;
		}

		if (likely(*ptr == '\r'))
			EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_last_lf, http_msg_ood, state, HTTP_MSG_LAST_LF);
		goto http_msg_last_lf;

	case HTTP_MSG_LAST_LF:
	http_msg_last_lf:
		/* Assumes msg->sol points to the first of either CR or LF.
		 * Sets ->sov and ->next to the total header length, ->eoh to
		 * the last CRLF, and ->eol to the last CRLF length (1 or 2).
		 */
		EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_LAST_LF);
		ptr++;
		msg->sov = msg->next = ptr - buf->p;
		msg->eoh = msg->sol;
		msg->sol = 0;
		msg->eol = msg->sov - msg->eoh;
		msg->msg_state = HTTP_MSG_BODY;
		return;

	case HTTP_MSG_ERROR:
		/* this may only happen if we call http_msg_analyser() twice with an error */
		break;

	default:
#ifdef DEBUG_FULL
		fprintf(stderr, "FIXME !!!! impossible state at %s:%d = %d\n", __FILE__, __LINE__, state);
		exit(1);
#endif
		;
	}
 http_msg_ood:
	/* out of data */
	msg->msg_state = state;
	msg->next = ptr - buf->p;
	return;

 http_msg_invalid:
	/* invalid message */
	msg->err_state = state;
	msg->msg_state = HTTP_MSG_ERROR;
	msg->next = ptr - buf->p;
	return;
}

/* This function skips trailers in the buffer associated with HTTP message
 * <msg>. The first visited position is msg->next. If the end of the trailers is
 * found, the function returns >0. So, the caller can automatically schedul it
 * to be forwarded, and switch msg->msg_state to HTTP_MSG_DONE. If not enough
 * data are available, the function does not change anything except maybe
 * msg->sol if it could parse some lines, and returns zero.  If a parse error
 * is encountered, the function returns < 0 and does not change anything except
 * maybe msg->sol. Note that the message must already be in HTTP_MSG_TRAILERS
 * state before calling this function, which implies that all non-trailers data
 * have already been scheduled for forwarding, and that msg->next exactly
 * matches the length of trailers already parsed and not forwarded. It is also
 * important to note that this function is designed to be able to parse wrapped
 * headers at end of buffer.
 */
int http_forward_trailers(struct http_msg *msg)
{
	const struct buffer *buf = msg->chn->buf;

	/* we have msg->next which points to next line. Look for CRLF. But
	 * first, we reset msg->sol */
	msg->sol = 0;
	while (1) {
		const char *p1 = NULL, *p2 = NULL;
		const char *start = b_ptr(buf, msg->next + msg->sol);
		const char *stop  = bi_end(buf);
		const char *ptr   = start;
		int bytes = 0;

		/* scan current line and stop at LF or CRLF */
		while (1) {
			if (ptr == stop)
				return 0;

			if (*ptr == '\n') {
				if (!p1)
					p1 = ptr;
				p2 = ptr;
				break;
			}

			if (*ptr == '\r') {
				if (p1) {
					msg->err_pos = buffer_count(buf, buf->p, ptr);
					return -1;
				}
				p1 = ptr;
			}

			ptr++;
			if (ptr >= buf->data + buf->size)
				ptr = buf->data;
		}

		/* after LF; point to beginning of next line */
		p2++;
		if (p2 >= buf->data + buf->size)
			p2 = buf->data;

		bytes = p2 - start;
		if (bytes < 0)
			bytes += buf->size;
		msg->sol += bytes;

		/* LF/CRLF at beginning of line => end of trailers at p2.
		 * Everything was scheduled for forwarding, there's nothing left
		 * from this message. */
		if (p1 == start)
			return 1;

		/* OK, next line then */
	}
}