RepoMirrors/haproxy
1
0
Fork 0
mirror of http://git.haproxy.org/git/haproxy.git/ synced 2025-04-19 13:35:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

MEDIUM: h1: port to new buffer API.

Browse Source 
The parser now uses the channel exclusively to access the data. In order
to avoid the cost of indirection, a local variable "input" was added to
the function that replaces buf->p. Given that this part is on the critical
path, it will have to be tested again for any visible performance loss.
This commit is contained in:
Willy Tarreau 2018-06-19 08:03:19 +02:00
parent fc0785d26c
commit 5e74b0ba3b
1 changed files with 33 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

73
src/h1.c
View File

@ -182,7 +182,7 @@ const char *http_parse_stsline(struct http_msg *msg,
enum h1_state state, const char *ptr, const char *end, enum h1_state state, const char *ptr, const char *end,
unsigned int *ret_ptr, enum h1_state *ret_state) unsigned int *ret_ptr, enum h1_state *ret_state)
{ {
const char *msg_start = msg->chn->buf->p; const char *msg_start = ci_head(msg->chn);
switch (state) { switch (state) {
case HTTP_MSG_RPVER: case HTTP_MSG_RPVER:
@ -295,7 +295,7 @@ const char *http_parse_reqline(struct http_msg *msg,
enum h1_state state, const char *ptr, const char *end, enum h1_state state, const char *ptr, const char *end,
unsigned int *ret_ptr, enum h1_state *ret_state) unsigned int *ret_ptr, enum h1_state *ret_state)
{ {
const char *msg_start = msg->chn->buf->p; const char *msg_start = ci_head(msg->chn);
switch (state) { switch (state) {
case HTTP_MSG_RQMETH: case HTTP_MSG_RQMETH:
@ -456,13 +456,14 @@ const char *http_parse_reqline(struct http_msg *msg,
void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx) void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
{ {
enum h1_state state; /* updated only when leaving the FSM */ enum h1_state state; /* updated only when leaving the FSM */
register char *ptr, *end; /* request pointers, to avoid dereferences */ register const char *ptr, *end; /* request pointers, to avoid dereferences */
char *input = (char *)ci_head(msg->chn);
struct buffer *buf; struct buffer *buf;
state = msg->msg_state; state = msg->msg_state;
buf = msg->chn->buf; buf = msg->chn->buf;
ptr = buf->p + msg->next; ptr = input + msg->next;
end = buf->p + buf->i; end = b_stop(buf);
if (unlikely(ptr >= end)) if (unlikely(ptr >= end))
goto http_msg_ood; goto http_msg_ood;
@ -480,11 +481,11 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
* first if we need to remove some CRLF. We can only * first if we need to remove some CRLF. We can only
* do this when o=0. * do this when o=0.
*/ */
if (unlikely(ptr != buf->p)) { if (unlikely(ptr != input)) {
if (buf->o) if (co_data(msg->chn))
goto http_msg_ood; goto http_msg_ood;
/* Remove empty leading lines, as recommended by RFC2616. */ /* Remove empty leading lines, as recommended by RFC2616. */
bi_fast_delete(buf, ptr - buf->p); bi_fast_delete(buf, ptr - input);
} }
msg->sol = 0; msg->sol = 0;
msg->sl.st.l = 0; /* used in debug mode */ msg->sl.st.l = 0; /* used in debug mode */
@ -526,7 +527,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
*/ */
hdr_idx_set_start(idx, msg->sl.st.l, *ptr == '\r'); hdr_idx_set_start(idx, msg->sl.st.l, *ptr == '\r');
msg->sol = ptr - buf->p; msg->sol = ptr - input;
if (likely(*ptr == '\r')) if (likely(*ptr == '\r'))
EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpline_end, http_msg_ood, state, HTTP_MSG_RPLINE_END); EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rpline_end, http_msg_ood, state, HTTP_MSG_RPLINE_END);
goto http_msg_rpline_end; goto http_msg_rpline_end;
@ -548,11 +549,11 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
* first if we need to remove some CRLF. We can only * first if we need to remove some CRLF. We can only
* do this when o=0. * do this when o=0.
*/ */
if (likely(ptr != buf->p)) { if (likely(ptr != input)) {
if (buf->o) if (co_data(msg->chn))
goto http_msg_ood; goto http_msg_ood;
/* Remove empty leading lines, as recommended by RFC2616. */ /* Remove empty leading lines, as recommended by RFC2616. */
bi_fast_delete(buf, ptr - buf->p); bi_fast_delete(buf, ptr - input);
} }
msg->sol = 0; msg->sol = 0;
msg->sl.rq.l = 0; /* used in debug mode */ msg->sl.rq.l = 0; /* used in debug mode */
@ -593,7 +594,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
*/ */
hdr_idx_set_start(idx, msg->sl.rq.l, *ptr == '\r'); hdr_idx_set_start(idx, msg->sl.rq.l, *ptr == '\r');
msg->sol = ptr - buf->p; msg->sol = ptr - input;
if (likely(*ptr == '\r')) if (likely(*ptr == '\r'))
EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqline_end, http_msg_ood, state, HTTP_MSG_RQLINE_END); EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_rqline_end, http_msg_ood, state, HTTP_MSG_RQLINE_END);
goto http_msg_rqline_end; goto http_msg_rqline_end;
@ -615,7 +616,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
*/ */
case HTTP_MSG_HDR_FIRST: case HTTP_MSG_HDR_FIRST:
http_msg_hdr_first: http_msg_hdr_first:
msg->sol = ptr - buf->p; msg->sol = ptr - input;
if (likely(!HTTP_IS_CRLF(*ptr))) { if (likely(!HTTP_IS_CRLF(*ptr))) {
goto http_msg_hdr_name; goto http_msg_hdr_name;
} }
@ -639,7 +640,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
} }
if (msg->err_pos == -1) /* capture error pointer */ if (msg->err_pos == -1) /* capture error pointer */
msg->err_pos = ptr - buf->p; /* >= 0 now */ msg->err_pos = ptr - input; /* >= 0 now */
/* and we still accept this non-token character */ /* 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); EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_name, http_msg_ood, state, HTTP_MSG_HDR_NAME);
@ -651,7 +652,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_l1_sp, http_msg_ood, state, HTTP_MSG_HDR_L1_SP); 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 */ /* header value can be basically anything except CR/LF */
msg->sov = ptr - buf->p; msg->sov = ptr - input;
if (likely(!HTTP_IS_CRLF(*ptr))) { if (likely(!HTTP_IS_CRLF(*ptr))) {
goto http_msg_hdr_val; goto http_msg_hdr_val;
@ -670,8 +671,8 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
http_msg_hdr_l1_lws: http_msg_hdr_l1_lws:
if (likely(HTTP_IS_SPHT(*ptr))) { if (likely(HTTP_IS_SPHT(*ptr))) {
/* replace HT,CR,LF with spaces */ /* replace HT,CR,LF with spaces */
for (; buf->p + msg->sov < ptr; msg->sov++) for (; input + msg->sov < ptr; msg->sov++)
buf->p[msg->sov] = ' '; input[msg->sov] = ' ';
goto http_msg_hdr_l1_sp; goto http_msg_hdr_l1_sp;
} }
/* we had a header consisting only in spaces ! */ /* we had a header consisting only in spaces ! */
@ -713,7 +714,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
if (likely(!HTTP_IS_CRLF(*ptr))) 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); EAT_AND_JUMP_OR_RETURN(ptr, end, http_msg_hdr_val2, http_msg_ood, state, HTTP_MSG_HDR_VAL);
msg->eol = ptr - buf->p; msg->eol = ptr - input;
/* Note: we could also copy eol into ->eoh so that we have the /* 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 * real header end in case it ends with lots of LWS, but is this
* really needed ? * really needed ?
@ -731,8 +732,8 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
http_msg_hdr_l2_lws: http_msg_hdr_l2_lws:
if (unlikely(HTTP_IS_SPHT(*ptr))) { if (unlikely(HTTP_IS_SPHT(*ptr))) {
/* LWS: replace HT,CR,LF with spaces */ /* LWS: replace HT,CR,LF with spaces */
for (; buf->p + msg->eol < ptr; msg->eol++) for (; input + msg->eol < ptr; msg->eol++)
buf->p[msg->eol] = ' '; input[msg->eol] = ' ';
goto http_msg_hdr_val; goto http_msg_hdr_val;
} }
http_msg_complete_header: http_msg_complete_header:
@ -743,13 +744,13 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
* first CR or LF so we know how the line ends. We insert last * first CR or LF so we know how the line ends. We insert last
* header into the index. * header into the index.
*/ */
if (unlikely(hdr_idx_add(msg->eol - msg->sol, buf->p[msg->eol] == '\r', if (unlikely(hdr_idx_add(msg->eol - msg->sol, input[msg->eol] == '\r',
idx, idx->tail) < 0)) { idx, idx->tail) < 0)) {
state = HTTP_MSG_HDR_L2_LWS; state = HTTP_MSG_HDR_L2_LWS;
goto http_msg_invalid; goto http_msg_invalid;
} }
msg->sol = ptr - buf->p; msg->sol = ptr - input;
if (likely(!HTTP_IS_CRLF(*ptr))) { if (likely(!HTTP_IS_CRLF(*ptr))) {
goto http_msg_hdr_name; goto http_msg_hdr_name;
} }
@ -766,7 +767,7 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
*/ */
EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_LAST_LF); EXPECT_LF_HERE(ptr, http_msg_invalid, state, HTTP_MSG_LAST_LF);
ptr++; ptr++;
msg->sov = msg->next = ptr - buf->p; msg->sov = msg->next = ptr - input;
msg->eoh = msg->sol; msg->eoh = msg->sol;
msg->sol = 0; msg->sol = 0;
msg->eol = msg->sov - msg->eoh; msg->eol = msg->sov - msg->eoh;
@ -787,14 +788,14 @@ void http_msg_analyzer(struct http_msg *msg, struct hdr_idx *idx)
http_msg_ood: http_msg_ood:
/* out of data */ /* out of data */
msg->msg_state = state; msg->msg_state = state;
msg->next = ptr - buf->p; msg->next = ptr - input;
return; return;
http_msg_invalid: http_msg_invalid:
/* invalid message */ /* invalid message */
msg->err_state = state; msg->err_state = state;
msg->msg_state = HTTP_MSG_ERROR; msg->msg_state = HTTP_MSG_ERROR;
msg->next = ptr - buf->p; msg->next = ptr - input;
return; return;
} }
@ -1276,6 +1277,8 @@ int h1_measure_trailers(const struct buffer *buf, unsigned int ofs, unsigned int
int http_forward_trailers(struct http_msg *msg) int http_forward_trailers(struct http_msg *msg)
{ {
const struct buffer *buf = msg->chn->buf; const struct buffer *buf = msg->chn->buf;
const char *parse = ci_head(msg->chn);
const char *stop = b_tail(buf);
/* we have msg->next which points to next line. Look for CRLF. But /* we have msg->next which points to next line. Look for CRLF. But
* first, we reset msg->sol */ * first, we reset msg->sol */
@ -1283,9 +1286,7 @@ int http_forward_trailers(struct http_msg *msg)
while (1) { while (1) {
const char *p1 = NULL, *p2 = NULL; const char *p1 = NULL, *p2 = NULL;
const char *start = c_ptr(msg->chn, msg->next + msg->sol); const char *start = c_ptr(msg->chn, msg->next + msg->sol);
const char *stop = b_tail(buf);
const char *ptr = start; const char *ptr = start;
int bytes = 0;
/* scan current line and stop at LF or CRLF */ /* scan current line and stop at LF or CRLF */
while (1) { while (1) {
@ -1301,26 +1302,18 @@ int http_forward_trailers(struct http_msg *msg)
if (*ptr == '\r') { if (*ptr == '\r') {
if (p1) { if (p1) {
msg->err_pos = buffer_count(buf, buf->p, ptr); msg->err_pos = b_dist(buf, parse, ptr);
return -1; return -1;
} }
p1 = ptr; p1 = ptr;
} }
ptr++; ptr = b_next(buf, ptr);
if (ptr >= buf->data + buf->size)
ptr = buf->data;
} }
/* after LF; point to beginning of next line */ /* after LF; point to beginning of next line */
p2++; p2 = b_next(buf, p2);
if (p2 >= buf->data + buf->size) msg->sol += b_dist(buf, start, p2);
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. /* LF/CRLF at beginning of line => end of trailers at p2.
* Everything was scheduled for forwarding, there's nothing left * Everything was scheduled for forwarding, there's nothing left