blob: 91e4f8e0c095a3db92fde13274e091a349b99c82 [file] [log] [blame]
Christopher Faulet3d97c902015-12-09 14:59:38 +01001/*
2 * Stream filters related variables and functions.
3 *
4 * Copyright (C) 2015 Qualys Inc., Christopher Faulet <cfaulet@qualys.com>
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 *
11 */
12
Willy Tarreau4c7e4b72020-05-27 12:58:42 +020013#include <haproxy/api.h>
Willy Tarreau6be78492020-06-05 00:00:29 +020014#include <haproxy/cfgparse.h>
Willy Tarreau0a3bd392020-06-04 08:52:38 +020015#include <haproxy/compression.h>
Willy Tarreau2741c8c2020-06-02 11:28:02 +020016#include <haproxy/dynbuf.h>
Willy Tarreauc7babd82020-06-04 21:29:29 +020017#include <haproxy/filters.h>
Willy Tarreaucd72d8c2020-06-02 19:11:26 +020018#include <haproxy/http.h>
Willy Tarreauc2b1ff02020-06-04 21:21:03 +020019#include <haproxy/http_ana-t.h>
Willy Tarreau87735332020-06-04 09:08:41 +020020#include <haproxy/http_htx.h>
Willy Tarreau16f958c2020-06-03 08:44:35 +020021#include <haproxy/htx.h>
Willy Tarreau853b2972020-05-27 18:01:47 +020022#include <haproxy/list.h>
Willy Tarreaua264d962020-06-04 22:29:18 +020023#include <haproxy/proxy-t.h>
Willy Tarreaue6ce10b2020-06-04 15:33:47 +020024#include <haproxy/sample.h>
Willy Tarreaudfd3de82020-06-04 23:46:14 +020025#include <haproxy/stream.h>
Willy Tarreau48fbcae2020-06-03 18:09:46 +020026#include <haproxy/tools.h>
Christopher Faulet3d97c902015-12-09 14:59:38 +010027
Christopher Faulet4288f7d2021-06-09 17:12:44 +020028#define COMP_STATE_PROCESSING 0x01
29
Christopher Fauletf4a4ef72018-12-07 17:39:53 +010030const char *http_comp_flt_id = "compression filter";
Christopher Faulet92d36382015-11-05 13:35:03 +010031
32struct flt_ops comp_ops;
33
Christopher Faulet92d36382015-11-05 13:35:03 +010034struct comp_state {
35 struct comp_ctx *comp_ctx; /* compression context */
36 struct comp_algo *comp_algo; /* compression algorithm if not NULL */
Christopher Faulet4288f7d2021-06-09 17:12:44 +020037 unsigned int flags; /* COMP_STATE_* */
Christopher Faulet92d36382015-11-05 13:35:03 +010038};
39
Willy Tarreau8ceae722018-11-26 11:58:30 +010040/* Pools used to allocate comp_state structs */
41DECLARE_STATIC_POOL(pool_head_comp_state, "comp_state", sizeof(struct comp_state));
42
43static THREAD_LOCAL struct buffer tmpbuf;
44static THREAD_LOCAL struct buffer zbuf;
Willy Tarreau8ceae722018-11-26 11:58:30 +010045
Christopher Faulet92d36382015-11-05 13:35:03 +010046static int select_compression_request_header(struct comp_state *st,
47 struct stream *s,
48 struct http_msg *msg);
49static int select_compression_response_header(struct comp_state *st,
50 struct stream *s,
51 struct http_msg *msg);
Christopher Faulet27d93c32018-12-15 22:32:02 +010052static int set_compression_response_header(struct comp_state *st,
53 struct stream *s,
54 struct http_msg *msg);
Christopher Faulet92d36382015-11-05 13:35:03 +010055
Christopher Faulete6902cd2018-11-30 22:29:48 +010056static int htx_compression_buffer_init(struct htx *htx, struct buffer *out);
57static int htx_compression_buffer_add_data(struct comp_state *st, const char *data, size_t len,
58 struct buffer *out);
59static int htx_compression_buffer_end(struct comp_state *st, struct buffer *out, int end);
60
Christopher Faulet92d36382015-11-05 13:35:03 +010061/***********************************************************************/
62static int
Christopher Faulete6902cd2018-11-30 22:29:48 +010063comp_flt_init(struct proxy *px, struct flt_conf *fconf)
64{
Christopher Faulet6e540952018-12-03 22:43:41 +010065 fconf->flags |= FLT_CFG_FL_HTX;
Christopher Faulete6902cd2018-11-30 22:29:48 +010066 return 0;
67}
68
69static int
Christopher Faulet8ca3b4b2017-07-25 11:07:15 +020070comp_flt_init_per_thread(struct proxy *px, struct flt_conf *fconf)
Christopher Faulet92d36382015-11-05 13:35:03 +010071{
Willy Tarreauc9fa0482018-07-10 17:43:27 +020072 if (!tmpbuf.size && b_alloc(&tmpbuf) == NULL)
Christopher Fauletb77c5c22015-12-07 16:48:42 +010073 return -1;
Willy Tarreauc9fa0482018-07-10 17:43:27 +020074 if (!zbuf.size && b_alloc(&zbuf) == NULL)
Christopher Fauletb77c5c22015-12-07 16:48:42 +010075 return -1;
Christopher Faulet92d36382015-11-05 13:35:03 +010076 return 0;
77}
78
79static void
Christopher Faulet8ca3b4b2017-07-25 11:07:15 +020080comp_flt_deinit_per_thread(struct proxy *px, struct flt_conf *fconf)
Christopher Faulet92d36382015-11-05 13:35:03 +010081{
Willy Tarreauc9fa0482018-07-10 17:43:27 +020082 if (tmpbuf.size)
Christopher Faulet92d36382015-11-05 13:35:03 +010083 b_free(&tmpbuf);
Willy Tarreauc9fa0482018-07-10 17:43:27 +020084 if (zbuf.size)
Christopher Fauletb77c5c22015-12-07 16:48:42 +010085 b_free(&zbuf);
Christopher Faulet92d36382015-11-05 13:35:03 +010086}
87
88static int
Christopher Faulet5e896512020-03-06 14:59:05 +010089comp_strm_init(struct stream *s, struct filter *filter)
Christopher Faulet92d36382015-11-05 13:35:03 +010090{
Christopher Faulet5e896512020-03-06 14:59:05 +010091 struct comp_state *st;
Christopher Faulet8ca3b4b2017-07-25 11:07:15 +020092
Christopher Faulet5e896512020-03-06 14:59:05 +010093 st = pool_alloc_dirty(pool_head_comp_state);
94 if (st == NULL)
95 return -1;
Christopher Faulet92d36382015-11-05 13:35:03 +010096
Christopher Faulet5e896512020-03-06 14:59:05 +010097 st->comp_algo = NULL;
98 st->comp_ctx = NULL;
Christopher Faulet4288f7d2021-06-09 17:12:44 +020099 st->flags = 0;
Christopher Faulet5e896512020-03-06 14:59:05 +0100100 filter->ctx = st;
Christopher Faulet3dc860d2017-09-15 11:39:36 +0200101
Christopher Faulet5e896512020-03-06 14:59:05 +0100102 /* Register post-analyzer on AN_RES_WAIT_HTTP because we need to
103 * analyze response headers before http-response rules execution
104 * to be sure we can use res.comp and res.comp_algo sample
105 * fetches */
106 filter->post_analyzers |= AN_RES_WAIT_HTTP;
Christopher Faulet92d36382015-11-05 13:35:03 +0100107 return 1;
108}
109
Christopher Faulet5e896512020-03-06 14:59:05 +0100110static void
111comp_strm_deinit(struct stream *s, struct filter *filter)
Christopher Faulet92d36382015-11-05 13:35:03 +0100112{
113 struct comp_state *st = filter->ctx;
Christopher Faulet92d36382015-11-05 13:35:03 +0100114
Christopher Fauletd60b3cf2017-06-26 11:47:13 +0200115 if (!st)
Christopher Faulet5e896512020-03-06 14:59:05 +0100116 return;
Christopher Faulet92d36382015-11-05 13:35:03 +0100117
Christopher Faulet92d36382015-11-05 13:35:03 +0100118 /* release any possible compression context */
Christopher Fauletd60b3cf2017-06-26 11:47:13 +0200119 if (st->comp_algo)
120 st->comp_algo->end(&st->comp_ctx);
Willy Tarreaubafbe012017-11-24 17:34:44 +0100121 pool_free(pool_head_comp_state, st);
Christopher Faulet92d36382015-11-05 13:35:03 +0100122 filter->ctx = NULL;
Christopher Faulet92d36382015-11-05 13:35:03 +0100123}
124
125static int
Christopher Faulet1339d742016-05-11 16:48:33 +0200126comp_http_headers(struct stream *s, struct filter *filter, struct http_msg *msg)
127{
128 struct comp_state *st = filter->ctx;
129
130 if (!strm_fe(s)->comp && !s->be->comp)
131 goto end;
132
133 if (!(msg->chn->flags & CF_ISRESP))
134 select_compression_request_header(st, s, msg);
135 else {
Christopher Faulet3dc860d2017-09-15 11:39:36 +0200136 /* Response headers have already been checked in
137 * comp_http_post_analyze callback. */
Christopher Faulet1339d742016-05-11 16:48:33 +0200138 if (st->comp_algo) {
Christopher Faulet27d93c32018-12-15 22:32:02 +0100139 if (!set_compression_response_header(st, s, msg))
140 goto end;
Christopher Faulet1339d742016-05-11 16:48:33 +0200141 register_data_filter(s, msg->chn, filter);
Christopher Faulet4288f7d2021-06-09 17:12:44 +0200142 st->flags |= COMP_STATE_PROCESSING;
Christopher Faulet1339d742016-05-11 16:48:33 +0200143 }
144 }
145
146 end:
147 return 1;
148}
149
150static int
Christopher Faulet3dc860d2017-09-15 11:39:36 +0200151comp_http_post_analyze(struct stream *s, struct filter *filter,
152 struct channel *chn, unsigned an_bit)
153{
154 struct http_txn *txn = s->txn;
155 struct http_msg *msg = &txn->rsp;
156 struct comp_state *st = filter->ctx;
157
158 if (an_bit != AN_RES_WAIT_HTTP)
159 goto end;
160
161 if (!strm_fe(s)->comp && !s->be->comp)
162 goto end;
163
164 select_compression_response_header(st, s, msg);
165
166 end:
167 return 1;
168}
169
170static int
Christopher Faulete6902cd2018-11-30 22:29:48 +0100171comp_http_payload(struct stream *s, struct filter *filter, struct http_msg *msg,
172 unsigned int offset, unsigned int len)
173{
174 struct comp_state *st = filter->ctx;
Christopher Faulet27ba2dc2018-12-05 11:53:24 +0100175 struct htx *htx = htxbuf(&msg->chn->buf);
Christopher Faulete6a62bf2020-03-02 16:20:05 +0100176 struct htx_ret htxret = htx_find_offset(htx, offset);
Christopher Faulete6902cd2018-11-30 22:29:48 +0100177 struct htx_blk *blk;
Christopher Faulete6902cd2018-11-30 22:29:48 +0100178 int ret, consumed = 0, to_forward = 0;
179
Christopher Faulete6a62bf2020-03-02 16:20:05 +0100180 blk = htxret.blk;
181 offset = htxret.ret;
182 for (; blk && len; blk = htx_get_next_blk(htx, blk)) {
Christopher Faulete6902cd2018-11-30 22:29:48 +0100183 enum htx_blk_type type = htx_get_blk_type(blk);
184 uint32_t sz = htx_get_blksz(blk);
185 struct ist v;
186
187 switch (type) {
188 case HTX_BLK_UNUSED:
189 break;
190
191 case HTX_BLK_DATA:
192 v = htx_get_blk_value(htx, blk);
193 v.ptr += offset;
194 v.len -= offset;
195 if (v.len > len)
196 v.len = len;
197 if (htx_compression_buffer_init(htx, &trash) < 0) {
198 msg->chn->flags |= CF_WAKE_WRITE;
199 goto end;
200 }
201 ret = htx_compression_buffer_add_data(st, v.ptr, v.len, &trash);
202 if (ret < 0)
203 goto error;
204 if (htx_compression_buffer_end(st, &trash, 0) < 0)
205 goto error;
206 len -= ret;
207 consumed += ret;
208 to_forward += b_data(&trash);
209 if (ret == sz && !b_data(&trash)) {
210 offset = 0;
211 blk = htx_remove_blk(htx, blk);
212 continue;
213 }
214 v.len = ret;
215 blk = htx_replace_blk_value(htx, blk, v, ist2(b_head(&trash), b_data(&trash)));
216 break;
217
Christopher Faulete6902cd2018-11-30 22:29:48 +0100218 case HTX_BLK_TLR:
Christopher Faulet2d7c5392019-06-03 10:41:26 +0200219 case HTX_BLK_EOT:
Christopher Faulete6902cd2018-11-30 22:29:48 +0100220 case HTX_BLK_EOM:
Christopher Faulet4288f7d2021-06-09 17:12:44 +0200221 if (st->flags & COMP_STATE_PROCESSING) {
Christopher Faulete6902cd2018-11-30 22:29:48 +0100222 if (htx_compression_buffer_init(htx, &trash) < 0) {
223 msg->chn->flags |= CF_WAKE_WRITE;
224 goto end;
225 }
226 if (htx_compression_buffer_end(st, &trash, 1) < 0)
227 goto error;
Christopher Fauletd238ae32018-12-21 15:10:25 +0100228 if (b_data(&trash)) {
Christopher Faulet86bc8df2019-06-11 10:38:38 +0200229 struct htx_blk *last = htx_add_last_data(htx, ist2(b_head(&trash), b_data(&trash)));
230 if (!last)
231 goto error;
232 blk = htx_get_next_blk(htx, last);
Christopher Fauletd238ae32018-12-21 15:10:25 +0100233 if (!blk)
234 goto error;
235 to_forward += b_data(&trash);
236 }
Christopher Faulete6902cd2018-11-30 22:29:48 +0100237 /* We let the mux add last empty chunk and empty trailers */
238 }
Christopher Faulet4288f7d2021-06-09 17:12:44 +0200239 st->flags &= ~COMP_STATE_PROCESSING;
Christopher Faulete6902cd2018-11-30 22:29:48 +0100240 /* fall through */
241
242 default:
243 sz -= offset;
244 if (sz > len)
245 sz = len;
246 consumed += sz;
247 to_forward += sz;
248 len -= sz;
249 break;
250 }
251
252 offset = 0;
Christopher Faulete6902cd2018-11-30 22:29:48 +0100253 }
254
255 end:
256 if (to_forward != consumed)
257 flt_update_offsets(filter, msg->chn, to_forward - consumed);
258
259 if (st->comp_ctx && st->comp_ctx->cur_lvl > 0) {
Willy Tarreauef6fd852019-02-04 11:48:03 +0100260 update_freq_ctr(&global.comp_bps_in, consumed);
Olivier Houchard43da3432019-03-08 18:50:27 +0100261 _HA_ATOMIC_ADD(&strm_fe(s)->fe_counters.comp_in, consumed);
262 _HA_ATOMIC_ADD(&s->be->be_counters.comp_in, consumed);
Christopher Faulete6902cd2018-11-30 22:29:48 +0100263 update_freq_ctr(&global.comp_bps_out, to_forward);
Olivier Houchard43da3432019-03-08 18:50:27 +0100264 _HA_ATOMIC_ADD(&strm_fe(s)->fe_counters.comp_out, to_forward);
265 _HA_ATOMIC_ADD(&s->be->be_counters.comp_out, to_forward);
Willy Tarreauef6fd852019-02-04 11:48:03 +0100266 } else {
Olivier Houchard43da3432019-03-08 18:50:27 +0100267 _HA_ATOMIC_ADD(&strm_fe(s)->fe_counters.comp_byp, consumed);
268 _HA_ATOMIC_ADD(&s->be->be_counters.comp_byp, consumed);
Christopher Faulete6902cd2018-11-30 22:29:48 +0100269 }
270 return to_forward;
271
272 error:
273 return -1;
274}
275
Christopher Faulet2fb28802015-12-01 10:40:57 +0100276
Christopher Faulet92d36382015-11-05 13:35:03 +0100277static int
Christopher Fauletd60b3cf2017-06-26 11:47:13 +0200278comp_http_end(struct stream *s, struct filter *filter,
279 struct http_msg *msg)
280{
281 struct comp_state *st = filter->ctx;
282
283 if (!(msg->chn->flags & CF_ISRESP) || !st || !st->comp_algo)
284 goto end;
285
286 if (strm_fe(s)->mode == PR_MODE_HTTP)
Olivier Houchard43da3432019-03-08 18:50:27 +0100287 _HA_ATOMIC_ADD(&strm_fe(s)->fe_counters.p.http.comp_rsp, 1);
Christopher Fauletd60b3cf2017-06-26 11:47:13 +0200288 if ((s->flags & SF_BE_ASSIGNED) && (s->be->mode == PR_MODE_HTTP))
Olivier Houchard43da3432019-03-08 18:50:27 +0100289 _HA_ATOMIC_ADD(&s->be->be_counters.p.http.comp_rsp, 1);
Christopher Fauletd60b3cf2017-06-26 11:47:13 +0200290 end:
291 return 1;
292}
Christopher Faulet27d93c32018-12-15 22:32:02 +0100293
Christopher Faulet89f2b162019-07-15 21:16:04 +0200294/***********************************************************************/
Christopher Faulet27d93c32018-12-15 22:32:02 +0100295static int
Christopher Faulet89f2b162019-07-15 21:16:04 +0200296set_compression_response_header(struct comp_state *st, struct stream *s, struct http_msg *msg)
Christopher Faulet27d93c32018-12-15 22:32:02 +0100297{
298 struct htx *htx = htxbuf(&msg->chn->buf);
Tim Duesterhusb229f012019-01-29 16:38:56 +0100299 struct http_hdr_ctx ctx;
Christopher Faulet27d93c32018-12-15 22:32:02 +0100300
301 /*
302 * Add Content-Encoding header when it's not identity encoding.
303 * RFC 2616 : Identity encoding: This content-coding is used only in the
304 * Accept-Encoding header, and SHOULD NOT be used in the Content-Encoding
305 * header.
306 */
307 if (st->comp_algo->cfg_name_len != 8 || memcmp(st->comp_algo->cfg_name, "identity", 8) != 0) {
308 struct ist v = ist2(st->comp_algo->ua_name, st->comp_algo->ua_name_len);
309
310 if (!http_add_header(htx, ist("Content-Encoding"), v))
311 goto error;
312 }
313
314 /* remove Content-Length header */
315 if (msg->flags & HTTP_MSGF_CNT_LEN) {
Christopher Faulet27d93c32018-12-15 22:32:02 +0100316 ctx.blk = NULL;
317 while (http_find_header(htx, ist("Content-Length"), &ctx, 1))
318 http_remove_header(htx, &ctx);
319 }
320
321 /* add "Transfer-Encoding: chunked" header */
322 if (!(msg->flags & HTTP_MSGF_TE_CHNK)) {
323 if (!http_add_header(htx, ist("Transfer-Encoding"), ist("chunked")))
324 goto error;
325 }
326
Tim Duesterhusb229f012019-01-29 16:38:56 +0100327 /* convert "ETag" header to a weak ETag */
328 ctx.blk = NULL;
329 if (http_find_header(htx, ist("ETag"), &ctx, 1)) {
330 if (ctx.value.ptr[0] == '"') {
331 /* This a strong ETag. Convert it to a weak one. */
332 struct ist v = ist2(trash.area, 0);
333 if (istcat(&v, ist("W/"), trash.size) == -1 || istcat(&v, ctx.value, trash.size) == -1)
334 goto error;
335
336 if (!http_replace_header_value(htx, &ctx, v))
337 goto error;
338 }
339 }
340
Tim Duesterhus721d6862019-06-17 16:10:07 +0200341 if (!http_add_header(htx, ist("Vary"), ist("Accept-Encoding")))
342 goto error;
343
Christopher Faulet27d93c32018-12-15 22:32:02 +0100344 return 1;
345
346 error:
347 st->comp_algo->end(&st->comp_ctx);
348 st->comp_algo = NULL;
349 return 0;
350}
351
Christopher Faulet3d97c902015-12-09 14:59:38 +0100352/*
353 * Selects a compression algorithm depending on the client request.
354 */
Christopher Faulete6902cd2018-11-30 22:29:48 +0100355static int
Christopher Faulet89f2b162019-07-15 21:16:04 +0200356select_compression_request_header(struct comp_state *st, struct stream *s, struct http_msg *msg)
Christopher Faulete6902cd2018-11-30 22:29:48 +0100357{
Christopher Faulet27ba2dc2018-12-05 11:53:24 +0100358 struct htx *htx = htxbuf(&msg->chn->buf);
Christopher Faulete6902cd2018-11-30 22:29:48 +0100359 struct http_hdr_ctx ctx;
360 struct comp_algo *comp_algo = NULL;
361 struct comp_algo *comp_algo_back = NULL;
362
363 /* Disable compression for older user agents announcing themselves as "Mozilla/4"
364 * unless they are known good (MSIE 6 with XP SP2, or MSIE 7 and later).
365 * See http://zoompf.com/2012/02/lose-the-wait-http-compression for more details.
366 */
367 ctx.blk = NULL;
368 if (http_find_header(htx, ist("User-Agent"), &ctx, 1) &&
369 ctx.value.len >= 9 &&
370 memcmp(ctx.value.ptr, "Mozilla/4", 9) == 0 &&
371 (ctx.value.len < 31 ||
372 memcmp(ctx.value.ptr + 25, "MSIE ", 5) != 0 ||
373 *(ctx.value.ptr + 30) < '6' ||
374 (*(ctx.value.ptr + 30) == '6' &&
375 (ctx.value.len < 54 || memcmp(ctx.value.ptr + 51, "SV1", 3) != 0)))) {
376 st->comp_algo = NULL;
377 return 0;
378 }
379
380 /* search for the algo in the backend in priority or the frontend */
381 if ((s->be->comp && (comp_algo_back = s->be->comp->algos)) ||
382 (strm_fe(s)->comp && (comp_algo_back = strm_fe(s)->comp->algos))) {
383 int best_q = 0;
384
385 ctx.blk = NULL;
386 while (http_find_header(htx, ist("Accept-Encoding"), &ctx, 0)) {
387 const char *qval;
388 int q;
389 int toklen;
390
391 /* try to isolate the token from the optional q-value */
392 toklen = 0;
393 while (toklen < ctx.value.len && HTTP_IS_TOKEN(*(ctx.value.ptr + toklen)))
394 toklen++;
395
396 qval = ctx.value.ptr + toklen;
397 while (1) {
398 while (qval < ctx.value.ptr + ctx.value.len && HTTP_IS_LWS(*qval))
399 qval++;
400
401 if (qval >= ctx.value.ptr + ctx.value.len || *qval != ';') {
402 qval = NULL;
403 break;
404 }
405 qval++;
406
407 while (qval < ctx.value.ptr + ctx.value.len && HTTP_IS_LWS(*qval))
408 qval++;
409
410 if (qval >= ctx.value.ptr + ctx.value.len) {
411 qval = NULL;
412 break;
413 }
414 if (strncmp(qval, "q=", MIN(ctx.value.ptr + ctx.value.len - qval, 2)) == 0)
415 break;
416
417 while (qval < ctx.value.ptr + ctx.value.len && *qval != ';')
418 qval++;
419 }
420
421 /* here we have qval pointing to the first "q=" attribute or NULL if not found */
422 q = qval ? http_parse_qvalue(qval + 2, NULL) : 1000;
423
424 if (q <= best_q)
425 continue;
426
427 for (comp_algo = comp_algo_back; comp_algo; comp_algo = comp_algo->next) {
428 if (*(ctx.value.ptr) == '*' ||
429 word_match(ctx.value.ptr, toklen, comp_algo->ua_name, comp_algo->ua_name_len)) {
430 st->comp_algo = comp_algo;
431 best_q = q;
432 break;
433 }
434 }
435 }
436 }
437
438 /* remove all occurrences of the header when "compression offload" is set */
439 if (st->comp_algo) {
440 if ((s->be->comp && s->be->comp->offload) ||
441 (strm_fe(s)->comp && strm_fe(s)->comp->offload)) {
442 http_remove_header(htx, &ctx);
443 ctx.blk = NULL;
444 while (http_find_header(htx, ist("Accept-Encoding"), &ctx, 1))
445 http_remove_header(htx, &ctx);
446 }
Christopher Faulet3d97c902015-12-09 14:59:38 +0100447 return 1;
448 }
449
450 /* identity is implicit does not require headers */
Christopher Faulet92d36382015-11-05 13:35:03 +0100451 if ((s->be->comp && (comp_algo_back = s->be->comp->algos)) ||
452 (strm_fe(s)->comp && (comp_algo_back = strm_fe(s)->comp->algos))) {
Christopher Faulet3d97c902015-12-09 14:59:38 +0100453 for (comp_algo = comp_algo_back; comp_algo; comp_algo = comp_algo->next) {
454 if (comp_algo->cfg_name_len == 8 && memcmp(comp_algo->cfg_name, "identity", 8) == 0) {
Christopher Faulet92d36382015-11-05 13:35:03 +0100455 st->comp_algo = comp_algo;
Christopher Faulet3d97c902015-12-09 14:59:38 +0100456 return 1;
457 }
458 }
459 }
460
Christopher Faulet92d36382015-11-05 13:35:03 +0100461 st->comp_algo = NULL;
Christopher Faulet3d97c902015-12-09 14:59:38 +0100462 return 0;
463}
464
465/*
Ilya Shipitsin46a030c2020-07-05 16:36:08 +0500466 * Selects a compression algorithm depending of the server response.
Christopher Faulet3d97c902015-12-09 14:59:38 +0100467 */
Christopher Faulet92d36382015-11-05 13:35:03 +0100468static int
Christopher Faulet89f2b162019-07-15 21:16:04 +0200469select_compression_response_header(struct comp_state *st, struct stream *s, struct http_msg *msg)
Christopher Faulete6902cd2018-11-30 22:29:48 +0100470{
Christopher Faulet27ba2dc2018-12-05 11:53:24 +0100471 struct htx *htx = htxbuf(&msg->chn->buf);
Christopher Faulete6902cd2018-11-30 22:29:48 +0100472 struct http_txn *txn = s->txn;
473 struct http_hdr_ctx ctx;
474 struct comp_type *comp_type;
475
476 /* no common compression algorithm was found in request header */
477 if (st->comp_algo == NULL)
478 goto fail;
479
Christopher Faulet1d3613a2019-01-07 14:41:59 +0100480 /* compression already in progress */
481 if (msg->flags & HTTP_MSGF_COMPRESSING)
482 goto fail;
483
Christopher Faulete6902cd2018-11-30 22:29:48 +0100484 /* HTTP < 1.1 should not be compressed */
485 if (!(msg->flags & HTTP_MSGF_VER_11) || !(txn->req.flags & HTTP_MSGF_VER_11))
486 goto fail;
487
488 if (txn->meth == HTTP_METH_HEAD)
489 goto fail;
490
491 /* compress 200,201,202,203 responses only */
492 if ((txn->status != 200) &&
493 (txn->status != 201) &&
494 (txn->status != 202) &&
495 (txn->status != 203))
496 goto fail;
497
Christopher Fauletc963eb22018-12-21 14:53:54 +0100498 if (!(msg->flags & HTTP_MSGF_XFER_LEN) || msg->flags & HTTP_MSGF_BODYLESS)
Christopher Faulete6902cd2018-11-30 22:29:48 +0100499 goto fail;
500
501 /* content is already compressed */
502 ctx.blk = NULL;
503 if (http_find_header(htx, ist("Content-Encoding"), &ctx, 1))
504 goto fail;
505
506 /* no compression when Cache-Control: no-transform is present in the message */
507 ctx.blk = NULL;
508 while (http_find_header(htx, ist("Cache-Control"), &ctx, 0)) {
509 if (word_match(ctx.value.ptr, ctx.value.len, "no-transform", 12))
510 goto fail;
511 }
512
Tim Duesterhusb229f012019-01-29 16:38:56 +0100513 /* no compression when ETag is malformed */
514 ctx.blk = NULL;
515 if (http_find_header(htx, ist("ETag"), &ctx, 1)) {
Tim Duesterhus6414cd12020-09-01 18:32:35 +0200516 if (http_get_etag_type(ctx.value) == ETAG_INVALID)
Tim Duesterhusb229f012019-01-29 16:38:56 +0100517 goto fail;
Tim Duesterhusb229f012019-01-29 16:38:56 +0100518 }
519 /* no compression when multiple ETags are present
520 * Note: Do not reset ctx.blk!
521 */
522 if (http_find_header(htx, ist("ETag"), &ctx, 1))
523 goto fail;
524
Christopher Faulete6902cd2018-11-30 22:29:48 +0100525 comp_type = NULL;
526
527 /* we don't want to compress multipart content-types, nor content-types that are
528 * not listed in the "compression type" directive if any. If no content-type was
529 * found but configuration requires one, we don't compress either. Backend has
530 * the priority.
531 */
532 ctx.blk = NULL;
533 if (http_find_header(htx, ist("Content-Type"), &ctx, 1)) {
534 if (ctx.value.len >= 9 && strncasecmp("multipart", ctx.value.ptr, 9) == 0)
535 goto fail;
536
537 if ((s->be->comp && (comp_type = s->be->comp->types)) ||
538 (strm_fe(s)->comp && (comp_type = strm_fe(s)->comp->types))) {
539 for (; comp_type; comp_type = comp_type->next) {
540 if (ctx.value.len >= comp_type->name_len &&
541 strncasecmp(ctx.value.ptr, comp_type->name, comp_type->name_len) == 0)
542 /* this Content-Type should be compressed */
543 break;
544 }
545 /* this Content-Type should not be compressed */
546 if (comp_type == NULL)
547 goto fail;
548 }
549 }
550 else { /* no content-type header */
551 if ((s->be->comp && s->be->comp->types) ||
552 (strm_fe(s)->comp && strm_fe(s)->comp->types))
553 goto fail; /* a content-type was required */
554 }
555
556 /* limit compression rate */
557 if (global.comp_rate_lim > 0)
558 if (read_freq_ctr(&global.comp_bps_in) > global.comp_rate_lim)
559 goto fail;
560
561 /* limit cpu usage */
Willy Tarreau81036f22019-05-20 19:24:50 +0200562 if (ti->idle_pct < compress_min_idle)
Christopher Faulete6902cd2018-11-30 22:29:48 +0100563 goto fail;
564
565 /* initialize compression */
566 if (st->comp_algo->init(&st->comp_ctx, global.tune.comp_maxlevel) < 0)
567 goto fail;
Christopher Faulete6902cd2018-11-30 22:29:48 +0100568 msg->flags |= HTTP_MSGF_COMPRESSING;
569 return 1;
570
Christopher Faulete6902cd2018-11-30 22:29:48 +0100571 fail:
572 st->comp_algo = NULL;
573 return 0;
574}
575
Christopher Faulet3d97c902015-12-09 14:59:38 +0100576/***********************************************************************/
Christopher Faulete6902cd2018-11-30 22:29:48 +0100577static int
578htx_compression_buffer_init(struct htx *htx, struct buffer *out)
579{
580 /* output stream requires at least 10 bytes for the gzip header, plus
581 * at least 8 bytes for the gzip trailer (crc+len), plus a possible
582 * plus at most 5 bytes per 32kB block and 2 bytes to close the stream.
583 */
584 if (htx_free_space(htx) < 20 + 5 * ((htx->data + 32767) >> 15))
585 return -1;
586 b_reset(out);
587 return 0;
588}
589
Christopher Faulete6902cd2018-11-30 22:29:48 +0100590static int
591htx_compression_buffer_add_data(struct comp_state *st, const char *data, size_t len,
592 struct buffer *out)
593{
594 return st->comp_algo->add_data(st->comp_ctx, data, len, out);
595}
596
Christopher Faulete6902cd2018-11-30 22:29:48 +0100597static int
598htx_compression_buffer_end(struct comp_state *st, struct buffer *out, int end)
599{
600 if (end)
601 return st->comp_algo->finish(st->comp_ctx, out);
602 else
603 return st->comp_algo->flush(st->comp_ctx, out);
604}
605
Christopher Faulet3d97c902015-12-09 14:59:38 +0100606
607/***********************************************************************/
Christopher Faulet92d36382015-11-05 13:35:03 +0100608struct flt_ops comp_ops = {
Christopher Faulete6902cd2018-11-30 22:29:48 +0100609 .init = comp_flt_init,
Christopher Faulet8ca3b4b2017-07-25 11:07:15 +0200610 .init_per_thread = comp_flt_init_per_thread,
611 .deinit_per_thread = comp_flt_deinit_per_thread,
Christopher Faulet92d36382015-11-05 13:35:03 +0100612
Christopher Faulet5e896512020-03-06 14:59:05 +0100613 .attach = comp_strm_init,
614 .detach = comp_strm_deinit,
615
Christopher Faulet3dc860d2017-09-15 11:39:36 +0200616 .channel_post_analyze = comp_http_post_analyze,
Christopher Faulet92d36382015-11-05 13:35:03 +0100617
Christopher Faulet1339d742016-05-11 16:48:33 +0200618 .http_headers = comp_http_headers,
Christopher Faulete6902cd2018-11-30 22:29:48 +0100619 .http_payload = comp_http_payload,
620 .http_end = comp_http_end,
Christopher Faulet92d36382015-11-05 13:35:03 +0100621};
622
Christopher Faulet3d97c902015-12-09 14:59:38 +0100623static int
624parse_compression_options(char **args, int section, struct proxy *proxy,
625 struct proxy *defpx, const char *file, int line,
626 char **err)
627{
Christopher Faulet92d36382015-11-05 13:35:03 +0100628 struct comp *comp;
Christopher Faulet3d97c902015-12-09 14:59:38 +0100629
630 if (proxy->comp == NULL) {
Vincent Bernat02779b62016-04-03 13:48:43 +0200631 comp = calloc(1, sizeof(*comp));
Christopher Faulet3d97c902015-12-09 14:59:38 +0100632 proxy->comp = comp;
633 }
634 else
635 comp = proxy->comp;
636
637 if (!strcmp(args[1], "algo")) {
638 struct comp_ctx *ctx;
639 int cur_arg = 2;
640
641 if (!*args[cur_arg]) {
642 memprintf(err, "parsing [%s:%d] : '%s' expects <algorithm>\n",
643 file, line, args[0]);
644 return -1;
645 }
646 while (*(args[cur_arg])) {
Remi Tricot-Le Breton46bd5dc2021-05-17 10:35:08 +0200647 int retval = comp_append_algo(comp, args[cur_arg]);
648 if (retval) {
649 if (retval < 0)
650 memprintf(err, "'%s' : '%s' is not a supported algorithm.\n",
651 args[0], args[cur_arg]);
652 else
653 memprintf(err, "'%s' : out of memory while parsing algo '%s'.\n",
654 args[0], args[cur_arg]);
Christopher Faulet3d97c902015-12-09 14:59:38 +0100655 return -1;
656 }
Remi Tricot-Le Breton46bd5dc2021-05-17 10:35:08 +0200657
Christopher Faulet3d97c902015-12-09 14:59:38 +0100658 if (proxy->comp->algos->init(&ctx, 9) == 0)
659 proxy->comp->algos->end(&ctx);
660 else {
661 memprintf(err, "'%s' : Can't init '%s' algorithm.\n",
662 args[0], args[cur_arg]);
663 return -1;
664 }
665 cur_arg++;
666 continue;
667 }
668 }
669 else if (!strcmp(args[1], "offload"))
670 comp->offload = 1;
671 else if (!strcmp(args[1], "type")) {
672 int cur_arg = 2;
673
674 if (!*args[cur_arg]) {
675 memprintf(err, "'%s' expects <type>\n", args[0]);
676 return -1;
677 }
678 while (*(args[cur_arg])) {
Remi Tricot-Le Breton46bd5dc2021-05-17 10:35:08 +0200679 if (comp_append_type(comp, args[cur_arg])) {
680 memprintf(err, "'%s': out of memory.", args[0]);
681 return -1;
682 }
Christopher Faulet3d97c902015-12-09 14:59:38 +0100683 cur_arg++;
684 continue;
685 }
686 }
687 else {
688 memprintf(err, "'%s' expects 'algo', 'type' or 'offload'\n",
689 args[0]);
690 return -1;
691 }
692
693 return 0;
694}
695
Christopher Faulet92d36382015-11-05 13:35:03 +0100696static int
697parse_http_comp_flt(char **args, int *cur_arg, struct proxy *px,
Thierry Fournier3610c392016-04-13 18:27:51 +0200698 struct flt_conf *fconf, char **err, void *private)
Christopher Faulet92d36382015-11-05 13:35:03 +0100699{
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100700 struct flt_conf *fc, *back;
Christopher Faulet92d36382015-11-05 13:35:03 +0100701
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100702 list_for_each_entry_safe(fc, back, &px->filter_configs, list) {
703 if (fc->id == http_comp_flt_id) {
Christopher Faulet92d36382015-11-05 13:35:03 +0100704 memprintf(err, "%s: Proxy supports only one compression filter\n", px->id);
705 return -1;
706 }
707 }
708
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100709 fconf->id = http_comp_flt_id;
710 fconf->conf = NULL;
711 fconf->ops = &comp_ops;
Christopher Faulet92d36382015-11-05 13:35:03 +0100712 (*cur_arg)++;
713
714 return 0;
715}
716
717
718int
Christopher Fauletc9df7f72018-12-10 16:14:04 +0100719check_implicit_http_comp_flt(struct proxy *proxy)
Christopher Faulet92d36382015-11-05 13:35:03 +0100720{
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100721 struct flt_conf *fconf;
Christopher Faulet27d93c32018-12-15 22:32:02 +0100722 int explicit = 0;
723 int comp = 0;
Christopher Faulet92d36382015-11-05 13:35:03 +0100724 int err = 0;
725
726 if (proxy->comp == NULL)
727 goto end;
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100728 if (!LIST_ISEMPTY(&proxy->filter_configs)) {
729 list_for_each_entry(fconf, &proxy->filter_configs, list) {
730 if (fconf->id == http_comp_flt_id)
Christopher Faulet27d93c32018-12-15 22:32:02 +0100731 comp = 1;
732 else if (fconf->id == cache_store_flt_id) {
733 if (comp) {
734 ha_alert("config: %s '%s': unable to enable the compression filter "
735 "before any cache filter.\n",
736 proxy_type_str(proxy), proxy->id);
737 err++;
738 goto end;
739 }
740 }
Christopher Faulet78fbb9f2019-08-11 23:11:03 +0200741 else if (fconf->id == fcgi_flt_id)
742 continue;
Christopher Faulet27d93c32018-12-15 22:32:02 +0100743 else
744 explicit = 1;
Christopher Faulet92d36382015-11-05 13:35:03 +0100745 }
Christopher Faulet27d93c32018-12-15 22:32:02 +0100746 }
747 if (comp)
748 goto end;
749 else if (explicit) {
750 ha_alert("config: %s '%s': require an explicit filter declaration to use "
751 "HTTP compression\n", proxy_type_str(proxy), proxy->id);
Christopher Faulet92d36382015-11-05 13:35:03 +0100752 err++;
753 goto end;
754 }
755
Christopher Faulet27d93c32018-12-15 22:32:02 +0100756 /* Implicit declaration of the compression filter is always the last
757 * one */
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100758 fconf = calloc(1, sizeof(*fconf));
759 if (!fconf) {
Christopher Faulet767a84b2017-11-24 16:50:31 +0100760 ha_alert("config: %s '%s': out of memory\n",
761 proxy_type_str(proxy), proxy->id);
Christopher Faulet92d36382015-11-05 13:35:03 +0100762 err++;
763 goto end;
764 }
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100765 fconf->id = http_comp_flt_id;
766 fconf->conf = NULL;
767 fconf->ops = &comp_ops;
768 LIST_ADDQ(&proxy->filter_configs, &fconf->list);
Christopher Faulet92d36382015-11-05 13:35:03 +0100769 end:
770 return err;
771}
772
773/*
774 * boolean, returns true if compression is used (either gzip or deflate) in the
775 * response.
776 */
Christopher Faulet3d97c902015-12-09 14:59:38 +0100777static int
Christopher Faulet92d36382015-11-05 13:35:03 +0100778smp_fetch_res_comp(const struct arg *args, struct sample *smp, const char *kw,
779 void *private)
Christopher Faulet3d97c902015-12-09 14:59:38 +0100780{
Willy Tarreaube508f12016-03-10 11:47:01 +0100781 struct http_txn *txn = smp->strm ? smp->strm->txn : NULL;
Christopher Faulet92d36382015-11-05 13:35:03 +0100782
Christopher Faulet3d97c902015-12-09 14:59:38 +0100783 smp->data.type = SMP_T_BOOL;
Christopher Faulet92d36382015-11-05 13:35:03 +0100784 smp->data.u.sint = (txn && (txn->rsp.flags & HTTP_MSGF_COMPRESSING));
Christopher Faulet3d97c902015-12-09 14:59:38 +0100785 return 1;
786}
787
Christopher Faulet92d36382015-11-05 13:35:03 +0100788/*
789 * string, returns algo
790 */
Christopher Faulet3d97c902015-12-09 14:59:38 +0100791static int
Christopher Faulet92d36382015-11-05 13:35:03 +0100792smp_fetch_res_comp_algo(const struct arg *args, struct sample *smp,
793 const char *kw, void *private)
Christopher Faulet3d97c902015-12-09 14:59:38 +0100794{
Willy Tarreaube508f12016-03-10 11:47:01 +0100795 struct http_txn *txn = smp->strm ? smp->strm->txn : NULL;
Christopher Faulet92d36382015-11-05 13:35:03 +0100796 struct filter *filter;
797 struct comp_state *st;
798
Christopher Faulet03d85532017-09-15 10:14:43 +0200799 if (!txn || !(txn->rsp.flags & HTTP_MSGF_COMPRESSING))
Christopher Faulet3d97c902015-12-09 14:59:38 +0100800 return 0;
801
Christopher Fauletfcf035c2015-12-03 11:48:03 +0100802 list_for_each_entry(filter, &strm_flt(smp->strm)->filters, list) {
Christopher Faulet443ea1a2016-02-04 13:40:26 +0100803 if (FLT_ID(filter) != http_comp_flt_id)
Christopher Faulet92d36382015-11-05 13:35:03 +0100804 continue;
805
806 if (!(st = filter->ctx))
807 break;
808
809 smp->data.type = SMP_T_STR;
810 smp->flags = SMP_F_CONST;
Willy Tarreau843b7cb2018-07-13 10:54:26 +0200811 smp->data.u.str.area = st->comp_algo->cfg_name;
812 smp->data.u.str.data = st->comp_algo->cfg_name_len;
Christopher Faulet92d36382015-11-05 13:35:03 +0100813 return 1;
814 }
815 return 0;
Christopher Faulet3d97c902015-12-09 14:59:38 +0100816}
817
818/* Declare the config parser for "compression" keyword */
819static struct cfg_kw_list cfg_kws = {ILH, {
820 { CFG_LISTEN, "compression", parse_compression_options },
821 { 0, NULL, NULL },
822 }
823};
824
Willy Tarreau0108d902018-11-25 19:14:37 +0100825INITCALL1(STG_REGISTER, cfg_register_keywords, &cfg_kws);
826
Christopher Faulet92d36382015-11-05 13:35:03 +0100827/* Declare the filter parser for "compression" keyword */
828static struct flt_kw_list filter_kws = { "COMP", { }, {
Thierry Fournier3610c392016-04-13 18:27:51 +0200829 { "compression", parse_http_comp_flt, NULL },
830 { NULL, NULL, NULL },
Christopher Faulet92d36382015-11-05 13:35:03 +0100831 }
832};
833
Willy Tarreau0108d902018-11-25 19:14:37 +0100834INITCALL1(STG_REGISTER, flt_register_keywords, &filter_kws);
835
Christopher Faulet3d97c902015-12-09 14:59:38 +0100836/* Note: must not be declared <const> as its list will be overwritten */
837static struct sample_fetch_kw_list sample_fetch_keywords = {ILH, {
Christopher Faulet92d36382015-11-05 13:35:03 +0100838 { "res.comp", smp_fetch_res_comp, 0, NULL, SMP_T_BOOL, SMP_USE_HRSHP },
839 { "res.comp_algo", smp_fetch_res_comp_algo, 0, NULL, SMP_T_STR, SMP_USE_HRSHP },
840 { /* END */ },
841 }
842};
Christopher Faulet3d97c902015-12-09 14:59:38 +0100843
Willy Tarreau0108d902018-11-25 19:14:37 +0100844INITCALL1(STG_REGISTER, sample_register_fetches, &sample_fetch_keywords);