Elliott Hughes | 82be86d | 2017-09-20 17:00:17 -0700 | [diff] [blame] | 1 | _ _ ____ _ |
| 2 | ___| | | | _ \| | |
| 3 | / __| | | | |_) | | |
| 4 | | (__| |_| | _ <| |___ |
| 5 | \___|\___/|_| \_\_____| |
Lucas Eckels | 9bd90e6 | 2012-08-06 15:07:02 -0700 | [diff] [blame] | 6 | |
Elliott Hughes | 82be86d | 2017-09-20 17:00:17 -0700 | [diff] [blame] | 7 | Changelog |
Lucas Eckels | 9bd90e6 | 2012-08-06 15:07:02 -0700 | [diff] [blame] | 8 | |
Haibo Huang | b5a52b9 | 2020-10-28 22:18:23 -0700 | [diff] [blame] | 9 | Version 7.73.0 (14 Oct 2020) |
| 10 | |
| 11 | Daniel Stenberg (14 Oct 2020) |
| 12 | - RELEASE-NOTES: synced |
| 13 | |
| 14 | for 7.73.0 |
| 15 | |
| 16 | - THANKS: from 7.73.0 and .mailmap fixes |
| 17 | |
| 18 | - mailmap: fixups of some contributors |
| 19 | |
| 20 | - projects/build-wolfssl.bat: fix the copyright year range |
| 21 | |
| 22 | Marc Hoersken (14 Oct 2020) |
| 23 | - [Sergei Nikulov brought this change] |
| 24 | |
| 25 | CI/tests: fix invocation of tests for CMake builds |
| 26 | |
| 27 | Update appveyor.yml to set env variable TFLAGS and run tests |
| 28 | Remove curly braces due to CMake error (${TFLAGS} -> $TFLAGS) |
| 29 | Move testdeps build to build step (per review comments) |
| 30 | |
| 31 | Reviewed-by: Marc Hörsken |
| 32 | |
| 33 | Closes #6066 |
| 34 | Fixes #6052 |
| 35 | |
| 36 | - tests/server/util.c: fix support for Windows Unicode builds |
| 37 | |
| 38 | Detected via #6066 |
| 39 | Closes #6070 |
| 40 | |
| 41 | Daniel Stenberg (13 Oct 2020) |
| 42 | - [Jay Satiro brought this change] |
| 43 | |
| 44 | strerror: Revert to local codepage for Windows error string |
| 45 | |
| 46 | - Change get_winapi_error() to return the error string in the local |
| 47 | codepage instead of UTF-8 encoding. |
| 48 | |
| 49 | Two weeks ago bed5f84 fixed get_winapi_error() to work on xbox, but it |
| 50 | also changed the error string's encoding from local codepage to UTF-8. |
| 51 | |
| 52 | We return the local codepage version of the error string because if it |
| 53 | is output to the user's terminal it will likely be with functions which |
| 54 | expect the local codepage (eg fprintf, failf, infof). |
| 55 | |
| 56 | This is essentially a partial revert of bed5f84. The support for xbox |
| 57 | remains but the error string is reverted back to local codepage. |
| 58 | |
| 59 | Ref: https://github.com/curl/curl/pull/6005 |
| 60 | |
| 61 | Reviewed-by: Marcel Raad |
| 62 | Closes #6065 |
| 63 | |
| 64 | Marc Hoersken (13 Oct 2020) |
| 65 | - CI/tests: use verification curl for test reporting APIs |
| 66 | |
| 67 | Avoid using our own, potentially installed, curl for |
| 68 | the test reporting APIs in case it is broken. |
| 69 | |
| 70 | Reviewed-by: Daniel Stenberg |
| 71 | |
| 72 | Preparation for #6049 |
| 73 | Closes #6063 |
| 74 | |
| 75 | Viktor Szakats (12 Oct 2020) |
| 76 | - windows: fix comparison of mismatched types warning |
| 77 | |
| 78 | clang 10, mingw-w64: |
| 79 | ``` |
| 80 | vtls/openssl.c:2917:33: warning: comparison of integers of different signs: 'DWORD' (aka 'unsigned long') and 'HRESULT' (aka 'long') |
| 81 | [-Wsign-compare] |
| 82 | if(GetLastError() != CRYPT_E_NOT_FOUND) |
| 83 | ~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~ |
| 84 | ``` |
| 85 | |
| 86 | Approved-by: Daniel Stenberg |
| 87 | Closes #6062 |
| 88 | |
| 89 | Daniel Stenberg (11 Oct 2020) |
| 90 | - [Viktor Szakats brought this change] |
| 91 | |
| 92 | src/Makefile.m32: fix undefined curlx_dyn_* errors |
| 93 | |
| 94 | by linking `lib/dynbuf.c` when building a static curl binary. |
| 95 | Previously this source file was only included when building |
| 96 | a dynamic curl binary. This was likely possibly because no |
| 97 | functions from the `src/Makefile.inc` / `CURLX_CFILES` sources |
| 98 | were actually required for a curl tool build. This has |
| 99 | recently changed with the introduction of `curlx_dyn_*()` |
| 100 | memory functions and their use by the tool sources. |
| 101 | |
| 102 | Closes #6060 |
| 103 | |
| 104 | - HISTORY: curl verifies SSL certs by default since version 7.10 |
| 105 | |
| 106 | Marc Hoersken (8 Oct 2020) |
| 107 | - runtests.pl: use $LIBDIR variable instead of hardcoded path |
| 108 | |
| 109 | Reviewed-by: Daniel Stenberg |
| 110 | Closes #6051 |
| 111 | |
| 112 | Daniel Stenberg (7 Oct 2020) |
| 113 | - checksrc: detect // comments on column 0 |
| 114 | |
| 115 | Spotted while working on #6045 |
| 116 | |
| 117 | Closes #6048 |
| 118 | |
| 119 | - [Frederik Wedel-Heinen brought this change] |
| 120 | |
| 121 | mbedtls: add missing header when defining MBEDTLS_DEBUG |
| 122 | |
| 123 | Closes #6045 |
| 124 | |
| 125 | - curl: make sure setopt CURLOPT_IPRESOLVE passes on a long |
| 126 | |
| 127 | Previously, it would pass on a define (int) which could make libcurl |
| 128 | read junk as a value - which prevented the CURLOPT_IPRESOLVE option to |
| 129 | "take". This could then make test 2100 do two DoH requests instead of |
| 130 | one! |
| 131 | |
| 132 | Fixes #6042 |
| 133 | Closes #6043 |
| 134 | |
| 135 | - RELEASE-NOTES: synced |
| 136 | |
| 137 | - scripts/release-notes.pl: don't "embed" $ in format string for printf() |
| 138 | |
| 139 | ... since they might contain %-codes that mess up the output! |
| 140 | |
| 141 | Jay Satiro (5 Oct 2020) |
| 142 | - [M.R.T brought this change] |
| 143 | |
| 144 | build-wolfssl: fix build with Visual Studio 2019 |
| 145 | |
| 146 | Closes https://github.com/curl/curl/pull/6033 |
| 147 | |
| 148 | Daniel Stenberg (4 Oct 2020) |
| 149 | - runtests: add %repeat[]% for test files |
| 150 | |
| 151 | ... and use this new keywords in all the test files larger than 50K to reduce |
| 152 | their sizes and make them a lot easier to read and understand. |
| 153 | |
| 154 | Closes #6040 |
| 155 | |
| 156 | - [Emil Engler brought this change] |
| 157 | |
| 158 | --help: move two options from the misc category |
| 159 | |
| 160 | The cmdline opts delegation and suppress-connect-headers |
| 161 | fit better into auth and proxy rather than misc. |
| 162 | |
| 163 | Follow-up to aa8777f63febc |
| 164 | Closes #6038 |
| 165 | |
| 166 | - [Samanta Navarro brought this change] |
| 167 | |
| 168 | docs/opts: fix typos in two manual pages |
| 169 | |
| 170 | Closes #6039 |
| 171 | |
| 172 | - ldap: reduce the amount of #ifdefs needed |
| 173 | |
| 174 | Closes #6035 |
| 175 | |
| 176 | - runtests: provide curl's version string as %VERSION for tests |
| 177 | |
| 178 | ... so that we can check HTTP requests for User-Agent: curl/%VERSION |
| 179 | |
| 180 | Update 600+ test cases accordingly. |
| 181 | |
| 182 | Closes #6037 |
| 183 | |
| 184 | - checksrc: warn on space after exclamation mark |
| 185 | |
| 186 | Closes #6034 |
| 187 | |
| 188 | - test1465: verify --libcurl with binary POST data |
| 189 | |
| 190 | - runtests: allow generating a binary sequence from hex |
| 191 | |
| 192 | - tool_setopt: escape binary data to hex, not octal |
| 193 | |
| 194 | - curl: make --libcurl show binary posts correctly |
| 195 | |
| 196 | Reported-by: Stephan Mühlstrasser |
| 197 | Fixes #6031 |
| 198 | Closes #6032 |
| 199 | |
| 200 | Jay Satiro (1 Oct 2020) |
| 201 | - strerror: fix null deref on winapi out-of-memory |
| 202 | |
| 203 | Follow-up to bed5f84 from several days ago. |
| 204 | |
| 205 | Ref: https://github.com/curl/curl/pull/6005 |
| 206 | |
| 207 | Daniel Stenberg (1 Oct 2020) |
| 208 | - [Kamil Dudka brought this change] |
| 209 | |
| 210 | vtls: deduplicate some DISABLE_PROXY ifdefs |
| 211 | |
| 212 | ... in the code of gtls, nss, and openssl |
| 213 | |
| 214 | Closes #5735 |
| 215 | |
| 216 | - RELEASE-NOTES: synced |
| 217 | |
| 218 | - [Emil Engler brought this change] |
| 219 | |
| 220 | TODO: Add OpenBSD libtool notice |
| 221 | |
| 222 | See #5862 |
| 223 | Closes #6030 |
| 224 | |
| 225 | - tests/unit/README: convert to markdown |
| 226 | |
| 227 | ... and add to dist! |
| 228 | |
| 229 | Closes #6028 |
| 230 | |
| 231 | - tests/README: convert to markdown |
| 232 | |
| 233 | Closes #6028 |
| 234 | |
| 235 | - include/README: convert to markdown |
| 236 | |
| 237 | Closes #6028 |
| 238 | |
| 239 | - examples/README: convert to markdown |
| 240 | |
| 241 | Closes #6028 |
| 242 | |
| 243 | - configure: don't say HTTPS-proxy is enabled when disabled! |
| 244 | |
| 245 | Reported-by: Kamil Dudka |
| 246 | Reviewed-by: Kamil Dudka |
| 247 | Bug: https://github.com/curl/curl/pull/5735#issuecomment-701376388 |
| 248 | Closes #6029 |
| 249 | |
| 250 | Daniel Gustafsson (30 Sep 2020) |
| 251 | - src: Consistently spell whitespace without whitespace |
| 252 | |
| 253 | Whitespace is spelled without a space between white and space, so |
| 254 | make sure to consistently spell it that way across the codebase. |
| 255 | |
| 256 | Closes #6023 |
| 257 | Reviewed-by: Daniel Stenberg <daniel@haxx.se> |
| 258 | Reviewed-by: Emil Engler <me@emilengler.com> |
| 259 | |
| 260 | - MANUAL: update examples to resolve without redirects |
| 261 | |
| 262 | www.netscape.com is redirecting to a cookie consent form on Aol, and |
| 263 | cool.haxx.se isn't responding to FTP anymore. Replace with examples |
| 264 | that resolves in case users try out the commands when reading the |
| 265 | manual. |
| 266 | |
| 267 | Closes #6024 |
| 268 | Reviewed-by: Daniel Stenberg <daniel@haxx.se> |
| 269 | Reviewed-by: Emil Engler <me@emilengler.com> |
| 270 | |
| 271 | Daniel Stenberg (30 Sep 2020) |
| 272 | - HISTORY: add some 2020 events |
| 273 | |
| 274 | - sectransp: make it build with --disable-proxy |
| 275 | |
| 276 | Follow-up from #5466 and f3d501dc678d80 |
| 277 | Reported-by: Javier Navarro |
| 278 | Fixes #6025 |
| 279 | Closes #6026 |
| 280 | |
| 281 | - ECH: renamed from ESNI in docs and configure |
| 282 | |
| 283 | Encrypted Client Hello (ECH) is the current name. |
| 284 | |
| 285 | Closes #6022 |
| 286 | |
| 287 | - configure: use "no" instead of "disabled" for the end summary |
| 288 | |
| 289 | ... for consistency but also to make them more distinctly stand out next |
| 290 | to the "enabled" lines. |
| 291 | |
| 292 | - TODO: SSH over HTTPS proxy with more backends |
| 293 | |
| 294 | ... as right now only the libssh2 backend supports it. |
| 295 | |
| 296 | - libssh2: handle the SSH protocols done over HTTPS proxy |
| 297 | |
| 298 | Reported-by: Robin Douine |
| 299 | Fixes #4295 |
| 300 | Closes #6021 |
| 301 | |
| 302 | - [Emil Engler brought this change] |
| 303 | |
| 304 | memdebug: remove 9 year old unused debug function |
| 305 | |
| 306 | There used to be a way to have memdebug fill allocated memory. 9 years |
| 307 | later this has no value there (valgrind and ASAN etc are way better). If |
| 308 | people need to know about it they can have a look at VCS logs. |
| 309 | |
| 310 | Closes #5973 |
| 311 | |
| 312 | - sendf: move Curl_sendf to dict.c and make it static |
| 313 | |
| 314 | ... as the only remaining user of that function. Also fix gopher.c to |
| 315 | instead use Curl_write() |
| 316 | |
| 317 | Closes #6020 |
| 318 | |
| 319 | - ROADMAP: updates and cleanups |
| 320 | |
| 321 | Fix the HSTS PR |
| 322 | |
| 323 | Remove DoT, thread-safe init and hard-coded localhost. I feel very |
| 324 | little interest for these with users so I downgrade them to plain "TODO" |
| 325 | entries again. |
| 326 | |
| 327 | - schannel: return CURLE_PEER_FAILED_VERIFICATION for untrusted root |
| 328 | |
| 329 | This matches what is returned in other TLS backends in the same |
| 330 | situation. |
| 331 | |
| 332 | Reviewed-by: Jay Satiro |
| 333 | Reviewed-by: Emil Engler |
| 334 | Follow-up to 5a3efb1 |
| 335 | Reported-by: iammrtau on github |
| 336 | Fixes #6003 |
| 337 | Closes #6018 |
| 338 | |
| 339 | - RELEASE-NOTES: synced |
| 340 | |
| 341 | - ftp: make a 552 response return CURLE_REMOTE_DISK_FULL |
| 342 | |
| 343 | Added test 348 to verify. Added a 'STOR' command to the test FTP |
| 344 | server to enable test 348. Documented the command in FILEFORMAT.md |
| 345 | |
| 346 | Reported-by: Duncan Wilcox |
| 347 | Fixes #6016 |
| 348 | Closes #6017 |
| 349 | |
| 350 | - pause: only trigger a reread if the unpause sticks |
| 351 | |
| 352 | As an unpause might itself get paused again and then triggering another |
| 353 | reread doesn't help. |
| 354 | |
| 355 | Follow-up from e040146f22608fd9 (shipped since 7.69.1) |
| 356 | |
| 357 | Bug: https://curl.haxx.se/mail/lib-2020-09/0081.html |
| 358 | Patch-by: Kunal Chandarana |
| 359 | Fixes #5988 |
| 360 | Closes #6013 |
| 361 | |
| 362 | - test163[12]: require http to be built-in to run |
| 363 | |
| 364 | ... as speaking over an HTTPS proxy implies http! |
| 365 | |
| 366 | Closes #6014 |
| 367 | |
| 368 | - ngtcp2: adapt to new NGTCP2_PROTO_VER_MAX define |
| 369 | |
| 370 | Closes #6012 |
| 371 | |
| 372 | - [Javier Blazquez brought this change] |
| 373 | |
| 374 | strerror: honor Unicode API choice on Windows |
| 375 | |
| 376 | Closes #6005 |
| 377 | |
| 378 | - imap: make imap_send use dynbuf for the send buffer management |
| 379 | |
| 380 | Reuses the buffer and thereby reduces number of mallocs over a transfer. |
| 381 | |
| 382 | Closes #6010 |
| 383 | |
| 384 | - Curl_send: return error when pre_receive_plain can't malloc |
| 385 | |
| 386 | ... will probably trigger some false DEAD CODE positives on non-windows |
| 387 | code analyzers for the conditional code. |
| 388 | |
| 389 | Closes #6011 |
| 390 | |
| 391 | - ftp: separate FTPS from FTP over "HTTPS proxy" |
| 392 | |
| 393 | When using HTTPS proxy, SSL is used but not in the view of the FTP |
| 394 | protocol handler itself so separate the connection's use of SSL from the |
| 395 | FTP control connection's sue. |
| 396 | |
| 397 | Reported-by: Mingtao Yang |
| 398 | Fixes #5523 |
| 399 | Closes #6006 |
| 400 | |
| 401 | Dan Fandrich (23 Sep 2020) |
| 402 | - tests/data: Fix some mismatched XML tags in test cases |
| 403 | |
| 404 | This allows these test files to pass xmllint. |
| 405 | |
| 406 | Daniel Stenberg (23 Sep 2020) |
| 407 | - pingpong: use a dynbuf for the *_pp_sendf() function |
| 408 | |
| 409 | ... reuses the same dynamic buffer instead of doing repeated malloc/free |
| 410 | cycles. |
| 411 | |
| 412 | Test case 100 (FTP dir list PASV) does 7 fewer memory allocation calls |
| 413 | after this change in my test setup (132 => 125), curl 7.72.0 needed 140 |
| 414 | calls for this. |
| 415 | |
| 416 | Test case 103 makes 9 less allocations now (130). Down from 149 in |
| 417 | 7.72.0. |
| 418 | |
| 419 | Closes #6004 |
| 420 | |
| 421 | - dynbuf: add Curl_dyn_vaddf |
| 422 | |
| 423 | Closes #6004 |
| 424 | |
| 425 | - dynbuf: make *addf() not require extra mallocs |
| 426 | |
| 427 | ... by introducing a printf() function that appends directly into a |
| 428 | dynbuf: Curl_dyn_vprintf(). This avoids the mandatory extra malloc so if |
| 429 | the buffer is already big enough it can just printf directly into it. |
| 430 | |
| 431 | Since this less-malloc version requires tthe use of a library internal |
| 432 | printf function, we only provide this version when building libcurl and |
| 433 | not for the dynbuf code that is used when building the curl tool. |
| 434 | |
| 435 | Closes #5998 |
| 436 | |
| 437 | - KNOWN_BUGS: Unable to use PKCS12 certificate with Secure Transport |
| 438 | |
| 439 | Closes #5403 |
| 440 | |
| 441 | - pingpong: remove a malloc per Curl_pp_vsendf call |
| 442 | |
| 443 | This typically makes 7-9 fewer mallocs per FTP transfer. |
| 444 | |
| 445 | Closes #5997 |
| 446 | |
| 447 | - symbian: drop support |
| 448 | |
| 449 | The OS is deprecated. I see no traces of anyone having actually built |
| 450 | curl for Symbian after 2012. |
| 451 | |
| 452 | The public headers are unmodified. |
| 453 | |
| 454 | Closes #5989 |
| 455 | |
| 456 | - RELEASE-NOTES: synced |
| 457 | |
| 458 | - curl_krb5.h: rename from krb5.h |
| 459 | |
| 460 | Follow-up from f4873ebd0be32cf |
| 461 | |
| 462 | Turns out some older openssl installations go bananas otherwise. |
| 463 | Reported-by: Tom van der Woerdt |
| 464 | Fixes #5995 |
| 465 | Closes #5996 |
| 466 | |
| 467 | - test1297: verify GOT_NOTHING with http proxy tunnel |
| 468 | |
| 469 | - http_proxy: do not count proxy headers in the header bytecount |
| 470 | |
| 471 | ... as that counter is subsequently used to detect if nothing was |
| 472 | returned from the peer. This made curl return CURLE_OK when it should |
| 473 | have returned CURLE_GOT_NOTHING. |
| 474 | |
| 475 | Fixes #5992 |
| 476 | Reported-by: Tom van der Woerdt |
| 477 | Closes #5994 |
| 478 | |
| 479 | - setopt: return CURLE_BAD_FUNCTION_ARGUMENT on bad argument |
| 480 | |
| 481 | Fixed two return code mixups. CURLE_UNKNOWN_OPTION is saved for when the |
| 482 | option is, yeah, not known. Clarified this in the setopt man page too. |
| 483 | |
| 484 | Closes #5993 |
| 485 | |
| 486 | - krb5: merged security.c and krb specific FTP functions in here |
| 487 | |
| 488 | These two files were always tightly connected and it was hard to |
| 489 | understand what went into which. This also allows us to make the |
| 490 | ftpsend() function static (moved from ftp.c). |
| 491 | |
| 492 | Removed security.c |
| 493 | Renamed curl_sec.h to krb5.h |
| 494 | |
| 495 | Closes #5987 |
| 496 | |
| 497 | - Curl_handler: add 'family' to each protocol |
| 498 | |
| 499 | Makes get_protocol_family() faster and it moves the knowledge about the |
| 500 | "families" to each protocol handler, where it belongs. |
| 501 | |
| 502 | Closes #5986 |
| 503 | |
| 504 | - parsedate: tune the date to epoch conversion |
| 505 | |
| 506 | By avoiding an unnecessary error check and the temp use of the tm |
| 507 | struct, the time2epoch conversion function gets a little bit faster. |
| 508 | When repeating test 517, the updated version is perhaps 1% faster (on |
| 509 | one particular build on one particular architecture). |
| 510 | |
| 511 | Closes #5985 |
| 512 | |
| 513 | - cmake: remove scary warning |
| 514 | |
| 515 | Remove the text saying |
| 516 | |
| 517 | "the curl cmake build system is poorly maintained. Be aware" |
| 518 | |
| 519 | ... not because anything changed just now, but to encourage users to use |
| 520 | it and subsequently improve it. |
| 521 | |
| 522 | Closes #5984 |
| 523 | |
| 524 | - docs/MQTT: remove outdated paaragraphs |
| 525 | |
| 526 | - docs/MQTT: not experimental anymore |
| 527 | |
| 528 | Follow-up to e37e4468688d8f |
| 529 | |
| 530 | - docs/RESOURCES: remove |
| 531 | |
| 532 | This document is not maintained and rather than trying to refresh it, |
| 533 | let's kill it. A more up-to-date document with relevant RFCs is this |
| 534 | page on the curl website: https://curl.haxx.se/rfc/ |
| 535 | |
| 536 | Closes #5980 |
| 537 | |
| 538 | - docs/TheArtOfHttpScripting: convert to markdown |
| 539 | |
| 540 | Makes it easier to browse on github etc. Offers (better) links. |
| 541 | |
| 542 | It should be noted that this document is already mostly outdated and |
| 543 | "Everything curl" at https://ec.haxx.se/ is a better resource and |
| 544 | tutorial. |
| 545 | |
| 546 | Closes #5981 |
| 547 | |
| 548 | - BUGS: convert document to markdown |
| 549 | |
| 550 | Closes #5979 |
| 551 | |
| 552 | - --help: strdup the category |
| 553 | |
| 554 | ... since it is converted and the original pointer is freed on Windows |
| 555 | unicode handling. |
| 556 | |
| 557 | Follow-up to aa8777f63febc |
| 558 | Fixes #5977 |
| 559 | Closes #5978 |
| 560 | Reported-by: xwxbug on github |
| 561 | |
| 562 | - CHECKSRC: document two missing warnings |
| 563 | |
| 564 | - RELEASE-NOTES: synced |
| 565 | |
| 566 | - ftp: avoid risk of reading uninitialized integers |
| 567 | |
| 568 | If the received PASV response doesn't match the expected pattern, we |
| 569 | could end up reading uninitialized integers for IP address and port |
| 570 | number. |
| 571 | |
| 572 | Issue pointed out by muse.dev |
| 573 | Closes #5972 |
| 574 | |
| 575 | - [Quentin Balland brought this change] |
| 576 | |
| 577 | easy_reset: clear retry counter |
| 578 | |
| 579 | Closes #5975 |
| 580 | Fixes #5974 |
| 581 | |
| 582 | - ftp: get rid of the PPSENDF macro |
| 583 | |
| 584 | The use of such a macro hides some of what's actually going on to the |
| 585 | reader and is generally disapproved of in the project. |
| 586 | |
| 587 | Closes #5971 |
| 588 | |
| 589 | - man pages: switch to https://example.com URLs |
| 590 | |
| 591 | Since HTTPS is "the new normal", this update changes a lot of man page |
| 592 | examples to use https://example.com instead of the previous "http://..." |
| 593 | |
| 594 | Closes #5969 |
| 595 | |
| 596 | - github: remove the duplicate "Security vulnerability" entry |
| 597 | |
| 598 | ... since github adds an entry automatically by itself. |
| 599 | |
| 600 | Closes #5970 |
| 601 | |
| 602 | - [Emil Engler brought this change] |
| 603 | |
| 604 | github: use new issue template feature |
| 605 | |
| 606 | This helps us to avoid getting feature requests as well as security |
| 607 | bugs reported into the issue tracker. |
| 608 | |
| 609 | Closes #5936 |
| 610 | |
| 611 | - [Emil Engler brought this change] |
| 612 | |
| 613 | urlapi: use more Curl_safefree |
| 614 | |
| 615 | Closes #5968 |
| 616 | |
| 617 | Marc Hoersken (17 Sep 2020) |
| 618 | - multi: align WinSock mask variables in Curl_multi_wait |
| 619 | |
| 620 | Also skip pre-checking sockets to set timeout_ms to 0 |
| 621 | after the first socket has been detected to be ready. |
| 622 | |
| 623 | Reviewed-by: rcombs on github |
| 624 | Reviewed-by: Daniel Stenberg |
| 625 | |
| 626 | Follow up to #5886 |
| 627 | |
| 628 | - multi: reuse WinSock events variable in Curl_multi_wait |
| 629 | |
| 630 | Since the struct is quite large (1 long and 10 ints) we |
| 631 | declare it once at the beginning of the function instead |
| 632 | of multiple times inside loops to avoid stack movements. |
| 633 | |
| 634 | Reviewed-by: Viktor Szakats |
| 635 | Reviewed-by: Daniel Stenberg |
| 636 | |
| 637 | Closes #5886 |
| 638 | |
| 639 | Daniel Stenberg (16 Sep 2020) |
| 640 | - TODO: dynamically decide to use socketpair |
| 641 | |
| 642 | Suggested-by: Anders Bakken |
| 643 | |
| 644 | Closes #4829 |
| 645 | |
| 646 | - TODO: add PR reference for native IDN support on macOS |
| 647 | |
| 648 | As there was work started on this that never got completed. |
| 649 | |
| 650 | Closes #5371 |
| 651 | |
| 652 | - tool_help.h: update copyright year range |
| 653 | |
| 654 | Follow-up from aa8777f63febca |
| 655 | |
| 656 | - CI/azure: disable test 571 in the msys2 builds |
| 657 | |
| 658 | It's just too flaky there |
| 659 | |
| 660 | Reviewed-by: Marc Hoersken |
| 661 | Closes #5954 |
| 662 | |
| 663 | - tool_writeout: protect fputs() from NULL |
| 664 | |
| 665 | When the code was changed to do fputs() instead of fprintf() it got |
| 666 | sensitive for NULL pointers; add checks for that. |
| 667 | |
| 668 | Follow-up from 0c1e767e83ec66 |
| 669 | |
| 670 | Closes #5963 |
| 671 | |
| 672 | - test3015: verify stdout "as text" |
| 673 | |
| 674 | Follow-up from 0c1e767e83e to please win32 tests |
| 675 | |
| 676 | Closes #5962 |
| 677 | |
| 678 | - travis: use libressl v3.1.4 instead of master |
| 679 | |
| 680 | ... as their git master seems too fragile to use (and 3.2.1 which is the |
| 681 | latest has a build failure). |
| 682 | |
| 683 | Closes #5964 |
| 684 | |
| 685 | - tests/FILEFORMAT: document type=shell for <command> |
| 686 | |
| 687 | - tests/FILEFORMAT: document nonewline support for <file> |
| 688 | |
| 689 | The one in <client>, that creates files. |
| 690 | |
| 691 | Follow-up from b83947c8df7 |
| 692 | |
| 693 | - [anio brought this change] |
| 694 | |
| 695 | tool_writeout: add new writeout variable, %{num_headers} |
| 696 | |
| 697 | This variable gives the number of headers. |
| 698 | |
| 699 | Closes #5947 |
| 700 | |
| 701 | - tool_urlglob: fix compiler warning "unreachable code" |
| 702 | |
| 703 | (On Windows builds.) |
| 704 | |
| 705 | Follow-up to 70a3b003d9 |
| 706 | |
| 707 | - [Gergely Nagy brought this change] |
| 708 | |
| 709 | vtls: deduplicate client certificates in ssl_config_data |
| 710 | |
| 711 | Closes #5629 |
| 712 | |
| 713 | - ftp: a 550 response to SIZE returns CURLE_REMOTE_FILE_NOT_FOUND |
| 714 | |
| 715 | This is primarily interesting for cases where CURLOPT_NOBODY is set as |
| 716 | previously curl would not return an error for this case. |
| 717 | |
| 718 | MDTM getting 550 now also returns this error (it returned |
| 719 | CURLE_FTP_COULDNT_RETR_FILE before) in order to unify return codes for |
| 720 | missing files across protocols and specific FTP commands. |
| 721 | |
| 722 | libcurl already returns error on a 550 as a MDTM response (when |
| 723 | CURLOPT_FILETIME is set). If CURLOPT_NOBODY is not set, an error would |
| 724 | happen subsequently anyway since the RETR command would fail. |
| 725 | |
| 726 | Add test 1913 and 1914 to verify. Updated several tests accordingly due |
| 727 | to the updated SIZE behavior. |
| 728 | |
| 729 | Reported-by: Tomas Berger |
| 730 | Fixes #5953 |
| 731 | Closes #5957 |
| 732 | |
| 733 | - curl: make checkpasswd use dynbuf |
| 734 | |
| 735 | Closes #5952 |
| 736 | |
| 737 | - curl: make glob_match_url use dynbuf |
| 738 | |
| 739 | Closes #5952 |
| 740 | |
| 741 | - curl: make file2memory use dynbuf |
| 742 | |
| 743 | Closes #5952 |
| 744 | |
| 745 | - curl: make file2string use dynbuf |
| 746 | |
| 747 | Closes #5952 |
| 748 | |
| 749 | - [Antarpreet Singh brought this change] |
| 750 | |
| 751 | imap: set cselect_bits to CURL_CSELECT_IN initially |
| 752 | |
| 753 | ... when continuing a transfer from a FETCH response. |
| 754 | |
| 755 | When the size of the file was small enough that the entirety of the |
| 756 | transfer happens in a single go and schannel buffers holds the entire |
| 757 | data. However, it wasn't completely read in Curl_pp_readresp since a |
| 758 | line break was found before that could happen. So, by the time we are in |
| 759 | imap_state_fetch_resp - there's data in buffers that needs to be read |
| 760 | via Curl_read but nothing to read from the socket. After we setup a |
| 761 | transfer (Curl_setup_transfer), curl just waits on the socket state to |
| 762 | change - which doesn't happen since no new data ever comes. |
| 763 | |
| 764 | Closes #5961 |
| 765 | |
| 766 | - RELEASE-NOTES: synced |
| 767 | |
| 768 | - test434: test -K use in a single line without newline |
| 769 | |
| 770 | Closes #5946 |
| 771 | |
| 772 | - runtests: allow creating files without newlines |
| 773 | |
| 774 | Closes #5946 |
| 775 | |
| 776 | - curl: use curlx_dynbuf for realloc when loading config files |
| 777 | |
| 778 | ... fixes an integer overflow at the same time. |
| 779 | |
| 780 | Reported-by: ihsinme on github |
| 781 | Assisted-by: Jay Satiro |
| 782 | |
| 783 | Closes #5946 |
| 784 | |
| 785 | - dynbuf: provide curlx_ names for reuse by the curl tool |
| 786 | |
| 787 | Closes #5946 |
| 788 | |
| 789 | - dynbuf: make sure Curl_dyn_tail() zero terminates |
| 790 | |
| 791 | Closes #5959 |
| 792 | |
| 793 | - tests: add test1912 to the dist |
| 794 | |
| 795 | Follow-up to 70984ce1be4cab6c |
| 796 | |
| 797 | - docs/LICENSE-MIXING: remove |
| 798 | |
| 799 | This document is not maintained and I feel that it doesn't provide much |
| 800 | value to users anymore (if it ever did). |
| 801 | |
| 802 | Closes #5955 |
| 803 | |
| 804 | - [Laramie Leavitt brought this change] |
| 805 | |
| 806 | http: consolidate nghttp2_session_mem_recv() call paths |
| 807 | |
| 808 | Previously there were several locations that called |
| 809 | nghttp2_session_mem_recv and handled responses slightly differently. |
| 810 | Those have been converted to call the existing |
| 811 | h2_process_pending_input() function. |
| 812 | |
| 813 | Moved the end-of-session check to h2_process_pending_input() since the |
| 814 | only place the end-of-session state can change is after nghttp2 |
| 815 | processes additional input frames. |
| 816 | |
| 817 | This will likely fix the fuzzing error. While I don't have a root cause |
| 818 | the out-of-bounds read seems like a use after free, so moving the |
| 819 | nghttp2_session_check_request_allowed() call to a location with a |
| 820 | guaranteed nghttp2 session seems reasonable. |
| 821 | |
| 822 | Also updated a few nghttp2 callsites to include error messages and added |
| 823 | a few additional error checks. |
| 824 | |
| 825 | Closes #5648 |
| 826 | |
| 827 | - HISTORY: mention alt-svc added in 2019 |
| 828 | |
| 829 | ... and make 1996 the first year subtitle |
| 830 | |
| 831 | - base64: also build for pop3 and imap |
| 832 | |
| 833 | Follow-up to the fix in 20417a13fb8f83 |
| 834 | |
| 835 | Reported-by: Michael Olbrich |
| 836 | Fixes #5937 |
| 837 | Closes #5948 |
| 838 | |
| 839 | - base64: enable in build with SMTP |
| 840 | |
| 841 | The oauth2 support is used with SMTP and it uses base64 functions. |
| 842 | |
| 843 | Reported-by: Michael Olbrich |
| 844 | Fixes #5937 |
| 845 | Closes #5938 |
| 846 | |
| 847 | - curl_mime_headers.3: fix the example's use of curl_slist_append |
| 848 | |
| 849 | Reported-by: sofaboss on github |
| 850 | Fixes #5942 |
| 851 | Closes #5943 |
| 852 | |
| 853 | - lib583: fix enum mixup |
| 854 | |
| 855 | grrr the previous follow-up to 17fcdf6a31 was wrong |
| 856 | |
| 857 | - libtest: fix build errors |
| 858 | |
| 859 | Follow-up from 17fcdf6a310d4c8076 |
| 860 | |
| 861 | - lib: fix -Wassign-enum warnings |
| 862 | |
| 863 | configure --enable-debug now enables -Wassign-enum with clang, |
| 864 | identifying several enum "abuses" also fixed. |
| 865 | |
| 866 | Reported-by: Gisle Vanem |
| 867 | Bug: https://github.com/curl/curl/commit/879007f8118771f4896334731aaca5850a154675#commitcomment-42087553 |
| 868 | |
| 869 | Closes #5929 |
| 870 | |
| 871 | - RELEASE-NOTES: synced |
| 872 | |
| 873 | - [Diven Qi brought this change] |
| 874 | |
| 875 | url: use blank credentials when using proxy w/o username and password |
| 876 | |
| 877 | Fixes proxy regression brought in commit ad829b21ae (7.71.0) |
| 878 | |
| 879 | Fixed #5911 |
| 880 | Closes #5914 |
| 881 | |
| 882 | - travis: add a build using libressl (from git master) |
| 883 | |
| 884 | The v3.2.1 tag (latest release atm) results in a broken build. |
| 885 | |
| 886 | Closes #5932 |
| 887 | |
| 888 | - configure: let --enable-debug set -Wenum-conversion with gcc >= 10 |
| 889 | |
| 890 | Unfortunately, this option is not detecting the same issues as clang's |
| 891 | -Wassign-enum flag, but should still be useful to detect future |
| 892 | mistakes. |
| 893 | |
| 894 | Closes #5930 |
| 895 | |
| 896 | - openssl: consider ALERT_CERTIFICATE_EXPIRED a failed verification |
| 897 | |
| 898 | If the error reason from the lib is |
| 899 | SSL_R_SSLV3_ALERT_CERTIFICATE_EXPIRED, libcurl will return |
| 900 | CURLE_PEER_FAILED_VERIFICATION and not CURLE_SSL_CONNECT_ERROR. |
| 901 | |
| 902 | This unifies the libcurl return code and makes libressl run test 313 |
| 903 | (CRL testing) fine. |
| 904 | |
| 905 | Closes #5934 |
| 906 | |
| 907 | - FAQ: refreshed some very old language |
| 908 | |
| 909 | - cmake: make HTTP_ONLY also disable MQTT |
| 910 | |
| 911 | ... and alphasort the order of disabling protocols to make it easier to |
| 912 | browse. |
| 913 | |
| 914 | Closes #5931 |
| 915 | |
| 916 | - libtest: remove lib1541 leftovers |
| 917 | |
| 918 | Caused automake errors. |
| 919 | |
| 920 | Follow-up to 8ca54a03ea08a |
| 921 | |
| 922 | - tests/libtests: remove test 1900 and 2033 |
| 923 | |
| 924 | We already remove the test files, now remove the libtest codes as well. |
| 925 | |
| 926 | Follow-up to e50a877df74 |
| 927 | |
| 928 | Marc Hoersken (7 Sep 2020) |
| 929 | - CI/azure: add test number to title for display in analytics |
| 930 | |
| 931 | To ease identification of tests the test number is added to |
| 932 | the test case title in order to have it on the Azure DevOps |
| 933 | Analytics pages and reports which currently do not show it. |
| 934 | |
| 935 | Bump test case revision to make Azure DevOps update titles. |
| 936 | |
| 937 | Closes #5927 |
| 938 | |
| 939 | Daniel Stenberg (6 Sep 2020) |
| 940 | - altsvc: clone setting in curl_easy_duphandle |
| 941 | |
| 942 | The cache content is not duplicated, like other caches, but the setting |
| 943 | and specified file name are. |
| 944 | |
| 945 | Test 1908 is extended to verify this somewhat. Since the duplicated |
| 946 | handle gets the same file name, the test unfortunately overwrites the |
| 947 | same file twice (with different contents) which makes it hard to check |
| 948 | automatically. |
| 949 | |
| 950 | Closes #5923 |
| 951 | |
| 952 | - test1541: remove since it is a known bug |
| 953 | |
| 954 | A shared connection cache is not thread-safe is a known issue. Stop |
| 955 | testing this until we believe this issue is addressed. Reduces |
| 956 | occasional test failures we don't care about. |
| 957 | |
| 958 | The test code in lib1541.c is left in git to allow us to restore it when |
| 959 | we get to fix this. |
| 960 | |
| 961 | Closes #5922 |
| 962 | |
| 963 | - tests: remove pipelining tests |
| 964 | |
| 965 | Remove the tests 530, 584, 1900, 1901, 1902, 1903 and 2033. They were |
| 966 | previously disabled. |
| 967 | |
| 968 | The Pipelining code was removed from curl in commit 2f44e94efb3df8e, |
| 969 | April 2019. |
| 970 | |
| 971 | Closes #5921 |
| 972 | |
| 973 | - curl: retry delays in parallel mode no longer sleeps blocking |
| 974 | |
| 975 | The previous sleep for retries would block all other concurrent |
| 976 | transfers. Starting now, the retry will instead be properly marked to |
| 977 | not get restarted until after the delay time but other transfers can |
| 978 | still continue in the mean time. |
| 979 | |
| 980 | Closes #5917 |
| 981 | |
| 982 | - curl:parallel_transfers: make sure retry readds the transfer |
| 983 | |
| 984 | Reported-by: htasta on github |
| 985 | Fixes #5905 |
| 986 | Closes #5917 |
| 987 | |
| 988 | - build: drop support for building with Watcom |
| 989 | |
| 990 | These files are not maintained, they seem to have no users, Watcom |
| 991 | compilers look like not having users nor releases anymore. |
| 992 | |
| 993 | Closes #5918 |
| 994 | |
| 995 | - winbuild/rundebug.cmd: remove |
| 996 | |
| 997 | Seems to have been added by mistake? Not included in dists. |
| 998 | |
| 999 | Closes #5919 |
| 1000 | |
| 1001 | - curl: in retry output don't call all problems "transient" |
| 1002 | |
| 1003 | ... because when --retry-all-errors is used, the error isn't necessarily |
| 1004 | transient at all. |
| 1005 | |
| 1006 | Closes #5916 |
| 1007 | |
| 1008 | - easygetopt: pass a valid enum to avoid compiler warning |
| 1009 | |
| 1010 | "integer constant not in range of enumerated type 'CURLoption'" |
| 1011 | |
| 1012 | Reported-by: Gisle Vanem |
| 1013 | Bug: https://github.com/curl/curl/commit/6ebe63fac23f38df911edc348e8ccc72280f9434#commitcomment-42042843 |
| 1014 | |
| 1015 | Closes #5915 |
| 1016 | |
| 1017 | - [Emil Engler brought this change] |
| 1018 | |
| 1019 | tests: Add tests for new --help |
| 1020 | |
| 1021 | This commit is a part of "--help me if you can" |
| 1022 | |
| 1023 | Closes #5680 |
| 1024 | |
| 1025 | - [Emil Engler brought this change] |
| 1026 | |
| 1027 | tool: update --help with categories |
| 1028 | |
| 1029 | This commit is a part of "--help me if you can" |
| 1030 | |
| 1031 | Closes #5680 |
| 1032 | |
| 1033 | - [Emil Engler brought this change] |
| 1034 | |
| 1035 | docs: add categories to all cmdline opts |
| 1036 | |
| 1037 | Adapted gen.pl with 'listcats' |
| 1038 | |
| 1039 | This commit is a part of "--help me if you can" |
| 1040 | |
| 1041 | Closes #5680 |
| 1042 | |
| 1043 | - RELEASE-NOTES: synced |
| 1044 | |
| 1045 | - [ihsinme brought this change] |
| 1046 | |
| 1047 | connect.c: remove superfluous 'else' in Curl_getconnectinfo |
| 1048 | |
| 1049 | Closes #5912 |
| 1050 | |
| 1051 | - [Samuel Marks brought this change] |
| 1052 | |
| 1053 | CMake: remove explicit `CMAKE_ANSI_CFLAGS` |
| 1054 | |
| 1055 | This variable was removed from cmake in commit |
| 1056 | https://gitlab.kitware.com/cmake/cmake/commit/5a834b0bb0bc288. A later |
| 1057 | CMake commit removes the variable from the tests, claiming that it was |
| 1058 | removed in CMake 2.6 |
| 1059 | |
| 1060 | Reviewed-By: Peter Wu |
| 1061 | Closes #5439 |
| 1062 | |
| 1063 | - [cbe brought this change] |
| 1064 | |
| 1065 | libssh2: pass on the error from ssh_force_knownhost_key_type |
| 1066 | |
| 1067 | Closes #5909 |
| 1068 | |
| 1069 | - scripts/delta: add diffstat summary |
| 1070 | |
| 1071 | ... and make output more table-like |
| 1072 | |
| 1073 | - [Martin Bašti brought this change] |
| 1074 | |
| 1075 | http_proxy: do not crash with HTTPS_PROXY and NO_PROXY set |
| 1076 | |
| 1077 | ... in case NO_PROXY takes an effect |
| 1078 | |
| 1079 | Without this patch, the following command crashes: |
| 1080 | |
| 1081 | $ GIT_CURL_VERBOSE=1 NO_PROXY=github.com HTTPS_PROXY=https://example.com \ |
| 1082 | git clone https://github.com/curl/curl.git |
| 1083 | |
| 1084 | Minimal libcurl-based reproducer: |
| 1085 | |
| 1086 | #include <curl/curl.h> |
| 1087 | |
| 1088 | int main() { |
| 1089 | CURL *curl = curl_easy_init(); |
| 1090 | if(curl) { |
| 1091 | CURLcode ret; |
| 1092 | curl_easy_setopt(curl, CURLOPT_URL, "https://github.com/"); |
| 1093 | curl_easy_setopt(curl, CURLOPT_PROXY, "example.com"); |
| 1094 | /* set the proxy type */ |
| 1095 | curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_HTTPS); |
| 1096 | curl_easy_setopt(curl, CURLOPT_NOPROXY, "github.com"); |
| 1097 | curl_easy_setopt(curl, CURLOPT_VERBOSE, 1L); |
| 1098 | ret = curl_easy_perform(curl); |
| 1099 | curl_easy_cleanup(curl); |
| 1100 | return ret; |
| 1101 | } |
| 1102 | return -1; |
| 1103 | } |
| 1104 | |
| 1105 | Assisted-by: Kamil Dudka |
| 1106 | Bug: https://bugzilla.redhat.com/1873327 |
| 1107 | Closes #5902 |
| 1108 | |
| 1109 | - travis: add a CI job with openssl3 (from git master) |
| 1110 | |
| 1111 | Closes #5908 |
| 1112 | |
| 1113 | - openssl: avoid error conditions when importing native CA |
| 1114 | |
| 1115 | The code section that is OpenSSL 3+ specific now uses the same logic as |
| 1116 | is used in the version < 3 section. It caused a compiler error without |
| 1117 | it. |
| 1118 | |
| 1119 | Closes #5907 |
| 1120 | |
| 1121 | - setopt: avoid curl_ on local variable |
| 1122 | |
| 1123 | Closes #5906 |
| 1124 | |
| 1125 | - mqtt.c: avoid curl_ prefix on local variable |
| 1126 | |
| 1127 | Closes #5906 |
| 1128 | |
| 1129 | - wildcard: strip "curl_" prefix from private symbols |
| 1130 | |
| 1131 | Closes #5906 |
| 1132 | |
| 1133 | - vtls: make it 'struct Curl_ssl_session' |
| 1134 | |
| 1135 | Use uppercase C for internal symbols. |
| 1136 | |
| 1137 | Closes #5906 |
| 1138 | |
| 1139 | - curl_threads: make it 'struct Curl_actual_call' |
| 1140 | |
| 1141 | Internal names should not be prefixed "curl_" |
| 1142 | |
| 1143 | Closes #5906 |
| 1144 | |
| 1145 | - schannel: make it 'struct Curl_schannel*' |
| 1146 | |
| 1147 | As internal global names should use captical C. |
| 1148 | |
| 1149 | Closes #5906 |
| 1150 | |
| 1151 | - hash: make it 'struct Curl_hash' |
| 1152 | |
| 1153 | As internal global names should use captical C. |
| 1154 | |
| 1155 | Closes #5906 |
| 1156 | |
| 1157 | - llist: make it "struct Curl_llist" |
| 1158 | |
| 1159 | As internal global names should use captical C. |
| 1160 | |
| 1161 | Closes #5906 |
| 1162 | |
| 1163 | Marc Hoersken (2 Sep 2020) |
| 1164 | - telnet.c: depend on static requirement of WinSock version 2 |
| 1165 | |
| 1166 | Drop dynamic loading of ws2_32.dll and instead rely on the |
| 1167 | imported version which is now required to be at least 2.2. |
| 1168 | |
| 1169 | Reviewed-by: Marcel Raad |
| 1170 | Reviewed-by: Jay Satiro |
| 1171 | Reviewed-by: Daniel Stenberg |
| 1172 | Reviewed-by: Viktor Szakats |
| 1173 | |
| 1174 | Closes #5854 |
| 1175 | |
| 1176 | - win32: drop support for WinSock version 1, require version 2 |
| 1177 | |
| 1178 | IPv6, telnet and now also the multi API require WinSock |
| 1179 | version 2 which is available starting with Windows 95. |
| 1180 | |
| 1181 | Therefore we think it is time to drop support for version 1. |
| 1182 | |
| 1183 | Reviewed-by: Marcel Raad |
| 1184 | Reviewed-by: Jay Satiro |
| 1185 | Reviewed-by: Daniel Stenberg |
| 1186 | Reviewed-by: Viktor Szakats |
| 1187 | |
| 1188 | Follow up to #5634 |
| 1189 | Closes #5854 |
| 1190 | |
| 1191 | - select: align poll emulation to return all relevant events |
| 1192 | |
| 1193 | The poll emulation via select already consumes POLLRDNORM, |
| 1194 | POLLWRNORM and POLLRDBAND as input events. Therefore it |
| 1195 | should also return them as output events if signaled. |
| 1196 | |
| 1197 | Also fix indentation in input event handling block. |
| 1198 | |
| 1199 | Assisted-by: Jay Satiro |
| 1200 | Reviewed-by: Daniel Stenberg |
| 1201 | |
| 1202 | Replaces #5852 |
| 1203 | Closes #5883 |
| 1204 | |
| 1205 | - CI/azure: MQTT is now enabled by default |
| 1206 | |
| 1207 | Reviewed-by: Daniel Stenberg |
| 1208 | |
| 1209 | Follow up to #5858 |
| 1210 | Closes #5903 |
| 1211 | |
| 1212 | Daniel Stenberg (2 Sep 2020) |
| 1213 | - copyright.pl: ignore buildconf |
| 1214 | |
| 1215 | - test971: show test mismatches "inline" |
| 1216 | |
| 1217 | - lib/Makefile.am: bump VERSIONINFO due to new functions |
| 1218 | |
| 1219 | ... we're generally bad at this, but we are adding new functions for |
| 1220 | this release. |
| 1221 | |
| 1222 | Closes #5899 |
| 1223 | |
| 1224 | - optiontable: use DEBUGBUILD |
| 1225 | |
| 1226 | Follow-up to commit 6e18568ba38 (#5877) |
| 1227 | |
| 1228 | - cmdline-opts/gen.pl: generate nicer "See Also" in curl.1 |
| 1229 | |
| 1230 | If there are more than two items in the list, use commas for all but the |
| 1231 | last separator which is set to 'and'. Reads better. |
| 1232 | |
| 1233 | Closes #5898 |
| 1234 | |
| 1235 | - curl.1: add see also no-progress-meter on two spots |
| 1236 | |
| 1237 | Ref: #5894 |
| 1238 | |
| 1239 | Closes #5897 |
| 1240 | |
| 1241 | - RELEASE-NOTES: synced |
| 1242 | |
| 1243 | - mqtt: enable by default |
| 1244 | |
| 1245 | No longer considered experimental. |
| 1246 | |
| 1247 | Closes #5858 |
| 1248 | |
| 1249 | - [Michael Baentsch brought this change] |
| 1250 | |
| 1251 | tls: add CURLOPT_SSL_EC_CURVES and --curves |
| 1252 | |
| 1253 | Closes #5892 |
| 1254 | |
| 1255 | - url: remove funny embedded comments in Curl_disonnect calls |
| 1256 | |
| 1257 | - [Chris Paulson-Ellis brought this change] |
| 1258 | |
| 1259 | conn: check for connection being dead before reuse |
| 1260 | |
| 1261 | Prevents incorrect reuse of an HTTP connection that has been prematurely |
| 1262 | shutdown() by the server. |
| 1263 | |
| 1264 | Partial revert of 755083d00deb16 |
| 1265 | |
| 1266 | Fixes #5884 |
| 1267 | Closes #5893 |
| 1268 | |
| 1269 | Marc Hoersken (29 Aug 2020) |
| 1270 | - buildconf: exec autoreconf to avoid additional process |
| 1271 | |
| 1272 | Also make buildconf exit with the return code of autoreconf. |
| 1273 | |
| 1274 | Reviewed-by: Daniel Stenberg |
| 1275 | |
| 1276 | Follow up to #5853 |
| 1277 | Closes #5890 |
| 1278 | |
| 1279 | - CI/azure: no longer ignore results of test 1013 |
| 1280 | |
| 1281 | Follow up to #5771 |
| 1282 | Closes #5889 |
| 1283 | |
| 1284 | - docs: add description about CI platforms to CONTRIBUTE.md |
| 1285 | |
| 1286 | Reviewed-by: Daniel Stenberg |
| 1287 | Reviewed-by: Marcel Raad |
| 1288 | Reviewed-by: Jay Satiro |
| 1289 | |
| 1290 | Closes #5882 |
| 1291 | |
| 1292 | Daniel Stenberg (29 Aug 2020) |
| 1293 | - tests/getpart: use MIME::Base64 instead of home-cooked |
| 1294 | |
| 1295 | Since we already use the base64 package since a while back, we can just |
| 1296 | as well switch to that here too. |
| 1297 | |
| 1298 | It also happens to use the exact same function name, which otherwise |
| 1299 | causes a run-time warning. |
| 1300 | |
| 1301 | Reported-by: Marc Hörsken |
| 1302 | Fixes #5885 |
| 1303 | Closes #5887 |
| 1304 | |
| 1305 | Marcel Raad (29 Aug 2020) |
| 1306 | - ntlm: fix condition for curl_ntlm_core usage |
| 1307 | |
| 1308 | `USE_WINDOWS_SSPI` without `USE_WIN32_CRYPTO` but with any other DES |
| 1309 | backend is fine, but was excluded before. |
| 1310 | |
| 1311 | This also fixes test 1013 as the condition for SMB support in |
| 1312 | configure.ac didn't match the condition in the source code. Now it |
| 1313 | does. |
| 1314 | |
| 1315 | Fixes https://github.com/curl/curl/issues/1262 |
| 1316 | Closes https://github.com/curl/curl/pull/5771 |
| 1317 | |
| 1318 | - AppVeyor: switch 64-bit Schannel Debug CMake builds to Unicode |
| 1319 | |
| 1320 | The Schannel builds are the most useful to verify as they make the most |
| 1321 | use of the Windows API. Classic MinGW doesn't support Unicode at all, |
| 1322 | only MinGW-w64 and MSVC do. |
| 1323 | |
| 1324 | Closes https://github.com/curl/curl/pull/5843 |
| 1325 | |
| 1326 | - CMake: add option to enable Unicode on Windows |
| 1327 | |
| 1328 | As already existing for winbuild. |
| 1329 | |
| 1330 | Closes https://github.com/curl/curl/pull/5843 |
| 1331 | |
| 1332 | Marc Hoersken (29 Aug 2020) |
| 1333 | - select: simplify return code handling for poll and select |
| 1334 | |
| 1335 | poll and select already return -1 on error according to POSIX, |
| 1336 | so there is no need to perform a <0 to -1 conversion in code. |
| 1337 | |
| 1338 | Also we can just use one check with <= 0 on the return code. |
| 1339 | |
| 1340 | Assisted-by: Daniel Stenberg |
| 1341 | Reviewed-by: Jay Satiro |
| 1342 | |
| 1343 | Replaces #5852 |
| 1344 | Closes #5880 |
| 1345 | |
| 1346 | Daniel Stenberg (28 Aug 2020) |
| 1347 | - RELEASE-NOTES: synced |
| 1348 | |
| 1349 | - [Jeroen Ooms brought this change] |
| 1350 | |
| 1351 | tests: add test1912 with typechecks |
| 1352 | |
| 1353 | Validates that gcc-typecheck macros match the new option type API. |
| 1354 | |
| 1355 | Closes #5873 |
| 1356 | |
| 1357 | - easyoptions: provide debug function when DEBUGBUILD |
| 1358 | |
| 1359 | ... not CURLDEBUG as they're not always set in conjunction. |
| 1360 | |
| 1361 | Follow-up to 6ebe63fac23f38df |
| 1362 | |
| 1363 | Fixes #5877 |
| 1364 | Closes #5878 |
| 1365 | |
| 1366 | Marc Hoersken (28 Aug 2020) |
| 1367 | - sockfilt: handle FD_CLOSE winsock event on write socket |
| 1368 | |
| 1369 | Learn from the way Cygwin handles and maps the WinSock events |
| 1370 | to simulate correct and complete poll and select behaviour |
| 1371 | according to Richard W. Stevens Network Programming book. |
| 1372 | |
| 1373 | Follow up to #5867 |
| 1374 | Closes #5879 |
| 1375 | |
| 1376 | - multi: handle connection state winsock events |
| 1377 | |
| 1378 | Learn from the way Cygwin handles and maps the WinSock events |
| 1379 | to simulate correct and complete poll and select behaviour |
| 1380 | according to Richard W. Stevens Network Programming book. |
| 1381 | |
| 1382 | Reviewed-by: Jay Satiro |
| 1383 | Reviewed-by: Marcel Raad |
| 1384 | |
| 1385 | Follow up to #5634 |
| 1386 | Closes #5867 |
| 1387 | |
| 1388 | Daniel Stenberg (28 Aug 2020) |
| 1389 | - Curl_pgrsTime - return new time to avoid timeout integer overflow |
| 1390 | |
| 1391 | Setting a timeout to INT_MAX could cause an immediate error to get |
| 1392 | returned as timeout because of an overflow when different values of |
| 1393 | 'now' were used. |
| 1394 | |
| 1395 | This is primarily fixed by having Curl_pgrsTime() return the "now" when |
| 1396 | TIMER_STARTSINGLE is set so that the parent function will continue using |
| 1397 | that time. |
| 1398 | |
| 1399 | Reported-by: Ionuț-Francisc Oancea |
| 1400 | Fixes #5583 |
| 1401 | Closes #5847 |
| 1402 | |
| 1403 | - TLS: fix SRP detection by using the proper #ifdefs |
| 1404 | |
| 1405 | USE_TLS_SRP will be true if *any* selected TLS backend can use SRP |
| 1406 | |
| 1407 | HAVE_OPENSSL_SRP is defined when OpenSSL can use it |
| 1408 | |
| 1409 | HAVE_GNUTLS_SRP is defined when GnuTLS can use it |
| 1410 | |
| 1411 | Clarify in the curl_verison_info docs that CURL_VERSION_TLSAUTH_SRP is |
| 1412 | set if at least one of the supported backends offers SRP. |
| 1413 | |
| 1414 | Reported-by: Stefan Strogin |
| 1415 | Fixes #5865 |
| 1416 | Closes #5870 |
| 1417 | |
| 1418 | - [Dan Kenigsberg brought this change] |
| 1419 | |
| 1420 | docs: SSLCERTS: fix English syntax |
| 1421 | |
| 1422 | Signed-off-by: Dan Kenigsberg <danken@redhat.com> |
| 1423 | |
| 1424 | Closes #5876 |
| 1425 | |
| 1426 | - [Alessandro Ghedini brought this change] |
| 1427 | |
| 1428 | docs: non-existing macros in man pages |
| 1429 | |
| 1430 | As reported by man(1) when invoked as: |
| 1431 | |
| 1432 | man --warnings -E UTF-8 -l -Tutf8 -Z <file> >/dev/null |
| 1433 | |
| 1434 | Closes #5846 |
| 1435 | |
| 1436 | - [Alessandro Ghedini brought this change] |
| 1437 | |
| 1438 | curl.1: fix typo invokved -> invoked |
| 1439 | |
| 1440 | Closes #5846 |
| 1441 | |
| 1442 | - buildconf: invoke 'autoreconf -fi' instead |
| 1443 | |
| 1444 | The custom script isn't necessary anymore - but remains for simplicity |
| 1445 | and just invokes autoreconf. |
| 1446 | |
| 1447 | Closes #5853 |
| 1448 | |
| 1449 | - [Emil Engler brought this change] |
| 1450 | |
| 1451 | lib: make Curl_gethostname accept a const pointer |
| 1452 | |
| 1453 | The address of that variable never gets changed, only the data in it so |
| 1454 | why not make it a "char * const"? |
| 1455 | |
| 1456 | Closes #5866 |
| 1457 | |
| 1458 | - docs/libcurl: update "Added in" version for curl_easy_option* |
| 1459 | |
| 1460 | Follow-up to 6ebe63fac23f38 |
| 1461 | |
| 1462 | - scripts: improve the "get latest curl release tag" logic |
| 1463 | |
| 1464 | ... by insiting on it matching "^curl-". |
| 1465 | |
| 1466 | - configure: added --disable-get-easy-options |
| 1467 | |
| 1468 | To allow disabling of the curl_easy_option APIs in a build. |
| 1469 | |
| 1470 | Closes #5365 |
| 1471 | |
| 1472 | - options: API for meta-data about easy options |
| 1473 | |
| 1474 | const struct curl_easyoption *curl_easy_option_by_name(const char *name); |
| 1475 | |
| 1476 | const struct curl_easyoption *curl_easy_option_by_id (CURLoption id); |
| 1477 | |
| 1478 | const struct curl_easyoption * |
| 1479 | curl_easy_option_next(const struct curl_easyoption *prev); |
| 1480 | |
| 1481 | The purpose is to provide detailed enough information to allow for |
| 1482 | example libcurl bindings to get option information at run-time about |
| 1483 | what easy options that exist and what arguments they expect. |
| 1484 | |
| 1485 | Assisted-by: Jeroen Ooms |
| 1486 | Closes #5365 |
| 1487 | |
| 1488 | - [Eric Curtin brought this change] |
| 1489 | |
| 1490 | HTTP/3: update to OpenSSL_1_1_1g-quic-draft-29 |
| 1491 | |
| 1492 | Closes #5871 |
| 1493 | |
| 1494 | - RELEASE-NOTES: synced |
| 1495 | |
| 1496 | Jay Satiro (26 Aug 2020) |
| 1497 | - openssl: Fix wincrypt symbols conflict with BoringSSL |
| 1498 | |
| 1499 | OpenSSL undefines the conflicting symbols but BoringSSL does not so we |
| 1500 | must do it ourselves. |
| 1501 | |
| 1502 | Reported-by: Samuel Tranchet |
| 1503 | Assisted-by: Javier Blazquez |
| 1504 | |
| 1505 | Ref: https://bugs.chromium.org/p/boringssl/issues/detail?id=371 |
| 1506 | Ref: https://github.com/openssl/openssl/blob/OpenSSL_1_1_1g/include/openssl/ossl_typ.h#L66-L73 |
| 1507 | |
| 1508 | Fixes https://github.com/curl/curl/issues/5669 |
| 1509 | Closes https://github.com/curl/curl/pull/5857 |
| 1510 | |
| 1511 | Daniel Stenberg (26 Aug 2020) |
| 1512 | - socketpair: allow CURL_DISABLE_SOCKETPAIR |
| 1513 | |
| 1514 | ... to completely disable the use of socketpair |
| 1515 | |
| 1516 | Closes #5850 |
| 1517 | |
| 1518 | - curl_get_line: build only if cookies or alt-svc are enabled |
| 1519 | |
| 1520 | Closes #5851 |
| 1521 | |
| 1522 | - [fullincome brought this change] |
| 1523 | |
| 1524 | schannel: fix memory leak when using get_cert_location |
| 1525 | |
| 1526 | The get_cert_location function allocates memory only on success. |
| 1527 | Previously get_cert_location was able to allocate memory and return |
| 1528 | error. It wasn't obvious and in this case the memory wasn't |
| 1529 | released. |
| 1530 | |
| 1531 | Fixes #5855 |
| 1532 | Closes #5860 |
| 1533 | |
| 1534 | - [Emil Engler brought this change] |
| 1535 | |
| 1536 | git: ignore libtests in 3XXX area |
| 1537 | |
| 1538 | Currently the file tests/libtest/lib3010 is not getting |
| 1539 | ignored by git. This fixes it by adding the 3XXX area to |
| 1540 | the according .gitignore file. |
| 1541 | |
| 1542 | Closes #5859 |
| 1543 | |
| 1544 | - [Emil Engler brought this change] |
| 1545 | |
| 1546 | doh: add error message for DOH_DNS_NAME_TOO_LONG |
| 1547 | |
| 1548 | When this error code was introduced in b6a53fff6c1d07e8a9, it was |
| 1549 | forgotten to be added in the errors array and doh_strerror function. |
| 1550 | |
| 1551 | Closes #5863 |
| 1552 | |
| 1553 | - ngtcp2: adapt to the new pkt_info arguments |
| 1554 | |
| 1555 | Guidance-by: Tatsuhiro Tsujikawa |
| 1556 | |
| 1557 | Closes #5864 |
| 1558 | |
| 1559 | - winbuild/README.md: make <options> visible |
| 1560 | |
| 1561 | Follow-up to be753add31c2d8c |
| 1562 | |
| 1563 | - winbuild: convert the instruction text to README.md |
| 1564 | |
| 1565 | Closes #5861 |
| 1566 | |
| 1567 | - lib1560: verify "redirect" to double-slash leading URL |
| 1568 | |
| 1569 | Closes #5849 |
| 1570 | |
| 1571 | Marc Hoersken (25 Aug 2020) |
| 1572 | - multi: expand pre-check for socket readiness |
| 1573 | |
| 1574 | Check readiness of all sockets before waiting on them |
| 1575 | to avoid locking in case the one-time event FD_WRITE |
| 1576 | was already consumed by a previous wait operation. |
| 1577 | |
| 1578 | More information about WinSock network events: |
| 1579 | https://docs.microsoft.com/en-us/windows/win32/api/ |
| 1580 | winsock2/nf-winsock2-wsaeventselect#return-value |
| 1581 | |
| 1582 | Closes #5634 |
| 1583 | |
| 1584 | - [rcombs brought this change] |
| 1585 | |
| 1586 | multi: implement wait using winsock events |
| 1587 | |
| 1588 | This avoids using a pair of TCP ports to provide wakeup functionality |
| 1589 | for every multi instance on Windows, where socketpair() is emulated |
| 1590 | using a TCP socket on loopback which could in turn lead to socket |
| 1591 | resource exhaustion. |
| 1592 | |
| 1593 | A previous version of this patch failed to account for how in WinSock, |
| 1594 | FD_WRITE is set only once when writing becomes possible and not again |
| 1595 | until after a send has failed due to the buffer filling. This contrasts |
| 1596 | to how FD_READ and FD_OOB continue to be set until the conditions they |
| 1597 | refer to no longer apply. This meant that if a user wrote some data to |
| 1598 | a socket, but not enough data to completely fill its send buffer, then |
| 1599 | waited on that socket to become writable, we'd erroneously stall until |
| 1600 | their configured timeout rather than returning immediately. |
| 1601 | |
| 1602 | This version of the patch addresses that issue by checking each socket |
| 1603 | we're waiting on to become writable with select() before the wait, and |
| 1604 | zeroing the timeout if it's already writable. |
| 1605 | |
| 1606 | Assisted-by: Marc Hörsken |
| 1607 | Reviewed-by: Marcel Raad |
| 1608 | Reviewed-by: Daniel Stenberg |
| 1609 | Tested-by: Gergely Nagy |
| 1610 | Tested-by: Rasmus Melchior Jacobsen |
| 1611 | Tested-by: Tomas Berger |
| 1612 | |
| 1613 | Replaces #5397 |
| 1614 | Reverts #5632 |
| 1615 | Closes #5634 |
| 1616 | |
| 1617 | - select: reduce duplication of Curl_poll in Curl_socket_check |
| 1618 | |
| 1619 | Change Curl_socket_check to use select-fallback in Curl_poll |
| 1620 | instead of implementing it in Curl_socket_check and Curl_poll. |
| 1621 | |
| 1622 | Reviewed-by: Daniel Stenberg |
| 1623 | Reviewed-by: Jay Satiro |
| 1624 | |
| 1625 | Replaces #5262 and #5492 |
| 1626 | Closes #5707 |
| 1627 | |
| 1628 | - select: fix poll-based check not detecting connect failure |
| 1629 | |
| 1630 | This commit changes Curl_socket_check to use POLLPRI to |
| 1631 | check for connect failure on the write socket, because |
| 1632 | POLLPRI maps to fds_err. This is in line with select(2). |
| 1633 | |
| 1634 | The select-based socket check correctly checks for connect |
| 1635 | failures by adding the write socket also to fds_err. |
| 1636 | |
| 1637 | The poll-based implementation (which internally can itself |
| 1638 | fallback to select again) did not previously check for |
| 1639 | connect failure by using POLLPRI with the write socket. |
| 1640 | |
| 1641 | See the follow up commit to this for more information. |
| 1642 | |
| 1643 | This commit makes sure connect failures can be detected |
| 1644 | and handled if HAVE_POLL_FINE is defined, eg. on msys2-devel. |
| 1645 | |
| 1646 | Reviewed-by: Daniel Stenberg |
| 1647 | Reviewed-by: Jay Satiro |
| 1648 | |
| 1649 | Replaces #5509 |
| 1650 | Prepares #5707 |
| 1651 | |
| 1652 | - select.h: make socket validation macros test for INVALID_SOCKET |
| 1653 | |
| 1654 | With Winsock the valid range is [0..INVALID_SOCKET-1] according to |
| 1655 | https://docs.microsoft.com/en-us/windows/win32/winsock/socket-data-type-2 |
| 1656 | |
| 1657 | Reviewed-by: Jay Satiro |
| 1658 | Reviewed-by: Marcel Raad |
| 1659 | Reviewed-by: Daniel Stenberg |
| 1660 | |
| 1661 | Closes #5760 |
| 1662 | |
| 1663 | Daniel Stenberg (24 Aug 2020) |
| 1664 | - docs: --output-dir is added in 7.73.0, nothing else |
| 1665 | |
| 1666 | Follow-up to 5620d2cc78c0 |
| 1667 | |
| 1668 | - curl: add --output-dir |
| 1669 | |
| 1670 | Works with --create-dirs and with -J |
| 1671 | |
| 1672 | Add test 3008, 3009, 3011, 3012 and 3013 to verify. |
| 1673 | |
| 1674 | Closes #5637 |
| 1675 | |
| 1676 | - configure: fix pkg-config detecting wolfssl |
| 1677 | |
| 1678 | When amending the include path with "/wolfssl", this now properly strips |
| 1679 | off all whitespace from the path variable! Previously this would lead to |
| 1680 | pkg-config builds creating bad command lines. |
| 1681 | |
| 1682 | Closes #5848 |
| 1683 | |
| 1684 | - [Michael Musset brought this change] |
| 1685 | |
| 1686 | sftp: add the option CURLKHSTAT_FINE_REPLACE |
| 1687 | |
| 1688 | Replace the old fingerprint of the host with a new. |
| 1689 | |
| 1690 | Closes #5685 |
| 1691 | |
| 1692 | - RELEASE-NOTES: synced |
| 1693 | |
| 1694 | The next release is now to become 7.73.0 |
| 1695 | |
| 1696 | - checksrc: verify do-while and spaces between the braces |
| 1697 | |
| 1698 | Updated mprintf.c to comply |
| 1699 | |
| 1700 | Closes #5845 |
| 1701 | |
| 1702 | - curl: support XDG_CONFIG_HOME to find .curlrc |
| 1703 | |
| 1704 | Added test433 to verify. Updated documentation. |
| 1705 | |
| 1706 | Reviewed-by: Jay Satiro |
| 1707 | Suggested-by: Eli Schwartz |
| 1708 | Fixes #5829 |
| 1709 | Closes #5837 |
| 1710 | |
| 1711 | - etag: save and use the full received contents |
| 1712 | |
| 1713 | ... which makes it support weak tags and non-standard etags too! |
| 1714 | |
| 1715 | Added test case 347 to verify blank incoming ETag: |
| 1716 | |
| 1717 | Fixes #5610 |
| 1718 | Closes #5833 |
| 1719 | |
| 1720 | - setopt: if the buffer exists, refuse the new BUFFERSIZE |
| 1721 | |
| 1722 | The buffer only exists during transfer and then we shouldn't change the |
| 1723 | size (the setopt is not documented to work then). |
| 1724 | |
| 1725 | Reported-by: Harry Sintonen |
| 1726 | Closes #5842 |
| 1727 | |
| 1728 | - [COFFEETALES brought this change] |
| 1729 | |
| 1730 | sftp: add new quote commands 'atime' and 'mtime' |
| 1731 | |
| 1732 | Closes #5810 |
| 1733 | |
| 1734 | - CURLE_PROXY: new error code |
| 1735 | |
| 1736 | Failures clearly returned from a (SOCKS) proxy now causes this return |
| 1737 | code. Previously the situation was not very clear as what would be |
| 1738 | returned and when. |
| 1739 | |
| 1740 | In addition: when this error code is returned, an application can use |
| 1741 | CURLINFO_PROXY_ERROR to query libcurl for the detailed error, which then |
| 1742 | returns a value from the new 'CURLproxycode' enum. |
| 1743 | |
| 1744 | Closes #5770 |
| 1745 | |
| 1746 | - runtests: make cleardir() erase dot files too |
| 1747 | |
| 1748 | Because test cases might use dot files. |
| 1749 | |
| 1750 | Closes #5838 |
| 1751 | |
| 1752 | - KNOWN_BUGS: 'no_proxy' string-matches IPv6 numerical addreses |
| 1753 | |
| 1754 | Also: the current behavior is now documented in the curl.1 and |
| 1755 | CURLOPT_NOPROXY.3 man pages. |
| 1756 | |
| 1757 | Reported-by: Andrew Barnes |
| 1758 | Closes #5745 |
| 1759 | Closes #5841 |
| 1760 | |
| 1761 | Viktor Szakats (22 Aug 2020) |
| 1762 | - Makefile.m32: add ability to override zstd libs [ci skip] |
| 1763 | |
| 1764 | Similarly to brotli, where this was already possible. |
| 1765 | E.g. it allows to link zstd statically to libcurl.dll. |
| 1766 | |
| 1767 | Ref: https://github.com/curl/curl-for-win/issues/12 |
| 1768 | Ref: https://github.com/curl/curl-for-win/commit/d9b266afd2e5d3f5604483010ef62340b5918c89 |
| 1769 | |
| 1770 | Closes https://github.com/curl/curl/pull/5840 |
| 1771 | |
| 1772 | Daniel Stenberg (21 Aug 2020) |
| 1773 | - runtests: avoid 'fail to start' repeated messages in attempt loops |
| 1774 | |
| 1775 | Closes #5834 |
| 1776 | |
| 1777 | - runtests: clear pid variables when failing to start a server |
| 1778 | |
| 1779 | ... as otherwise the parent doesn't detect the failure and believe it |
| 1780 | actually worked to start. |
| 1781 | |
| 1782 | Reported-by: Christian Weisgerber |
| 1783 | Bug: https://curl.haxx.se/mail/lib-2020-08/0018.html |
| 1784 | Closes #5834 |
| 1785 | |
| 1786 | - TODO: Virtual external sockets |
| 1787 | |
| 1788 | Closes #5835 |
| 1789 | |
| 1790 | - [Don J Olmstead brought this change] |
| 1791 | |
| 1792 | dist: add missing CMake Find modules to the distribution |
| 1793 | |
| 1794 | Closes #5836 |
| 1795 | |
| 1796 | - RELEASE-NOTES: synced |
| 1797 | |
| 1798 | ... and version bumped to 7.72.1 |
| 1799 | |
| 1800 | - tls: provide the CApath verbose log on its own line |
| 1801 | |
| 1802 | ... not newline separated from the previous line. This makes it output |
| 1803 | asterisk prefixed properly like other verbose putput! |
| 1804 | |
| 1805 | Reported-by: jmdavitt on github |
| 1806 | Fixes #5826 |
| 1807 | Closes #5827 |
| 1808 | |
Haibo Huang | c3c04f4 | 2020-08-19 13:00:07 -0700 | [diff] [blame] | 1809 | Version 7.72.0 (19 Aug 2020) |
| 1810 | |
| 1811 | Daniel Stenberg (19 Aug 2020) |
| 1812 | - RELEASE-NOTES: synced |
| 1813 | |
| 1814 | The curl 7.72.0 release |
| 1815 | |
| 1816 | - THANKS: add names from curl 7.72.0 release |
| 1817 | |
| 1818 | Jay Satiro (18 Aug 2020) |
| 1819 | - KNOWN_BUGS: Schannel TLS 1.2 handshake bug in old Windows versions |
| 1820 | |
| 1821 | Reported-by: plujon@users.noreply.github.com |
| 1822 | |
| 1823 | Closes https://github.com/curl/curl/issues/5488 |
| 1824 | |
| 1825 | Daniel Stenberg (17 Aug 2020) |
| 1826 | - Curl_easy: remember last connection by id, not by pointer |
| 1827 | |
| 1828 | CVE-2020-8231 |
| 1829 | |
| 1830 | Bug: https://curl.haxx.se/docs/CVE-2020-8231.html |
| 1831 | |
| 1832 | Reported-by: Marc Aldorasi |
| 1833 | Closes #5824 |
| 1834 | |
| 1835 | - examples/rtsp.c: correct the copyright year |
| 1836 | |
| 1837 | - RELEASE-PROCEDURE.md: add more future release dates |
| 1838 | |
| 1839 | - [H3RSKO brought this change] |
| 1840 | |
| 1841 | docs: change "web site" to "website" |
| 1842 | |
| 1843 | According to wikipedia: |
| 1844 | |
| 1845 | While "web site" was the original spelling, this variant has become |
| 1846 | rarely used, and "website" has become the standard spelling |
| 1847 | |
| 1848 | Closes #5822 |
| 1849 | |
| 1850 | - [Bevan Weiss brought this change] |
| 1851 | |
| 1852 | CMake: don't complain about missing nroff |
| 1853 | |
| 1854 | The curl_nroff_check() was always being called, and complaining if |
| 1855 | *NROFF wasn't found, even when not making the manual. |
| 1856 | |
| 1857 | Only check for nroff (and complain) if actually making the manual |
| 1858 | |
| 1859 | Closes #5817 |
| 1860 | |
| 1861 | - [Brian Inglis brought this change] |
| 1862 | |
| 1863 | libtest/Makefile.am: add -no-undefined for libstubgss for Cygwin |
| 1864 | |
| 1865 | copy the LDFLAGS approach for adding same option with `libhostname` in |
| 1866 | `libtest/Makefile.am`: |
| 1867 | |
| 1868 | - init `libstubgss_la_LDFLAGS_EXTRA` variable, |
| 1869 | - add option to variable inside conditional, |
| 1870 | - use variable in `libstubgss_la_LDFLAGS` |
| 1871 | |
| 1872 | Fixes #5819 |
| 1873 | Closes #5820 |
| 1874 | |
| 1875 | - docs: clarify MAX_SEND/RECV_SPEED functionality |
| 1876 | |
| 1877 | ... in particular what happens if the maximum speed limit is set to a |
| 1878 | value that's smaller than the transfer buffer size in use. |
| 1879 | |
| 1880 | Reported-by: Tomas Berger |
| 1881 | Fixes #5788 |
| 1882 | Closes #5813 |
| 1883 | |
| 1884 | - test1140: compare stdout |
| 1885 | |
| 1886 | To make problems more immediately obvious when tests fail. |
| 1887 | |
| 1888 | Closes #5814 |
| 1889 | |
| 1890 | - asyn-ares: correct some bad comments |
| 1891 | |
| 1892 | Closes #5812 |
| 1893 | |
| 1894 | - [Emil Engler brought this change] |
| 1895 | |
| 1896 | docs: Add video link to docs/CONTRIBUTE.md |
| 1897 | |
| 1898 | Closes #5811 |
| 1899 | |
| 1900 | - curl-config: ignore REQUIRE_LIB_DEPS in --libs output |
| 1901 | |
| 1902 | Fixes a curl-config issue on cygwin by making sure REQUIRE_LIB_DEPS is |
| 1903 | not considered for the --libs output. |
| 1904 | |
| 1905 | Reported-by: ramsay-jones on github |
| 1906 | Assisted-by: Brian Inglis and Ken Brown |
| 1907 | Fixes #5793 |
| 1908 | Closes #5808 |
| 1909 | |
| 1910 | - copyright: update/correct the year range on a few files |
| 1911 | |
| 1912 | - scripts/copyright.pl: ignore .muse files |
| 1913 | |
| 1914 | - [Emil Engler brought this change] |
| 1915 | |
| 1916 | multi: Remove 10-year old out-commented code |
| 1917 | |
| 1918 | The code hasn't been touched since 2010-08-18 |
| 1919 | |
| 1920 | Closes #5805 |
| 1921 | |
| 1922 | - KNOWN_BUGS: A shared connection cache is not thread-safe |
| 1923 | |
| 1924 | Closes #4915 |
| 1925 | Closes #5802 |
| 1926 | |
| 1927 | - CONTRIBUTE: extend git commit message description |
| 1928 | |
| 1929 | In particular how the first line works. |
| 1930 | |
| 1931 | Closes #5803 |
| 1932 | |
| 1933 | - RELEASE-NOTES: synced |
| 1934 | |
| 1935 | - [Stefan Yohansson brought this change] |
| 1936 | |
| 1937 | transfer: move retrycount from connect struct to easy handle |
| 1938 | |
| 1939 | This flag was applied to the connection struct that is released on |
| 1940 | retry. These changes move the retry counter into Curl_easy struct that |
| 1941 | lives across retries and retains the new connection. |
| 1942 | |
| 1943 | Reported-by: Cherish98 on github |
| 1944 | Fixes #5794 |
| 1945 | Closes #5800 |
| 1946 | |
| 1947 | - libssh2: s/ssherr/sftperr/ |
| 1948 | |
| 1949 | The debug output used ssherr instead of sftperr which not only outputs |
| 1950 | the wrong error code but also casues a warning on Windows. |
| 1951 | |
| 1952 | Follow-up to 7370b4e39f1 |
| 1953 | |
| 1954 | Reported-by: Gisle Vanem |
| 1955 | Bug: https://github.com/curl/curl/commit/7370b4e39f1390e701f5b68d910c619151daf72b#r41334700 |
| 1956 | Closes #5799 |
| 1957 | |
| 1958 | - ftp: don't do ssl_shutdown instead of ssl_close |
| 1959 | |
| 1960 | The shutdown function is for downgrading a connection from TLS to plain, |
| 1961 | and this is not requested here. |
| 1962 | |
| 1963 | Have ssl_close reset the TLS connection state. |
| 1964 | |
| 1965 | This partially reverts commit f002c850d98d |
| 1966 | |
| 1967 | Reported-by: Rasmus Melchior Jacobsen |
| 1968 | Reported-by: Denis Goleshchikhin |
| 1969 | Fixes #5797 |
| 1970 | |
| 1971 | Marc Hoersken (9 Aug 2020) |
| 1972 | - CI/azure: fix test outcome values and use latest API version |
| 1973 | |
| 1974 | This makes sure that tests ignored or skipped are not shown |
| 1975 | just in the category "Other", but with their correct state. |
| 1976 | |
| 1977 | Closes #5796 |
| 1978 | |
| 1979 | - CI/azure: show runtime stats to investigate slowness |
| 1980 | |
| 1981 | Also avoid naming conflict of TFLAGS env and tflags variables. |
| 1982 | |
| 1983 | Closes #5776 |
| 1984 | |
| 1985 | Daniel Stenberg (8 Aug 2020) |
| 1986 | - TLS naming: fix more Winssl and Darwinssl leftovers |
| 1987 | |
| 1988 | The CMake option is now called CMAKE_USE_SCHANNEL |
| 1989 | |
| 1990 | The winbuild flag is USE_SCHANNEL |
| 1991 | |
| 1992 | The CI jobs and build scripts only use the new names and the new name |
| 1993 | options |
| 1994 | |
| 1995 | Tests now require 'Schannel' (when necessary) |
| 1996 | |
| 1997 | Closes #5795 |
| 1998 | |
| 1999 | - smtp_parse_address: handle blank input string properly |
| 2000 | |
| 2001 | Closes #5792 |
| 2002 | |
| 2003 | - runtests: run the DICT server on a random port number |
| 2004 | |
| 2005 | Removed support for -b (base port number) |
| 2006 | |
| 2007 | Closes #5783 |
| 2008 | |
| 2009 | - RELEASE-NOTES: synced |
| 2010 | |
| 2011 | - runtests: move the TELNET server to a dynamic port |
| 2012 | |
| 2013 | Rename the port variable to TELNETPORT to better match the existing |
| 2014 | pattern. |
| 2015 | |
| 2016 | Closes #5785 |
| 2017 | |
| 2018 | - ngtcp2: adapt to error code rename |
| 2019 | |
| 2020 | Closes #5786 |
| 2021 | |
| 2022 | - runtests: move the smbserver to use a dynamic port number |
| 2023 | |
| 2024 | Closes #5782 |
| 2025 | |
| 2026 | - runtests: run the http2 tests on a random port number |
| 2027 | |
| 2028 | Closes #5779 |
| 2029 | |
| 2030 | - gtls: survive not being able to get name/issuer |
| 2031 | |
| 2032 | Closes #5778 |
| 2033 | |
| 2034 | - runtests: move the gnutls-serv tests to a dynamic port |
| 2035 | |
| 2036 | Affects test 320, 321, 322 and 324. |
| 2037 | |
| 2038 | Closes #5778 |
| 2039 | |
| 2040 | - runtests: support dynamicly base64 encoded sections in tests |
| 2041 | |
| 2042 | This allows us to make test cases to use base64 at run-time and still |
| 2043 | use and verify information determined at run-time, such as the IMAP test |
| 2044 | server's port number in test 842. |
| 2045 | |
| 2046 | This change makes 12 tests run again that basically never ran since we |
| 2047 | moved to dynamic port numbers. |
| 2048 | |
| 2049 | ftpserver.pl is adjusted to load test instructions and test number from |
| 2050 | the preprocessed test file. |
| 2051 | |
| 2052 | FILEFORMAT.md now documents the new base64 encoding syntax. |
| 2053 | |
| 2054 | Reported-by: Marcel Raad |
| 2055 | Fixes #5761 |
| 2056 | Closes #5775 |
| 2057 | |
| 2058 | - curl.1: add a few missing valid exit codes |
| 2059 | |
| 2060 | 93 - 96 can be returned as well. |
| 2061 | |
| 2062 | Closes #5777 |
| 2063 | |
| 2064 | - TODO: Use multiple parallel transfers for a single download |
| 2065 | |
| 2066 | Closes #5774 |
| 2067 | |
| 2068 | - TODO: Set the modification date on an uploaded file |
| 2069 | |
| 2070 | Closes #5768 |
| 2071 | |
| 2072 | - [Thomas M. DuBuisson brought this change] |
| 2073 | |
| 2074 | CI: Add muse CI config |
| 2075 | |
| 2076 | Closes #5772 |
| 2077 | |
| 2078 | - [Thomas M. DuBuisson brought this change] |
| 2079 | |
| 2080 | travis/script.sh: fix use of `-n' with unquoted envvar |
| 2081 | |
| 2082 | Shellcheck tells us "-n doesn't work with unquoted arguments. quote or |
| 2083 | use [[ ]]." |
| 2084 | |
| 2085 | And testing shows: |
| 2086 | |
| 2087 | ``` |
| 2088 | docker run --rm -it ubuntu bash |
| 2089 | root@fe85ce156856:/# [ -n $DOES_NOT_EXIST ] && echo "I ran" |
| 2090 | I ran |
| 2091 | root@fe85ce156856:/# [ -n "$DOES_NOT_EXIST" ] && echo "I ran" |
| 2092 | root@fe85ce156856:/# |
| 2093 | ``` |
| 2094 | |
| 2095 | Closes #5773 |
| 2096 | |
| 2097 | - h2: repair trailer handling |
| 2098 | |
| 2099 | The previous h2 trailer fix in 54a2b63 was wrong and caused a |
| 2100 | regression: it cannot deal with trailers immediately when read since |
| 2101 | they may be read off the connection by the wrong 'data' owner. |
| 2102 | |
| 2103 | This change reverts the logic back to gathering all trailers into a |
| 2104 | single buffer, like before 54a2b63. |
| 2105 | |
| 2106 | Reported-by: Tadej Vengust |
| 2107 | Fixes #5663 |
| 2108 | Closes #5769 |
| 2109 | |
| 2110 | Viktor Szakats (3 Aug 2020) |
| 2111 | - windows: disable Unix Sockets for old mingw |
| 2112 | |
| 2113 | Classic mingw and 10y+ old versions of mingw-w64 don't ship with |
| 2114 | Windows headers having the typedef necessary for Unix Sockets |
| 2115 | support, so try detecting these environments to disable this |
| 2116 | feature. |
| 2117 | |
| 2118 | Ref: https://sourceforge.net/p/mingw-w64/mingw-w64/ci/cf6afc57179a5910621215f8f4037d406892072c/ |
| 2119 | |
| 2120 | Reviewed-by: Daniel Stenberg |
| 2121 | |
| 2122 | Fixes #5674 |
| 2123 | Closes #5758 |
| 2124 | |
| 2125 | Marcel Raad (3 Aug 2020) |
| 2126 | - test1908: treat file as text |
| 2127 | |
| 2128 | Fixes the line endings on Windows. |
| 2129 | |
| 2130 | Closes https://github.com/curl/curl/pull/5767 |
| 2131 | |
| 2132 | - TrackMemory tests: ignore realloc and free in getenv.c |
| 2133 | |
| 2134 | These are only called for WIN32. |
| 2135 | |
| 2136 | Closes https://github.com/curl/curl/pull/5767 |
| 2137 | |
| 2138 | Daniel Stenberg (3 Aug 2020) |
| 2139 | - tests/FILEFORMAT.md: mention %HTTP2PORT |
| 2140 | |
| 2141 | - RELEASE-NOTES: synced |
| 2142 | |
| 2143 | - tlsv1.3.d. only for TLS-using connections |
| 2144 | |
| 2145 | ... and rephrase that "not all" TLS backends support it. |
| 2146 | |
| 2147 | Closes #5764 |
| 2148 | |
| 2149 | - tls-max.d: this option is only for TLS-using connections |
| 2150 | |
| 2151 | Ref: #5763 |
| 2152 | Closes #5764 |
| 2153 | |
| 2154 | Marcel Raad (2 Aug 2020) |
| 2155 | - [Cameron Cawley brought this change] |
| 2156 | |
| 2157 | tool_doswin: Simplify Windows version detection |
| 2158 | |
| 2159 | Closes https://github.com/curl/curl/pull/5754 |
| 2160 | |
| 2161 | - [Cameron Cawley brought this change] |
| 2162 | |
| 2163 | win32: Add Curl_verify_windows_version() to curlx |
| 2164 | |
| 2165 | Closes https://github.com/curl/curl/pull/5754 |
| 2166 | |
| 2167 | - runtests.pl: treat LibreSSL and BoringSSL as OpenSSL |
| 2168 | |
| 2169 | This makes the tests that require the OpenSSL feature also run for |
| 2170 | those two compatible libraries. |
| 2171 | |
| 2172 | Closes https://github.com/curl/curl/pull/5762 |
| 2173 | |
| 2174 | Daniel Stenberg (1 Aug 2020) |
| 2175 | - multi: Condition 'extrawait' is always true |
| 2176 | |
| 2177 | Reported by Codacy. |
| 2178 | |
| 2179 | Reviewed-by: Marcel Raad |
| 2180 | Closes #5759 |
| 2181 | |
| 2182 | Marcel Raad (1 Aug 2020) |
| 2183 | - openssl: fix build with LibreSSL < 2.9.1 |
| 2184 | |
| 2185 | `SSL_CTX_add0_chain_cert` and `SSL_CTX_clear_chain_certs` were |
| 2186 | introduced in LibreSSL 2.9.1 [0]. |
| 2187 | |
| 2188 | [0] https://github.com/libressl-portable/openbsd/commit/0db809ee178457c8170abfae3931d7bd13abf3ef |
| 2189 | |
| 2190 | Closes https://github.com/curl/curl/pull/5757 |
| 2191 | |
| 2192 | Daniel Stenberg (1 Aug 2020) |
| 2193 | - [Marc Aldorasi brought this change] |
| 2194 | |
| 2195 | multi_remove_handle: close unused connect-only connections |
| 2196 | |
| 2197 | Previously any connect-only connections in a multi handle would be kept |
| 2198 | alive until the multi handle was closed. Since these connections cannot |
| 2199 | be re-used, they can be marked for closure when the associated easy |
| 2200 | handle is removed from the multi handle. |
| 2201 | |
| 2202 | Closes #5749 |
| 2203 | |
| 2204 | - checksrc: invoke script with -D to find .checksrc proper |
| 2205 | |
| 2206 | Without the -D command line option, checksrc.pl won't know which |
| 2207 | directory to load the ".checksrc" file from when building out of the |
| 2208 | source tree. |
| 2209 | |
| 2210 | Reported-by: Marcel Raad |
| 2211 | Fixes #5715 |
| 2212 | Closes #5755 |
| 2213 | |
| 2214 | - [Carlo Marcelo Arenas Belón brought this change] |
| 2215 | |
| 2216 | buildconf: retire ares buildconf invocation |
| 2217 | |
| 2218 | no longer needed after 4259d2df7dd95637a4b1e3fb174fe5e5aef81069 |
| 2219 | |
| 2220 | - [Carlo Marcelo Arenas Belón brought this change] |
| 2221 | |
| 2222 | buildconf: excempt defunct reference to ACLOCAL_FLAGS |
| 2223 | |
| 2224 | retired with 09f278121e815028adb24d228d8092fc6cb022aa but kept around as |
| 2225 | the name is generic enough that it might be in use and relied upon from |
| 2226 | the environment. |
| 2227 | |
| 2228 | - [Carlo Marcelo Arenas Belón brought this change] |
| 2229 | |
| 2230 | buildconf: avoid array concatenation in die() |
| 2231 | |
| 2232 | reported as error SC2145[1] by shellcheck, but not expected to cause |
| 2233 | any behavioural differences otherwise. |
| 2234 | |
| 2235 | [1] https://github.com/koalaman/shellcheck/wiki/SC2145 |
| 2236 | |
| 2237 | Closes #5701 |
| 2238 | |
| 2239 | - travis: add ppc64le and s390x builds |
| 2240 | |
| 2241 | Closes #5752 |
| 2242 | |
| 2243 | Marc Hoersken (31 Jul 2020) |
| 2244 | - connect: remove redundant message about connect failure |
| 2245 | |
| 2246 | Reviewed-by: Daniel Stenberg |
| 2247 | |
| 2248 | Closes #5708 |
| 2249 | |
| 2250 | - tests/sshserver.pl: fix compatibility with OpenSSH for Windows |
| 2251 | |
| 2252 | Follow up to #5721 |
| 2253 | |
| 2254 | - CI/azure: install libssh2 for use with msys2-based builds |
| 2255 | |
| 2256 | This enables building and running the SFTP tests. |
| 2257 | Unfortunately OpenSSH for Windows does not support SCP (yet). |
| 2258 | |
| 2259 | Reviewed-by: Daniel Stenberg |
| 2260 | |
| 2261 | Closes #5721 |
| 2262 | |
| 2263 | - CI/azure: increase Windows job timeout once again |
| 2264 | |
| 2265 | Avoid aborted jobs due to performance issues on Azure DevOps. |
| 2266 | |
| 2267 | Reviewed-by: Daniel Stenberg |
| 2268 | Reviewed-by: Jay Satiro |
| 2269 | |
| 2270 | Closes #5738 |
| 2271 | |
| 2272 | Jay Satiro (30 Jul 2020) |
| 2273 | - TODO: Schannel: 'Add option to allow abrupt server closure' |
| 2274 | |
| 2275 | We should offer an option to allow abrupt server closures (server closes |
| 2276 | SSL transfer without sending a known termination point such as length of |
| 2277 | transfer or close_notify alert). Abrupt server closures are usually |
| 2278 | because of misconfigured or very old servers. |
| 2279 | |
| 2280 | Closes https://github.com/curl/curl/issues/4427 |
| 2281 | |
| 2282 | - url: fix CURLU and location following |
| 2283 | |
| 2284 | Prior to this change if the user set a URL handle (CURLOPT_CURLU) it was |
| 2285 | incorrectly used for the location follow, resulting in infinite requests |
| 2286 | to the original location. |
| 2287 | |
| 2288 | Reported-by: sspiri@users.noreply.github.com |
| 2289 | |
| 2290 | Fixes https://github.com/curl/curl/issues/5709 |
| 2291 | Closes https://github.com/curl/curl/pull/5713 |
| 2292 | |
| 2293 | Daniel Stenberg (30 Jul 2020) |
| 2294 | - RELEASE-NOTES: synced |
| 2295 | |
| 2296 | - [divinity76 brought this change] |
| 2297 | |
| 2298 | docs: add date of 7.20 to CURLM_CALL_MULTI_PERFORM mentions |
| 2299 | |
| 2300 | it helps make it obvious that most developers don't have to care about |
| 2301 | the CURLM_CALL_MULTI_PERFORM value (last release using it is nearly 11 |
| 2302 | years old, November 4 2009) |
| 2303 | |
| 2304 | Closes #5744 |
| 2305 | |
| 2306 | Jay Satiro (29 Jul 2020) |
| 2307 | - tool_cb_wrt: fix outfile mode flags for Windows |
| 2308 | |
| 2309 | - Use S_IREAD and S_IWRITE mode permission flags to create the file |
| 2310 | on Windows instead of S_IRUSR, S_IWUSR, etc. |
| 2311 | |
| 2312 | Windows only accepts a combination of S_IREAD and S_IWRITE. It does not |
| 2313 | acknowledge other combinations, for which it may generate an assertion. |
| 2314 | |
| 2315 | This is a follow-up to 81b4e99 from yesterday, which improved the |
| 2316 | existing file check with -J. |
| 2317 | |
| 2318 | Ref: https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/open-wopen#remarks |
| 2319 | Ref: https://github.com/curl/curl/pull/5731 |
| 2320 | |
| 2321 | Closes https://github.com/curl/curl/pull/5742 |
| 2322 | |
| 2323 | Daniel Stenberg (28 Jul 2020) |
| 2324 | - checksrc: ban gmtime/localtime |
| 2325 | |
| 2326 | They're not thread-safe so they should not be used in libcurl code. |
| 2327 | |
| 2328 | Explictly enabled when deemed necessary and in examples and tests |
| 2329 | |
| 2330 | Reviewed-by: Nicolas Sterchele |
| 2331 | Closes #5732 |
| 2332 | |
| 2333 | - transfer: fix data_pending for builds with both h2 and h3 enabled |
| 2334 | |
| 2335 | Closes #5734 |
| 2336 | |
| 2337 | - curl_multi_setopt: fix compiler warning "result is always false" |
| 2338 | |
| 2339 | On systems with 32 bit long the expression is always false. Avoid |
| 2340 | the warning. |
| 2341 | |
| 2342 | Reported-by: Gisle Vanem |
| 2343 | Bug: https://github.com/curl/curl/commit/61a08508f6a458fe21bbb18cd2a9bac2f039452b#commitcomment-40941232 |
| 2344 | Closes #5736 |
| 2345 | |
| 2346 | - curl: improve the existing file check with -J |
| 2347 | |
| 2348 | Previously a file that isn't user-readable but is user-writable would |
| 2349 | not be properly avoided and would get overwritten. |
| 2350 | |
| 2351 | Reported-by: BrumBrum on hackerone |
| 2352 | Assisted-by: Jay Satiro |
| 2353 | Bug: https://hackerone.com/reports/926638 |
| 2354 | Closes #5731 |
| 2355 | |
| 2356 | - [Jonathan Nieder brought this change] |
| 2357 | |
| 2358 | multi: update comment to say easyp list is linear |
| 2359 | |
| 2360 | Since 09b9fc900 (multi: remove 'Curl_one_easy' struct, phase 1, |
| 2361 | 2013-08-02), the easy handle list is not circular but ends with |
| 2362 | ->next pointing to NULL. |
| 2363 | |
| 2364 | Reported-by: Masaya Suzuki <masayasuzuki@google.com> |
| 2365 | Closes #5737 |
| 2366 | |
| 2367 | - CURLOPT_NOBODY.3: fix the syntax for referring to options |
| 2368 | |
| 2369 | As test 1140 fails otherwise! |
| 2370 | |
| 2371 | Follow-up to e1bac81cc815 |
| 2372 | |
| 2373 | - ngtcp2: store address in sockaddr_storage |
| 2374 | |
| 2375 | Reported-by: Tatsuhiro Tsujikawa |
| 2376 | Closes #5733 |
| 2377 | |
| 2378 | - CURLOPT_NOBODY.3: clarify what setting to 0 means |
| 2379 | |
| 2380 | ... and mention that HTTP with other methods than HEAD might get a body and |
| 2381 | there's no option available to stop that. |
| 2382 | |
| 2383 | Closes #5729 |
| 2384 | |
| 2385 | - setopt: unset NOBODY switches to GET if still HEAD |
| 2386 | |
| 2387 | Unsetting CURLOPT_NOBODY with 0L when doing HTTP has no documented |
| 2388 | action but before 7.71.0 that used to switch back to GET and with this |
| 2389 | change (assuming the method is still set to HEAD) this behavior is |
| 2390 | brought back. |
| 2391 | |
| 2392 | Reported-by: causal-agent on github |
| 2393 | Fixes #5725 |
| 2394 | Closes #5728 |
| 2395 | |
| 2396 | - [Ehren Bendler brought this change] |
| 2397 | |
| 2398 | configure: cleanup wolfssl + pkg-config conflicts when cross compiling. |
| 2399 | |
| 2400 | Also choose a different wolfSSL function to test for NTLM support. |
| 2401 | |
| 2402 | Fixes #5605 |
| 2403 | Closes #5682 |
| 2404 | |
| 2405 | - configure: show zstd "no" in summary when built without it |
| 2406 | |
| 2407 | Reported-by: Marc Hörsken |
| 2408 | Fixes #5720 |
| 2409 | Closes #5730 |
| 2410 | |
| 2411 | - quiche: handle calling disconnect twice |
| 2412 | |
| 2413 | Reported-by: lilongyan-huawei on github |
| 2414 | Fixes #5726 |
| 2415 | Closes #5727 |
| 2416 | |
| 2417 | - [Nicolas Sterchele brought this change] |
| 2418 | |
| 2419 | getinfo: reset retry-after value in initinfo |
| 2420 | |
| 2421 | - Avoid re-using retry_after value from preceding request |
| 2422 | - Add libtest 3010 to verify |
| 2423 | |
| 2424 | Reported-by: joey-l-us on github |
| 2425 | Fixes #5661 |
| 2426 | Closes #5672 |
| 2427 | |
| 2428 | Marcel Raad (27 Jul 2020) |
| 2429 | - WIN32: stop forcing narrow-character API |
| 2430 | |
| 2431 | Except where the results are only used for character output. |
| 2432 | getenv is not touched because it's part of the public API, and having |
| 2433 | it return UTF-8 instead of ANSI would be a breaking change. |
| 2434 | |
| 2435 | Fixes https://github.com/curl/curl/issues/5658 |
| 2436 | Fixes https://github.com/curl/curl/issues/5712 |
| 2437 | Closes https://github.com/curl/curl/pull/5718 |
| 2438 | |
| 2439 | Jay Satiro (27 Jul 2020) |
| 2440 | - [Tobias Stoeckmann brought this change] |
| 2441 | |
| 2442 | mprintf: Fix stack overflows |
| 2443 | |
| 2444 | Stack overflows can occur with precisions for integers and floats. |
| 2445 | |
| 2446 | Proof of concepts: |
| 2447 | - curl_mprintf("%d, %.*1$d", 500, 1); |
| 2448 | - curl_mprintf("%d, %+0500.*1$f", 500, 1); |
| 2449 | |
| 2450 | Ideally, compile with -fsanitize=address which makes this undefined |
| 2451 | behavior a bit more defined for debug purposes. |
| 2452 | |
| 2453 | The format strings are valid. The overflows occur due to invalid |
| 2454 | arguments. If these arguments are variables with contents controlled |
| 2455 | by an attacker, the function's stack can be corrupted. |
| 2456 | |
| 2457 | Also see CVE-2016-9586 which partially fixed the float aspect. |
| 2458 | |
| 2459 | Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org> |
| 2460 | |
| 2461 | Closes https://github.com/curl/curl/pull/5722 |
| 2462 | |
| 2463 | - [Tobias Stoeckmann brought this change] |
| 2464 | |
| 2465 | mprintf: Fix dollar string handling |
| 2466 | |
| 2467 | Verify that specified parameters are in range. If parameters are too |
| 2468 | large, fail early on and avoid out of boundary accesses. |
| 2469 | |
| 2470 | Also do not read behind boundaries of illegal format strings. |
| 2471 | |
| 2472 | These are defensive measures since it is expected that format strings |
| 2473 | are well-formed. Format strings should not be modifiable by user |
| 2474 | input due to possible generic format string attacks. |
| 2475 | |
| 2476 | Closes https://github.com/curl/curl/pull/5722 |
| 2477 | |
| 2478 | Daniel Stenberg (26 Jul 2020) |
| 2479 | - ntlm: free target_info before (re-)malloc |
| 2480 | |
| 2481 | OSS-Fuzz found a way this could get called again with the pointer still |
| 2482 | pointing to a malloc'ed memory, leading to a leak. |
| 2483 | |
| 2484 | Bug: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=24379 |
| 2485 | |
| 2486 | Closes #5724 |
| 2487 | |
| 2488 | Marcel Raad (26 Jul 2020) |
| 2489 | - CI/macos: set minimum macOS version |
| 2490 | |
| 2491 | This enables some deprecation warnings. |
| 2492 | Previously, autotools defaulted to 10.8. |
| 2493 | |
| 2494 | Closes https://github.com/curl/curl/pull/5723 |
| 2495 | |
| 2496 | Daniel Stenberg (26 Jul 2020) |
| 2497 | - RELEASE-NOTES: synced |
| 2498 | |
| 2499 | Marcel Raad (25 Jul 2020) |
| 2500 | - CI/macos: enable warnings as errors for CMake builds |
| 2501 | |
| 2502 | Closes https://github.com/curl/curl/pull/5716 |
| 2503 | |
| 2504 | - CMake: fix test for warning suppressions |
| 2505 | |
| 2506 | GCC doesn't warn for unknown `-Wno-` options, except if there are other |
| 2507 | warnings or errors [0]. This was problematic with `CURL_WERROR` as that |
| 2508 | warning-as-error cannot be suppressed. Notably, this always happened |
| 2509 | with `-Wno-pedantic-ms-format` when not targeting Windows. So test for |
| 2510 | the positive form of the warning instead, which should always result in |
| 2511 | a diagnostic if unknown. |
| 2512 | |
| 2513 | [0] https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html |
| 2514 | |
| 2515 | Closes https://github.com/curl/curl/pull/5714 |
| 2516 | |
| 2517 | Jay Satiro (23 Jul 2020) |
| 2518 | - curl.h: update CURLINFO_LASTONE |
| 2519 | |
| 2520 | CURLINFO_LASTONE should have been updated when |
| 2521 | CURLINFO_EFFECTIVE_METHOD was added. |
| 2522 | |
| 2523 | Reported-by: xwxbug@users.noreply.github.com |
| 2524 | |
| 2525 | Fixes https://github.com/curl/curl/issues/5711 |
| 2526 | |
| 2527 | Marc Hoersken (22 Jul 2020) |
| 2528 | - CI/azure: unconditionally enable warnings-as-errors with autotools |
| 2529 | |
| 2530 | Reviewed-by: Marcel Raad |
| 2531 | |
| 2532 | Follow up to #5694 |
| 2533 | Closes #5706 |
| 2534 | |
| 2535 | Marcel Raad (21 Jul 2020) |
| 2536 | - doh: remove redundant cast |
| 2537 | |
| 2538 | Closes https://github.com/curl/curl/pull/5704 |
| 2539 | |
| 2540 | - CI/macos: unconditionally enable warnings-as-errors with autotools |
| 2541 | |
| 2542 | Previously, warnings were only visible in the output for most jobs. |
| 2543 | |
| 2544 | Closes https://github.com/curl/curl/pull/5694 |
| 2545 | |
| 2546 | - util: silence conversion warnings |
| 2547 | |
| 2548 | timeval::tv_usec might be a 32-bit integer and timespec::tv_nsec might |
| 2549 | be a 64-bit integer. This is the case when building for recent macOS |
| 2550 | versions, for example. Just treat tv_usec as an int, which should |
| 2551 | hopefully always be sufficient on systems with |
| 2552 | `HAVE_CLOCK_GETTIME_MONOTONIC`. |
| 2553 | |
| 2554 | Closes https://github.com/curl/curl/pull/5695 |
| 2555 | |
| 2556 | - md(4|5): don't use deprecated macOS functions |
| 2557 | |
| 2558 | They are marked as deprecated for -mmacosx-version-min >= 10.15, |
| 2559 | which might result in warnings-as-errors. |
| 2560 | |
| 2561 | Closes https://github.com/curl/curl/pull/5695 |
| 2562 | |
| 2563 | Daniel Stenberg (18 Jul 2020) |
| 2564 | - strdup: remove the odd strlen check |
| 2565 | |
| 2566 | It confuses code analyzers with its use of -1 for unsigned value. Also, |
| 2567 | a check that's not normally used in strdup() code - and not necessary. |
| 2568 | |
| 2569 | Closes #5697 |
| 2570 | |
| 2571 | - [Alessandro Ghedini brought this change] |
| 2572 | |
| 2573 | travis: update quiche builds for new boringssl layout |
| 2574 | |
| 2575 | This is required after https://github.com/cloudflare/quiche/pull/593 |
| 2576 | moved BoringSSL around slightly. |
| 2577 | |
| 2578 | This also means that Go is not needed to build BoringSSL anymore (the |
| 2579 | one provided by quiche anyway). |
| 2580 | |
| 2581 | Closes #5691 |
| 2582 | |
| 2583 | Marcel Raad (17 Jul 2020) |
| 2584 | - configure: allow disabling warnings |
| 2585 | |
| 2586 | When using `--enable-warnings`, it was not possible to disable warnings |
| 2587 | via CFLAGS that got explicitly enabled. Now warnings are not enabled |
| 2588 | anymore if they are explicitly disabled (or enabled) in CFLAGS. This |
| 2589 | works for at least GCC, clang, and TCC as they have corresponding |
| 2590 | `-Wno-` options for every warning. |
| 2591 | |
| 2592 | Closes https://github.com/curl/curl/pull/5689 |
| 2593 | |
| 2594 | Daniel Stenberg (16 Jul 2020) |
| 2595 | - ngtcp2: adjust to recent sockaddr updates |
| 2596 | |
| 2597 | Closes #5690 |
| 2598 | |
| 2599 | - page-header: provide protocol details in the curl.1 man page |
| 2600 | |
| 2601 | Add protocol and version specific information about all protocols curl |
| 2602 | supports. |
| 2603 | |
| 2604 | Fixes #5679 |
| 2605 | Reported-by: tbugfinder on github |
| 2606 | Closes #5686 |
| 2607 | |
| 2608 | Daniel Gustafsson (16 Jul 2020) |
| 2609 | - docs: Update a few leftover mentions of DarwinSSL |
| 2610 | |
| 2611 | Commit 76a9c3c4be10b3d4d379d5b23ca76806bbae536a renamed DarwinSSL to the |
| 2612 | more correct/common name Secure Transport, but a few mentions in the docs |
| 2613 | remained. |
| 2614 | |
| 2615 | Closes #5688 |
| 2616 | Reviewed-by: Daniel Stenberg <daniel@haxx.se> |
| 2617 | |
| 2618 | Daniel Stenberg (16 Jul 2020) |
| 2619 | - file2memory: use a define instead of -1 unsigned value |
| 2620 | |
| 2621 | ... to use the maximum value for 'size_t' when detecting integer overflow. |
| 2622 | Changed the limit to max/4 as already that seems unreasonably large. |
| 2623 | |
| 2624 | Codacy didn't like the previous approach. |
| 2625 | |
| 2626 | Closes #5683 |
| 2627 | |
| 2628 | - CURL_PUSH_ERROROUT: allow the push callback to fail the parent stream |
| 2629 | |
| 2630 | ... by adding support for a new dedicated return code. |
| 2631 | |
| 2632 | Suggested-by: Jonathan Cardoso |
| 2633 | Assisted-by: Erik Johansson |
| 2634 | URL: https://curl.haxx.se/mail/lib-2020-06/0099.html |
| 2635 | Closes #5636 |
| 2636 | |
| 2637 | - [Baruch Siach brought this change] |
| 2638 | |
| 2639 | nss: fix build with disabled proxy support |
| 2640 | |
| 2641 | Avoid reference to fields that do not exist when CURL_DISABLE_PROXY is |
| 2642 | defined. |
| 2643 | |
| 2644 | Closes #5667 |
| 2645 | |
| 2646 | - test1139: make it display the difference on test failures |
| 2647 | |
| 2648 | - test1119: verify stdout in the test |
| 2649 | |
| 2650 | So that failures will be displayed in the terminal, as it makes test failures |
| 2651 | visually displayed easier and faster. |
| 2652 | |
| 2653 | Closes #5644 |
| 2654 | |
| 2655 | - curl: add %{method} to the -w variables |
| 2656 | |
| 2657 | Gets the CURLINFO_EFFECTIVE_METHOD from libcurl. |
| 2658 | |
| 2659 | Added test 1197 to verify. |
| 2660 | |
| 2661 | - CURLINFO_EFFECTIVE_METHOD: added |
| 2662 | |
| 2663 | Provide the HTTP method that was used on the latest request, which might |
| 2664 | be relevant for users when there was one or more redirects involved. |
| 2665 | |
| 2666 | Closes #5511 |
| 2667 | |
| 2668 | Viktor Szakats (14 Jul 2020) |
| 2669 | - windows: add unicode to feature list |
| 2670 | |
| 2671 | Reviewed-by: Marcel Raad |
| 2672 | Reviewed-by: Marc Hörsken |
| 2673 | |
| 2674 | Closes #5491 |
| 2675 | |
| 2676 | Daniel Stenberg (14 Jul 2020) |
| 2677 | - multi: remove two checks always true |
| 2678 | |
| 2679 | Detected by Codacy |
| 2680 | Closes #5676 |
| 2681 | |
| 2682 | Marc Hoersken (13 Jul 2020) |
| 2683 | - workflows: limit what branches to run CodeQL on |
| 2684 | |
| 2685 | Align CodeQL action with existing CI actions: |
| 2686 | - Update branch filter to avoid duplicate CI runs. |
| 2687 | - Shorten workflow name due to informative job name. |
| 2688 | |
| 2689 | Reviewed-by: Daniel Stenberg |
| 2690 | |
| 2691 | Closes #5660 |
| 2692 | |
| 2693 | - appveyor: collect libcurl.dll variants with prefix or suffix |
| 2694 | |
| 2695 | On some platforms libcurl is build with a platform-specific |
| 2696 | prefix and/or a version number suffix. |
| 2697 | |
| 2698 | Assisted-by: Jay Satiro |
| 2699 | |
| 2700 | Closes #5659 |
| 2701 | |
| 2702 | Daniel Stenberg (12 Jul 2020) |
| 2703 | - [ihsinme brought this change] |
| 2704 | |
| 2705 | socks: use size_t for size variable |
| 2706 | |
| 2707 | Use the unsigned type (size_t) in the arithmetic of pointers. In this |
| 2708 | context, the signed type (ssize_t) is used unnecessarily. |
| 2709 | |
| 2710 | Authored-by: ihsinme on github |
| 2711 | Closes #5654 |
| 2712 | |
| 2713 | - RELEASE-NOTES: synced |
| 2714 | |
| 2715 | ... and bumped to 7.72.0 as the next release version number |
| 2716 | |
| 2717 | - [Gilles Vollant brought this change] |
| 2718 | |
| 2719 | content_encoding: add zstd decoding support |
| 2720 | |
| 2721 | include zstd curl patch for Makefile.m32 from vszakats |
| 2722 | and include Add CMake support for zstd from Peter Wu |
| 2723 | |
| 2724 | Helped-by: Viktor Szakats |
| 2725 | Helped-by: Peter Wu |
| 2726 | Closes #5453 |
| 2727 | |
| 2728 | - asyn.h: remove the Curl_resolver_getsock define |
| 2729 | |
| 2730 | - not used |
| 2731 | - used the wrong number of arguments |
| 2732 | - confused the Codeacy code analyzer |
| 2733 | |
| 2734 | Closes #5647 |
| 2735 | |
| 2736 | - [Nicolas Sterchele brought this change] |
| 2737 | |
| 2738 | configure.ac: Sort features name in summary |
| 2739 | |
| 2740 | - Same as protocols |
| 2741 | |
| 2742 | Closes #5656 |
| 2743 | |
| 2744 | - [Matthias Naegler brought this change] |
| 2745 | |
| 2746 | cmake: fix windows xp build |
| 2747 | |
| 2748 | Reviewed-by: Marcel Raad |
| 2749 | Closes #5662 |
| 2750 | |
| 2751 | - ngtcp2: update to modified qlog callback prototype |
| 2752 | |
| 2753 | Closes #5675 |
| 2754 | |
| 2755 | - transfer: fix memory-leak with CURLOPT_CURLU in a duped handle |
| 2756 | |
| 2757 | Added test case 674 to reproduce and verify the bug report. |
| 2758 | |
| 2759 | Fixes #5665 |
| 2760 | Reported-by: NobodyXu on github |
| 2761 | Closes #5673 |
| 2762 | |
| 2763 | - [Baruch Siach brought this change] |
| 2764 | |
| 2765 | bearssl: fix build with disabled proxy support |
| 2766 | |
| 2767 | Avoid reference to fields that do not exist when CURL_DISABLE_PROXY is |
| 2768 | defined. |
| 2769 | |
| 2770 | Reviewed-by: Nicolas Sterchele |
| 2771 | Closes #5666 |
| 2772 | |
| 2773 | - RELEASE-NOTES: synced |
| 2774 | |
| 2775 | Jay Satiro (11 Jul 2020) |
| 2776 | - [Carlo Marcelo Arenas Belón brought this change] |
| 2777 | |
| 2778 | cirrus-ci: upgrade 11-STABLE to 11.4 |
| 2779 | |
| 2780 | Meant to be the last of the 11 series and so make sure that all |
| 2781 | other references reflect all 11 versions so they can be retired |
| 2782 | together later. |
| 2783 | |
| 2784 | Closes https://github.com/curl/curl/pull/5668 |
| 2785 | |
| 2786 | - [Filip Salomonsson brought this change] |
| 2787 | |
| 2788 | CURLINFO_CERTINFO.3: fix typo |
| 2789 | |
| 2790 | Closes https://github.com/curl/curl/pull/5655 |
| 2791 | |
| 2792 | Daniel Stenberg (4 Jul 2020) |
| 2793 | - http2: only do the *done() cleanups for HTTP |
| 2794 | |
| 2795 | Follow-up to ef86daf4d3 |
| 2796 | |
| 2797 | Closes #5650 |
| 2798 | Fixes #5646 |
| 2799 | |
| 2800 | - [Alex Kiernan brought this change] |
| 2801 | |
| 2802 | gnutls: repair the build with `CURL_DISABLE_PROXY` |
| 2803 | |
| 2804 | `http_proxy`/`proxy_ssl`/`tunnel_proxy` will not be available in `conn` |
| 2805 | if `CURL_DISABLE_PROXY` is enabled. Repair the build with that |
| 2806 | configuration. |
| 2807 | |
| 2808 | Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com> |
| 2809 | Closes #5645 |
| 2810 | |
| 2811 | Alex Kiernan (3 Jul 2020) |
| 2812 | - gnutls: Fetch backend when using proxy |
| 2813 | |
| 2814 | Fixes: 89865c149 ("gnutls: remove the BACKEND define kludge") |
| 2815 | Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com> |
| 2816 | |
| 2817 | Daniel Stenberg (3 Jul 2020) |
| 2818 | - [Laramie Leavitt brought this change] |
| 2819 | |
| 2820 | http2: close the http2 connection when no more requests may be sent |
| 2821 | |
| 2822 | Well-behaving HTTP2 servers send two GOAWAY messages. The first |
| 2823 | message is a warning that indicates that the server is going to |
| 2824 | stop accepting streams. The second one actually closes the stream. |
| 2825 | |
| 2826 | nghttp2 reports this state (and the other state of no more stream |
| 2827 | identifiers) via the call nghttp2_session_check_request_allowed(). |
| 2828 | In this state the client should not create more streams on the |
| 2829 | session (tcp connection), and in curl this means that the server |
| 2830 | has requested that the connection is closed. |
| 2831 | |
| 2832 | It would be also be possible to put the connclose() call into the |
| 2833 | on_http2_frame_recv() function that triggers on the GOAWAY message. |
| 2834 | |
| 2835 | This fixes a bug seen when the client sees the following sequence of |
| 2836 | frames: |
| 2837 | |
| 2838 | // advisory GOAWAY |
| 2839 | HTTP2 GOAWAY [stream-id = 0, promised-stream-id = -1] |
| 2840 | ... some additional frames |
| 2841 | |
| 2842 | // final GOAWAY |
| 2843 | HTTP2 GOAWAY [stream-id = 0, promised-stream-id = N ] |
| 2844 | |
| 2845 | Before this change, curl will attempt to reuse the connection even |
| 2846 | after the last stream, will encounter this error: |
| 2847 | |
| 2848 | * Found bundle for host localhost: 0x5595f0a694e0 [can multiplex] |
| 2849 | * Re-using existing connection! (#0) with host localhost |
| 2850 | * Connected to localhost (::1) port 10443 (#0) |
| 2851 | * Using Stream ID: 9 (easy handle 0x5595f0a72e30) |
| 2852 | > GET /index.html?5 HTTP/2 |
| 2853 | > Host: localhost:10443 |
| 2854 | > user-agent: curl/7.68.0 |
| 2855 | > accept: */* |
| 2856 | > |
| 2857 | * stopped the pause stream! |
| 2858 | * Connection #0 to host localhost left intact |
| 2859 | curl: (16) Error in the HTTP2 framing layer |
| 2860 | |
| 2861 | This error may posion the connection cache, causing future requests |
| 2862 | which resolve to the same curl connection to go through the same error |
| 2863 | path. |
| 2864 | |
| 2865 | Closes #5643 |
| 2866 | |
| 2867 | - ftpserver: don't verify SMTP MAIL FROM names |
| 2868 | |
| 2869 | Rely on tests asking the names to get refused instead - test servers |
| 2870 | should be as dumb as possible. Edited test 914, 955 and 959 accordingly. |
| 2871 | |
| 2872 | Closes #5639 |
| 2873 | |
| 2874 | - curl_version_info.3: CURL_VERSION_KERBEROS4 is deprecated |
| 2875 | |
| 2876 | This came up in #5640. It make sense to clarify this in the docs! |
| 2877 | |
| 2878 | Reminded-by: Kamil Dudka |
| 2879 | Closes #5642 |
| 2880 | |
| 2881 | Kamil Dudka (3 Jul 2020) |
| 2882 | - tool_getparam: make --krb option work again |
| 2883 | |
| 2884 | It was disabled by mistake in commit curl-7_37_1-23-ge38ba4301. |
| 2885 | |
| 2886 | Bug: https://bugzilla.redhat.com/1833193 |
| 2887 | Closes #5640 |
| 2888 | |
| 2889 | Daniel Stenberg (2 Jul 2020) |
| 2890 | - [Jeremy Maitin-Shepard brought this change] |
| 2891 | |
| 2892 | http2: fix nghttp2_strerror -> nghttp2_http2_strerror in debug messages |
| 2893 | |
| 2894 | Confusingly, nghttp2 has two different error code enums: |
| 2895 | |
| 2896 | - nghttp2_error, to be used with nghttp2_strerror |
| 2897 | - nghttp2_error_code, to be used with nghttp2_http2_strerror |
| 2898 | |
| 2899 | Closes #5641 |
| 2900 | |
| 2901 | Marcel Raad (2 Jul 2020) |
| 2902 | - url: silence MSVC warning |
| 2903 | |
| 2904 | Since commit f3d501dc678, if proxy support is disabled, MSVC warns: |
| 2905 | url.c : warning C4701: potentially uninitialized local variable |
| 2906 | 'hostaddr' used |
| 2907 | url.c : error C4703: potentially uninitialized local pointer variable |
| 2908 | 'hostaddr' used |
| 2909 | |
| 2910 | That could actually only happen if both `conn->bits.proxy` and |
| 2911 | `CURL_DISABLE_PROXY` were enabled. |
| 2912 | Initialize it to NULL to silence the warning. |
| 2913 | |
| 2914 | Closes https://github.com/curl/curl/pull/5638 |
| 2915 | |
| 2916 | Daniel Stenberg (1 Jul 2020) |
| 2917 | - RELEASE-NOTES: synced |
| 2918 | |
Haibo Huang | ca2a802 | 2020-07-10 20:17:42 -0700 | [diff] [blame] | 2919 | Version 7.71.1 (30 Jun 2020) |
| 2920 | |
| 2921 | Daniel Stenberg (30 Jun 2020) |
| 2922 | - RELEASE-NOTES: curl 7.71.1 |
| 2923 | |
| 2924 | - THANKS: add contributors to 7.71.1 |
| 2925 | |
| 2926 | - scripts/copyright.pl: skip .dcignore |
| 2927 | |
| 2928 | - Revert "multi: implement wait using winsock events" |
| 2929 | |
| 2930 | This reverts commit 8bc25c590e530de87595d1bb3577f699eb1309b9. |
| 2931 | |
| 2932 | That commit (from #5397) introduced a regression in 7.71.0. |
| 2933 | |
| 2934 | Reported-by: tmkk on github |
| 2935 | Fixes #5631 |
| 2936 | Closes #5632 |
| 2937 | |
| 2938 | - TODO: Add flag to specify download directory |
| 2939 | |
| 2940 | - TODO: return code to CURLMOPT_PUSHFUNCTION to fail connection |
| 2941 | |
| 2942 | - cirrus-ci: disable FreeBSD 13 (again) |
| 2943 | |
| 2944 | It has been failing for a good while again. This time we better leave it |
| 2945 | disabled until we have more reason to believe it behaves. |
| 2946 | |
| 2947 | Closes #5628 |
| 2948 | |
| 2949 | - ngtcp2: sync with current master |
| 2950 | |
| 2951 | ngtcp2 added two new callbacks |
| 2952 | |
| 2953 | Reported-by: Lucien Zürcher |
| 2954 | Fixes #5624 |
| 2955 | Closes #5627 |
| 2956 | |
| 2957 | - examples/multithread.c: call curl_global_cleanup() |
| 2958 | |
| 2959 | Reported-by: qiandu2006 on github |
| 2960 | Fixes #5622 |
| 2961 | Closes #5623 |
| 2962 | |
| 2963 | - vtls: compare cert blob when finding a connection to reuse |
| 2964 | |
| 2965 | Reported-by: Gergely Nagy |
| 2966 | Fixes #5617 |
| 2967 | Closes #5619 |
| 2968 | |
| 2969 | - RELEASE-NOTES: synced |
| 2970 | |
| 2971 | - terminology: call them null-terminated strings |
| 2972 | |
| 2973 | Updated terminology in docs, comments and phrases to refer to C strings |
| 2974 | as "null-terminated". Done to unify with how most other C oriented docs |
| 2975 | refer of them and what users in general seem to prefer (based on a |
| 2976 | single highly unscientific poll on twitter). |
| 2977 | |
| 2978 | Reported-by: coinhubs on github |
| 2979 | Fixes #5598 |
| 2980 | Closes #5608 |
| 2981 | |
| 2982 | - http: fix proxy auth with blank password |
| 2983 | |
| 2984 | Regression in 7.71.0 |
| 2985 | |
| 2986 | Added test case 346 to verify. |
| 2987 | |
| 2988 | Reported-by: Kristoffer Gleditsch |
| 2989 | Fixes #5613 |
| 2990 | Closes #5616 |
| 2991 | |
| 2992 | - .dcignore: ignore tests and docs directories |
| 2993 | |
| 2994 | This is a config file for deepcode.ai, a static code analyzer. |
| 2995 | |
| 2996 | Jay Satiro (26 Jun 2020) |
| 2997 | - tool_cb_hdr: Fix etag warning output and return code |
| 2998 | |
| 2999 | - Return 'failure' on failure, to follow the existing style. |
| 3000 | |
| 3001 | - Put Warning: and the warning message on the same line. |
| 3002 | |
| 3003 | Ref: https://github.com/curl/curl/issues/5610 |
| 3004 | |
| 3005 | Closes https://github.com/curl/curl/pull/5612 |
| 3006 | |
| 3007 | Daniel Stenberg (26 Jun 2020) |
| 3008 | - CURLOPT_READFUNCTION.3: provide the upload data size up front |
| 3009 | |
| 3010 | Assisted-by: Jay Satiro |
| 3011 | Closes #5607 |
| 3012 | |
| 3013 | - test1539: do a HTTP 1.0 POST without a set size (fails) |
| 3014 | |
| 3015 | Attempt to reproduce #5593. Test case 1514 is very similar but uses |
| 3016 | HTTP/1.1 and thus switches to chunked. |
| 3017 | |
| 3018 | Closes #5595 |
| 3019 | |
| 3020 | - [Baruch Siach brought this change] |
| 3021 | |
| 3022 | mbedtls: fix build with disabled proxy support |
| 3023 | |
| 3024 | Don't reference fields that do not exist. Fixes build failure: |
| 3025 | |
| 3026 | vtls/mbedtls.c: In function 'mbed_connect_step1': |
| 3027 | vtls/mbedtls.c:249:54: error: 'struct connectdata' has no member named 'http_proxy' |
| 3028 | |
| 3029 | Closes #5615 |
| 3030 | |
| 3031 | - codeql-analysis.yml: fix the 'languages' setting |
| 3032 | |
| 3033 | It needs a 'with:' in front of it. |
| 3034 | |
| 3035 | GitHub (26 Jun 2020) |
| 3036 | - [Daniel Stenberg brought this change] |
| 3037 | |
| 3038 | gtihub: codeql-analysis.yml |
| 3039 | |
| 3040 | enables code security scanning with github actions |
| 3041 | |
| 3042 | Daniel Stenberg (25 Jun 2020) |
| 3043 | - tests: verify newline in username and password for HTTP |
| 3044 | |
| 3045 | test 1296 is a simply command line test |
| 3046 | |
| 3047 | test 1910 is a libcurl test including a redirect |
| 3048 | |
| 3049 | - url: allow user + password to contain "control codes" for HTTP(S) |
| 3050 | |
| 3051 | Reported-by: Jon Johnson Jr |
| 3052 | Fixes #5582 |
| 3053 | Closes #5592 |
| 3054 | |
| 3055 | - escape: make the URL decode able to reject only %00 bytes |
| 3056 | |
| 3057 | ... or all "control codes" or nothing. |
| 3058 | |
| 3059 | Assisted-by: Nicolas Sterchele |
| 3060 | |
| 3061 | - http2: set the correct URL in pushed transfers |
| 3062 | |
| 3063 | ...previously CURLINFO_EFFECTIVE_URL would report the URL of the |
| 3064 | original "mother transfer", not the actually pushed resource. |
| 3065 | |
| 3066 | Reported-by: Jonathan Cardoso Machado |
| 3067 | Fixes #5589 |
| 3068 | Closes #5591 |
| 3069 | |
| 3070 | Jay Satiro (25 Jun 2020) |
| 3071 | - [Javier Blazquez brought this change] |
| 3072 | |
| 3073 | openssl: Fix compilation on Windows when ngtcp2 is enabled |
| 3074 | |
| 3075 | - Include wincrypt before OpenSSL includes so that the latter can |
| 3076 | properly handle any conflicts between the two. |
| 3077 | |
| 3078 | Closes https://github.com/curl/curl/pull/5606 |
| 3079 | |
| 3080 | Daniel Stenberg (25 Jun 2020) |
| 3081 | - test543: extended to verify zero length input |
| 3082 | |
| 3083 | As was reported in #5601 |
| 3084 | |
| 3085 | - escape: zero length input should return a zero length output |
| 3086 | |
| 3087 | Regression added in 7.71.0. |
| 3088 | |
| 3089 | Fixes #5601 |
| 3090 | Reported-by: Kristoffer Gleditsch |
| 3091 | Closes #5602 |
| 3092 | |
| 3093 | - Curl_inet_ntop: always check the return code |
| 3094 | |
| 3095 | Reported-by: Siva Sivaraman |
| 3096 | Fixes #5412 |
| 3097 | Closes #5597 |
| 3098 | |
| 3099 | - sendf: improve the message on client write errors |
| 3100 | |
| 3101 | Replace "Failed writing body (X != Y)" with |
| 3102 | "Failure writing output to destination". Possibly slightly less cryptic. |
| 3103 | |
| 3104 | Reported-by: coinhubs on github |
| 3105 | Fixes #5594 |
| 3106 | Closes #5596 |
| 3107 | |
| 3108 | - RELEASE-NOTES: synced |
| 3109 | |
| 3110 | - curlver: start working on 7.71.1 |
| 3111 | |
| 3112 | - [Denis Baručić brought this change] |
| 3113 | |
| 3114 | DYNBUF.md: fix a typo: trail => tail |
| 3115 | |
| 3116 | Closes #5599 |
| 3117 | |
| 3118 | Version 7.71.0 (23 Jun 2020) |
| 3119 | |
| 3120 | Daniel Stenberg (23 Jun 2020) |
| 3121 | - RELEASE-NOTES: curl 7.71.0 release |
| 3122 | |
| 3123 | - THANKS: curl 7.71.0 additions |
| 3124 | |
| 3125 | - url: make sure pushed streams get an allocated download buffer |
| 3126 | |
| 3127 | Follow-up to c4e6968127e876b0 |
| 3128 | |
| 3129 | When a new transfer is created, as a resuly of an acknowledged push, |
| 3130 | that transfer needs a download buffer allocated. |
| 3131 | |
| 3132 | Closes #5590 |
| 3133 | |
| 3134 | Jay Satiro (22 Jun 2020) |
| 3135 | - openssl: Don't ignore CA paths when using Windows CA store |
| 3136 | |
| 3137 | This commit changes the behavior of CURLSSLOPT_NATIVE_CA so that it does |
| 3138 | not override CURLOPT_CAINFO / CURLOPT_CAPATH, or the hardcoded default |
| 3139 | locations. Instead the CA store can now be used at the same time. |
| 3140 | |
| 3141 | The change is due to the impending release. The issue is still being |
| 3142 | discussed. The behavior of CURLSSLOPT_NATIVE_CA is subject to change and |
| 3143 | is now documented as experimental. |
| 3144 | |
| 3145 | Ref: bc052cc (parent commit) |
| 3146 | Ref: https://github.com/curl/curl/issues/5585 |
| 3147 | |
| 3148 | - tool_operate: Don't use Windows CA store as a fallback |
| 3149 | |
| 3150 | Background: |
| 3151 | |
| 3152 | 148534d added CURLSSLOPT_NATIVE_CA to use the Windows OS certificate |
| 3153 | store in libcurl w/ OpenSSL on Windows. CURLSSLOPT_NATIVE_CA overrides |
| 3154 | CURLOPT_CAINFO if both are set. The curl tool will fall back to |
| 3155 | CURLSSLOPT_NATIVE_CA if it could not find a certificate bundle to set |
| 3156 | via CURLOPT_CAINFO. |
| 3157 | |
| 3158 | Problem: |
| 3159 | |
| 3160 | libcurl may be built with hardcoded paths to a certificate bundle or |
| 3161 | directory, and if CURLSSLOPT_NATIVE_CA is used then those paths are |
| 3162 | ignored. |
| 3163 | |
| 3164 | Solution: |
| 3165 | |
| 3166 | A solution is still being discussed but since there's an impending |
| 3167 | release this commit removes using CURLSSLOPT_NATIVE_CA in the curl tool. |
| 3168 | |
| 3169 | Ref: https://github.com/curl/curl/issues/5585 |
| 3170 | |
| 3171 | - openssl: Fix CA fallback logic for OpenSSL 3.0 build |
| 3172 | |
| 3173 | Prior to this change I assume a build error would occur when |
| 3174 | CURL_CA_FALLBACK was used. |
| 3175 | |
| 3176 | Closes https://github.com/curl/curl/pull/5587 |
| 3177 | |
| 3178 | Daniel Stenberg (22 Jun 2020) |
| 3179 | - copyright: update mismatched copyright years |
| 3180 | |
| 3181 | - test1460: verify that -Ji is not ok |
| 3182 | |
| 3183 | - tool_getparam: -i is not OK if -J is used |
| 3184 | |
| 3185 | Reported-by: sn on hackerone |
| 3186 | Bug: https://curl.haxx.se/docs/CVE-2020-8177.html |
| 3187 | |
| 3188 | - [Peter Wu brought this change] |
| 3189 | |
| 3190 | CMake: ignore INTERFACE_LIBRARY targets for pkg-config file |
| 3191 | |
| 3192 | Reviewed-by: Marcel Raad |
| 3193 | Fixes #5512 |
| 3194 | Closes #5517 |
| 3195 | |
| 3196 | - [Valentyn Korniienko brought this change] |
| 3197 | |
| 3198 | multibyte: Fixed access-> waccess to file for Windows Plarform |
| 3199 | |
| 3200 | Reviewed-by: Marcel Raad |
| 3201 | Closes #5580 |
| 3202 | |
| 3203 | - altsvc: bump to h3-29 |
| 3204 | |
| 3205 | Closes #5584 |
| 3206 | |
| 3207 | - urlglob: treat literal IPv6 addresses with zone IDs as a host name |
| 3208 | |
| 3209 | ... and not as a "glob". Now done by passing the supposed host to the |
| 3210 | URL parser which supposedly will do a better job at identifying "real" |
| 3211 | numerical IPv6 addresses. |
| 3212 | |
| 3213 | Reported-by: puckipedia on github |
| 3214 | Fixes #5576 |
| 3215 | Closes #5579 |
| 3216 | |
| 3217 | - test1179: verify error message for non-existing cmdline option |
| 3218 | |
| 3219 | - tool_getparam: repair the error message for unknown flag |
| 3220 | |
| 3221 | Follow-up to 9e5669f3880674 |
| 3222 | Detected by Coverity CID 1464582 ("Logically dead code") |
| 3223 | |
| 3224 | Closes #5577 |
| 3225 | |
| 3226 | - FILEFORMAT: describe verify/stderr |
| 3227 | |
| 3228 | - connect: improve happy eyeballs handling |
| 3229 | |
| 3230 | For QUIC but also for regular TCP when the second family runs out of IPs |
| 3231 | with a failure while the first family is still trying to connect. |
| 3232 | |
| 3233 | Separated the timeout handling for IPv4 and IPv6 connections when they |
| 3234 | both have a number of addresses to iterate over. |
| 3235 | |
| 3236 | - ngtcp2: never call fprintf() in lib code in release version |
| 3237 | |
| 3238 | - ngtcp2: fix happy eyeballs quic connect crash |
| 3239 | |
| 3240 | Reported-by: Peter Wu |
| 3241 | Fixes #5565 |
| 3242 | Closes #5568 |
| 3243 | |
| 3244 | - select: remove the unused ELAPSED_MS() macro |
| 3245 | |
| 3246 | Closes #5573 |
| 3247 | |
| 3248 | Marc Hoersken (17 Jun 2020) |
| 3249 | - [rcombs brought this change] |
| 3250 | |
| 3251 | multi: implement wait using winsock events |
| 3252 | |
| 3253 | This avoids using a pair of TCP ports to provide wakeup functionality |
| 3254 | for every multi instance on Windows, where socketpair() is emulated |
| 3255 | using a TCP socket on loopback which could in turn lead to socket |
| 3256 | resource exhaustion. |
| 3257 | |
| 3258 | Reviewed-by: Gergely Nagy |
| 3259 | Reviewed-by: Marc Hörsken |
| 3260 | |
| 3261 | Closes #5397 |
| 3262 | |
| 3263 | Daniel Stenberg (17 Jun 2020) |
| 3264 | - manpage: add three missing environment variables |
| 3265 | |
| 3266 | CURL_SSL_BACKEND, QLOGDIR and SSLKEYLOGFILE |
| 3267 | |
| 3268 | Closes #5571 |
| 3269 | |
| 3270 | - RELEASE-NOTES: synced |
| 3271 | |
| 3272 | - configure: for wolfSSL, check for the DES func needed for NTLM |
| 3273 | |
| 3274 | Also adds pkg-config support for the wolfSSL detection. |
| 3275 | |
| 3276 | - [Ruurd Beerstra brought this change] |
| 3277 | |
| 3278 | ntlm: enable NTLM support with wolfSSL |
| 3279 | |
| 3280 | When wolfSSL is built with its OpenSSL API layer, it fetures the same DES* |
| 3281 | functions that OpenSSL has. This change take advantage of that. |
| 3282 | |
| 3283 | Co-authored-by: Daniel Stenberg |
| 3284 | Closes #5556 |
| 3285 | Fixes #5548 |
| 3286 | |
| 3287 | - http: move header storage to Curl_easy from connectdata |
| 3288 | |
| 3289 | Since the connection can be used by many independent requests (using |
| 3290 | HTTP/2 or HTTP/3), things like user-agent and other transfer-specific |
| 3291 | data MUST NOT be kept connection oriented as it could lead to requests |
| 3292 | getting the wrong string for their requests. This struct data was |
| 3293 | lingering like this due to old HTTP1 legacy thinking where it didn't |
| 3294 | mattered.. |
| 3295 | |
| 3296 | Fixes #5566 |
| 3297 | Closes #5567 |
| 3298 | |
| 3299 | - CODE_REVIEW.md: how to do code reviews in curl |
| 3300 | |
| 3301 | Assisted-by: Daniel Gustafsson |
| 3302 | Assisted-by: Rich Salz |
| 3303 | Assisted-by: Hugo van Kemenade |
| 3304 | Assisted-by: James Fuller |
| 3305 | Assisted-by: Marc Hörsken |
| 3306 | Assisted-by: Jay Satiro |
| 3307 | |
| 3308 | Closes #5555 |
| 3309 | |
| 3310 | - altsvc: remove the num field from the altsvc struct |
| 3311 | |
| 3312 | It was superfluous since we have the list.size alredy |
| 3313 | |
| 3314 | Reported-by: Jay Satiro |
| 3315 | Fixes #5553 |
| 3316 | Closes #5563 |
| 3317 | |
| 3318 | - version.d: expanded and alpha-sorted |
| 3319 | |
| 3320 | Added a few missing features not previously mentioned. Ordered them |
| 3321 | alphabetically. |
| 3322 | |
| 3323 | Closes #5558 |
| 3324 | |
| 3325 | - ABI.md: rename to .md and polish the markdown |
| 3326 | |
| 3327 | Closes #5562 |
| 3328 | |
| 3329 | - HELP-US: add a section for "smaller tasks" |
| 3330 | |
| 3331 | The point of this section is to meet the CII Best Practices gold level |
| 3332 | critera: |
| 3333 | |
| 3334 | "The project MUST clearly identify small tasks that can be performed by |
| 3335 | new or casual contributors" |
| 3336 | |
| 3337 | Closes #5560 |
| 3338 | |
| 3339 | - TODO: retry on the redirected-to URL |
| 3340 | |
| 3341 | Closes #5462 |
| 3342 | |
| 3343 | - mailmap: Nicolas Sterchele |
| 3344 | |
| 3345 | - [Nicolas Sterchele brought this change] |
| 3346 | |
| 3347 | TODO: remove 19.3 section title |
| 3348 | |
| 3349 | Follow-up to ad6416986755e417c66e2c6, which caused wrong formatting on |
| 3350 | curl documentation website |
| 3351 | |
| 3352 | Closes #5561 |
| 3353 | |
| 3354 | - [Martin V brought this change] |
| 3355 | |
| 3356 | test1560: avoid possibly negative association in wording |
| 3357 | |
| 3358 | Closes #5549 |
| 3359 | |
| 3360 | - share: don't set the share flag it something fails |
| 3361 | |
| 3362 | When asking for a specific feature to be shared in the share object, |
| 3363 | that bit was previously set unconditionally even if the shared feature |
| 3364 | failed or otherwise wouldn't work. |
| 3365 | |
| 3366 | Closes #5554 |
| 3367 | |
| 3368 | - buildconf: remove -print from the find command that removes files |
| 3369 | |
| 3370 | It's just too annoying and unnecessary to get a long list of files shown |
| 3371 | |
| 3372 | - RELEASE-NOTES: synced |
| 3373 | |
| 3374 | - wording: avoid blacklist/whitelist stereotypes |
| 3375 | |
| 3376 | Instead of discussing if there's value or meaning (implied or not) in |
| 3377 | the colors, let's use words without the same possibly negative |
| 3378 | associations. |
| 3379 | |
| 3380 | Closes #5546 |
| 3381 | |
| 3382 | Jay Satiro (9 Jun 2020) |
| 3383 | - tool_getparam: fix memory leak in parse_args |
| 3384 | |
| 3385 | Prior to this change in Windows Unicode builds most parsed options would |
| 3386 | not be freed. |
| 3387 | |
| 3388 | Found using _CrtDumpMemoryLeaks(). |
| 3389 | |
| 3390 | Ref: https://github.com/curl/curl/issues/5545 |
| 3391 | |
| 3392 | Daniel Stenberg (8 Jun 2020) |
| 3393 | - socks: detect connection close during handshake |
| 3394 | |
| 3395 | The SOCKS4/5 state machines weren't properly terminated when the proxy |
| 3396 | connection got closed, leading to a busy-loop. |
| 3397 | |
| 3398 | Reported-By: zloi-user on github |
| 3399 | Fixes #5532 |
| 3400 | Closes #5542 |
| 3401 | |
| 3402 | - [James Fuller brought this change] |
| 3403 | |
| 3404 | multi: add defensive check on data->multi->num_alive |
| 3405 | |
| 3406 | Closes #5540 |
| 3407 | |
| 3408 | - Curl_addrinfo: use one malloc instead of three |
| 3409 | |
| 3410 | To reduce the amount of allocations needed for creating a Curl_addrinfo |
| 3411 | struct, make a single larger malloc instead of three separate smaller |
| 3412 | ones. |
| 3413 | |
| 3414 | Closes #5533 |
| 3415 | |
| 3416 | - [Alessandro Ghedini brought this change] |
| 3417 | |
| 3418 | quiche: update SSLKEYLOGFILE support |
| 3419 | |
| 3420 | quiche now requires the application to explicitly set the keylog path |
| 3421 | for each connection, rather than reading the environment variable |
| 3422 | itself. |
| 3423 | |
| 3424 | Closes #5541 |
| 3425 | |
| 3426 | - tests: add two simple tests for --login-options |
| 3427 | |
| 3428 | Test 895 and 896 - as a follow-up to a3e972313b |
| 3429 | |
| 3430 | Closes #5539 |
| 3431 | |
| 3432 | - ngtcp2: update with recent API changes |
| 3433 | |
| 3434 | Syncs with ngtcp2 commit 7e9a917d386d98 merged June 7 2020. |
| 3435 | |
| 3436 | Assisted-by: Tatsuhiro Tsujikawa |
| 3437 | Closes #5538 |
| 3438 | |
| 3439 | - [James Fuller brought this change] |
| 3440 | |
| 3441 | socks: remove unreachable breaks in socks.c and mime.c |
| 3442 | |
| 3443 | Closes #5537 |
| 3444 | |
| 3445 | - tool_cfgable: free login_options at exit |
| 3446 | |
| 3447 | Memory leak |
| 3448 | Reported-by: Geeknik Labs |
| 3449 | Fixes #5535 |
| 3450 | Closes #5536 |
| 3451 | |
| 3452 | - libssh2: keep sftp errors as 'unsigned long' |
| 3453 | |
| 3454 | Remove weird work-around for storing the SFTP errors as int instead of |
| 3455 | the "unsigned long" that libssh2 actually returns for SFTP errors. |
| 3456 | |
| 3457 | Closes #5534 |
| 3458 | |
| 3459 | Marc Hoersken (6 Jun 2020) |
| 3460 | - timeouts: move ms timeouts to timediff_t from int and long |
| 3461 | |
| 3462 | Now that all functions in select.[ch] take timediff_t instead |
| 3463 | of the limited int or long, we can remove type conversions |
| 3464 | and related preprocessor checks to silence compiler warnings. |
| 3465 | |
| 3466 | Avoiding conversions from time_t was already done in 842f73de. |
| 3467 | |
| 3468 | Based upon #5262 |
| 3469 | Supersedes #5214, #5220 and #5221 |
| 3470 | Follow up to #5343 and #5479 |
| 3471 | Closes #5490 |
| 3472 | |
| 3473 | Daniel Stenberg (6 Jun 2020) |
| 3474 | - [François Rigault brought this change] |
| 3475 | |
| 3476 | openssl: set FLAG_TRUSTED_FIRST unconditionally |
| 3477 | |
| 3478 | On some systems, openssl 1.0.x is still the default, but it has been |
| 3479 | patched to contain all the recent security fixes. As a result of this |
| 3480 | patching, it is possible for macro X509_V_FLAG_NO_ALT_CHAINS to be |
| 3481 | defined, while the previous behavior of openssl to not look at trusted |
| 3482 | chains first, remains. |
| 3483 | |
| 3484 | Fix it: ensure X509_V_FLAG_TRUSTED_FIRST is always set, do not try to |
| 3485 | probe for the behavior of openssl based on the existence ofmacros. |
| 3486 | |
| 3487 | Closes #5530 |
| 3488 | |
| 3489 | - server/util: fix logmsg format using curl_off_t argument |
| 3490 | |
| 3491 | ... this caused segfaults on armv7. |
| 3492 | |
| 3493 | Regression added in dd0365d560aea5a (7.70.0) |
| 3494 | |
| 3495 | Reviewed-by: Jay Satiro |
| 3496 | Closes #5529 |
| 3497 | |
| 3498 | - RELEASE-NOTES: synced |
| 3499 | |
| 3500 | - [Cherish98 brought this change] |
| 3501 | |
| 3502 | socks: fix expected length of SOCKS5 reply |
| 3503 | |
| 3504 | Commit 4a4b63d forgot to set the expected SOCKS5 reply length when the |
| 3505 | reply ATYP is X'01'. This resulted in erroneously expecting more bytes |
| 3506 | when the request length is greater than the reply length (e.g., when |
| 3507 | remotely resolving the hostname). |
| 3508 | |
| 3509 | Closes #5527 |
| 3510 | |
| 3511 | Marc Hoersken (5 Jun 2020) |
| 3512 | - .gitignore: add directory containing the stats repo |
| 3513 | |
| 3514 | Since the new curl/stats repository is designed to be |
| 3515 | checked out into the curl repository working tree as stats/ |
| 3516 | it should be on the ignore list to aid in commit staging. |
| 3517 | |
| 3518 | Daniel Stenberg (5 Jun 2020) |
| 3519 | - [Adnan Khan brought this change] |
| 3520 | |
| 3521 | HTTP3.md: clarify cargo build directory |
| 3522 | |
| 3523 | Cargo needs to be called from within the 'quiche' directory. |
| 3524 | |
| 3525 | Closes #5522 |
| 3526 | |
| 3527 | - user-agent.d: spell out what happens given a blank argument |
| 3528 | |
| 3529 | Closes #5525 |
| 3530 | |
| 3531 | - trailers: switch h1-trailer logic to use dynbuf |
| 3532 | |
| 3533 | In the continued effort to remove "manual" realloc schemes. |
| 3534 | |
| 3535 | Closes #5524 |
| 3536 | |
| 3537 | - CURLINFO_ACTIVESOCKET.3: clarify the description |
| 3538 | |
| 3539 | Reported-by: Jay Satiro |
| 3540 | Fixes #5299 |
| 3541 | Closes #5520 |
| 3542 | |
| 3543 | - mailmap: Don J Olmstead |
| 3544 | |
| 3545 | - configure: only strip first -L from LDFLAGS |
| 3546 | |
| 3547 | In the logic that works out if a given OpenSSL path works, it stripped |
| 3548 | off a possibly leading -L flag using an incorrect sed pattern which |
| 3549 | would remove all instances of -L in the string, including if the path |
| 3550 | itself contained that two-letter sequence! |
| 3551 | |
| 3552 | The same pattern was used and is now updated in multiple places. Now it |
| 3553 | only removes -L if it starts the strings. |
| 3554 | |
| 3555 | Reported-by: Mohamed Osama |
| 3556 | Fixes #5519 |
| 3557 | Closes #5521 |
| 3558 | |
| 3559 | Peter Wu (4 Jun 2020) |
| 3560 | - quiche: advertise draft 28 support |
| 3561 | |
| 3562 | Fix the verbose message while at it, quiche currently supports draft |
| 3563 | 27 and draft 28 simultaneously. |
| 3564 | |
| 3565 | Closes #5518 |
| 3566 | |
| 3567 | Daniel Stenberg (4 Jun 2020) |
| 3568 | - KNOWN_BUGS: RTSP authentication breaks without redirect support |
| 3569 | |
| 3570 | Closes #4750 |
| 3571 | |
| 3572 | Jay Satiro (4 Jun 2020) |
| 3573 | - projects: Add crypt32.lib to dependencies for all OpenSSL configs |
| 3574 | |
| 3575 | Windows project configurations that use OpenSSL with USE_WIN32_CRYPTO |
| 3576 | need crypt32. |
| 3577 | |
| 3578 | Follow-up to 148534d which added CURLSSLOPT_NATIVE_CA for 7.71.0. |
| 3579 | |
| 3580 | The changes that are in this commit were made by script. |
| 3581 | |
| 3582 | Ref: https://gist.github.com/jay/a1861b50ecce2b32931237180f856e28 |
| 3583 | |
| 3584 | Closes https://github.com/curl/curl/pull/5516 |
| 3585 | |
| 3586 | Marc Hoersken (3 Jun 2020) |
| 3587 | - CI/macos: fix 'is already installed' errors by using bundle |
| 3588 | |
| 3589 | Avoid failing CI builds due to nghttp2 being already installed. |
| 3590 | |
| 3591 | Closes #5513 |
| 3592 | |
| 3593 | Daniel Stenberg (3 Jun 2020) |
| 3594 | - altsvc: fix 'dsthost' may be used uninitialized in this function |
| 3595 | |
| 3596 | - RELEASE-NOTES: synced |
| 3597 | |
| 3598 | - urldata: let the HTTP method be in the set.* struct |
| 3599 | |
| 3600 | When the method is updated inside libcurl we must still not change the |
| 3601 | method as set by the user as then repeated transfers with that same |
| 3602 | handle might not execute the same operation anymore! |
| 3603 | |
| 3604 | This fixes the libcurl part of #5462 |
| 3605 | |
| 3606 | Test 1633 added to verify. |
| 3607 | |
| 3608 | Closes #5499 |
| 3609 | |
| 3610 | - hostip: fix the memory-leak introduced in 67d2802 |
| 3611 | |
| 3612 | Fixes #5503 |
| 3613 | Closes #5504 |
| 3614 | |
| 3615 | - test970: make it require proxy support |
| 3616 | |
| 3617 | This test verifies the -w %json output and the test case includes a full |
| 3618 | generated "blob". If there's no proxy support built into libcurl, it |
| 3619 | will return an error for proxy related info variables and they will not |
| 3620 | be included in the json, thus causing a mismatch and this test fails. |
| 3621 | |
| 3622 | Reported-by: Marc Hörsken |
| 3623 | Fixes #5501 |
| 3624 | Closes #5502 |
| 3625 | |
| 3626 | - [Radoslav Georgiev brought this change] |
| 3627 | |
| 3628 | examples/http2-down/upload: add error checks |
| 3629 | |
| 3630 | If `index.html` does not exist in the directory from which the example |
| 3631 | is invoked, the fopen(upload, "rb") invocation in `setup` would fail, |
| 3632 | returning NULL. This value is subsequently passed as the FILE* argument |
| 3633 | of the `fread` invocation in the `read_callback` function, which is the |
| 3634 | actual cause of the crash (apparently `fread` assumes that argument to |
| 3635 | be non-null). |
| 3636 | |
| 3637 | In addition, mitigate some possible crashes of similar origin. |
| 3638 | |
| 3639 | Closes #5463 |
| 3640 | |
| 3641 | - [kotoriのねこ brought this change] |
| 3642 | |
| 3643 | examples/ephiperfifo: turn off interval when setting timerfd |
| 3644 | |
| 3645 | Reported-by: therealhirudo on github |
| 3646 | Fixes #5485 |
| 3647 | Closes #5497 |
| 3648 | |
| 3649 | - [Saleem Abdulrasool brought this change] |
| 3650 | |
| 3651 | vtls: repair the build with `CURL_DISABLE_PROXY` |
| 3652 | |
| 3653 | `http_proxy` will not be available in `conndata` if `CURL_DISABLE_PROXY` |
| 3654 | is enabled. Repair the build with that configuration. |
| 3655 | |
| 3656 | Follow-up to f3d501dc67 |
| 3657 | |
| 3658 | Closes #5498 |
| 3659 | |
| 3660 | - transfer: remove k->str NULL check |
| 3661 | |
| 3662 | "Null-checking k->str suggests that it may be null, but it has already |
| 3663 | been dereferenced on all paths leading to the check" - and it can't |
| 3664 | legally be NULL at this point. Remove check. |
| 3665 | |
| 3666 | Detected by Coverity CID 1463884 |
| 3667 | |
| 3668 | Closes #5495 |
| 3669 | |
| 3670 | Marc Hoersken (1 Jun 2020) |
| 3671 | - select: always use Sleep in Curl_wait_ms on Win32 |
| 3672 | |
| 3673 | Since Win32 almost always will also have USE_WINSOCK, |
| 3674 | we can reduce complexity and always use Sleep there. |
| 3675 | |
| 3676 | Assisted-by: Jay Satiro |
| 3677 | Reviewed-by: Daniel Stenberg |
| 3678 | |
| 3679 | Follow up to #5343 |
| 3680 | Closes #5489 |
| 3681 | |
| 3682 | Daniel Stenberg (31 May 2020) |
| 3683 | - conncache: download buffer needs +1 size for trailing zero |
| 3684 | |
| 3685 | Follow-up to c4e6968127e |
| 3686 | Detected by OSS-Fuzz: https://oss-fuzz.com/testcase-detail/5727799779524608 |
| 3687 | |
| 3688 | Marc Hoersken (31 May 2020) |
| 3689 | - azure: use matrix strategy to avoid configuration redundancy |
| 3690 | |
| 3691 | This also includes the following changes: |
| 3692 | |
| 3693 | - Use the same timeout for all jobs on Linux (60 minutes) |
| 3694 | and Windows (90 minutes) |
| 3695 | - Use CLI stable apt-get install -y instead of apt install |
| 3696 | which warns about that and run apt-get update first |
| 3697 | - Enable MQTT for Windows msys2 builds instead of |
| 3698 | legacy msys1 builds |
| 3699 | - Add ./configure --prefix parameter to the msys2 builds |
| 3700 | - The MSYSTEM environment variable is now preset inside |
| 3701 | the container images for the msys2 builds |
| 3702 | |
| 3703 | Note: on Azure Pipelines the matrix strategy is basically |
| 3704 | just a simple list of job copies and not really a matrix. |
| 3705 | |
| 3706 | Closes #5468 |
| 3707 | |
| 3708 | Daniel Stenberg (30 May 2020) |
| 3709 | - build: disable more code/data when built without proxy support |
| 3710 | |
| 3711 | Added build to travis to verify |
| 3712 | |
| 3713 | Closes #5466 |
| 3714 | |
| 3715 | - url: alloc the download buffer at transfer start |
| 3716 | |
| 3717 | ... and free it as soon as the transfer is done. It removes the extra |
| 3718 | alloc when a new size is set with setopt() and reduces memory for unused |
| 3719 | easy handles. |
| 3720 | |
| 3721 | In addition: the closure_handle now doesn't use an allocated buffer at |
| 3722 | all but the smallest supported size as a stack based one. |
| 3723 | |
| 3724 | Closes #5472 |
| 3725 | |
| 3726 | - timeouts: change millisecond timeouts to timediff_t from time_t |
| 3727 | |
| 3728 | For millisecond timers we like timediff_t better. Also, time_t can be |
| 3729 | unsigned so returning a negative value doesn't work then. |
| 3730 | |
| 3731 | Closes #5479 |
| 3732 | |
| 3733 | Marc Hoersken (30 May 2020) |
| 3734 | - select: add overflow checks for timeval conversions |
| 3735 | |
| 3736 | Using time_t and suseconds_t if suseconds_t is available, |
| 3737 | long on Windows (maybe others in the future) and int elsewhere. |
| 3738 | |
| 3739 | Also handle case of ULONG_MAX being greater or equal to INFINITE. |
| 3740 | |
| 3741 | Assisted-by: Jay Satiro |
| 3742 | Reviewed-by: Daniel Stenberg |
| 3743 | |
| 3744 | Part of #5343 |
| 3745 | |
| 3746 | - select: use timediff_t instead of time_t and int for timeout_ms |
| 3747 | |
| 3748 | Make all functions in select.[ch] take timeout_ms as timediff_t |
| 3749 | which should always be large enough and signed on all platforms |
| 3750 | to take all possible timeout values and avoid type conversions. |
| 3751 | |
| 3752 | Reviewed-by: Jay Satiro |
| 3753 | Reviewed-by: Daniel Stenberg |
| 3754 | |
| 3755 | Replaces #5107 and partially #5262 |
| 3756 | Related to #5240 and #5286 |
| 3757 | Closes #5343 |
| 3758 | |
| 3759 | - unit1604.c: fix implicit conv from 'SANITIZEcode' to 'CURLcode' |
| 3760 | |
| 3761 | GCC 10 warns about this with warning: implicit conversion |
| 3762 | from 'SANITIZEcode' to 'CURLcode' [-Wenum-conversion] |
| 3763 | |
| 3764 | Since 'expected_result' is not really of type 'CURLcode' and |
| 3765 | it is not exposed in any way, we can just use 'SANITIZEcode'. |
| 3766 | |
| 3767 | Reviewed-by: Daniel Stenberg |
| 3768 | Reviewed-by: Marcel Raad |
| 3769 | |
| 3770 | Closes #5476 |
| 3771 | |
| 3772 | - tests/libtest: fix undefined reference to 'curlx_win32_fopen' |
| 3773 | |
| 3774 | Since curl_setup.h now makes use of curlx_win32_fopen for Win32 |
| 3775 | builds with USE_WIN32_LARGE_FILES or USE_WIN32_SMALL_FILES defined, |
| 3776 | we need to include the relevant files for tests using fopen, |
| 3777 | because the libtest sources are also including curl_setup.h |
| 3778 | |
| 3779 | Reviewed-by: Marcel Raad |
| 3780 | Reviewed-by: Daniel Stenberg |
| 3781 | |
| 3782 | Follow up to #3784 (ffdddb45d9) |
| 3783 | Closes #5475 |
| 3784 | |
| 3785 | - appveyor: add non-debug plain autotools-based build |
| 3786 | |
| 3787 | This should enable us to catch linking issues with the |
| 3788 | testsuite early, like the one described/fixed in #5475. |
| 3789 | |
| 3790 | Reviewed-by: Daniel Stenberg |
| 3791 | Reviewed-by: Marcel Raad |
| 3792 | |
| 3793 | Closes #5477 |
| 3794 | |
| 3795 | Daniel Stenberg (29 May 2020) |
| 3796 | - RELEASE-NOTES: synced |
| 3797 | |
| 3798 | - Revert "buildconf: use find -execdir" |
| 3799 | |
| 3800 | This partially reverts commit c712009838f44211958854de431315586995bc61. |
| 3801 | |
| 3802 | Keep the ares_ files removed but bring back the older way to run find, |
| 3803 | to make it work with busybox's find, as apparently that's being used. |
| 3804 | |
| 3805 | Reported-by: Max Peal |
| 3806 | Fixes #5483 |
| 3807 | Closes #5484 |
| 3808 | |
| 3809 | - server/sws: fix asan warning on use of uninitialized variable |
| 3810 | |
| 3811 | - libssh2: improved error output for wrong quote syntax |
| 3812 | |
| 3813 | Reported-by: Werner Stolz |
| 3814 | |
| 3815 | Closes #5474 |
| 3816 | |
| 3817 | - mk-lib1521: generate code for testing BLOB options as well |
| 3818 | |
| 3819 | Follow-up to cac5374298b3 |
| 3820 | |
| 3821 | Closes #5478 |
| 3822 | |
| 3823 | - configure: repair the check if argv can be written to |
| 3824 | |
| 3825 | Due to bad escaping of the test code, the test wouldn't build and thus |
| 3826 | result in a negative test result, which would lead to the unconditional |
| 3827 | assumption that overwriting the arguments doesn't work and thus curl |
| 3828 | would never hide credentials given in the command line, even when it |
| 3829 | would otherwise be possible. |
| 3830 | |
| 3831 | Regression from commit 2d4c2152c (7.60.0) |
| 3832 | |
| 3833 | Reported-by: huzunhao on github |
| 3834 | Fixes #5470 |
| 3835 | Closes #5471 |
| 3836 | |
| 3837 | Peter Wu (28 May 2020) |
| 3838 | - CMake: rebuild Makefile.inc.cmake when Makefile.inc changes |
| 3839 | |
| 3840 | Otherwise the build might fail due to missing source files, as |
| 3841 | demonstrated by the recent keylog.c addition on an existing build dir. |
| 3842 | |
| 3843 | Closes #5469 |
| 3844 | |
| 3845 | Daniel Stenberg (28 May 2020) |
| 3846 | - urldata: fix comments: Curl_done() is called multi_done() now |
| 3847 | |
| 3848 | ... since 575e885db |
| 3849 | |
| 3850 | Peter Wu (27 May 2020) |
| 3851 | - ngtcp2: use common key log routine for better thread-safety |
| 3852 | |
| 3853 | Tested with ngtcp2 built against the OpenSSL library. Additionally |
| 3854 | tested with MultiSSL (NSS for TLS and ngtcp2+OpenSSL for QUIC). |
| 3855 | |
| 3856 | The TLS backend (independent of QUIC) may or may not already have opened |
| 3857 | the keylog file before. Therefore Curl_tls_keylog_open is always called |
| 3858 | to ensure the file is open. |
| 3859 | |
| 3860 | - wolfssl: add SSLKEYLOGFILE support |
| 3861 | |
| 3862 | Tested following the same curl and tshark commands as in commit |
| 3863 | "vtls: Extract and simplify key log file handling from OpenSSL" using |
| 3864 | WolfSSL v4.4.0-stable-128-g5179503e8 from git master built with |
| 3865 | `./configure --enable-all --enable-debug CFLAGS=-DHAVE_SECRET_CALLBACK`. |
| 3866 | |
| 3867 | Full support for this feature requires certain wolfSSL build options, |
| 3868 | see "Availability note" in lib/vtls/wolfssl.c for details. |
| 3869 | |
| 3870 | Closes #5327 |
| 3871 | |
| 3872 | - vtls: Extract and simplify key log file handling from OpenSSL |
| 3873 | |
| 3874 | Create a set of routines for TLS key log file handling to enable reuse |
| 3875 | with other TLS backends. Simplify the OpenSSL backend as follows: |
| 3876 | |
| 3877 | - Drop the ENABLE_SSLKEYLOGFILE macro as it is unconditionally enabled. |
| 3878 | - Do not perform dynamic memory allocation when preparing a log entry. |
| 3879 | Unless the TLS specifications change we can suffice with a reasonable |
| 3880 | fixed-size buffer. |
| 3881 | - Simplify state tracking when SSL_CTX_set_keylog_callback is |
| 3882 | unavailable. My original sslkeylog.c code included this tracking in |
| 3883 | order to handle multiple calls to SSL_connect and detect new keys |
| 3884 | after renegotiation (via SSL_read/SSL_write). For curl however we can |
| 3885 | be sure that a single master secret eventually becomes available |
| 3886 | after SSL_connect, so a simple flag is sufficient. An alternative to |
| 3887 | the flag is examining SSL_state(), but this seems more complex and is |
| 3888 | not pursued. Capturing keys after server renegotiation was already |
| 3889 | unsupported in curl and remains unsupported. |
| 3890 | |
| 3891 | Tested with curl built against OpenSSL 0.9.8zh, 1.0.2u, and 1.1.1f |
| 3892 | (`SSLKEYLOGFILE=keys.txt curl -vkso /dev/null https://localhost:4433`) |
| 3893 | against an OpenSSL 1.1.1f server configured with: |
| 3894 | |
| 3895 | # Force non-TLSv1.3, use TLSv1.0 since 0.9.8 fails with 1.1 or 1.2 |
| 3896 | openssl s_server -www -tls1 |
| 3897 | # Likewise, but fail the server handshake. |
| 3898 | openssl s_server -www -tls1 -Verify 2 |
| 3899 | # TLS 1.3 test. No need to test the failing server handshake. |
| 3900 | openssl s_server -www -tls1_3 |
| 3901 | |
| 3902 | Verify that all secrets (1 for TLS 1.0, 4 for TLS 1.3) are correctly |
| 3903 | written using Wireshark. For the first and third case, expect four |
| 3904 | matches per connection (decrypted Server Finished, Client Finished, HTTP |
| 3905 | Request, HTTP Response). For the second case where the handshake fails, |
| 3906 | expect a decrypted Server Finished only. |
| 3907 | |
| 3908 | tshark -i lo -pf tcp -otls.keylog_file:keys.txt -Tfields \ |
| 3909 | -eframe.number -eframe.time -etcp.stream -e_ws.col.Info \ |
| 3910 | -dtls.port==4433,http -ohttp.desegment_body:FALSE \ |
| 3911 | -Y 'tls.handshake.verify_data or http' |
| 3912 | |
| 3913 | A single connection can easily be identified via the `tcp.stream` field. |
| 3914 | |
| 3915 | Daniel Stenberg (27 May 2020) |
| 3916 | - FILEFORMAT: add more features that tests can depend on |
| 3917 | |
| 3918 | - [Michael Kaufmann brought this change] |
| 3919 | |
| 3920 | transfer: close connection after excess data has been read |
| 3921 | |
| 3922 | For HTTP 1.x, it's a protocol error when the server sends more bytes |
| 3923 | than announced. If this happens, don't reuse the connection, because the |
| 3924 | start position of the next response is undefined. |
| 3925 | |
| 3926 | Closes #5440 |
| 3927 | |
| 3928 | - [Estanislau Augé-Pujadas brought this change] |
| 3929 | |
| 3930 | Revert "ssh: ignore timeouts during disconnect" |
| 3931 | |
| 3932 | This reverts commit f31760e63b4e9ef1eb25f8f211390f8239388515. Shipped in |
| 3933 | curl 7.54.1. |
| 3934 | |
| 3935 | Bug: https://curl.haxx.se/mail/lib-2020-05/0068.html |
| 3936 | Closes #5465 |
| 3937 | |
| 3938 | - urldata: connect related booleans live in struct ConnectBits |
| 3939 | |
| 3940 | And remove a few unused booleans! |
| 3941 | |
| 3942 | Closes #5461 |
| 3943 | |
| 3944 | - hostip: on macOS avoid DoH when given a numerical IP address |
| 3945 | |
| 3946 | When USE_RESOLVE_ON_IPS is set (defined on macOS), it means that |
| 3947 | numerical IP addresses still need to get "resolved" - but not with DoH. |
| 3948 | |
| 3949 | Reported-by: Viktor Szakats |
| 3950 | Fixes #5454 |
| 3951 | Closes #5459 |
| 3952 | |
| 3953 | - ngtcp2: cleanup memory when failing to connect |
| 3954 | |
| 3955 | Reported-by: Peter Wu |
| 3956 | Fixes #5447 (the ngtcp2 side of it) |
| 3957 | Closes #5451 |
| 3958 | |
| 3959 | - quiche: clean up memory properly when failing to connect |
| 3960 | |
| 3961 | Addresses the quiche side of #5447 |
| 3962 | Reported-by: Peter Wu |
| 3963 | Closes #5450 |
| 3964 | |
| 3965 | - cleanup: use a single space after equals sign in assignments |
| 3966 | |
| 3967 | - url: accept "any length" credentials for proxy auth |
| 3968 | |
| 3969 | They're only limited to the maximum string input restrictions, not to |
| 3970 | 256 bytes. |
| 3971 | |
| 3972 | Added test 1178 to verify |
| 3973 | |
| 3974 | Reported-by: Will Roberts |
| 3975 | Fixes #5448 |
| 3976 | Closes #5449 |
| 3977 | |
| 3978 | - [Maksim Stsepanenka brought this change] |
| 3979 | |
| 3980 | test1167: fixes in badsymbols.pl |
| 3981 | |
| 3982 | Closes #5442 |
| 3983 | |
| 3984 | - altsvc: fix parser for lines ending with CRLF |
| 3985 | |
| 3986 | Fixed the alt-svc parser to treat a newline as end of line. |
| 3987 | |
| 3988 | The unit tests in test 1654 were done without CRLF and thus didn't quite |
| 3989 | match the real world. Now they use CRLF as well. |
| 3990 | |
| 3991 | Reported-by: Peter Wu |
| 3992 | Assisted-by: Peter Wu |
| 3993 | Assisted-by: Jay Satiro |
| 3994 | Fixes #5445 |
| 3995 | Closes #5446 |
| 3996 | |
| 3997 | Viktor Szakats (25 May 2020) |
| 3998 | - all: fix codespell errors |
| 3999 | |
| 4000 | Reviewed-by: Jay Satiro |
| 4001 | Reviewed-by: Daniel Stenberg |
| 4002 | Closes https://github.com/curl/curl/pull/5452 |
| 4003 | |
| 4004 | Peter Wu (25 May 2020) |
| 4005 | - ngtcp2: fix build with current ngtcp2 master implementing draft 28 |
| 4006 | |
| 4007 | Based on client.cc changes from ngtcp2. Tested with current git master, |
| 4008 | ngtcp2 commit c77d5731ce92, nghttp3 commit 65ff479d4380. |
| 4009 | |
| 4010 | Fixes #5444 |
| 4011 | Closes #5443 |
| 4012 | |
| 4013 | Daniel Stenberg (25 May 2020) |
| 4014 | - RELEASE-NOTES: synced |
| 4015 | |
| 4016 | moved the new setopts up to a "change" |
| 4017 | |
| 4018 | - RELEASE-NOTES: synced |
| 4019 | |
| 4020 | - copyright: updated year ranges out of sync |
| 4021 | |
| 4022 | ... and whitelisted a few more files in the the copyright.pl script. |
| 4023 | |
| 4024 | - [Gilles Vollant brought this change] |
| 4025 | |
| 4026 | setopt: add CURLOPT_PROXY_ISSUERCERT(_BLOB) for coherency |
| 4027 | |
| 4028 | Closes #5431 |
| 4029 | |
| 4030 | - curl: remove -J "informational" written on stdout |
| 4031 | |
| 4032 | curl would previously show "curl: Saved to filename 'name from header'" |
| 4033 | if -J was used and a name was picked from the Content-Disposition |
| 4034 | header. That output could interfer with other stdout output, such as -w. |
| 4035 | |
| 4036 | This commit removes that output line. |
| 4037 | Bug: https://curl.haxx.se/mail/archive-2020-05/0044.html |
| 4038 | Reported-by: Коваленко Анатолий Викторович |
| 4039 | Closes #5435 |
| 4040 | |
| 4041 | Peter Wu (22 May 2020) |
| 4042 | - travis: simplify quiche build instructions wrt boringssl |
| 4043 | |
| 4044 | quiche builds boringssl as static library, reuse that instead of |
| 4045 | building another shared library. |
| 4046 | |
| 4047 | Closes #5438 |
| 4048 | |
| 4049 | - configure: fix pthread check with static boringssl |
| 4050 | |
| 4051 | A shared boringssl/OpenSSL library requires -lcrypto only for linking. |
| 4052 | A static build additionally requires `-ldl -lpthread`. In the latter |
| 4053 | case `-lpthread` is added to LIBS which prevented `-pthread` from being |
| 4054 | added to CFLAGS. Clear LIBS to fix linking failures for libtest tests. |
| 4055 | |
| 4056 | Daniel Stenberg (22 May 2020) |
| 4057 | - Revert "sendf: make failf() use the mvsnprintf() return code" |
| 4058 | |
| 4059 | This reverts commit 74623551f306990e70c7c5515b88972005604a74. |
| 4060 | |
| 4061 | Instead mark the function call with (void). Getting the return code and |
| 4062 | using it instead triggered Coverity warning CID 1463596 because |
| 4063 | snprintf() can return a negative value... |
| 4064 | |
| 4065 | Closes #5441 |
| 4066 | |
| 4067 | - typecheck-gcc.h: CURLINFO_PRIVATE does not need a 'char *' |
| 4068 | |
| 4069 | Reported-by: Billyzou0741326 on github |
| 4070 | Fixes #5432 |
| 4071 | Closes #5436 |
| 4072 | |
| 4073 | - tests/server/util.h: add extern to silence compiler warning |
| 4074 | |
| 4075 | Follow-up from a3b0699d5c1 |
| 4076 | |
| 4077 | - typecheck-gcc.h: fix the OFF_T check |
| 4078 | |
| 4079 | The option number also needs to be less than CURLOPTTYPE_BLOB. |
| 4080 | |
| 4081 | Follow-up to cac5374298 |
| 4082 | Reported-by: Jeroen Ooms |
| 4083 | Bug: https://github.com/curl/curl/pull/5365#issuecomment-631084114 |
| 4084 | |
| 4085 | - TODO: --dry-run |
| 4086 | |
| 4087 | Closes #5426 |
| 4088 | |
| 4089 | - TODO: Ratelimit or wait between serial requests |
| 4090 | |
| 4091 | Closes #5406 |
| 4092 | |
| 4093 | - tool_paramhlp: fixup C89 mistake |
| 4094 | |
| 4095 | Follow-up to c5f0a9db22. |
| 4096 | |
| 4097 | - [Siva Sivaraman brought this change] |
| 4098 | |
| 4099 | tool_paramhlp: fixed potentially uninitialized strtol() variable |
| 4100 | |
| 4101 | Seems highly unlikely to actually be possible, but better safe than |
| 4102 | sorry. |
| 4103 | |
| 4104 | Closes #5417 |
| 4105 | |
| 4106 | - [Siva Sivaraman brought this change] |
| 4107 | |
| 4108 | tool_operate: fixed potentially uninitialized variables |
| 4109 | |
| 4110 | ... in curl_easy_getinfo() calls. They're harmless but clearing the |
| 4111 | variables makes the code safer and comforts the reader. |
| 4112 | |
| 4113 | Closes #5416 |
| 4114 | |
| 4115 | - sha256: move assign to the declaration line |
| 4116 | |
| 4117 | Follow-up to fae30656. Should've been squashed with that commit... |
| 4118 | |
| 4119 | - [Siva Sivaraman brought this change] |
| 4120 | |
| 4121 | sha256: fixed potentially uninitialized variable |
| 4122 | |
| 4123 | Closes #5414 |
| 4124 | |
| 4125 | - sendf: make failf() use the mvsnprintf() return code |
| 4126 | |
| 4127 | ... and avoid a strlen() call. Fixes a MonocleAI warning. |
| 4128 | |
| 4129 | Reported-by: MonocleAI |
| 4130 | Fixes #5413 |
| 4131 | Closes #5420 |
| 4132 | |
| 4133 | - hostip: make Curl_printable_address not return anything |
| 4134 | |
| 4135 | It was not used much anyway and instead we let it store a blank buffer |
| 4136 | in case of failure. |
| 4137 | |
| 4138 | Reported-by: MonocleAI |
| 4139 | Fixes #5411 |
| 4140 | Closes #5418 |
| 4141 | |
| 4142 | - ftp: mark return-ignoring calls to Curl_GetFTPResponse with (void) |
| 4143 | |
| 4144 | They're done on purpose, make that visible in the code. |
| 4145 | Reported-by: MonocleAI |
| 4146 | Fixes #5412 |
| 4147 | Closes #549 |
| 4148 | |
| 4149 | - TODO: forbid TLS post-handshake auth and do TLS record padding |
| 4150 | |
| 4151 | Closes #5396 |
| 4152 | Closes #5398 |
| 4153 | |
| 4154 | - RELEASE-NOTES: synced |
| 4155 | |
| 4156 | - dynbuf: return NULL when there's no buffer length |
| 4157 | |
| 4158 | ... as returning a "" is not a good idea as the string is supposed to be |
| 4159 | allocated and returning a const string will cause issues. |
| 4160 | |
| 4161 | Reported-by: Brian Carpenter |
| 4162 | Follow-up to ed35d6590e72c |
| 4163 | Closes #5405 |
| 4164 | |
| 4165 | Peter Wu (16 May 2020) |
| 4166 | - travis: upgrade to bionic, clang-9, improve readability |
| 4167 | |
| 4168 | Changes, partially to reduce build failures from external dependencies: |
| 4169 | - Upgrade Ubuntu and drop unnecessary third-party repos. |
| 4170 | - Properly clone apt config to ensure retries. |
| 4171 | - Upgrade to clang-9 from the standard repos. |
| 4172 | - Use Ubuntu 20.04 focal for the libssh build, use of ssh_get_publickey |
| 4173 | fails on -Werror=deprecated-declarations in Ubuntu 18.04. Do not use |
| 4174 | focal everywhere yet since Travis CI has not documented this option. |
| 4175 | In focal, python-impacket (Py2.7) has been removed, leaving only |
| 4176 | python3-impacket. Since it is only needed for SMB tests and not SSH, |
| 4177 | skip it for the libssh job since it might need more work. |
| 4178 | - apt: Remove gcc-8 and libstdc++-8-dev, already installed via g++-8. |
| 4179 | |
| 4180 | Non-functional cleanups: |
| 4181 | - Simplify test matrix, drop redundant os and compiler keys. |
| 4182 | - Deprecation fixes: remove sudo, rename matrix -> jobs. |
| 4183 | - Every job has an 'env' key, put this key first in a list item. |
| 4184 | |
| 4185 | Closes #5370 |
| 4186 | |
| 4187 | - travis: whitespace-only changes for consistency |
| 4188 | |
| 4189 | Automatically apply a consistent indentation with: |
| 4190 | |
| 4191 | python3 -c 'from ruamel.yaml import YAML;y=YAML();d=y.load(open(".travis.yml"));y.width=500;y.dump(d,open(".travis.yml.new","w"))' |
| 4192 | |
| 4193 | followed by manually re-indenting three comments. |
| 4194 | |
| 4195 | Closes #5370 |
| 4196 | |
| 4197 | - CMake: add libssh build support |
| 4198 | |
| 4199 | Closes #5372 |
| 4200 | |
| 4201 | Daniel Stenberg (15 May 2020) |
| 4202 | - KNOWN_BUGS: wolfssh: publickey auth doesn't work |
| 4203 | |
| 4204 | Closes #4820 |
| 4205 | |
| 4206 | - KNOWN_BUGS: OS400 port requires deprecated IBM library |
| 4207 | |
| 4208 | Closes #5176 |
| 4209 | |
| 4210 | - [Vyron Tsingaras brought this change] |
| 4211 | |
| 4212 | http2: keep trying to send pending frames after req.upload_done |
| 4213 | |
| 4214 | Fixes #1410 |
| 4215 | Closes #5401 |
| 4216 | |
| 4217 | - [Gilles Vollant brought this change] |
| 4218 | |
| 4219 | setopt: support certificate options in memory with struct curl_blob |
| 4220 | |
| 4221 | This change introduces a generic way to provide binary data in setopt |
| 4222 | options, called BLOBs. |
| 4223 | |
| 4224 | This change introduces these new setopts: |
| 4225 | |
| 4226 | CURLOPT_ISSUERCERT_BLOB, CURLOPT_PROXY_SSLCERT_BLOB, |
| 4227 | CURLOPT_PROXY_SSLKEY_BLOB, CURLOPT_SSLCERT_BLOB and CURLOPT_SSLKEY_BLOB. |
| 4228 | |
| 4229 | Reviewed-by: Daniel Stenberg |
| 4230 | Closes #5357 |
| 4231 | |
| 4232 | - source cleanup: remove all custom typedef structs |
| 4233 | |
| 4234 | - Stick to a single unified way to use structs |
| 4235 | - Make checksrc complain on 'typedef struct {' |
| 4236 | - Allow them in tests, public headers and examples |
| 4237 | |
| 4238 | - Let MD4_CTX, MD5_CTX, and SHA256_CTX typedefs remain as they actually |
| 4239 | typedef different types/structs depending on build conditions. |
| 4240 | |
| 4241 | Closes #5338 |
| 4242 | |
| 4243 | - travis: remove the .checksrc fiddling |
| 4244 | |
| 4245 | - ftp: make domore_getsock() return the secondary socket properly |
| 4246 | |
| 4247 | Previously, after PASV and immediately after the data connection has |
| 4248 | connected, the function would only return the control socket to wait for |
| 4249 | which then made the data connection simply timeout and not get polled |
| 4250 | correctly. This become obvious when running test 1631 and 1632 event- |
| 4251 | based. |
| 4252 | |
| 4253 | - test1632: verify FTP through HTTPS-proxy with connection re-use |
| 4254 | |
| 4255 | - test1631: verify FTP download through HTTPS-proxy |
| 4256 | |
| 4257 | - sws: as last resort, get test number from server cmd file |
| 4258 | |
| 4259 | If it can't be found in the request. Also support --cmdfile to set it to |
| 4260 | a custom file name. |
| 4261 | |
| 4262 | runtests.pl always writes this file with the test number in it since a |
| 4263 | while back. |
| 4264 | |
| 4265 | - ftp: shut down the secondary connection properly when SSL is used |
| 4266 | |
| 4267 | Reported-by: Neal Poole |
| 4268 | Fixes #5340 |
| 4269 | Closes #5385 |
| 4270 | |
| 4271 | Marcel Raad (14 May 2020) |
| 4272 | - KNOWN_BUGS: adapt 5.5 to recent changes |
| 4273 | |
| 4274 | It only applies to non-Unicode builds now. |
| 4275 | Also merge 5.10 into it as it's effectively a duplicate. |
| 4276 | |
| 4277 | Closes https://github.com/curl/curl/pull/3784 |
| 4278 | |
| 4279 | - curl_setup: support Unicode functions to open files on Windows |
| 4280 | |
| 4281 | Use them only if `_UNICODE` is defined, in which case command-line |
| 4282 | arguments have been converted to UTF-8. |
| 4283 | |
| 4284 | Closes https://github.com/curl/curl/pull/3784 |
| 4285 | |
| 4286 | - tool: support UTF-16 command line on Windows |
| 4287 | |
| 4288 | - use `wmain` instead of `main` when `_UNICODE` is defined [0] |
| 4289 | - define `argv_item_t` as `wchar_t *` in this case |
| 4290 | - use the curl_multibyte gear to convert the command-line arguments to |
| 4291 | UTF-8 |
| 4292 | |
| 4293 | This makes it possible to pass parameters with characters outside of |
| 4294 | the current locale on Windows, which is required for some tests, e.g. |
| 4295 | the IDN tests. Out of the box, this currently only works with the |
| 4296 | Visual Studio project files, which default to Unicode, and winbuild |
| 4297 | with the `ENABLE_UNICODE` option. |
| 4298 | |
| 4299 | [0] https://devblogs.microsoft.com/oldnewthing/?p=40643 |
| 4300 | |
| 4301 | Ref: https://github.com/curl/curl/issues/3747 |
| 4302 | Closes https://github.com/curl/curl/pull/3784 |
| 4303 | |
| 4304 | - curl_multibyte: add to curlx |
| 4305 | |
| 4306 | This will also be needed in the tool and tests. |
| 4307 | |
| 4308 | Ref: https://github.com/curl/curl/pull/3758#issuecomment-482197512 |
| 4309 | Closes https://github.com/curl/curl/pull/3784 |
| 4310 | |
| 4311 | Daniel Stenberg (14 May 2020) |
| 4312 | - url: make the updated credentials URL-encoded in the URL |
| 4313 | |
| 4314 | Found-by: Gregory Jefferis |
| 4315 | Reported-by: Jeroen Ooms |
| 4316 | Added test 1168 to verify. Bug spotted when doing a redirect. |
| 4317 | Bug: https://github.com/jeroen/curl/issues/224 |
| 4318 | Closes #5400 |
| 4319 | |
| 4320 | - tests: add https-proxy support to the test suite |
| 4321 | |
| 4322 | Initial test 1630 added with basic HTTPS-proxy use. HTTPS-proxy is like |
| 4323 | HTTP proxy but with a full TLS connection to the proxy. |
| 4324 | |
| 4325 | Closes #5399 |
| 4326 | |
| 4327 | - mailmap: James Fuller |
| 4328 | |
| 4329 | - [Major_Tom brought this change] |
| 4330 | |
| 4331 | vauth/cleartext: fix theoretical integer overflow |
| 4332 | |
| 4333 | Fix theoretical integer overflow in Curl_auth_create_plain_message. |
| 4334 | |
| 4335 | The security impact of the overflow was discussed on hackerone. We |
| 4336 | agreed this is more of a theoretical vulnerability, as the integer |
| 4337 | overflow would only be triggerable on systems using 32-bits size_t with |
| 4338 | over 4GB of available memory space for the process. |
| 4339 | |
| 4340 | Closes #5391 |
| 4341 | |
| 4342 | Jay Satiro (13 May 2020) |
| 4343 | - curl.1: Quote globbed URLs |
| 4344 | |
| 4345 | - Quote the globbing example URLs that contain characters [] {} since |
| 4346 | otherwise they may be interpreted as shell metacharacters. |
| 4347 | |
| 4348 | Bug: https://github.com/curl/curl/issues/5388 |
| 4349 | Reported-by: John Simpson |
| 4350 | |
| 4351 | Closes https://github.com/curl/curl/pull/5394 |
| 4352 | |
| 4353 | Daniel Stenberg (14 May 2020) |
| 4354 | - checksrc: enhance the ASTERISKSPACE and update code accordingly |
| 4355 | |
| 4356 | Fine: "struct hello *world" |
| 4357 | |
| 4358 | Not fine: "struct hello* world" (and variations) |
| 4359 | |
| 4360 | Closes #5386 |
| 4361 | |
| 4362 | - docs/options-in-versions: which version added each cmdline option |
| 4363 | |
| 4364 | Added test 971 to verify that the list is in sync with the files in |
| 4365 | cmdline-opts. The check also verifies that .d-files that uses Added: |
| 4366 | specify the same version number as the options-in-versions file does. |
| 4367 | |
| 4368 | Closes #5381 |
| 4369 | |
| 4370 | - docs: unify protocol lists |
| 4371 | |
| 4372 | We boast support for 25 transfer protocols. Make sure the lists are |
| 4373 | consistent |
| 4374 | |
| 4375 | Closes #5384 |
| 4376 | |
| 4377 | - OpenSSL: have CURLOPT_CRLFILE imply CURLSSLOPT_NO_PARTIALCHAIN |
| 4378 | |
| 4379 | ... to avoid an OpenSSL bug that otherwise makes the CRL check to fail. |
| 4380 | |
| 4381 | Reported-by: Michael Kaufmann |
| 4382 | Fixes #5374 |
| 4383 | Closes #5376 |
| 4384 | |
| 4385 | - tls13-ciphers.d: shorten the Arg |
| 4386 | |
| 4387 | - sasl-authzid.d: add Arg: and shorten the desc |
| 4388 | |
| 4389 | - cert-type.d: mention the available types in the desc |
| 4390 | |
| 4391 | - tool: shorten 3 --help descriptions |
| 4392 | |
| 4393 | --happy-eyeballs-timeout-ms, --resolve and --ssl-revoke-best-effort |
| 4394 | |
| 4395 | gen.pl already warned about these lines but we didn't listen |
| 4396 | |
| 4397 | Closes #5379 |
| 4398 | |
| 4399 | - configure: the wolfssh backend does not provide SCP |
| 4400 | |
| 4401 | Closes #5387 |
| 4402 | |
| 4403 | - RELEASE-NOTES: synced |
| 4404 | |
| 4405 | - url: reject too long input when parsing credentials |
| 4406 | |
| 4407 | Since input passed to libcurl with CURLOPT_USERPWD and |
| 4408 | CURLOPT_PROXYUSERPWD circumvents the regular string length check we have |
| 4409 | in Curl_setstropt(), the input length limit is enforced in |
| 4410 | Curl_parse_login_details too, separately. |
| 4411 | |
| 4412 | Reported-by: Thomas Bouzerar |
| 4413 | Closes #5383 |
| 4414 | |
| 4415 | - list-only.d: this option existed already in 4.0 |
| 4416 | |
| 4417 | Jay Satiro (12 May 2020) |
| 4418 | - retry-all-errors.d: Shorten the summary line |
| 4419 | |
| 4420 | Follow-up to b995bb5 from a few moments ago. |
| 4421 | |
| 4422 | Reported-by: Daniel Stenberg |
| 4423 | |
| 4424 | Ref: https://github.com/curl/curl/commit/b995bb5#r39108929 |
| 4425 | |
| 4426 | - [denzor brought this change] |
| 4427 | |
| 4428 | easy: fix dangling pointer on easy_perform fail |
| 4429 | |
| 4430 | Closes https://github.com/curl/curl/pull/5363 |
| 4431 | |
| 4432 | - tool: Add option --retry-all-errors to retry on any error |
| 4433 | |
| 4434 | The "sledgehammer" of retrying. |
| 4435 | |
| 4436 | Closes https://github.com/curl/curl/pull/5185 |
| 4437 | |
| 4438 | Daniel Stenberg (12 May 2020) |
| 4439 | - [James Le Cuirot brought this change] |
| 4440 | |
| 4441 | libcurl.pc: Merge Libs.private into Libs for static-only builds |
| 4442 | |
| 4443 | A project being built entirely statically will call pkg-config with |
| 4444 | --static, which utilises the Libs.private field. Conversely it will |
| 4445 | not use --static when not being built entirely statically, even if |
| 4446 | there is only a static build of libcurl available. This will most |
| 4447 | likely cause the build to fail due to underlinking unless we merge the |
| 4448 | Libs fields. |
| 4449 | |
| 4450 | Consider that this is what the Meson build system does when it |
| 4451 | generates pkg-config files. |
| 4452 | |
| 4453 | I have also reflected this in the --libs argument of curl-config even |
| 4454 | though REQUIRE_LIB_DEPS always seems to be "yes" anyway. |
| 4455 | |
| 4456 | Closes #5373 |
| 4457 | |
| 4458 | - [Peter Wu brought this change] |
| 4459 | |
| 4460 | CMake: fix runtests.pl with CMake, add new test targets |
| 4461 | |
| 4462 | * runtests.pl: |
| 4463 | - Fix out-of-tree build under CMake when srcdir is not set. Default |
| 4464 | srcdir to the location of runtests.pl. |
| 4465 | - Add a hack to allow CMake to use the TFLAGS option as documented |
| 4466 | in tests/README and used in scripts/travis/script.sh. |
| 4467 | * Bump CMake version to 3.2 for USES_TERMINAL, dropping Debian Jessie |
| 4468 | support (no one should care, it is already EOL.). |
| 4469 | * Remove CTest since it defines its own 'test' target with no tests |
| 4470 | since all unittests are already broken and not built by default. |
| 4471 | * Add new test targets based on the options from Makefile.am. Since |
| 4472 | new test targets are rarely added, I opted for duplicating the |
| 4473 | runtests.pl options as opposed to creating a new Makefile.inc file. |
| 4474 | Use top-level target names (test-x) instead of x-test since that is |
| 4475 | used by CI and others. |
| 4476 | |
| 4477 | Closes #5358 |
| 4478 | |
| 4479 | - [Peter Wu brought this change] |
| 4480 | |
| 4481 | CMake: do not build test programs by default |
| 4482 | |
| 4483 | The default target should only build libcurl and curl. Add a dedicated |
| 4484 | 'testdeps' target which will be used later when running tests. Note that |
| 4485 | unittests are currently broken in CMake and already excluded. |
| 4486 | |
| 4487 | Closes #5368 |
| 4488 | |
| 4489 | - FILEFORMAT: moved up the variables section and further polished |
| 4490 | |
| 4491 | - runtests: remove ftp2 support, not used |
| 4492 | |
| 4493 | We once supported two separate ftp instances in the test suite. Has not |
| 4494 | been used the last decade. |
| 4495 | |
| 4496 | Closes #5375 |
| 4497 | |
| 4498 | - url: sort the protocol schemes in rough popularity order |
| 4499 | |
| 4500 | When looking for a protocol match among supported schemes, check the |
| 4501 | most "popular" schemes first. It has zero functionality difference and |
| 4502 | for all practical purposes a speed difference will not be measureable |
| 4503 | but it still think it makes sense to put the least likely matches last. |
| 4504 | |
| 4505 | "Popularity" based on the 2019 user survey. |
| 4506 | |
| 4507 | Closes #5377 |
| 4508 | |
| 4509 | Marc Hoersken (11 May 2020) |
| 4510 | - test1238: avoid tftpd being busy for tests shortly following |
| 4511 | |
| 4512 | The tftpd server may still be busy if the total timeout of |
| 4513 | 25 seconds has not been reached or no sread error was received |
| 4514 | during or after the execution of the timeout test 1238. |
| 4515 | |
| 4516 | Once the next TFTP test comes around (eg. 1242 or 1243), |
| 4517 | those will fail because the tftpd server is still waiting |
| 4518 | on data from curl due to the UDP protocol being stateless |
| 4519 | and having no connection close. On Linux this error may not |
| 4520 | happen, because ICMP errors generated due to a swrite error |
| 4521 | can also be returned async on the next sread call instead. |
| 4522 | |
| 4523 | Therefore we will now just kill the tftpd server after test |
| 4524 | 1238 to make sure that the following tests are not affected. |
| 4525 | |
| 4526 | This enables us to no longer ignore tests 1242, 1243, 2002 |
| 4527 | and 2003 on the CI platforms CirrusCI and AppVeyor. |
| 4528 | |
| 4529 | Assisted-by: Peter Wu |
| 4530 | Closes #5364 |
| 4531 | |
| 4532 | Daniel Stenberg (11 May 2020) |
| 4533 | - write-out.d: added "response_code" |
| 4534 | |
| 4535 | - KNOWN_BUGS: Build with staticly built dependency |
| 4536 | |
| 4537 | I rewrote the item 5.4 to be more generic about static dependencies. |
| 4538 | |
| 4539 | - ROADMAP: remove old entries |
| 4540 | |
| 4541 | MQTT - the start has already landed |
| 4542 | |
| 4543 | tiny-curl - also mostly landed and is a continuous work |
| 4544 | |
| 4545 | make menuconfig - basically no interest from users, not pushing there |
| 4546 | |
| 4547 | - [Peter Wu brought this change] |
| 4548 | |
| 4549 | travis: Add ngtcp2 and quiche tests for CMake |
| 4550 | |
| 4551 | To avoid an explosion of jobs, extend the existing CMake tests with |
| 4552 | ngtcp2 and quiche support. macOS was previously moved to GitHub actions, |
| 4553 | so the non-Linux case can be dropped. |
| 4554 | |
| 4555 | - [Peter Wu brought this change] |
| 4556 | |
| 4557 | CMake: add ENABLE_ALT_SVC option |
| 4558 | |
| 4559 | Tested alt-svc with quiche. While at it, add missing MultiSSL reporting |
| 4560 | (not tested). |
| 4561 | |
| 4562 | - [Peter Wu brought this change] |
| 4563 | |
| 4564 | CMake: add HTTP/3 support (ngtcp2+nghttp3, quiche) |
| 4565 | |
| 4566 | Add three new CMake Find modules (using the curl license, but I grant |
| 4567 | others the right to apply the CMake BSD license instead). |
| 4568 | |
| 4569 | This CMake config is simpler than the autotools one because it assumes |
| 4570 | ngtcp2 and nghttp3 to be used together. Another difference is that this |
| 4571 | CMake config checks whether QUIC is actually supported by the TLS |
| 4572 | library (patched OpenSSL or boringssl) since this can be a common |
| 4573 | configuration mistake that could result in build errors later. |
| 4574 | |
| 4575 | Unlike autotools, CMake does not warn you that the features are |
| 4576 | experimental. The user is supposed to already know that and read the |
| 4577 | documentation. It requires a very special build environment anyway. |
| 4578 | |
| 4579 | Tested with ngtcp2+OpenSSL+nghttp3 and quiche+boringssl, both built from |
| 4580 | current git master. Use `LD_DEBUG=files src/curl |& grep need` to figure |
| 4581 | out which features (libldap-2.4, libssh2) to disable due to conflicts |
| 4582 | with boringssl. |
| 4583 | |
| 4584 | Closes #5359 |
| 4585 | |
| 4586 | Marc Hoersken (10 May 2020) |
| 4587 | - tests/server/tftpd.c: fix include and enhance debug logging |
| 4588 | |
| 4589 | setjmp.h should only be included if HAVE_SETJMP_H is defined. |
| 4590 | |
| 4591 | Add additional log statements to see wether reads and writes |
| 4592 | are blocking or finishing before an alarm signal is received. |
| 4593 | |
| 4594 | Assisted-by: Peter Wu |
| 4595 | Part of #5364 |
| 4596 | |
| 4597 | Daniel Stenberg (10 May 2020) |
| 4598 | - tool_operate: only set CURLOPT_SSL_OPTIONS if SSL support is present |
| 4599 | |
| 4600 | Reported-by: Marcel Raad |
| 4601 | Follow-up to 148534db5 |
| 4602 | Fixes #5367 |
| 4603 | Closes #5369 |
| 4604 | |
| 4605 | Marc Hoersken (9 May 2020) |
| 4606 | - appveyor: update comments to be clear about toolchain |
| 4607 | |
| 4608 | - CMake-based MSYS builds use mingw-w64 to cross-compile. |
| 4609 | - autotools-based builds are compiled using msys2-devel. |
| 4610 | |
| 4611 | The difference is that the later ones are not cross-compiled |
| 4612 | to Windows and instead require the msys2 runtime to be present. |
| 4613 | |
| 4614 | At the moment only the Azure Pipelines CI builds actually |
| 4615 | run autotools-based cross-compilation builds for Windows. |
| 4616 | |
| 4617 | - TODO: update regarding missing Schannel features |
| 4618 | |
| 4619 | Some aspects have already been implemented over the years. |
| 4620 | |
| 4621 | 15.1 Client certificates are now supported: |
| 4622 | |
| 4623 | - System stores via e35b0256eb34f1fe562e3e2a2615beb50a391c52 |
| 4624 | - PKCS#12 files via 0fdf96512613574591f501d63fe49495ba40e1d5 |
| 4625 | |
| 4626 | 15.2 Ciphers can now be specified through: |
| 4627 | |
| 4628 | - Algorithms via 9aefbff30d280c60fc9d8cc3e0b2f19fc70a2f28 |
| 4629 | |
| 4630 | Reviewed-by: Daniel Stenberg and Marcel Raad |
| 4631 | Closes #5358 |
| 4632 | |
| 4633 | Daniel Stenberg (8 May 2020) |
| 4634 | - checksrc: close the .checksrc file handle when done reading |
| 4635 | |
| 4636 | - RELEASE-NOTES: synced |
| 4637 | |
| 4638 | And bumped next version to 7.71.0 |
| 4639 | |
| 4640 | - [Gilles Vollant brought this change] |
| 4641 | |
| 4642 | CURLOPT_SSL_OPTIONS: add *_NATIVE_CA to use Windows CA store (with openssl) |
| 4643 | |
| 4644 | Closes #4346 |
| 4645 | |
| 4646 | - TODO: native IDN support on macOS |
| 4647 | |
| 4648 | - urlapi: accept :: as a valid IPv6 address |
| 4649 | |
| 4650 | Text 1560 is extended to verify. |
| 4651 | |
| 4652 | Reported-by: Pavel Volgarev |
| 4653 | Fixes #5344 |
| 4654 | Closes #5351 |
| 4655 | |
| 4656 | - THANKS-filter: Peter Wang |
| 4657 | |
| 4658 | - [Peter Wang brought this change] |
| 4659 | |
| 4660 | *_sspi: fix bad uses of CURLE_NOT_BUILT_IN |
| 4661 | |
| 4662 | Return CURLE_AUTH_ERROR instead of CURLE_NOT_BUILT_IN for other |
| 4663 | instances of QuerySecurityPackageInfo failing, as in |
| 4664 | commit 2a81439553286f12cd04a4bdcdf66d8e026d8201. |
| 4665 | |
| 4666 | Closes #5355 |
| 4667 | |
| 4668 | - docs/HTTP3: add qlog to the quiche build instruction |
| 4669 | |
| 4670 | - ngtcp2: introduce qlog support |
| 4671 | |
| 4672 | If the QLOGDIR environment variable is set, enable qlogging. |
| 4673 | |
| 4674 | ... and create Curl_qlogdir() in the new generic vquic/vquic.c file for |
| 4675 | QUIC functions that are backend independent. |
| 4676 | |
| 4677 | Closes #5353 |
| 4678 | |
| 4679 | - ntlm_sspi: fix bad use of CURLE_NOT_BUILT_IN |
| 4680 | |
| 4681 | That return code is reserved for build-time conditional code not being |
| 4682 | present while this was a regular run-time error from a Windows API. |
| 4683 | |
| 4684 | Reported-by: wangp on github |
| 4685 | Fixes #5349 |
| 4686 | Closes #5350 |
| 4687 | |
| 4688 | - runtests: show elapsed test time with higher precision (ms) |
| 4689 | |
| 4690 | - RELEASE-NOTES: synced |
| 4691 | |
| 4692 | - http2: simplify and clean up trailer handling |
| 4693 | |
| 4694 | Triggered by a crash detected by OSS-Fuzz after the dynbuf introduction in |
| 4695 | ed35d6590e72. This should make the trailer handling more straight forward and |
| 4696 | hopefully less error-prone. |
| 4697 | |
| 4698 | Deliver the trailer header to the callback already at receive-time. No |
| 4699 | longer caches the trailers to get delivered at end of stream. |
| 4700 | |
| 4701 | Bug: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=22030 |
| 4702 | Closes #5348 |
| 4703 | |
| 4704 | Marc Hoersken (7 May 2020) |
| 4705 | - appveyor: disable test 1139 instead of ignoring it |
| 4706 | |
| 4707 | Spending time on manpage checking makes no sense |
| 4708 | for these builds due to lacking manpage support. |
| 4709 | |
| 4710 | - appveyor: disable flaky test 1501 and ignore broken 1056 |
| 4711 | |
| 4712 | Test 1501 is flaky on Windows CI due to being time sensitive |
| 4713 | and the testsuite relying on taskkill.exe to check for the |
| 4714 | existance of processes which can take to much time itself. |
| 4715 | |
| 4716 | Test 1056 is broken in autotools-based Windows builds due |
| 4717 | to scope ID support missing in these builds at the moment. |
| 4718 | |
| 4719 | - test613.pl: make tests 613 and 614 work with OpenSSH for Windows |
| 4720 | |
| 4721 | OpenSSH for Windows shows group and other/world permissions as *, |
| 4722 | because those concepts do not exist on Windows. It also does not |
| 4723 | show the current or parent directory, so we just ignore those. |
| 4724 | |
| 4725 | Reviewed-by: Daniel Stenberg |
| 4726 | Closes #5328 |
| 4727 | |
| 4728 | Daniel Stenberg (6 May 2020) |
| 4729 | - runtests: set +x mode again |
| 4730 | |
| 4731 | - libssh2: convert over to use dynbuf |
| 4732 | |
| 4733 | In my very basic test that lists sftp://127.0.0.1/tmp/, this patched |
| 4734 | code makes 161 allocations compared to 194 in git master. A 17% |
| 4735 | reduction. |
| 4736 | |
| 4737 | Closes #5336 |
| 4738 | |
| 4739 | - travis: add "qlog" as feature in the quiche build |
| 4740 | |
| 4741 | - quiche: enable qlog output |
| 4742 | |
| 4743 | quiche has the potential to log qlog files. To enable this, you must |
| 4744 | build quiche with the qlog feature enabled `cargo build --features |
| 4745 | qlog`. curl then passes a file descriptor to quiche, which takes |
| 4746 | ownership of the file. The FD transfer only works on UNIX. |
| 4747 | |
| 4748 | The convention is to enable logging when the QLOGDIR environment is |
| 4749 | set. This should be a path to a folder where files are written with the |
| 4750 | naming template <SCID>.qlog. |
| 4751 | |
| 4752 | Co-authored-by: Lucas Pardue |
| 4753 | Replaces #5337 |
| 4754 | Closes #5341 |
| 4755 | |
| 4756 | - urldata.h: remove #define HEADERSIZE, not used anymore |
| 4757 | |
| 4758 | Follow-up to ed35d6590e72c |
| 4759 | |
| 4760 | - ngtcp2: convert to dynbuf |
| 4761 | |
| 4762 | Closes #5335 |
| 4763 | |
| 4764 | - connect: make happy eyeballs work for QUIC (again) |
| 4765 | |
| 4766 | Follow-up from dbd16c3e256c6c (regression in 7.70.0) |
| 4767 | |
| 4768 | Closes #5334 |
| 4769 | |
| 4770 | - connect: add two asserts to clue code analyzers in a little |
| 4771 | |
| 4772 | - http_proxy: ported to use dynbuf instead of a static size buffer |
| 4773 | |
| 4774 | Removes a 16K static buffer from the easy handle. Simplifies the code. |
| 4775 | |
| 4776 | - dynbuf: introduce internal generic dynamic buffer functions |
| 4777 | |
| 4778 | A common set of functions instead of many separate implementations for |
| 4779 | creating buffers that can grow when appending data to them. Existing |
| 4780 | functionality has been ported over. |
| 4781 | |
| 4782 | In my early basic testing, the total number of allocations seem at |
| 4783 | roughly the same amount as before, possibly a few less. |
| 4784 | |
| 4785 | See docs/DYNBUF.md for a description of the API. |
| 4786 | |
| 4787 | Closes #5300 |
| 4788 | |
| 4789 | - runtests: remove sleep calls |
| 4790 | |
| 4791 | Remove many one second sleeps that were done *after* each newly started |
| 4792 | test server already has been verified. They should not have any purpose |
| 4793 | there. |
| 4794 | |
| 4795 | Closes #5323 |
| 4796 | |
| 4797 | - asyn-*: remove support for never-used NULL entry pointers |
| 4798 | |
| 4799 | ... and instead convert those to asserts to make sure they are truly |
| 4800 | never NULL. |
| 4801 | |
| 4802 | Closes #5324 |
| 4803 | |
| 4804 | - [Emil Engler brought this change] |
| 4805 | |
| 4806 | doc: Rename VERSIONS to VERSIONS.md as it already has Markdown syntax |
| 4807 | |
| 4808 | Closes #5325 |
| 4809 | |
| 4810 | Jay Satiro (2 May 2020) |
| 4811 | - asyn-thread: fix cppcheck warning |
| 4812 | |
| 4813 | - Check for NULL entry parameter before attempting to deref entry in |
| 4814 | Curl_resolver_is_resolved, like is already done in asyn-ares. |
| 4815 | |
| 4816 | This is to silence cppcheck which does not seem to understand that |
| 4817 | asyn-ares and asyn-thread have separate Curl_resolver_is_resolved |
| 4818 | and those units are mutually exclusive. Prior to this change it warned |
| 4819 | of a scenario where asyn-thread's Curl_resolver_is_resolved is called |
| 4820 | with a NULL entry from asyn-ares, but that couldn't happen. |
| 4821 | |
| 4822 | Reported-by: rl1987@users.noreply.github.com |
| 4823 | |
| 4824 | Fixes https://github.com/curl/curl/issues/5326 |
| 4825 | |
| 4826 | - select: fix overflow protection in Curl_socket_check |
| 4827 | |
| 4828 | Follow-up to a96c752 which changed the timeout_ms type from time_t to |
| 4829 | timediff_t. |
| 4830 | |
| 4831 | Ref: https://github.com/curl/curl/pull/5240 |
| 4832 | |
| 4833 | Closes https://github.com/curl/curl/pull/5286 |
| 4834 | |
| 4835 | Marc Hoersken (2 May 2020) |
| 4836 | - sockfilt: make select_ws stop waiting on exit signal event |
| 4837 | |
| 4838 | This makes sure that select_ws behaves similar to real select |
| 4839 | which stops waiting on a signal handler being triggered. |
| 4840 | |
| 4841 | This makes it possible to gracefully stop sockfilt.exe on |
| 4842 | Windows with taskkill /IM sockfilt.exe (without /F force flag). |
| 4843 | |
| 4844 | Reviewed-by: Jay Satiro |
| 4845 | Part of #5260 |
| 4846 | |
| 4847 | - tests/server/util.[ch]: add exit event to stop waiting on Windows |
| 4848 | |
| 4849 | This commit adds a global exit event to the test servers that |
| 4850 | Windows-specific wait routines can use to get triggered if the |
| 4851 | program was signaled to be terminated, eg. select_ws in sockfilt.c |
| 4852 | |
| 4853 | The exit event will be managed by the signal handling code and is |
| 4854 | set to not reset automatically to support multiple wait routines. |
| 4855 | |
| 4856 | Reviewed-by: Jay Satiro |
| 4857 | Closes #5260 |
| 4858 | |
| 4859 | - tests/server/util.c: fix thread handle not being closed |
| 4860 | |
| 4861 | Reviewed-by: Jay Satiro |
| 4862 | Part of #5260 |
| 4863 | |
| 4864 | - tests/server/util.c: use raise instead of calling signal handler |
| 4865 | |
| 4866 | Use raise to trigger signal handler instead of calling it |
| 4867 | directly and causing potential unexpected control flow. |
| 4868 | |
| 4869 | Reviewed-by: Jay Satiro |
| 4870 | Part of #5260 |
| 4871 | |
| 4872 | - tests: add support for SSH server variant specific transfer paths |
| 4873 | |
| 4874 | OpenSSH for Windows requires paths in the format of /C:/ |
| 4875 | instead of the pseudo-POSIX paths /cygdrive/c/ or just /c/ |
| 4876 | |
| 4877 | Reviewed-by: Daniel Stenberg |
| 4878 | Closes #5298 |
| 4879 | |
| 4880 | Daniel Stenberg (2 May 2020) |
| 4881 | - RELEASE-NOTES: synced |
| 4882 | |
| 4883 | - libssh2: set the expected total size in SCP upload init |
| 4884 | |
| 4885 | ... as otherwise the progress callback gets called without that |
| 4886 | information, making the progress meter have less info. |
| 4887 | |
| 4888 | Reported-by: Murugan Balraj |
| 4889 | Bug: https://curl.haxx.se/mail/archive-2020-05/0000.html |
| 4890 | Closes #5317 |
| 4891 | |
| 4892 | - runtests: make the logmsg from the ssh server only show in verbose |
| 4893 | |
| 4894 | - tests: make test 1248 + 1249 use %NOLISTENPORT |
| 4895 | |
| 4896 | ... instead of a port of a non-running server so that it works |
| 4897 | stand-alone. |
| 4898 | |
| 4899 | Closes #5318 |
| 4900 | |
| 4901 | - examples: remove asiohiper.cpp |
| 4902 | |
| 4903 | This example has repeatedly been reported to contain bugs, and as users |
| 4904 | copy and paste code from this into production, I now deem it better to |
| 4905 | not provide the example at all. |
| 4906 | |
| 4907 | Closes #5090 |
| 4908 | Closes #5322 |
| 4909 | |
| 4910 | - [Emil Engler brought this change] |
| 4911 | |
| 4912 | doc: add missing closing parenthesis in CURLINFO_SSL_VERIFYRESULT.3 |
| 4913 | |
| 4914 | Closes #5320 |
| 4915 | |
| 4916 | - [Emil Engler brought this change] |
| 4917 | |
| 4918 | KNOWN_BUGS: Remove "curl --upload-file . hang if delay in STDIN" |
| 4919 | |
| 4920 | It was fixed in 9a2cbf3 |
| 4921 | |
| 4922 | Closes #5319 |
| 4923 | |
| 4924 | - cirrus: disable SFTP and SCP tests |
| 4925 | |
| 4926 | ... as we can't seem to start the sshd server on it. Those problems |
| 4927 | existed before d1239b50bececd (running the SSH server on a random port), |
| 4928 | but they're more noticable now since there are more failed attempts in |
| 4929 | the logs. |
| 4930 | |
| 4931 | Closes #5315 |
| 4932 | |
| 4933 | - [Emil Engler brought this change] |
| 4934 | |
| 4935 | runtests: fix typo in the existence of disabled tests checker |
| 4936 | |
| 4937 | Closes #5316 |
| 4938 | |
| 4939 | Dan Fandrich (30 Apr 2020) |
| 4940 | - test75: Remove precheck test |
| 4941 | |
| 4942 | This has not been needed since commit 9fa42bed and often prevents it |
| 4943 | from running at all with dynamic test ports. |
| 4944 | |
| 4945 | - tests: Stop referring to server ports when they're not used |
| 4946 | |
| 4947 | Several tests referred to specific server ports even when the test |
| 4948 | didn't actually use that server or specify that it's needed. In such |
| 4949 | cases, the test harness substitutes the text "[not running]" as the port |
| 4950 | number which causes many such tests to fail due to the inability to |
| 4951 | parse the URL. These tests are changed to use %NOLISTENPORT which will |
| 4952 | always be substituted correctly. |
| 4953 | |
| 4954 | Daniel Stenberg (30 Apr 2020) |
| 4955 | - [Emil Engler brought this change] |
| 4956 | |
| 4957 | GnuTLS: Backend support for CURLINFO_SSL_VERIFYRESULT |
| 4958 | |
| 4959 | Closes #5287 |
| 4960 | |
| 4961 | - conncache: various concept cleanups |
| 4962 | |
| 4963 | More connection cache accesses are protected by locks. |
| 4964 | |
| 4965 | CONNCACHE_* is a beter prefix for the connection cache lock macros. |
| 4966 | |
| 4967 | Curl_attach_connnection: now called as soon as there's a connection |
| 4968 | struct available and before the connection is added to the connection |
| 4969 | cache. |
| 4970 | |
| 4971 | Curl_disconnect: now assumes that the connection is already removed from |
| 4972 | the connection cache. |
| 4973 | |
| 4974 | Ref: #4915 |
| 4975 | Closes #5009 |
| 4976 | |
| 4977 | - tests: tests: run stunnel for HTTPS and FTPS on dynamic ports |
| 4978 | |
| 4979 | As stunnel is an external tool and it has no specific option to export |
| 4980 | the actually used port number when asked to listen to 0, runtests |
| 4981 | instead iterates over ten randomly picked high number ports and sticks |
| 4982 | to the first one stunnel can listen to. |
| 4983 | |
| 4984 | Closes #5267 |
| 4985 | |
| 4986 | - tests: pick a random port number for SSH |
| 4987 | |
| 4988 | Since sshd doesn't have such an option by itself, we iterate over a |
| 4989 | series of random ports until one works. |
| 4990 | |
| 4991 | Closes #5273 |
| 4992 | |
| 4993 | - [Rikard Falkeborn brought this change] |
| 4994 | |
| 4995 | libtest/cmake: Remove commented code |
| 4996 | |
| 4997 | These were commented out in e9dd0998706a when Makefile.inc was included |
| 4998 | instead. 11 years have passed since then and the commented code is of |
| 4999 | course very outdated. Remove it to avoid confusion. |
| 5000 | |
| 5001 | Closes #5311 |
| 5002 | |
| 5003 | - schannel: source code reindent |
| 5004 | |
| 5005 | White space edits only. Conform better to standard curl source code |
| 5006 | indenting style. |
| 5007 | |
| 5008 | Closes #5305 |
| 5009 | |
| 5010 | Kamil Dudka (29 Apr 2020) |
| 5011 | - test1177: look for curl.h in source directory |
| 5012 | |
| 5013 | If we use a separate build directory, there is no copy of the header. |
| 5014 | |
| 5015 | Closes #5310 |
| 5016 | |
| 5017 | - tests: look for preprocessed tests in build directory |
| 5018 | |
| 5019 | ... which is not always the same directory as source directory |
| 5020 | |
| 5021 | Closes #5310 |
| 5022 | |
| 5023 | Daniel Stenberg (29 Apr 2020) |
| 5024 | - RELEASE-NOTES: synced |
| 5025 | |
| 5026 | ... and bumped curlver.h to 7.70.1 |
| 5027 | |
Haibo Huang | 24c77a1 | 2020-04-29 13:49:57 -0700 | [diff] [blame] | 5028 | Version 7.70.0 (29 Apr 2020) |
| 5029 | |
| 5030 | Daniel Stenberg (29 Apr 2020) |
| 5031 | - RELEASE-NOTES: 7.70.0 |
| 5032 | |
| 5033 | - THANKS: synced with the 7.70.0 release |
| 5034 | |
| 5035 | - headers: copyright range fix |
| 5036 | |
| 5037 | - [Rikard Falkeborn brought this change] |
| 5038 | |
| 5039 | doh: Constify some input pointers |
| 5040 | |
| 5041 | Closes #5306 |
| 5042 | |
| 5043 | - nss: check for PK11_CreateDigestContext() returning NULL |
| 5044 | |
| 5045 | ... to avoid crashes! |
| 5046 | |
| 5047 | Reported-by: Hao Wu |
| 5048 | Fixes #5302 |
| 5049 | Closes #5303 |
| 5050 | |
| 5051 | - travis: bump the wolfssl CI build to use 4.4.0 |
| 5052 | |
| 5053 | Closes #5301 |
| 5054 | |
| 5055 | - copyright updates: adjust year ranges |
| 5056 | |
| 5057 | Marc Hoersken (26 Apr 2020) |
| 5058 | - CI: do not include */ci branches in PR builds |
| 5059 | |
| 5060 | Align Azure Pipelines with GitHub Actions. |
| 5061 | |
| 5062 | Daniel Stenberg (25 Apr 2020) |
| 5063 | - runtests: check for the disabled tests relative srcdir |
| 5064 | |
| 5065 | To make it work correctly for out-of-tree builds. |
| 5066 | |
| 5067 | Follow-up to 75e8feb6fb08b |
| 5068 | |
| 5069 | Bug: https://github.com/curl/curl/pull/5288#issuecomment-619346389 |
| 5070 | Reported-by: Marcel Raad |
| 5071 | Closes #5297 |
| 5072 | |
| 5073 | - runtests: revert commenting out a line I did for debugging |
| 5074 | |
| 5075 | Follow-up to 11091cd4d. It was not meant to be pushed! |
| 5076 | |
| 5077 | - smtp: set auth correctly |
| 5078 | |
| 5079 | Regression since 7.69.0 and 68fb25fa3fcff. |
| 5080 | |
| 5081 | The code wrongly assigned 'from' instead of 'auth' which probably was a |
| 5082 | copy and paste mistake from other code, leading to that auth could |
| 5083 | remain NULL and later cause an error to be returned. |
| 5084 | |
| 5085 | Assisted-by: Eric Sauvageau |
| 5086 | Fixes #5294 |
| 5087 | Closes #5295 |
| 5088 | |
| 5089 | Marcel Raad (25 Apr 2020) |
| 5090 | - lib: clean up whitespace |
| 5091 | |
| 5092 | This fixes CodeFactor warnings. |
| 5093 | |
| 5094 | Daniel Stenberg (25 Apr 2020) |
| 5095 | - [Anderson Toshiyuki Sasaki brought this change] |
| 5096 | |
| 5097 | libssh: avoid options override by configuration files |
| 5098 | |
| 5099 | Previously, options set explicitly through command line options could be |
| 5100 | overridden by the configuration files parsed automatically when |
| 5101 | ssh_connect() was called. |
| 5102 | |
| 5103 | By calling ssh_options_parse_config() explicitly, the configuration |
| 5104 | files are parsed before setting the options, avoiding the options |
| 5105 | override. Once the configuration files are parsed, the automatic |
| 5106 | configuration parsing is not executed. |
| 5107 | |
| 5108 | Fixes #4972 |
| 5109 | Closes #5283 |
| 5110 | Signed-off-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com> |
| 5111 | |
| 5112 | - runtests: when <killserver> mentions http, kill http/2 too |
| 5113 | |
| 5114 | Since the http2 test server is a mere proxy that needs to know about the |
| 5115 | dynamic port the HTTP server is using, it too needs to get restarted |
| 5116 | when the http server is killed. |
| 5117 | |
| 5118 | A regression caused by 80d6515. |
| 5119 | |
| 5120 | Fixes #5289 |
| 5121 | Closes #5291 |
| 5122 | |
| 5123 | - [Yuri Slobodyanyuk brought this change] |
| 5124 | |
| 5125 | docs: fix two typos |
| 5126 | |
| 5127 | Closes #5292 |
| 5128 | |
| 5129 | - [Emil Engler brought this change] |
| 5130 | |
| 5131 | tests/git: ignore mqttd and port files |
| 5132 | |
| 5133 | Closes #5290 |
| 5134 | |
| 5135 | - tests: make runtests check that disabled tests exists |
| 5136 | |
| 5137 | ... and error out if so. Removed '536' from DISABLED as there is no such |
| 5138 | test file. |
| 5139 | |
| 5140 | Closes #5288 |
| 5141 | |
| 5142 | - test1154: set a proper name |
| 5143 | |
| 5144 | - select: make Curl_socket_check take timediff_t timeout |
| 5145 | |
| 5146 | Coverity found CID 1461718: |
| 5147 | |
| 5148 | Integer handling issues (CONSTANT_EXPRESSION_RESULT) "timeout_ms > |
| 5149 | 9223372036854775807L" is always false regardless of the values of its |
| 5150 | operands. This occurs as the logical second operand of "||". |
| 5151 | |
| 5152 | Closes #5240 |
| 5153 | |
| 5154 | - [i-ky brought this change] |
| 5155 | |
| 5156 | libcurl-multi.3: added missing full stop |
| 5157 | |
| 5158 | Closes #5285 |
| 5159 | |
| 5160 | Jay Satiro (22 Apr 2020) |
| 5161 | - transfer: Switch PUT to GET/HEAD on 303 redirect |
| 5162 | |
| 5163 | Prior to this change if there was a 303 reply to a PUT request then |
| 5164 | the subsequent request to respond to that redirect would also be a PUT. |
| 5165 | It was determined that was most likely incorrect based on the language |
| 5166 | of the RFCs. Basically 303 means "see other" resource, which implies it |
| 5167 | is most likely not the same resource, therefore we should not try to PUT |
| 5168 | to that different resource. |
| 5169 | |
| 5170 | Refer to the discussions in #5237 and #5248 for more information. |
| 5171 | |
| 5172 | Fixes https://github.com/curl/curl/issues/5237 |
| 5173 | Closes https://github.com/curl/curl/pull/5248 |
| 5174 | |
| 5175 | Daniel Stenberg (22 Apr 2020) |
| 5176 | - lib/mk-ca-bundle: skip empty certs |
| 5177 | |
| 5178 | Reviewed-by: Emil Engler |
| 5179 | Reported-by: Ashwin Metpalli |
| 5180 | Fixes #5278 |
| 5181 | Closes #5280 |
| 5182 | |
| 5183 | - version: skip idn2_check_version() check and add precaution |
| 5184 | |
| 5185 | A gcc-10's -fanalyze complaint made me spot and do these improvements. |
| 5186 | |
| 5187 | Closes #5281 |
| 5188 | |
| 5189 | - RELEASE-NOTES: synced |
| 5190 | |
| 5191 | - [Brian Bergeron brought this change] |
| 5192 | |
| 5193 | curl.h: update comment typo |
| 5194 | |
| 5195 | "routines with be invoked" -> "routines will be invoked" |
| 5196 | |
| 5197 | Closes #5279 |
| 5198 | |
| 5199 | - [Emil Engler brought this change] |
| 5200 | |
| 5201 | GnuTLS: Don't skip really long certificate fields |
| 5202 | |
| 5203 | Closes #5271 |
| 5204 | |
| 5205 | - gnutls: bump lowest supported version to 3.1.10 |
| 5206 | |
| 5207 | GnuTLS 3.1.10 added new functions we want to use. That version was |
| 5208 | released on Mar 22, 2013. Removing support for older versions also |
| 5209 | greatly simplifies the code. |
| 5210 | |
| 5211 | Ref: #5271 |
| 5212 | Closes #5276 |
| 5213 | |
| 5214 | - mqtt: make NOSTATE get within the debug name array |
| 5215 | |
| 5216 | - tests: run the RTSP test server on a dynamic port number |
| 5217 | |
| 5218 | To avoid port collisions. |
| 5219 | |
| 5220 | Closes #5272 |
| 5221 | |
| 5222 | - tests: add %NOLISTENPORT and use it |
| 5223 | |
| 5224 | The purpose with this variable is to provide a port number that is |
| 5225 | reasonably likely to not have a listener on the local host so that tests |
| 5226 | can try connect failures against it. It uses port 47 - "reserved" |
| 5227 | according to IANA. |
| 5228 | |
| 5229 | Updated six tests to use it instead of the previous different ports. |
| 5230 | |
| 5231 | Assisted-by: Emil Engler |
| 5232 | Closes #5270 |
| 5233 | |
| 5234 | - mqtt: remove code with no purpose |
| 5235 | |
| 5236 | Detected by Coverity. CID 1462319. |
| 5237 | |
| 5238 | "The same code is executed when the condition result is true or false, |
| 5239 | because the code in the if-then branch and after the if statement is |
| 5240 | identical." |
| 5241 | |
| 5242 | Closes #5275 |
| 5243 | |
| 5244 | - mqtt: fix Curl_read() error handling while reading remaining length |
| 5245 | |
| 5246 | Detected by Coverity. CID 1462320. |
| 5247 | |
| 5248 | Closes #5274 |
| 5249 | |
| 5250 | - server/tftpd: fix compiler warning |
| 5251 | |
| 5252 | Follow-up from 369ce38ac1d |
| 5253 | Reported-by: Marc Hörsken |
| 5254 | |
| 5255 | - http: free memory when Alt-Used header creation fails due to OOM |
| 5256 | |
| 5257 | Reported-by: James Fuller |
| 5258 | Fixes #5268 |
| 5259 | Closes #5269 |
| 5260 | |
| 5261 | Daniel Gustafsson (20 Apr 2020) |
| 5262 | - lib: fix typos in comments and errormessages |
| 5263 | |
| 5264 | This fixes a few randomly spotted typos in recently merged code, most |
| 5265 | notably one in a userfacing errormessage the schannel code. |
| 5266 | |
| 5267 | Daniel Stenberg (20 Apr 2020) |
| 5268 | - tests: run the SOCKS test server on a dynamic port number |
| 5269 | |
| 5270 | Closes #5266 |
| 5271 | |
| 5272 | - [Johannes Schindelin brought this change] |
| 5273 | |
| 5274 | multi-ssl: reset the SSL backend on `Curl_global_cleanup()` |
| 5275 | |
| 5276 | When cURL is compiled with support for multiple SSL backends, it is |
| 5277 | possible to configure an SSL backend via `curl_global_sslset()`, but |
| 5278 | only *before* `curl_global_init()` was called. |
| 5279 | |
| 5280 | If another SSL backend should be used after that, a user might be |
| 5281 | tempted to call `curl_global_cleanup()` to start over. However, we did |
| 5282 | not foresee that use case and forgot to reset the SSL backend in that |
| 5283 | cleanup. |
| 5284 | |
| 5285 | Let's allow that use case. |
| 5286 | |
| 5287 | Fixes #5255 |
| 5288 | Closes #5257 |
| 5289 | Reported-by: davidedec on github |
| 5290 | Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> |
| 5291 | |
| 5292 | - tests: run the TFTP test server on a dynamic port number |
| 5293 | |
| 5294 | Picking a dynamic unused port is better than a fixed to avoid the |
| 5295 | collision risk. |
| 5296 | |
| 5297 | Closes #5265 |
| 5298 | |
| 5299 | - mqtt: improve the state machine |
| 5300 | |
| 5301 | To handle PUBLISH before SUBACK and more. |
| 5302 | |
| 5303 | Updated the existing tests and added three new ones. |
| 5304 | |
| 5305 | Reported-by: Christoph Krey |
| 5306 | Bug: https://curl.haxx.se/mail/lib-2020-04/0021.html |
| 5307 | Closes #5246 |
| 5308 | |
| 5309 | - runtests: always put test number in servercmd file |
| 5310 | |
| 5311 | - RELEASE-NOTES: synced |
| 5312 | |
| 5313 | - release-notes.pl: fix parsing typo |
| 5314 | |
| 5315 | James Fuller (20 Apr 2020) |
Haibo Huang | ca2a802 | 2020-07-10 20:17:42 -0700 | [diff] [blame] | 5316 | - ensure all references to ports are replaced by vars |
Haibo Huang | 24c77a1 | 2020-04-29 13:49:57 -0700 | [diff] [blame] | 5317 | |
Haibo Huang | ca2a802 | 2020-07-10 20:17:42 -0700 | [diff] [blame] | 5318 | - add more alt-svc test coverage |
Haibo Huang | 24c77a1 | 2020-04-29 13:49:57 -0700 | [diff] [blame] | 5319 | |
| 5320 | Daniel Stenberg (20 Apr 2020) |
| 5321 | - test1247: use http server to get the port number set |
| 5322 | |
| 5323 | Follow-up to 0f5db7b263f |
| 5324 | |
| 5325 | - runtests: use a unix domain socket path with the pid in the name |
| 5326 | |
| 5327 | To make it impossible for test cases to access the file name without |
| 5328 | using the proper variable for the purpose. |
| 5329 | |
| 5330 | Closes #5264 |
| 5331 | |
| 5332 | Daniel Gustafsson (19 Apr 2020) |
Haibo Huang | b5a52b9 | 2020-10-28 22:18:23 -0700 | [diff] [blame] | 5333 | - [Mipsters on github brought this change] |
Haibo Huang | 24c77a1 | 2020-04-29 13:49:57 -0700 | [diff] [blame] | 5334 | |
| 5335 | src: Remove C99 constructs to ensure C89 compliance |
| 5336 | |
| 5337 | This fixes the error: 'for' loop initial declaration used outside C99 |
| 5338 | mode by declaring the loop increment variable in the beginning of the |
| 5339 | block instead of inside the for loop. |
| 5340 | |
| 5341 | Fixes #5254 |
| 5342 | Reviewed-by: Daniel Gustafsson <daniel@yesql.se> |
| 5343 | |
| 5344 | Daniel Stenberg (19 Apr 2020) |
| 5345 | - runtests: dummy init the ports variables to avoid warnings |
| 5346 | |
| 5347 | ... and generate something that can help debug test cases. |
| 5348 | |
| 5349 | - [Patrick Monnerat brought this change] |
| 5350 | |
| 5351 | mime: properly check Content-Type even if it has parameters |
| 5352 | |
| 5353 | New test 669 checks this fix is effective. |
| 5354 | |
| 5355 | Fixes #5256 |
| 5356 | Closes #5258 |
| 5357 | Reported-by: thanhchungbtc on github |
| 5358 | |
| 5359 | - tests/FILEFORMAT: converted to markdown and extended |
| 5360 | |
| 5361 | Closes #5261 |
| 5362 | |
| 5363 | - test1245: make it work with dynamic FTP server port |
| 5364 | |
| 5365 | - test1055: make it work with dynamic FTP port |
| 5366 | |
| 5367 | - test1028: make it run on dynamic FTP server port |
| 5368 | |
| 5369 | - tests: move pingpong server to dynamic listening port |
| 5370 | |
| 5371 | FTP, IMAP, POP3, SMTP and their IPv6 versions are now all on dynamic |
| 5372 | ports |
| 5373 | |
| 5374 | Test 842-845 are unfortunately a bit hard to move over to this concept |
| 5375 | right now and require "default port" still... |
| 5376 | |
| 5377 | - test1056: work with dynamic HTTP ipv6 port |
| 5378 | |
| 5379 | - test1448: work with dynamic HTTP server port |
| 5380 | |
| 5381 | - tests: introduce preprocessed test cases |
| 5382 | |
| 5383 | The runtests script now always performs variable replacement on the |
| 5384 | entire test source file before the test gets executed, and saves the |
| 5385 | updated version in a temporary file (log/test[num]) so that all test |
| 5386 | case readers/servers can use that version (if present) and thus enjoy |
| 5387 | the powers of test case variable substitution. |
| 5388 | |
| 5389 | This is necessary to allow complete port number freedom. |
| 5390 | |
| 5391 | Test 309 is updated to work with a non-fixed port number thanks to this. |
| 5392 | |
| 5393 | - tests: make 2006-2010 handle different port number lengths |
| 5394 | |
| 5395 | - tests: run the sws server on "any port" |
| 5396 | |
| 5397 | Makes the test servers for HTTP and Gopher pop up on a currently unused |
| 5398 | port and runtests adapts to that! |
| 5399 | |
| 5400 | Closes #5247 |
| 5401 | |
| 5402 | Marc Hoersken (18 Apr 2020) |
| 5403 | - sockfilt: tidy variable naming and data structure in select_ws |
| 5404 | |
| 5405 | This commit does not introduce any logical changes to the code. |
| 5406 | |
| 5407 | Reviewed-by: Jay Satiro and Marcel Raad |
| 5408 | Closes #5238 |
| 5409 | |
| 5410 | Daniel Stenberg (17 Apr 2020) |
| 5411 | - [Anderson Toshiyuki Sasaki brought this change] |
| 5412 | |
| 5413 | libssh: Use new ECDSA key types to check known hosts |
| 5414 | |
| 5415 | From libssh 0.9.0, ssh_key_type() returns different key types for ECDSA |
| 5416 | keys depending on the curve. |
| 5417 | |
| 5418 | Signed-off-by: Anderson Toshiyuki Sasaki <ansasaki@redhat.com> |
| 5419 | Fixes #5252 |
| 5420 | Closes #5253 |
| 5421 | |
| 5422 | Marcel Raad (17 Apr 2020) |
| 5423 | - appveyor: add Unicode winbuild jobs |
| 5424 | |
| 5425 | These are cheap as they don't build tests. |
| 5426 | |
| 5427 | Closes https://github.com/curl/curl/pull/5063 |
| 5428 | |
| 5429 | Daniel Stenberg (16 Apr 2020) |
| 5430 | - mqttd: s/errno/SOCKERRNO |
| 5431 | |
| 5432 | To behave proper on Windows |
| 5433 | Reported-by: Gisle Vanem |
| 5434 | Bug: https://github.com/curl/curl/commit/5e855bbd18f84a02c951be7cac6188276818cdac#r38507132 |
| 5435 | Closes #5241 |
| 5436 | |
| 5437 | - buildconf: use find -execdir instead, remove -print and the ares files |
| 5438 | |
| 5439 | Follow-up to 1e41bec96a6e |
| 5440 | |
| 5441 | Suggested-by: Marc Hörsken |
| 5442 | |
| 5443 | - [Alexander V. Tikhonov brought this change] |
| 5444 | |
| 5445 | buildconf: avoid using tempfile when removing files |
| 5446 | |
| 5447 | Closes #5213 |
| 5448 | |
| 5449 | - copyright: bump the copyright year range |
| 5450 | |
| 5451 | - scripts/release-notes.pl: accept colon after the Fixes/Closes keywords |
| 5452 | |
| 5453 | - [JP Mens brought this change] |
| 5454 | |
| 5455 | docs/MQTT: replace confusing 80 by 75 |
| 5456 | |
| 5457 | I was a bit surprised by the `80`: first thought: what's HTTP doing |
| 5458 | here? ;) |
| 5459 | |
| 5460 | Closes #5236 |
| 5461 | |
| 5462 | - [Brad King brought this change] |
| 5463 | |
| 5464 | cmake: Avoid MSVC C4273 warnings in send/recv checks |
| 5465 | |
| 5466 | We use `check_c_source_compiles` to check possible send/recv signatures |
| 5467 | by reproducing the forward declarations from system headers. On Windows |
| 5468 | the `winsock2.h` header adds dll linkage settings to its forward |
| 5469 | declaration. If ours does not match the compiler warns: |
| 5470 | |
| 5471 | warning C4273: 'recv': inconsistent dll linkage |
| 5472 | |
| 5473 | Add `WINSOCK_API_LINKAGE` to our test signatures when it is defined so |
| 5474 | that our linkage is consistent with that from `winsock2.h`. |
| 5475 | |
| 5476 | Fixes #4764 |
| 5477 | Closes #5232 |
| 5478 | |
| 5479 | Jay Satiro (14 Apr 2020) |
| 5480 | - KNOWN_BUGS: Add entry 'Blocking socket operations' |
| 5481 | |
| 5482 | - Add threaded resolver cleanup and GSSAPI for FTP to the TODO list of |
| 5483 | known blocking operations. |
| 5484 | |
| 5485 | - New known bugs entry 'Blocking socket operations in non-blocking API' |
| 5486 | that directs to the TODO's list of known blocking operations. |
| 5487 | |
| 5488 | Ref: https://github.com/curl/curl/pull/5214#issuecomment-612488021 |
| 5489 | |
| 5490 | Reported-by: Marc Hoersken |
| 5491 | |
| 5492 | Closes https://github.com/curl/curl/pull/5216 |
| 5493 | |
| 5494 | Marc Hoersken (14 Apr 2020) |
| 5495 | - test2043: use revoked.badssl.com instead of revoked.grc.com |
| 5496 | |
| 5497 | The certificate of revoked.grc.com has expired on 2020-04-13. |
| 5498 | |
| 5499 | Reviewed-by: Jay Satiro |
| 5500 | |
| 5501 | Closes #5233 |
| 5502 | |
| 5503 | - sockfilt: fix broken pipe on Windows to be ready in select_ws |
| 5504 | |
| 5505 | Closes #5228 |
| 5506 | |
| 5507 | Daniel Stenberg (14 Apr 2020) |
| 5508 | - RELEASE-NOTES: synced |
| 5509 | |
| 5510 | - scripts/release-notes: fix duplicate output header |
| 5511 | |
| 5512 | - github/workflow: enable MQTT in the macOS debug build |
| 5513 | |
| 5514 | - azure: add mqtt support to one of the Windows builds |
| 5515 | |
| 5516 | - travis: add mqtt job on Linux |
| 5517 | |
| 5518 | - tests: add four MQTT tests 1190 - 1193 |
| 5519 | |
| 5520 | - tests: add the mqtt test server mqttd |
| 5521 | |
| 5522 | - tests: support hex encoded data and mqtt server |
| 5523 | |
| 5524 | The mqtt server is started using a "random" port. |
| 5525 | |
| 5526 | - [Björn Stenberg brought this change] |
| 5527 | |
| 5528 | mqtt: add new experimental protocol |
| 5529 | |
| 5530 | Closes #5173 |
| 5531 | |
| 5532 | - TODO: Consider convenience options for JSON and XML? |
| 5533 | |
| 5534 | Closes #5203 |
| 5535 | |
| 5536 | - tool: do not declare functions with Curl_ prefix |
| 5537 | |
| 5538 | To avoid collision risks with private libcurl symbols when linked with |
| 5539 | static versions (or just versions not hiding internal symbols). |
| 5540 | |
| 5541 | Reported-by: hydra3333 on github |
| 5542 | Fixes #5219 |
| 5543 | Closes #5234 |
| 5544 | |
| 5545 | - [Nathaniel R. Lewis brought this change] |
| 5546 | |
| 5547 | cmake: add aliases so exported target names are available in tree |
| 5548 | |
| 5549 | Reviewed-by: Brad King |
| 5550 | Closes #5206 |
| 5551 | |
| 5552 | - version: increase buffer space for ssl version output |
| 5553 | |
| 5554 | To avoid it getting truncated, especially when several SSL backends are |
| 5555 | built-in. |
| 5556 | |
| 5557 | Reported-by: Gisle Vanem |
| 5558 | Fixes #5222 |
| 5559 | Closes #5226 |
| 5560 | |
| 5561 | Marc Hoersken (13 Apr 2020) |
| 5562 | - cirrus: no longer ignore test 504 which is working again |
| 5563 | |
| 5564 | The test is working again, because TCP blackholing is disabled. |
| 5565 | |
| 5566 | - appveyor: completely disable tests that fail to timeout early |
| 5567 | |
| 5568 | The tests changed from ignored to disabled are tests that are |
| 5569 | about connecting to non-listening socket. On AppVeyor these |
| 5570 | tests are not reliable, because for some unknown reason the |
| 5571 | connect is not timing out before the test time limit is reached. |
| 5572 | |
| 5573 | Daniel Stenberg (13 Apr 2020) |
| 5574 | - test1908: avoid using fixed port number in test data |
| 5575 | |
| 5576 | Closes #5225 |
| 5577 | |
| 5578 | Jay Satiro (12 Apr 2020) |
| 5579 | - [Andrew Kurushin brought this change] |
| 5580 | |
| 5581 | schannel: Fix blocking timeout logic |
| 5582 | |
| 5583 | - Fix schannel_send for the case when no timeout was set. |
| 5584 | |
| 5585 | Prior to this change schannel would error if the socket was not ready |
| 5586 | to send data and no timeout was set. |
| 5587 | |
| 5588 | This commit is similar to parent commit 89dc6e0 which recently made the |
| 5589 | same change for SOCKS, for the same reason. Basically it was not well |
| 5590 | understood that when Curl_timeleft returns 0 it is not a timeout of 0 ms |
| 5591 | but actually means no timeout. |
| 5592 | |
| 5593 | Fixes https://github.com/curl/curl/issues/5177 |
| 5594 | Closes https://github.com/curl/curl/pull/5221 |
| 5595 | |
| 5596 | - socks: Fix blocking timeout logic |
| 5597 | |
| 5598 | - Document in Curl_timeleft's comment block that returning 0 signals no |
| 5599 | timeout (ie there's infinite time left). |
| 5600 | |
| 5601 | - Fix SOCKS' Curl_blockread_all for the case when no timeout was set. |
| 5602 | |
| 5603 | Prior to this change if the timeout had a value of 0 and that was passed |
| 5604 | to SOCKET_READABLE it would return right away instead of blocking. That |
| 5605 | was likely because it was not well understood that when Curl_timeleft |
| 5606 | returns 0 it is not a timeout of 0 ms but actually means no timeout. |
| 5607 | |
| 5608 | Ref: https://github.com/curl/curl/pull/5214#issuecomment-612512360 |
| 5609 | |
| 5610 | Closes https://github.com/curl/curl/pull/5220 |
| 5611 | |
| 5612 | - [Marc Hoersken brought this change] |
| 5613 | |
| 5614 | gopher: check remaining time left during write busy loop |
| 5615 | |
| 5616 | Prior to this change gopher's blocking code would block forever, |
| 5617 | ignoring any set timeout value. |
| 5618 | |
| 5619 | Assisted-by: Jay Satiro |
| 5620 | Reviewed-by: Daniel Stenberg |
| 5621 | |
| 5622 | Similar to #5220 and #5221 |
| 5623 | Closes #5214 |
| 5624 | |
| 5625 | Daniel Stenberg (13 Apr 2020) |
| 5626 | - [Dirkjan Bussink brought this change] |
| 5627 | |
| 5628 | gnutls: ensure TLS 1.3 when SRP isn't requested |
| 5629 | |
| 5630 | When SRP is requested in the priority string, GnuTLS will disable |
| 5631 | support for TLS 1.3. Before this change, curl would always add +SRP to |
| 5632 | the priority list, effectively always disabling TLS 1.3 support. |
| 5633 | |
| 5634 | With this change, +SRP is only added to the priority list when SRP |
| 5635 | authentication is also requested. This also allows updating the error |
| 5636 | handling here to not have to retry without SRP. This is because SRP is |
| 5637 | only added when requested and in that case a retry is not needed. |
| 5638 | |
| 5639 | Closes #5223 |
| 5640 | |
| 5641 | Marc Hoersken (12 Apr 2020) |
| 5642 | - tests/server: add hidden window to gracefully handle WM_CLOSE |
| 5643 | |
| 5644 | Forward Window events as signals to existing signal event handler. |
| 5645 | |
| 5646 | - tests/server: add CTRL event handler for Win32 consoles |
| 5647 | |
| 5648 | Forward CTRL events as signals to existing signal event handler. |
| 5649 | |
| 5650 | - tests/server: move all signal handling routines to util.[ch] |
| 5651 | |
| 5652 | Avoid code duplication to prepare for portability enhancements. |
| 5653 | |
| 5654 | Daniel Stenberg (12 Apr 2020) |
| 5655 | - compressed.d: stress that the headers are not modified |
| 5656 | |
| 5657 | Suggested-by: Michael Osipov |
| 5658 | Assisted-by: Jay Satiro |
| 5659 | Bug: https://github.com/curl/curl/issues/5182#issuecomment-611638008 |
| 5660 | Closes #5217 |
| 5661 | |
| 5662 | Marc Hoersken (11 Apr 2020) |
| 5663 | - tests/server/util.c: use curl_off_t instead of long for pid |
| 5664 | |
| 5665 | Avoid potential overflow of huge PIDs on Windows. |
| 5666 | |
| 5667 | Related to #5188 |
| 5668 | Assisted-by: Marcel Raad |
| 5669 | |
| 5670 | - tests: use Cygwin/msys PIDs for stunnel and sshd on Windows |
| 5671 | |
| 5672 | Since the Windows versions of both programs would write Windows |
| 5673 | PIDs to their pidfiles which we cannot handle, we need to use |
| 5674 | our known perl.exe Cygwin/msys PID together with exec() in order |
| 5675 | to tie the spawned processes to the existance of our perl.exe |
| 5676 | |
| 5677 | The perl.exe that is executing secureserver.pl and sshserver.pl |
| 5678 | has a Cygwin/msys PID, because it is started inside Cygwin/msys. |
| 5679 | |
| 5680 | Related to #5188 |
| 5681 | |
| 5682 | - tests: add Windows compatible pidwait like pidkill and pidterm |
| 5683 | |
| 5684 | Related to #5188 |
| 5685 | |
| 5686 | - tests: fix conflict between Cygwin/msys and Windows PIDs |
| 5687 | |
| 5688 | Add 65536 to Windows PIDs to allow Windows specific treatment |
| 5689 | by having disjunct ranges for Cygwin/msys and Windows PIDs. |
| 5690 | |
| 5691 | See also: |
| 5692 | - https://cygwin.com/git/?p=newlib-cygwin.git;a=commit; ↵ |
| 5693 | h=b5e1003722cb14235c4f166be72c09acdffc62ea |
| 5694 | - https://cygwin.com/git/?p=newlib-cygwin.git;a=commit; ↵ |
| 5695 | h=448cf5aa4b429d5a9cebf92a0da4ab4b5b6d23fe |
| 5696 | |
| 5697 | Replaces #5178 |
| 5698 | Closes #5188 |
| 5699 | |
| 5700 | Daniel Stenberg (11 Apr 2020) |
| 5701 | - RELEASE-NOTES: synced |
| 5702 | |
| 5703 | - release-notes.pl: detect the start of the references in cleanup mode |
| 5704 | |
| 5705 | - Revert "file: on Windows, refuse paths that start with \\" |
| 5706 | |
| 5707 | This reverts commit 1b71bc532bde8621fd3260843f8197182a467ff2. |
| 5708 | |
| 5709 | Reminded-by: Chris Roberts |
| 5710 | Bug: https://curl.haxx.se/mail/archive-2020-04/0013.html |
| 5711 | |
| 5712 | Closes #5215 |
| 5713 | |
| 5714 | Jay Satiro (11 Apr 2020) |
| 5715 | - lib: fix conversion warnings for SOCKET_WRITABLE/READABLE |
| 5716 | |
| 5717 | - If loss of data may occur converting a timediff_t to time_t and |
| 5718 | the time value is > TIME_T_MAX then treat it as TIME_T_MAX. |
| 5719 | |
| 5720 | This is a follow-up to 8843678 which removed the (time_t) typecast |
| 5721 | from the macros so that conversion warnings could be identified. |
| 5722 | |
| 5723 | Closes https://github.com/curl/curl/pull/5199 |
| 5724 | |
| 5725 | - test1148: tolerate progress updates better (again) |
| 5726 | |
| 5727 | - Ignore intermediate progress updates. |
| 5728 | |
| 5729 | - Support locales that use a character other than period as decimal |
| 5730 | separator (eg 100,0%). |
| 5731 | |
| 5732 | test1148 checks that the progress finishes at 100% and has the right |
| 5733 | bar width. Prior to this change the test assumed that the only progress |
| 5734 | reported for such a quick transfer was 100%, however in rare instances |
| 5735 | (like in the CI where transfer time can slow considerably) there may be |
| 5736 | intermediate updates. For example, below is stderrlog1148 from a failed |
| 5737 | CI run with explicit \r and \n added (it is one line; broken up so that |
| 5738 | it's easier to understand). |
| 5739 | |
| 5740 | \r |
| 5741 | \r################################## 48.3% |
| 5742 | \r######################################################################## 100.0% |
| 5743 | \n |
| 5744 | |
| 5745 | Closes https://github.com/curl/curl/pull/5194 |
| 5746 | |
| 5747 | Marc Hoersken (10 Apr 2020) |
| 5748 | - sshserver.pl: use cached Win32 environment check variable |
| 5749 | |
| 5750 | - appveyor: partially revert 3413a110 to keep build without proxy |
| 5751 | |
| 5752 | Ref: #5211 and #4526 |
| 5753 | Reported-by: Marcel Raad |
| 5754 | |
| 5755 | - appveyor: ignore failing 'connect to non-listening proxy' tests |
| 5756 | |
| 5757 | Closes #5211 |
| 5758 | |
| 5759 | - CI/macos: convert CRLF to LF and align indentation |
| 5760 | |
| 5761 | Daniel Stenberg (9 Apr 2020) |
| 5762 | - url: allow non-HTTPS altsvc-matching for debug builds |
| 5763 | |
| 5764 | This is already partly supported but this part was missing. |
| 5765 | Reported-by: James Fuller |
| 5766 | |
| 5767 | Closes #5205 |
| 5768 | |
| 5769 | - server/resolve: remove AI_CANONNAME to make macos tell the truth |
| 5770 | |
| 5771 | With this bit set, my mac successfully resolves "ip6-localhost" when in |
| 5772 | fact there is no such host known to my machine! That in turn made test |
| 5773 | 241 wrongly execute and fail. |
| 5774 | |
| 5775 | Closes #5202 |
| 5776 | |
| 5777 | - runtests: fix warning about using an undefined variable |
| 5778 | |
| 5779 | Follow-up from 4d939ef6ceb2db1 |
| 5780 | |
| 5781 | - release-notes: fix the initial reference list output |
| 5782 | |
| 5783 | - github actions: run when pushed to master or */ci + PRs |
| 5784 | |
| 5785 | Avoid double-builds when using "local" branches for PRs. For both macos |
| 5786 | and fuzz jobs. |
| 5787 | |
| 5788 | Closes #5201 |
| 5789 | |
| 5790 | - runtests: provide nicer errormsg when protocol "dump" file is empty |
| 5791 | |
| 5792 | - [Gilles Vollant brought this change] |
| 5793 | |
| 5794 | schannel: support .P12 or .PFX client certificates |
| 5795 | |
| 5796 | Used with curl command line option like this: --cert |
| 5797 | <filename>:<password> --cert-type p12 |
| 5798 | |
| 5799 | Closes #5193 |
| 5800 | |
| 5801 | - tests: verify split initial HTTP requests with CURL_SMALLREQSEND |
| 5802 | |
| 5803 | test1294: "split request" being when the entire request isn't sent in |
| 5804 | the first go, and the remainder is sent in the PERFORM state. A GET |
| 5805 | request is otherwise not sending anything during PERFORM. |
| 5806 | |
| 5807 | test1295: same kind of split but with POST |
| 5808 | |
| 5809 | Closes #5197 |
| 5810 | |
| 5811 | - http: don't consider upload done if the request isn't completely sent off |
| 5812 | |
| 5813 | Fixes #4919 |
| 5814 | Closes #5197 |
| 5815 | |
| 5816 | - http: allow Curl_add_buffer_send() to do a short first send by force |
| 5817 | |
| 5818 | In a debug build, settting the environment variable "CURL_SMALLREQSEND" |
| 5819 | will make the first HTTP request send not send more bytes than the set |
| 5820 | amount, thus ending up verifying that the logic for handling a split |
| 5821 | HTTP request send works correctly. |
| 5822 | |
| 5823 | - connect: store connection info for QUIC connections |
| 5824 | |
| 5825 | Restores the --head functionality to the curl utility which extracts |
| 5826 | 'protocol' that is stored that way. |
| 5827 | |
| 5828 | Reported-by: James Fuller |
| 5829 | Fixes #5196 |
| 5830 | Closes #5198 |
| 5831 | |
| 5832 | - tests/README: update the port numbers list |
| 5833 | |
| 5834 | Since the pipelining server is long gone. |
| 5835 | Reported-by: James Fuller |
| 5836 | |
| 5837 | - select: remove typecast from SOCKET_WRITABLE/READABLE macros |
| 5838 | |
| 5839 | So that they don't hide conversions-by-mistake |
| 5840 | |
| 5841 | Reviewed-by: Jay Satiro |
| 5842 | Closes #5190 |
| 5843 | |
| 5844 | - CURLOPT_WRITEFUNCTION.3: add inline example and new see-also |
| 5845 | |
| 5846 | Closes #5192 |
| 5847 | |
| 5848 | - release-notes: output trailing references sorted numerically |
| 5849 | |
| 5850 | - cleanup: correct copyright year range on a few files |
| 5851 | |
| 5852 | - configure: remove use of -vec-report0 from CFLAGS with icc |
| 5853 | |
| 5854 | ... as it apparently isn't (always) supported. |
| 5855 | Reported-by: Alain Miniussi |
| 5856 | Fixes #5096 |
| 5857 | Closes #5191 |
| 5858 | |
| 5859 | - warnless: remove code block for icc that didn't work |
| 5860 | |
| 5861 | Reported-by: Alain Miniussi |
| 5862 | Fixes #5096 |
| 5863 | |
| 5864 | Marc Hoersken (6 Apr 2020) |
| 5865 | - dist: add missing setup-win32.h |
| 5866 | |
| 5867 | Follow up to d820224b8b |
| 5868 | |
| 5869 | Daniel Stenberg (6 Apr 2020) |
| 5870 | - RELEASE-NOTES: synced |
| 5871 | |
| 5872 | - scripts/release-notes.pl: add helper script for RELEASE-NOTES maintenance |
| 5873 | |
| 5874 | This script helps putting entries in the RELEASE-NOTES using a coherent |
| 5875 | style and sorting with a minimal human editing effort - as long as the |
| 5876 | first line in the commit message is good enough! There's a short howto |
| 5877 | at the top of the file. |
| 5878 | |
| 5879 | - [Dennis Felsing brought this change] |
| 5880 | |
| 5881 | configure: don't check for Security.framework when cross-compiling |
| 5882 | |
| 5883 | Since it checks for the local file, not the cross-compiled one. |
| 5884 | |
| 5885 | Closes #5189 |
| 5886 | |
| 5887 | - TODO: Option to make -Z merge lined based outputs on stdout |
| 5888 | |
| 5889 | Closes #5175 |
| 5890 | |
| 5891 | - lib: never define CURL_CA_BUNDLE with a getenv |
| 5892 | |
| 5893 | - it breaks the build (since 6de756c9b1de34b7a1) |
| 5894 | - it's not documented and not consistent across platforms |
| 5895 | - the curl tool does that getenv magic |
| 5896 | |
| 5897 | Bug: https://github.com/curl/curl/commit/6de756c#r38127030 |
| 5898 | Reported-by: Gisle Vanem |
| 5899 | |
| 5900 | Closes #5187 |
| 5901 | |
| 5902 | Marc Hoersken (5 Apr 2020) |
| 5903 | - lib670: use the same Win32 API check as all other lib tests |
| 5904 | |
| 5905 | - appveyor: use random test server ports based upon APPVEYOR_API_URL |
| 5906 | |
| 5907 | Avoid conflicts of test server ports with AppVeyor API on localhost. |
| 5908 | |
| 5909 | Closes #5034 |
| 5910 | |
| 5911 | - appveyor: sort builds by type and add two new variants |
| 5912 | |
| 5913 | Related to #5034 and #5063 |
| 5914 | |
| 5915 | - appveyor: show failed tests in log even if test is ignored |
| 5916 | |
| 5917 | And print API response with newline only if there is one |
| 5918 | |
| 5919 | - appveyor: turn disabled tests into ignored result tests |
| 5920 | |
| 5921 | Daniel Stenberg (5 Apr 2020) |
| 5922 | - KNOWN_BUGS: fixed "USE_UNIX_SOCKETS on Windows" |
| 5923 | |
| 5924 | Fixed with #5170 (commit 23a870f2fd041278) |
| 5925 | |
| 5926 | - test1566: verify --etag-compare that gets a 304 back |
| 5927 | |
| 5928 | Verifies the fix in #5183 |
| 5929 | |
| 5930 | Closes #5186 |
| 5931 | |
| 5932 | - [Kwon-Young Choi brought this change] |
| 5933 | |
| 5934 | CURLINFO_CONDITION_UNMET: return true for 304 http status code |
| 5935 | |
| 5936 | In libcurl, CURLINFO_CONDITION_UNMET is used to avoid writing to the |
| 5937 | output file if the server did not transfered a file based on time |
| 5938 | condition. In the same manner, getting a 304 HTTP response back from the |
| 5939 | server, for example after passing a custom If-Match-* header, also |
| 5940 | fulfill this condition. |
| 5941 | |
| 5942 | Fixes #5181 |
| 5943 | Closes #5183 |
| 5944 | |
| 5945 | - [Kwon-Young Choi brought this change] |
| 5946 | |
| 5947 | curl: allow both --etag-compare and --etag-save with same file name |
| 5948 | |
| 5949 | This change inverse the order of processing for the --etag-compare and |
| 5950 | --etag-save option to process first --etag-compare. This in turn allows |
| 5951 | to use the same file name to compare and save an etag. |
| 5952 | |
| 5953 | The original behavior of not failing if the etag file does not exists is |
| 5954 | conserved. |
| 5955 | |
| 5956 | Fixes #5179 |
| 5957 | Closes #5180 |
| 5958 | |
| 5959 | Viktor Szakats (4 Apr 2020) |
| 5960 | - windows: enable UnixSockets with all build toolchains |
| 5961 | |
| 5962 | Extend existing unix socket support in Windows builds to be |
| 5963 | enabled for all toolchain vendors or versions. (Previously |
| 5964 | it was only supported with certain MSVC versions + more recent |
| 5965 | Windows 10 SDKs) |
| 5966 | |
| 5967 | Ref: https://devblogs.microsoft.com/commandline/af_unix-comes-to-windows/ |
| 5968 | Ref: https://github.com/curl/curl/issues/5162 |
| 5969 | Closes: https://github.com/curl/curl/pull/5170 |
| 5970 | |
| 5971 | Daniel Stenberg (4 Apr 2020) |
| 5972 | - KNOWN_BUGS: Store TLS context per transfer instead of per connection |
| 5973 | |
| 5974 | Closes #5102 |
| 5975 | |
| 5976 | Marc Hoersken (3 Apr 2020) |
| 5977 | - sockfilt: remove redundancy in timeout handling |
| 5978 | |
| 5979 | And update other logmsg output in select_ws on Windows. |
| 5980 | |
| 5981 | - sockfilt: fix handling of ready closed sockets on Windows |
| 5982 | |
| 5983 | Replace the incomplete workaround regarding FD_CLOSE |
| 5984 | only signalling once by instead doing a pre-check with |
| 5985 | standard select and storing the result for later use. |
| 5986 | |
| 5987 | select keeps triggering on closed sockets on Windows while |
| 5988 | WSAEventSelect fires only once with data still available. |
| 5989 | By doing the pre-check we do not run in a deadlock |
| 5990 | due to waiting forever for another FD_CLOSE event. |
| 5991 | |
| 5992 | - sockfilt: fix race-condition of waiting threads and event handling |
| 5993 | |
| 5994 | Fix race-condition of waiting threads finishing while events are |
| 5995 | already being processed which lead to invalid or skipped events. |
| 5996 | |
| 5997 | Use mutex to check for one event at a time or do post-processing. |
| 5998 | In addition to mutex-based locking use specific event as signal. |
| 5999 | |
| 6000 | Closes #5156 |
| 6001 | |
| 6002 | Daniel Stenberg (2 Apr 2020) |
| 6003 | - [Leo Neat brought this change] |
| 6004 | |
| 6005 | CI-fuzz: increase fuzz time to 40 minutes |
| 6006 | |
| 6007 | Closes #5174 |
| 6008 | |
| 6009 | Marc Hoersken (2 Apr 2020) |
| 6010 | - CI: increase Azure Pipelines timeouts due to performance issues |
| 6011 | |
| 6012 | The current demand on Azure negatively impacts the CI performance. |
| 6013 | |
| 6014 | - runtests.pl: log host OS as detected by Perl environment |
| 6015 | |
| 6016 | - ftpserver.pl: log before and after data connection is closed |
| 6017 | |
| 6018 | Daniel Stenberg (1 Apr 2020) |
| 6019 | - RELEASE-NOTES: synced |
| 6020 | |
| 6021 | - RELEASE-PROCEDURE.md: run the copyright.pl script! |
| 6022 | |
| 6023 | - vquic/ngtcp2.h: update copyright year range |
| 6024 | |
| 6025 | Follow-up to 0736ee73d346a52 |
| 6026 | |
| 6027 | - [Daiki Ueno brought this change] |
| 6028 | |
| 6029 | CI: add build with ngtcp2 + gnutls on Travis CI |
| 6030 | |
| 6031 | - [Daiki Ueno brought this change] |
| 6032 | |
| 6033 | vquic: add support for GnuTLS backend of ngtcp2 |
| 6034 | |
| 6035 | Currently, the TLS backend used by vquic/ngtcp2.c is selected at compile |
| 6036 | time. Therefore OpenSSL support needs to be explicitly disabled. |
| 6037 | |
| 6038 | Signed-off-by: Daiki Ueno <dueno@redhat.com> |
| 6039 | Closes #5148 |
| 6040 | |
| 6041 | - [Gisle Vanem brought this change] |
| 6042 | |
| 6043 | examples/sessioninfo.c: add include to fix compiler warning |
| 6044 | |
| 6045 | Fixes #5171 |
| 6046 | |
| 6047 | - misc: copyright year updates |
| 6048 | |
| 6049 | Follow-up to 7a71965e9 |
| 6050 | |
| 6051 | - [Harry Sintonen brought this change] |
| 6052 | |
| 6053 | build: fixed build for systems with select() in unistd.h |
| 6054 | |
| 6055 | Closes #5169 |
| 6056 | |
| 6057 | - memdebug: don't log free(NULL) |
| 6058 | |
| 6059 | ... it serves no purpose and fills up the log. |
| 6060 | |
| 6061 | - cleanup: insert newline after if() conditions |
| 6062 | |
| 6063 | Our code style mandates we put the conditional block on a separate |
| 6064 | line. These mistakes are now detected by the updated checksrc. |
| 6065 | |
| 6066 | - checksrc: warn on obvious conditional blocks on the same line as if() |
| 6067 | |
| 6068 | Closes #5164 |
| 6069 | |
| 6070 | - [Roger Orr brought this change] |
| 6071 | |
| 6072 | cmake: add CMAKE_MSVC_RUNTIME_LIBRARY |
| 6073 | |
| 6074 | Fixes #5165 |
| 6075 | Closes #5167 |
| 6076 | |
| 6077 | - [Daiki Ueno brought this change] |
| 6078 | |
| 6079 | ngtcp2: update to git master for the key installation API change |
| 6080 | |
| 6081 | This updates the ngtcp2 OpenSSL backend to follow the API change in |
| 6082 | commit 32e703164 of ngtcp2. |
| 6083 | |
| 6084 | Notable changes are: |
| 6085 | - ngtcp2_crypto_derive_and_install_{rx,tx}_key have been added to replace |
| 6086 | ngtcp2_crypto_derive_and_install_key |
| 6087 | - the 'side' argument of ngtcp2_crypto_derive_and_install_initial_key |
| 6088 | has been removed |
| 6089 | |
| 6090 | Fixes #5166 |
| 6091 | Closes #5168 |
| 6092 | |
| 6093 | - [Cyrus brought this change] |
| 6094 | |
| 6095 | SECURITY.md: minor rephrase |
| 6096 | |
| 6097 | Closes #5158 |
| 6098 | |
| 6099 | - output.d: quote the URL when globbing |
| 6100 | |
| 6101 | Some shells do globbing of their own unless the URL is quoted, so maybe |
| 6102 | encourage this. |
| 6103 | |
| 6104 | Co-authored-by: Jay Satiro |
| 6105 | Closes #5160 |
| 6106 | |
| 6107 | - dist: add tests/version-scan.pl to tarball |
| 6108 | |
| 6109 | ... used in test 1177. |
| 6110 | |
| 6111 | Follow-up to a97d826f6de3 |
| 6112 | |
| 6113 | - test1177: verify that all the CURL_VERSION_ bits are documented |
| 6114 | |
| 6115 | - curl.h: remnove CURL_VERSION_ESNI. Never supported nor documented |
| 6116 | |
| 6117 | Considered experimental and therefore we can do this. |
| 6118 | |
| 6119 | Closes #5157 |
| 6120 | |
| 6121 | - KNOWN_BUGS: DoH doesn't inherit all transfer options |
| 6122 | |
| 6123 | Closes #4578 |
| 6124 | Closes #4579 |
| 6125 | |
| 6126 | - KNOWN_BUGS: DoH leaks memory after followlocation |
| 6127 | |
| 6128 | Closes #4592 |
| 6129 | |
| 6130 | - KNOWN_BUGS: "FTPS needs session reuse" |
| 6131 | |
| 6132 | Closes #4654 |
| 6133 | |
| 6134 | - KNOWN_BUGS: "stick to same family over SOCKS pro" is presumed fixed |
| 6135 | |
| 6136 | - TODO: Set custom client ip when using haproxy protocol |
| 6137 | |
| 6138 | Closes #5125 |
| 6139 | |
| 6140 | Michael Kaufmann (27 Mar 2020) |
| 6141 | - writeout_json: Fix data type issues |
| 6142 | |
| 6143 | Load long values correctly (e.g. for http_code). |
| 6144 | |
| 6145 | Use curl_off_t (not long) for: |
| 6146 | - size_download (CURLINFO_SIZE_DOWNLOAD_T) |
| 6147 | - size_upload (CURLINFO_SIZE_UPLOAD_T) |
| 6148 | |
| 6149 | The unit for these values is bytes/second, not microseconds: |
| 6150 | - speed_download (CURLINFO_SPEED_DOWNLOAD_T) |
| 6151 | - speed_upload (CURLINFO_SPEED_UPLOAD_T) |
| 6152 | |
| 6153 | Fixes #5131 |
| 6154 | Closes #5152 |
| 6155 | |
| 6156 | Daniel Stenberg (27 Mar 2020) |
| 6157 | - mailmap: fixup a few author names/fields |
| 6158 | |
| 6159 | Douglas Steinwand, Gökhan Şengün, Jessa Chandler, Julian Z and |
| 6160 | Svyatoslav Mishyn |
| 6161 | |
| 6162 | - version: add 'cainfo' and 'capath' to version info struct |
| 6163 | |
| 6164 | Suggested-by: Timothe Litt |
| 6165 | URL: https://curl.haxx.se/mail/lib-2020-03/0090.html |
| 6166 | Reviewed-by: Jay Satiro |
| 6167 | |
| 6168 | Closes #5150 |
| 6169 | |
| 6170 | - RELEASE-NOTES: synced |
| 6171 | |
| 6172 | Jay Satiro (26 Mar 2020) |
| 6173 | - SSLCERTS.md: Fix example code for setting CA cert file |
| 6174 | |
| 6175 | Prior to this change the documentation erroneously said use |
| 6176 | CURLOPT_CAPATH to set a CA cert file. |
| 6177 | |
| 6178 | Bug: https://curl.haxx.se/mail/lib-2020-03/0121.html |
| 6179 | Reported-by: Timothe Litt |
| 6180 | |
| 6181 | Closes https://github.com/curl/curl/pull/5151 |
| 6182 | |
| 6183 | Marc Hoersken (26 Mar 2020) |
| 6184 | - sockfilt: add logmsg output to select_ws_wait_thread on Windows |
| 6185 | |
| 6186 | Assisted-by: Jay Satiro |
| 6187 | Reviewed-by: Daniel Stenberg |
| 6188 | |
| 6189 | Closes #5086 |
| 6190 | |
| 6191 | Daniel Stenberg (26 Mar 2020) |
| 6192 | - docs/make: generate curl.1 from listed files only |
| 6193 | |
| 6194 | Previously it rendered the page from files matching "*.d" in the correct |
| 6195 | directory, which worked fine in git builds when the files were added but |
| 6196 | made it easy to forget adding the files to the dist. |
| 6197 | |
| 6198 | Now, only man page sections listed in DPAGES in Makefile.inc will be |
| 6199 | used, thus "forcing" us to update this to get the man page right and get |
| 6200 | it included in the dist at the same time. |
| 6201 | |
| 6202 | Ref: #5146 |
| 6203 | Closes #5149 |
| 6204 | |
| 6205 | - openssl: adapt to functions marked as deprecated since version 3 |
| 6206 | |
| 6207 | OpenSSL 3 deprecates SSL_CTX_load_verify_locations and the MD4, DES |
| 6208 | functions we use. |
| 6209 | |
| 6210 | Fix the MD4 and SSL_CTX_load_verify_locations warnings. |
| 6211 | |
| 6212 | In configure, detect OpenSSL v3 and if so, inhibit the deprecation |
| 6213 | warnings. OpenSSL v3 deprecates the DES functions we use for NTLM and |
| 6214 | until we rewrite the code to use non-deprecated functions we better |
| 6215 | ignore these warnings as they don't help us. |
| 6216 | |
| 6217 | Closes #5139 |
| 6218 | |
| 6219 | - dist: add mail-rcpt-allowfails.d to the tarball |
| 6220 | |
| 6221 | Reported-by: Maksim Stsepanenka |
| 6222 | Reviewed-by: Jat Satiro |
| 6223 | |
| 6224 | Closes #5146 |
| 6225 | |
| 6226 | - travis: update the ngtcp2 build to use the latest OpenSSL patch |
| 6227 | |
| 6228 | ... which also makes it OpenSSL 1.1.1d based and not v3. |
| 6229 | |
| 6230 | Marc Hoersken (24 Mar 2020) |
| 6231 | - CI: remove default Ubuntu build from GitHub Actions |
| 6232 | |
| 6233 | We are already running a very similar Ubuntu build on Travis CI. |
| 6234 | The macOS variant of this default build is kept on Github Actions. |
| 6235 | |
| 6236 | - CI: bring GitHub Actions fuzzing job in line with macOS jobs |
| 6237 | |
| 6238 | Update YAML formatting, job naming and triggers. |
| 6239 | |
| 6240 | - CI: migrate macOS jobs from Azure and Travis CI to GitHub Actions |
| 6241 | |
| 6242 | Reduce workload on Azure Pipelines and Travis CI while |
| 6243 | consolidating macOS jobs onto less utilized GitHub Actions. |
| 6244 | |
| 6245 | Reviewed-by: Daniel Stenberg |
| 6246 | |
| 6247 | Closes #5124 |
| 6248 | |
| 6249 | Daniel Stenberg (24 Mar 2020) |
| 6250 | - config: remove all defines of HAVE_DES_H |
| 6251 | |
| 6252 | As there's no code using it. |
| 6253 | |
| 6254 | Closes #5144 |
| 6255 | |
| 6256 | - copyright: fix out-of-date copyright ranges and missing headers |
| 6257 | |
| 6258 | Reported by the new script 'scripts/copyright.pl'. The script has a |
| 6259 | regex whitelist for the files that don't need copyright headers. |
| 6260 | |
| 6261 | Removed three (mostly usesless) README files from docs/ |
| 6262 | |
| 6263 | Closes #5141 |
| 6264 | |
| 6265 | - packages: add OS400/chkstrings.c to the dist |
| 6266 | |
| 6267 | Reported-by: Jon Rumsey |
| 6268 | Fixes #5142 |
| 6269 | Closes #5143 |
| 6270 | |
| 6271 | - [Clément Notin brought this change] |
| 6272 | |
| 6273 | nghttp2: 1.12.0 required |
| 6274 | |
| 6275 | since nghttp2_session_set_local_window_size is needed |
| 6276 | |
| 6277 | Closes #5140 |
| 6278 | |
| 6279 | - RELEASE-NOTES: synced |
| 6280 | |
| 6281 | - [Calvin Buckley brought this change] |
| 6282 | |
| 6283 | OS400: Update strings for ccsid-ifier |
| 6284 | |
| 6285 | Fixes build. |
| 6286 | |
| 6287 | Closes #5132 |
| 6288 | |
| 6289 | - cirrus: make freebsd ignore the tests instead of skipping |
| 6290 | |
| 6291 | To allow us to see in the CI logs how they actually behave |
| 6292 | |
| 6293 | Closes #5091 |
| 6294 | |
| 6295 | - cirrus: move the sanitizer build from freebsd 13 to freebsd 12 |
| 6296 | |
| 6297 | - Revert "cirrus-ci: disable the FreeBSD 13 builds" |
| 6298 | |
| 6299 | This reverts commit 691b71be930f0e285c8f7a76efd56bbe0576cda6. |
| 6300 | |
| 6301 | - getinfo: provide CURLINFO_HEADER_SIZE and CURLINFO_REQUEST_SIZE override |
| 6302 | |
| 6303 | To let debug-builds return fake values, like in test 970. |
| 6304 | |
| 6305 | Ref: #5131 |
| 6306 | Closes #5136 |
| 6307 | |
| 6308 | - test970: improve the test |
| 6309 | |
| 6310 | - send more data to make problems more obvious |
| 6311 | - don't start the data with minus, it makes diffs harder to read |
| 6312 | - skip the headers in the stdout comparison |
| 6313 | - save to a file name to also verify 'filename_effective' |
| 6314 | |
| 6315 | Ref: #5131 |
| 6316 | |
| 6317 | - CURLINFO_NUM_CONNECTS: improve accuracy |
| 6318 | |
| 6319 | The counter was not bumped in all cases correctly. |
| 6320 | |
| 6321 | Reported-by: Marcel Raad |
| 6322 | Ref: #5131 |
| 6323 | Closes #5135 |
| 6324 | |
| 6325 | - TODO: Use "random" ports for the test servers |
| 6326 | |
| 6327 | - lib/curl_setup: adjust the copyright year range |
| 6328 | |
| 6329 | Follow-up from d820224b8 |
| 6330 | |
| 6331 | Jay Satiro (21 Mar 2020) |
| 6332 | - curl_setup: define _WIN32_WINNT_[OS] symbols |
| 6333 | |
| 6334 | .. because not all Windows build systems have those symbols, and even |
| 6335 | those that do may be missing newer symbols (eg the Windows 7 SDK does |
| 6336 | not define _WIN32_WINNT_WIN10). |
| 6337 | |
| 6338 | Those symbols are used in build-time logic to decide which API to use |
| 6339 | and prior to this change if the symbols were missing it would have |
| 6340 | resulted in deprecated API being used when more recent functions were |
| 6341 | available (eg GetVersionEx used instead of VerifyVersionInfo). |
| 6342 | |
| 6343 | Reported-by: FuccDucc@users.noreply.github.com |
| 6344 | |
| 6345 | Probably fixes https://github.com/curl/curl/issues/4995 |
| 6346 | Closes https://github.com/curl/curl/pull/5057 |
| 6347 | |
| 6348 | - [Ross Burton brought this change] |
| 6349 | |
| 6350 | curl-functions.m4: remove inappropriate AC_REQUIRE |
| 6351 | |
| 6352 | AC_REQUIRE means "if this macro hasn't been executed already, execute |
| 6353 | it". So in a wrapper around AC_RUN_IFELSE, AC_REQUIRE(AC_RUN_IFELSE) |
| 6354 | isn't correct at that will execute AC_RUN_IFELSE without any arguments. |
| 6355 | |
| 6356 | With autoconf 2.69 this is basically a no-op, but with autoconf 2.70, |
| 6357 | AC_RUN_IFELSE without a default value when cross-compiling is fatal. |
| 6358 | The result is that curl with autoconf 2.70 cannot cross-compile. |
| 6359 | |
| 6360 | Fixes https://github.com/curl/curl/issues/5126 |
| 6361 | Closes https://github.com/curl/curl/pull/5130 |
| 6362 | |
| 6363 | Marc Hoersken (20 Mar 2020) |
| 6364 | - ci/tests: fix Azure Pipelines not running Windows containers |
| 6365 | |
| 6366 | Workaround posted here: microsoft/azure-pipelines-agent#2864 |
| 6367 | |
| 6368 | Assisted-by: Simon Chalifoux |
| 6369 | Assisted-by: Tommy Petty |
| 6370 | |
| 6371 | Fixes #5117 |
| 6372 | Closes #5129 |
| 6373 | |
| 6374 | Daniel Stenberg (20 Mar 2020) |
| 6375 | - tests: add test 430, 431 and 432 to verify the --config fix |
| 6376 | |
| 6377 | Verify the fixes in 4e0b4fee4 |
| 6378 | |
| 6379 | - [Rici Lake brought this change] |
| 6380 | |
| 6381 | cmdline: fix handling of OperationConfig linked list (--next) |
| 6382 | |
| 6383 | Ensures that -K/--config inserts new items at the end of the list |
| 6384 | instead of overwriting the second item, and that after a -K/--config |
| 6385 | option has been parsed, the option parser's view of the current config |
| 6386 | is update. |
| 6387 | |
| 6388 | Fixes #5120 |
| 6389 | Closes #5123 |
| 6390 | |
| 6391 | Marc Hoersken (20 Mar 2020) |
| 6392 | - test2100: fix static port instead of dynamic value being used |
| 6393 | |
| 6394 | - test970: fix static ip:port instead of dynamic values being used |
| 6395 | |
| 6396 | Daniel Stenberg (19 Mar 2020) |
| 6397 | - secure transport: remove the BACKEND define kludge |
| 6398 | |
| 6399 | Closes #5122 |
| 6400 | |
| 6401 | - mbedtls: remove the BACKEND define kludge |
| 6402 | |
| 6403 | - bearssl: remove the BACKEND define kludge |
| 6404 | |
| 6405 | - wolfssl: remove the BACKEND define kludge |
| 6406 | |
| 6407 | - nss: remove the BACKEND define kludge |
| 6408 | |
| 6409 | - gnutls: remove the BACKEND define kludge |
| 6410 | |
| 6411 | - openssl: remove the BACKEND define kludge |
| 6412 | |
| 6413 | Use a proper variable instead to make it easier to use a debugger and |
| 6414 | read the code. |
| 6415 | |
| 6416 | Marc Hoersken (19 Mar 2020) |
| 6417 | - tests: make Python-based servers compatible with Python 2 and 3 |
| 6418 | |
| 6419 | Update smbserver.py and negtelnetserver.py to be compatible with |
| 6420 | Python 3 while staying backwards-compatible to support Python 2. |
| 6421 | |
| 6422 | Fix string encoding and handling of echoed and transferred data. |
| 6423 | |
| 6424 | Tested with both Python 2.7.17 and Python 3.7.7 |
| 6425 | |
| 6426 | Reported-by: Daniel Stenberg |
| 6427 | Assisted-by: Kamil Dudka |
| 6428 | Reviewed-by: Marcel Raad |
| 6429 | |
| 6430 | Fixes #5104 |
| 6431 | Closes #5110 |
| 6432 | |
| 6433 | Daniel Stenberg (18 Mar 2020) |
| 6434 | - writeout_json: use curl_off_t printf() option for the time output |
| 6435 | |
| 6436 | Follow-up to: 04c03416e68fd635a15 |
| 6437 | |
| 6438 | Closes #5115 |
| 6439 | |
| 6440 | - RELEASE-NOTES: synced |
| 6441 | |
| 6442 | Uh, I missed this in 1a46b218db |
| 6443 | |
| 6444 | - RELEASE-NOTES: synced |
| 6445 | |
| 6446 | ... and bumped curlver.h to 7.70.0 |
| 6447 | |
| 6448 | Jay Satiro (18 Mar 2020) |
| 6449 | - http2: Fix erroneous debug message that h2 connection closed |
| 6450 | |
| 6451 | Prior to this change in libcurl debug builds http2 stream closure was |
| 6452 | erroneously referred to as connection closure. |
| 6453 | |
| 6454 | Before: |
| 6455 | * nread <= 0, server closed connection, bailing |
| 6456 | |
| 6457 | After: |
| 6458 | * nread == 0, stream closed, bailing |
| 6459 | |
| 6460 | Closes https://github.com/curl/curl/pull/5118 |
| 6461 | |
| 6462 | Daniel Stenberg (18 Mar 2020) |
| 6463 | - tool_setopt: correct the copyright year range |
| 6464 | |
| 6465 | Follow-up to 5450428491 |
| 6466 | |
| 6467 | Jay Satiro (18 Mar 2020) |
| 6468 | - [Johannes Schindelin brought this change] |
| 6469 | |
| 6470 | schannel: add "best effort" revocation check option |
| 6471 | |
| 6472 | - Implement new option CURLSSLOPT_REVOKE_BEST_EFFORT and |
| 6473 | --ssl-revoke-best-effort to allow a "best effort" revocation check. |
| 6474 | |
| 6475 | A best effort revocation check ignores errors that the revocation check |
| 6476 | was unable to take place. The reasoning is described in detail below and |
| 6477 | discussed further in the PR. |
| 6478 | |
| 6479 | --- |
| 6480 | |
| 6481 | When running e.g. with Fiddler, the schannel backend fails with an |
| 6482 | unhelpful error message: |
| 6483 | |
| 6484 | Unknown error (0x80092012) - The revocation function was unable |
| 6485 | to check revocation for the certificate. |
| 6486 | |
| 6487 | Sadly, many enterprise users who are stuck behind MITM proxies suffer |
| 6488 | the very same problem. |
| 6489 | |
| 6490 | This has been discussed in plenty of issues: |
| 6491 | https://github.com/curl/curl/issues/3727, |
| 6492 | https://github.com/curl/curl/issues/264, for example. |
| 6493 | |
| 6494 | In the latter, a Microsoft Edge developer even made the case that the |
| 6495 | common behavior is to ignore issues when a certificate has no recorded |
| 6496 | distribution point for revocation lists, or when the server is offline. |
| 6497 | This is also known as "best effort" strategy and addresses the Fiddler |
| 6498 | issue. |
| 6499 | |
| 6500 | Unfortunately, this strategy was not chosen as the default for schannel |
| 6501 | (and is therefore a backend-specific behavior: OpenSSL seems to happily |
| 6502 | ignore the offline servers and missing distribution points). |
| 6503 | |
| 6504 | To maintain backward-compatibility, we therefore add a new flag |
| 6505 | (`CURLSSLOPT_REVOKE_BEST_EFFORT`) and a new option |
| 6506 | (`--ssl-revoke-best-effort`) to select the new behavior. |
| 6507 | |
| 6508 | Due to the many related issues Git for Windows and GitHub Desktop, the |
| 6509 | plan is to make this behavior the default in these software packages. |
| 6510 | |
| 6511 | The test 2070 was added to verify this behavior, adapted from 310. |
| 6512 | |
| 6513 | Based-on-work-by: georgeok <giorgos.n.oikonomou@gmail.com> |
| 6514 | Co-authored-by: Markus Olsson <j.markus.olsson@gmail.com> |
| 6515 | Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> |
| 6516 | |
| 6517 | Closes https://github.com/curl/curl/pull/4981 |
| 6518 | |
| 6519 | - multi: Improve parameter check for curl_multi_remove_handle |
| 6520 | |
| 6521 | - If an easy handle is owned by a multi different from the one specified |
| 6522 | then return CURLM_BAD_EASY_HANDLE. |
| 6523 | |
| 6524 | Prior to this change I assume user error could cause corruption. |
| 6525 | |
| 6526 | Closes https://github.com/curl/curl/pull/5116 |
| 6527 | |
| 6528 | Viktor Szakats (17 Mar 2020) |
| 6529 | - windows: suppress UI in all CryptAcquireContext() calls |
| 6530 | |
| 6531 | Ref: https://docs.microsoft.com/windows/win32/api/wincrypt/nf-wincrypt-cryptacquirecontexta#parameters |
| 6532 | Reviewed-by: Marc Hörsken |
| 6533 | Closes https://github.com/curl/curl/pull/5088 |
| 6534 | |
| 6535 | Daniel Stenberg (17 Mar 2020) |
| 6536 | - writeout_json: add missing comma to fix the HTTP version |
| 6537 | |
| 6538 | Follow-up to 04c03416e68fd635a15 |
| 6539 | |
| 6540 | - test 970: verify --write-out '%{json}' |
| 6541 | |
| 6542 | Makes curl_easy_getinfo() of "variable" numerical content instead return |
| 6543 | the number set in the env variable `CURL_TIME`. |
| 6544 | |
| 6545 | Makes curl_version() of "variable" textual content. This guarantees a |
| 6546 | stable version string which can be tested against. Environment variable |
| 6547 | `CURL_VERSION` defines the content. |
| 6548 | |
| 6549 | Assisted-by: Mathias Gumz |
| 6550 | |
| 6551 | - [Mathias Gumz brought this change] |
| 6552 | |
| 6553 | writeout: support to generate JSON output |
| 6554 | |
| 6555 | This commit adds support to generate JSON via the writeout feature: |
| 6556 | |
| 6557 | -w "%{json}" |
| 6558 | |
| 6559 | It leverages the existing infrastructure as much as possible. Thus, |
| 6560 | generating the JSON on STDERR is possible by: |
| 6561 | |
| 6562 | -w "%{stderr}%{json}" |
| 6563 | |
| 6564 | This implements a variant of |
| 6565 | https://github.com/curl/curl/wiki/JSON#--write-out-json. |
| 6566 | |
| 6567 | Closes #4870 |
| 6568 | |
| 6569 | - CI: stop ignoring 323, it is disabled |
| 6570 | |
| 6571 | - DISABLED: disable test 323 |
| 6572 | |
| 6573 | The test uses SRP to "a server not supporting it" but modern stunnel |
| 6574 | versions will silently accept it and remain happy. The test is therefore |
| 6575 | faulty. |
| 6576 | |
| 6577 | I haven't figured out how to make stunnel explicitly reject SRP-using |
| 6578 | connects. |
| 6579 | |
| 6580 | Reported-by: Marc Hörsken |
| 6581 | Fixes #5105 |
| 6582 | Closes #5113 |
| 6583 | |
| 6584 | Marc Hoersken (17 Mar 2020) |
| 6585 | - ci/tests: increase timeouts for torture builds on Azure Pipelines |
| 6586 | |
| 6587 | For some reason the torture builds have slowed down recently. |
| 6588 | |
| 6589 | Reported-by: Daniel Stenberg |
| 6590 | |
| 6591 | Daniel Stenberg (16 Mar 2020) |
| 6592 | - cmake: add support for building with wolfSSL |
| 6593 | |
| 6594 | My working build cmdline: |
| 6595 | |
| 6596 | $ cmake -DCMAKE_PREFIX_PATH=$HOME/build-wolfssl -DCMAKE_USE_WOLFSSL=ON . |
| 6597 | |
| 6598 | Assisted-by: Brad King |
| 6599 | Closes #5095 |
| 6600 | |
| 6601 | - tool_operate: fix add_parallel_transfers when more are in queue |
| 6602 | |
| 6603 | Trying to return early from the function if no new transfers were added |
| 6604 | would break the "morep" argument and cause issues. This could lead to |
| 6605 | zero content "transfers" (within quotes since they would never be |
| 6606 | started) when parallel-max was reduced. |
| 6607 | |
| 6608 | Reported-by: Gavin Wong |
| 6609 | Analyzed-by: Jay Satiro |
| 6610 | Fixes #4937 |
| 6611 | Closes #5112 |
| 6612 | |
| 6613 | - vtls: free ssl_config leftovers on out-of-memory |
| 6614 | |
| 6615 | Torture testing 2034 and 2037 found this. |
| 6616 | |
| 6617 | Reported-by: Marc Hörsken |
| 6618 | Fixes #5108 |
| 6619 | Closes #5109 |
| 6620 | |
| 6621 | Marc Hoersken (16 Mar 2020) |
| 6622 | - ci/tests: fix Azure Pipelines not running for pull requests |
| 6623 | |
| 6624 | Closes #5111 |
| 6625 | |
| 6626 | Daniel Stenberg (15 Mar 2020) |
| 6627 | - gskit: update the copyright year range |
| 6628 | |
| 6629 | Follow-up from 083603c63a3 |
| 6630 | |
| 6631 | Marc Hoersken (15 Mar 2020) |
| 6632 | - gskit: use our internal select wrapper for portability |
| 6633 | |
| 6634 | Follow up to c52b342 |
| 6635 | Closes #5106 |
| 6636 | |
| 6637 | - tests: fix verification of stdout in test 1452 due to newline |
| 6638 | |
| 6639 | Fixes test1452:41:1: error: missing </stdout> tag before </verify> |
| 6640 | |
| 6641 | - ci/tests: install impacket for SMB tests on FreeBSD using CirrusCI |
| 6642 | |
| 6643 | Also force the package index/cache to be updated before installing. |
| 6644 | |
| 6645 | Closes #5103 |
| 6646 | |
| 6647 | - tests/README: add note about manually installing python-impacket |
| 6648 | |
| 6649 | Follow up to 4be2560 |
| 6650 | |
| 6651 | Daniel Stenberg (15 Mar 2020) |
| 6652 | - transfer: cap retries of "dead connections" to 5 |
| 6653 | |
| 6654 | When libcurl retries a connection due to it being "seemingly dead" or by |
| 6655 | REFUSED_STREAM, it will now only do it up five times before giving up, |
| 6656 | to avoid never-ending loops. |
| 6657 | |
| 6658 | Reported-by: Dima Tisnek |
| 6659 | Bug: https://curl.haxx.se/mail/lib-2020-03/0044.html |
| 6660 | Closes #5074 |
| 6661 | |
| 6662 | - TODO: TLS-PSK with OpenSSL |
| 6663 | |
| 6664 | Closes #5081 |
| 6665 | |
| 6666 | Marc Hoersken (15 Mar 2020) |
| 6667 | - select: add 'timeout_ms' wrap-around precaution to Curl_select |
| 6668 | |
| 6669 | - select: fix 'pending_ms' is assigned a value that is never used |
| 6670 | |
| 6671 | Detected by Codacy |
| 6672 | |
| 6673 | - select: move duplicate select preparation code into Curl_select |
| 6674 | |
| 6675 | Reviewed by Daniel Stenberg |
| 6676 | Reviewed by Marcel Raad |
| 6677 | Closes #5078 |
| 6678 | |
| 6679 | Daniel Stenberg (15 Mar 2020) |
| 6680 | - connect: happy eyeballs cleanup |
| 6681 | |
| 6682 | Make sure each separate index in connn->tempaddr[] is used for a fixed |
| 6683 | family (and only that family) during the connection process. |
| 6684 | |
| 6685 | If family one takes a long time and family two fails immediately, the |
| 6686 | previous logic could misbehave and retry the same family two address |
| 6687 | repeatedly. |
| 6688 | |
| 6689 | Reported-by: Paul Vixie |
| 6690 | Reported-by: Jay Satiro |
| 6691 | Fixes #5083 |
| 6692 | Fixes #4954 |
| 6693 | Closes #5089 |
| 6694 | |
| 6695 | Marc Hoersken (15 Mar 2020) |
| 6696 | - ci/tests: fix and align setting TFLAGS for make test-nonflaky |
| 6697 | |
| 6698 | - ci/tests: install test suite dependencies stunnel and impacket |
| 6699 | |
| 6700 | - tests: remove python_dependencies for smbserver from our tree |
| 6701 | |
| 6702 | Users of the SMB tests will have to install impacket manually. |
| 6703 | |
| 6704 | Reasoning: our in-tree version of impacket was quite outdated |
| 6705 | and only compatible with Python 2 which is already end-of-life. |
| 6706 | Upgrading to Python 3 and a compatible impacket version would |
| 6707 | require to import additional Python-only and CPython-extension |
| 6708 | dependencies. This would have hindered portability enormously. |
| 6709 | |
| 6710 | Closes #5094 |
| 6711 | |
| 6712 | Jay Satiro (14 Mar 2020) |
| 6713 | - Makefile.m32: Improve windres parameter compatibility |
| 6714 | |
| 6715 | - s/COFF/coff/ |
| 6716 | |
| 6717 | Some versions of windres do not recognize uppercase COFF as a valid |
| 6718 | way to specify the COFF output format. |
| 6719 | |
| 6720 | Reported-by: Steven Penny |
| 6721 | |
| 6722 | Fixes https://github.com/curl/curl/issues/5099 |
| 6723 | Closes https://github.com/curl/curl/pull/5101 |
| 6724 | |
| 6725 | - easy: Fix curl_easy_duphandle for builds missing IPv6 that use c-ares |
| 6726 | |
| 6727 | - Ignore CURLE_NOT_BUILT_IN errors returned by c-ares functions in |
| 6728 | curl_easy_duphandle. |
| 6729 | |
| 6730 | Prior to this change if c-ares was used as the resolver backend and |
| 6731 | either it was too old or libcurl was built without IPv6 support then |
| 6732 | some of our resolver functions could return CURLE_NOT_BUILT_IN to |
| 6733 | curl_easy_duphandle causing it to fail. |
| 6734 | |
| 6735 | Caused by c8f086b which shipped in 7.69.1. |
| 6736 | |
| 6737 | Reported-by: Karl Chen |
| 6738 | |
| 6739 | Fixes https://github.com/curl/curl/issues/5097 |
| 6740 | Closes https://github.com/curl/curl/pull/5100 |
| 6741 | |
| 6742 | Daniel Stenberg (13 Mar 2020) |
| 6743 | - docs: add warnings about FILE: URLs on Windows |
| 6744 | |
| 6745 | - --url man page section |
| 6746 | - libcurl-security.3 gets the full text |
| 6747 | - CURLOPT_URL.3 |
| 6748 | |
| 6749 | Reported-by: Tim Sedlmeyer |
| 6750 | |
| 6751 | - server/getpart: make the "XML-parser" stricter |
| 6752 | |
| 6753 | When extracting a <section> <part> and there's no </part> before |
| 6754 | </section>, this now outputs an error and returns a wrong string to |
| 6755 | make users spot the mistake. |
| 6756 | |
| 6757 | Ref: #5070 |
| 6758 | Closes #5071 |
| 6759 | |
| 6760 | Marc Hoersken (13 Mar 2020) |
| 6761 | - impacket: some more Python 3 code compatibility updates |
| 6762 | |
| 6763 | This makes smbserver load on Python 3, but still not work completely. |
| 6764 | |
| 6765 | - smbserver: pin Python version to 2 since we are not yet 3 compatible |
| 6766 | |
| 6767 | Even though the existing code can be fixed to run on Python 3, the |
| 6768 | tests will fail due to the Unicode transition the protocol is invalid. |
| 6769 | |
| 6770 | Follow up to ee63837 |
| 6771 | Closes #5085 |
| 6772 | |
| 6773 | Daniel Stenberg (12 Mar 2020) |
| 6774 | - [Viktor Szakats brought this change] |
| 6775 | |
| 6776 | cleanup: fix some text/comment typos |
| 6777 | |
| 6778 | Closes #5087 |
| 6779 | |
| 6780 | Marc Hoersken (12 Mar 2020) |
| 6781 | - smbserver: fix Python version specific ConfigParser import |
| 6782 | |
| 6783 | Follow up to ee63837 and 8c7c4a6 |
| 6784 | Fixes #5077 |
| 6785 | |
| 6786 | Daniel Stenberg (11 Mar 2020) |
| 6787 | - RELEASE-NOTES: synced |
| 6788 | |
| 6789 | bumped to 7.69.2 |
| 6790 | |
| 6791 | Dan Fandrich (11 Mar 2020) |
| 6792 | - tests/data: Fix some XML formatting issues in test cases |
| 6793 | |
| 6794 | This allows these test files to pass xmllint. |
| 6795 | |
| 6796 | Daniel Stenberg (11 Mar 2020) |
| 6797 | - [Muhammad Herdiansyah brought this change] |
| 6798 | |
| 6799 | Makefile: run the cd commands in a subshell |
| 6800 | |
| 6801 | In bmake, if the directory is changed (with cd or anything else), bmake |
| 6802 | won't return to the "root directory" on the next command (in the same |
| 6803 | Makefile rule). This commit runs the cd command in a subshell so it |
| 6804 | would work in bmake. |
| 6805 | |
| 6806 | Closes #5073 |
| 6807 | |
| 6808 | - configure: convert -I to -isystem as a last step |
| 6809 | |
| 6810 | As all the -I uses in CFLAGS at that point are for system headers and |
| 6811 | third party libraries this helps us remove/ignore warnings on those! |
| 6812 | |
| 6813 | Closes #5060 |
| 6814 | |
| 6815 | - configure: fix -pedantic-errors for GCC 5 and later |
| 6816 | |
| 6817 | If --enable-werror is used. |
| 6818 | |
| 6819 | Follow-up to d5c0351055d5709da which added it too early in the configure |
| 6820 | script before $compiler_num was set correctly and thus this option was |
| 6821 | never used. |
| 6822 | |
| 6823 | Reported-by: Stepan Efremov |
| 6824 | Fixes #5067 |
| 6825 | Closes #5068 |
| 6826 | |
| 6827 | - configure: document 'compiler_num' for gcc |
| 6828 | |
| 6829 | The CURL_CHECK_COMPILER_GNU_C function sets the number to MAJOR*100 + |
| 6830 | MINOR and ignores the patch version, and since gcc version 7 it only |
| 6831 | sets it to MAJOR*100. |
| 6832 | |
| 6833 | Reported-by: Stepan Efremov |
| 6834 | Ref: #5067 |
| 6835 | Closes #5069 |
| 6836 | |
Haibo Huang | ee03b1a | 2020-03-11 01:33:32 -0700 | [diff] [blame] | 6837 | Version 7.69.1 (11 Mar 2020) |
| 6838 | |
| 6839 | Daniel Stenberg (11 Mar 2020) |
| 6840 | - RELEASE-NOTES: 7.69.1 |
| 6841 | |
| 6842 | - THANKS: from the 7.69.1 release |
| 6843 | |
| 6844 | - [Marc Hoersken brought this change] |
| 6845 | |
| 6846 | test1129: fix invalid case of closing XML-tag and Content-Length |
| 6847 | |
| 6848 | Fixes #5070 |
| 6849 | Closes #5072 |
| 6850 | |
| 6851 | Marc Hoersken (10 Mar 2020) |
| 6852 | - tests/data: fix static ip instead of dynamic value being used |
| 6853 | |
| 6854 | Follow up to 94ced8e |
| 6855 | |
| 6856 | - tests/data: fix static ip:port instead of dynamic values being used |
| 6857 | |
| 6858 | Closes #5065 |
| 6859 | |
| 6860 | - tests/server: fix missing use of exe_ext helper function |
| 6861 | |
| 6862 | Follow up to 9819984 and 3dce984 |
| 6863 | Reviewed-By: Daniel Stenberg |
| 6864 | Closes #5064 |
| 6865 | |
| 6866 | - runtests: log minimal and maximal used port numbers |
| 6867 | |
| 6868 | Daniel Stenberg (9 Mar 2020) |
Haibo Huang | ca2a802 | 2020-07-10 20:17:42 -0700 | [diff] [blame] | 6869 | - [James Fuller brought this change] |
Haibo Huang | ee03b1a | 2020-03-11 01:33:32 -0700 | [diff] [blame] | 6870 | |
| 6871 | sftp: fix segfault regression introduced by #4747 |
| 6872 | |
| 6873 | This fix adds a defensive check for the case where the char *name in |
| 6874 | struct libssh2_knownhost is NULL |
| 6875 | |
| 6876 | Fixes #5041 |
| 6877 | Closes #5062 |
| 6878 | |
| 6879 | - RELEASE-NOTES: synced |
| 6880 | |
| 6881 | - socks4: fix host resolve regression |
| 6882 | |
| 6883 | 1. The socks4 state machine was broken in the host resolving phase |
| 6884 | |
| 6885 | 2. The code now insists on IPv4-only when using SOCKS4 as the protocol |
| 6886 | only supports that. |
| 6887 | |
| 6888 | Regression from #4907 and 4a4b63d, shipped in 7.69.0 |
| 6889 | |
| 6890 | Reported-by: amishmm on github |
| 6891 | Bug: https://github.com/curl/curl/issues/5053#issuecomment-596191594 |
| 6892 | Closes #5061 |
| 6893 | |
| 6894 | - [Patrick Monnerat brought this change] |
| 6895 | |
| 6896 | silly web server: silent a compilation warning |
| 6897 | |
| 6898 | Recent gcc warns when byte count of strncpy() equals the destination |
| 6899 | buffer size. Since the destination buffer is previously cleared and |
| 6900 | the source string is always shorter, reducing the byte count by one |
| 6901 | silents the warning without affecting the result. |
| 6902 | |
| 6903 | Closes #5059 |
| 6904 | |
| 6905 | - [Patrick Monnerat brought this change] |
| 6906 | |
| 6907 | cookie: get_top_domain() sets zero length for null domains |
| 6908 | |
| 6909 | This silents a compilation warning with gcc -O3. |
| 6910 | |
| 6911 | - [Patrick Monnerat brought this change] |
| 6912 | |
| 6913 | test 1560: avoid valgrind false positives |
| 6914 | |
| 6915 | When using maximum code optimization level (-O3), valgrind wrongly |
| 6916 | detects uses of uninitialized values in strcmp(). |
| 6917 | |
| 6918 | Preset buffers with all zeroes to avoid that. |
| 6919 | |
| 6920 | Steve Holme (8 Mar 2020) |
| 6921 | - sha256: Added WinCrypt implementation |
| 6922 | |
| 6923 | Closed #5030 |
| 6924 | |
| 6925 | - sha256: Added SecureTransport implementation |
| 6926 | |
| 6927 | Daniel Stenberg (7 Mar 2020) |
| 6928 | - lib1564: reduce number of mid-wait wakeup calls |
| 6929 | |
| 6930 | This test does A LOT of *wakeup() calls and then calls curl_multi_poll() |
| 6931 | twice. The first *poll() is then expected to return early and the second |
| 6932 | not - as the first is supposed to drain the socketpair pipe. |
| 6933 | |
| 6934 | It turns out however that when given "excessive" amounts of writes to |
| 6935 | the pipe, some operating systems (the Solaris based are known) will |
| 6936 | return EAGAIN before the pipe is drained, which in our test case causes |
| 6937 | the second *poll() call to also abort early. |
| 6938 | |
| 6939 | This change attempts to avoid the OS-specific behaviors in the test by |
| 6940 | reducing the amount of wakeup calls from 1234567 to 10. |
| 6941 | |
| 6942 | Reported-by: Andy Fiddaman |
| 6943 | Fixes #5037 |
| 6944 | Closes #5058 |
| 6945 | |
| 6946 | - [Patrick Monnerat brought this change] |
| 6947 | |
| 6948 | mime: fix the binary encoder to handle large data properly |
| 6949 | |
| 6950 | New test 666 checks this is effective. |
| 6951 | As upload buffer size is significant in this kind of tests, shorten it |
| 6952 | in similar test 652. |
| 6953 | |
| 6954 | Fixes #4860 |
| 6955 | Closes #4833 |
| 6956 | Reported-by: RuurdBeerstra on github |
| 6957 | |
| 6958 | - [Patrick Monnerat brought this change] |
| 6959 | |
| 6960 | mime: do not perform more than one read in a row |
| 6961 | |
| 6962 | Input buffer filling may delay the data sending if data reads are slow. |
| 6963 | To overcome this problem, file and callback data reads do not accumulate |
| 6964 | in buffer anymore. All other data (memory data and mime framing) are |
| 6965 | considered as fast and still concatenated in buffer. |
| 6966 | As this may highly impact performance in terms of data overhead, an early |
| 6967 | end of part data check is added to spare a read call. |
| 6968 | When encoding a part's data, an encoder may require more bytes than made |
| 6969 | available by a single read. In this case, the above rule does not apply |
| 6970 | and reads are performed until the encoder is able to deliver some data. |
| 6971 | |
| 6972 | Tests 643, 644, 645, 650 and 654 have been adapted to the output data |
| 6973 | changes, with test data size reduced to avoid the boredom of long lists of |
| 6974 | 1-byte chunks in verification data. |
| 6975 | New test 667 checks mimepost using single-byte read callback with encoder. |
| 6976 | New test 668 checks the end of part data early detection. |
| 6977 | |
| 6978 | Fixes #4826 |
| 6979 | Reported-by: MrdUkk on github |
| 6980 | |
| 6981 | - [Patrick Monnerat brought this change] |
| 6982 | |
| 6983 | mime: latch last read callback status. |
| 6984 | |
| 6985 | In case a read callback returns a status (pause, abort, eof, |
| 6986 | error) instead of a byte count, drain the bytes read so far but |
| 6987 | remember this status for further processing. |
| 6988 | Takes care of not losing data when pausing, and properly resume a |
| 6989 | paused mime structure when requested. |
| 6990 | New tests 670-673 check unpausing cases, with easy or multi |
| 6991 | interface and mime or form api. |
| 6992 | |
| 6993 | Fixes #4813 |
| 6994 | Reported-by: MrdUkk on github |
| 6995 | |
| 6996 | Marc Hoersken (7 Mar 2020) |
| 6997 | - runtests: fix missing use of exe_ext helper function |
| 6998 | |
| 6999 | Daniel Stenberg (7 Mar 2020) |
| 7000 | - [Ernst Sjöstrand brought this change] |
| 7001 | |
| 7002 | ares: store dns parameters for duphandle |
| 7003 | |
| 7004 | With c-ares the dns parameters lives in ares_channel. Store them in the |
| 7005 | curl handle and set them again in easy_duphandle. |
| 7006 | |
| 7007 | Regression introduced in #3228 (6765e6d), shipped in curl 7.63.0. |
| 7008 | |
| 7009 | Fixes #4893 |
| 7010 | Closes #5020 |
| 7011 | Signed-off-by: Ernst Sjöstrand <ernst.sjostrand@verisure.com> |
| 7012 | |
| 7013 | - version: make curl_version* thread-safe without using global context |
| 7014 | |
| 7015 | Closes #5010 |
| 7016 | |
| 7017 | - RELEASE-NOTES: synced |
| 7018 | |
| 7019 | Marc Hoersken (7 Mar 2020) |
| 7020 | - tests: use native Sleep function as fallback on Windows |
| 7021 | |
| 7022 | Reviewed-By: Daniel Stenberg |
| 7023 | Closes #5054 |
| 7024 | |
| 7025 | - perl: align order and completeness of Windows OS checks |
| 7026 | |
| 7027 | Daniel Stenberg (7 Mar 2020) |
| 7028 | - tool_cb_see: set correct copyright year range |
| 7029 | |
| 7030 | Follow-up to a39e5bfb9 |
| 7031 | |
| 7032 | Marc Hoersken (7 Mar 2020) |
| 7033 | - seek: fix fallback for missing ftruncate on Windows |
| 7034 | |
| 7035 | This fixes test 198 on versions of MinGW-w64 without ftruncate |
| 7036 | |
| 7037 | Reviewed-By: Daniel Stenberg |
| 7038 | Reviewed-By: Marcel Raad |
| 7039 | Closes #5055 |
| 7040 | |
| 7041 | - config-win32: Windows does not have ftruncate |
| 7042 | |
| 7043 | Daniel Stenberg (7 Mar 2020) |
| 7044 | - pause: force a connection (re-)check after unpausing |
| 7045 | |
| 7046 | There might be data available that was already read off the socket, for |
| 7047 | example in the TLS layer. |
| 7048 | |
| 7049 | Reported-by: Anders Berg |
| 7050 | Fixes #4966 |
| 7051 | Closes #5049 |
| 7052 | |
| 7053 | - socks5: switch state properly when the resolve is done |
| 7054 | |
| 7055 | Regression from 4a4b63d (and #4907) |
| 7056 | Reported-by: vitaha85 on github |
| 7057 | Fixes #5053 |
| 7058 | Closes #5056 |
| 7059 | |
| 7060 | Jay Satiro (7 Mar 2020) |
| 7061 | - libssh: Fix matching user-specified MD5 hex key |
| 7062 | |
| 7063 | Prior to this change a match would never be successful because it |
| 7064 | was mistakenly coded to compare binary data from libssh to a |
| 7065 | user-specified hex string (ie CURLOPT_SSH_HOST_PUBLIC_KEY_MD5). |
| 7066 | |
| 7067 | Reported-by: fds242@users.noreply.github.com |
| 7068 | |
| 7069 | Fixes https://github.com/curl/curl/issues/4971 |
| 7070 | Closes https://github.com/curl/curl/pull/4974 |
| 7071 | |
| 7072 | Daniel Stenberg (6 Mar 2020) |
| 7073 | - pause: bail out on bad input |
| 7074 | |
| 7075 | A NULL easy handle or an easy handle without an associated connection |
| 7076 | cannot be paused or unpaused. |
| 7077 | |
| 7078 | Closes #5050 |
| 7079 | |
| 7080 | Steve Holme (6 Mar 2020) |
| 7081 | - unit1612: fixed the inclusion and compilation of the HMAC unit test |
| 7082 | |
| 7083 | Follow up to 3f74e5e6 to fix: |
| 7084 | |
| 7085 | - A typo in Makefile.inc where unit1611 was used instead |
| 7086 | - Some compilation issues in unit1612.c |
| 7087 | |
| 7088 | Closes #5024 |
| 7089 | |
| 7090 | Daniel Stenberg (6 Mar 2020) |
| 7091 | - pause: return early for calls that don't change pause state |
| 7092 | |
| 7093 | Reviewed-by: Patrick Monnerat |
| 7094 | Ref: #4833 |
| 7095 | Closes #5026 |
| 7096 | |
| 7097 | Jay Satiro (6 Mar 2020) |
| 7098 | - curl_share_setopt.3: Note sharing cookies doesn't enable the engine |
| 7099 | |
| 7100 | Follow-up to d0a7ee3 which fixed a bug in 7.66.0 that caused |
| 7101 | CURL_LOCK_DATA_COOKIE to enable the easy handle's cookie engine. |
| 7102 | |
| 7103 | Bug: https://curl.haxx.se/mail/lib-2020-03/0019.html |
| 7104 | Reported-by: Felipe Gasper |
| 7105 | |
| 7106 | Closes https://github.com/curl/curl/pull/5048 |
| 7107 | |
| 7108 | - multi: skip EINTR check on wakeup socket if it was closed |
| 7109 | |
| 7110 | - Don't check errno on wakeup socket if sread returned 0 since sread |
| 7111 | doesn't set errno in that case. |
| 7112 | |
| 7113 | This is a follow-up to cf7760a from several days ago which fixed |
| 7114 | Curl_multi_wait to stop busy looping sread on the non-blocking wakeup |
| 7115 | socket if it was closed (ie sread returns 0). Due to a logic error it |
| 7116 | was still possible to busy loop in that case if errno == EINTR. |
| 7117 | |
| 7118 | Closes https://github.com/curl/curl/pull/5047 |
| 7119 | |
| 7120 | Daniel Stenberg (6 Mar 2020) |
| 7121 | - transfer: set correct copyright year range |
| 7122 | |
| 7123 | - urldata: remove the 'stream_was_rewound' connectdata struct member |
| 7124 | |
| 7125 | ... as it is never set anywhere. |
| 7126 | |
| 7127 | Follow-up to 2f44e94ef |
| 7128 | Closes #5046 |
| 7129 | |
| 7130 | - Revert "pause: force-drain the transfer on unpause" |
| 7131 | |
| 7132 | This reverts commit fa0216b294af4c7113a9040ca65eefc7fc18ac1c (from #5000) |
| 7133 | |
| 7134 | Clearly that didn't solve the problem correctly. |
| 7135 | |
| 7136 | Reported-by: Christopher Reid |
| 7137 | Reopens #4966 |
| 7138 | Fixes #5044 |
| 7139 | |
| 7140 | - RELEASE-NOTES: synced |
| 7141 | |
| 7142 | and bumped curlver.h |
| 7143 | |
| 7144 | - MANUAL: update a dict-using command line |
| 7145 | |
| 7146 | The 'web1913' database is now invalid, use 'gcide' instead. |
| 7147 | |
| 7148 | - KNOWN_BUGS: configure --with-gssapi with Heimdal is ignored on macOS |
| 7149 | |
| 7150 | Closes #3841 |
| 7151 | |
| 7152 | - polarssl: remove more references and mentions |
| 7153 | |
| 7154 | Assisted-by: Jay Satiro |
| 7155 | Follow-up to 6357a19ff29dac04 |
| 7156 | Closes #5036 |
| 7157 | |
| 7158 | Marc Hoersken (4 Mar 2020) |
| 7159 | - tests: wrap ignored test failures in braces |
| 7160 | |
| 7161 | - tests: align some Windows sleep defines with each other |
| 7162 | |
| 7163 | - tests: try to make sleeping portable by avoiding select |
| 7164 | |
| 7165 | select does not support just waiting on Windows: |
| 7166 | https://perldoc.perl.org/perlport.html#select |
| 7167 | |
| 7168 | Reviewed-By: Daniel Stenberg |
| 7169 | Closes #5035 |
| 7170 | |
| 7171 | Daniel Stenberg (4 Mar 2020) |
| 7172 | - runtests.1: rephrase how to specify what tests to run |
| 7173 | |
| 7174 | Also mention the new tilde-prefixed way to ignore test results. |
| 7175 | |
| 7176 | Reviewed-By: Marc Hoersken |
| 7177 | Closes #5033 |
| 7178 | |
| 7179 | - cirrus-ci: disable the FreeBSD 13 builds |
| 7180 | |
| 7181 | FreeBSD 13.0 is apparently close to a year away from a stable release |
| 7182 | and has proven to cause intermittent builds failures recently. |
| 7183 | |
| 7184 | Assisted-by: Dan Fandrich |
| 7185 | Assisted-by: Fedor Korotkov |
| 7186 | Fixes #5028 |
| 7187 | Closes #5029 |
| 7188 | |
Haibo Huang | b51266f | 2020-03-04 02:22:48 -0800 | [diff] [blame] | 7189 | Version 7.69.0 (4 Mar 2020) |
| 7190 | |
| 7191 | Daniel Stenberg (4 Mar 2020) |
| 7192 | - RELEASE-NOTES: 7.69.0 |
| 7193 | |
| 7194 | - THANKS: from 7.69.0 |
| 7195 | |
| 7196 | Now sorted case insensitive |
| 7197 | |
| 7198 | Marc Hoersken (3 Mar 2020) |
| 7199 | - ci/tests: fix escaping of testnames and disable proxy for CI APIs |
| 7200 | |
| 7201 | Follow up to ada581f and c0d8b96 |
| 7202 | Closes #5031 |
| 7203 | |
| 7204 | Jay Satiro (3 Mar 2020) |
| 7205 | - cmake: Show HTTPS-proxy in the features output |
| 7206 | |
| 7207 | - Show HTTPS-proxy in the features output for those backends that |
| 7208 | support it: OpenSSL, GnuTLS and NSS. |
| 7209 | |
| 7210 | Prior to this change HTTPS-proxy was missing from the cmake features |
| 7211 | output even if curl was built with it. Only cmake output was affected. |
| 7212 | Both the library and tool correctly reported the feature. |
| 7213 | |
| 7214 | Bug: https://curl.haxx.se/mail/lib-2020-03/0008.html |
| 7215 | Reported-by: David Lopes |
| 7216 | |
| 7217 | Closes https://github.com/curl/curl/pull/5025 |
| 7218 | |
| 7219 | Marc Hoersken (3 Mar 2020) |
| 7220 | - ci/tests: Make it possible to still run but ignore failing tests |
| 7221 | |
| 7222 | This enables the development of a solution for the failing tests by |
| 7223 | running them on CI while ignoring their result for the overall status. |
| 7224 | |
| 7225 | Closes #4994 |
| 7226 | |
| 7227 | - README.md: add Azure DevOps Pipelines build status badge |
| 7228 | |
| 7229 | - ci/tests: Move CI test result creation above environment setup |
| 7230 | |
| 7231 | This avoids using our test servers as proxy to the AppVeyor API. |
| 7232 | |
| 7233 | Closes #5022 |
| 7234 | |
| 7235 | - ci/tests: Send test results to AppVeyor for status overview |
| 7236 | |
| 7237 | Closes #5021 |
| 7238 | |
| 7239 | Daniel Stenberg (3 Mar 2020) |
| 7240 | - Revert "sha256: Added SecureTransport implementation" |
| 7241 | |
| 7242 | This reverts commit 4feb38deed33fed14ff7c370a6a9153c661dbb9c (from #4956) |
| 7243 | |
| 7244 | That commit broke test 1610 on macos builds without TLS. |
| 7245 | |
| 7246 | Closes #5027 |
| 7247 | |
| 7248 | - dist: include tests/azure.pm in the tarball |
| 7249 | |
| 7250 | Bug: https://github.com/curl/curl/commit/ada581f2cc32f48c1629b729707ac19208435b27#commitcomment-37601589 |
| 7251 | Reported-by: Marcel Raad |
| 7252 | |
| 7253 | Steve Holme (3 Mar 2020) |
| 7254 | - configure.ac: Disable metalink if mbedTLS is specified |
| 7255 | |
| 7256 | Follow up to cdcc9df1 and #5006. Even though I mentioned mbedTLS as |
| 7257 | being one of the backends that metalink needs to be disabled for, I |
| 7258 | seem to have included it in the list of allowed SSL/TLS backends in |
| 7259 | comnfigure.ac :( |
| 7260 | |
| 7261 | Closes #5013 |
| 7262 | |
| 7263 | - sha256: Tidy up following recent changes |
| 7264 | |
| 7265 | Reviewed-by: Daniel Stenberg |
| 7266 | Closes #4956 |
| 7267 | |
| 7268 | - sha256: Added WinCrypt implementation |
| 7269 | |
| 7270 | - sha256: Added SecureTransport implementation |
| 7271 | |
| 7272 | - sha256: Added mbedtls implementation |
| 7273 | |
| 7274 | - sha256: Added GNU TLS gcrypt implementation |
| 7275 | |
| 7276 | - sha256: Added GNU TLS Nettle implementation |
| 7277 | |
| 7278 | Jay Satiro (2 Mar 2020) |
| 7279 | - curl_escape.3: Add a link to curl_free |
| 7280 | |
| 7281 | Ref: https://github.com/curl/curl/pull/5016#issuecomment-593628582 |
| 7282 | |
| 7283 | - curl_getenv.3: Fix the memory handling description |
| 7284 | |
| 7285 | - Tell the user to call curl_free() to free the pointer returned by |
| 7286 | curl_getenv(). |
| 7287 | |
| 7288 | Prior to this change the user was directed to call free(), but that |
| 7289 | would not work in cases where the library and application use separate C |
| 7290 | runtimes and therefore have separate heap memory management. |
| 7291 | |
| 7292 | Closes https://github.com/curl/curl/pull/5016 |
| 7293 | |
| 7294 | Daniel Stenberg (2 Mar 2020) |
| 7295 | - [Nick Zitzmann brought this change] |
| 7296 | |
| 7297 | md4: use init/update/final functions in Secure Transport |
| 7298 | |
| 7299 | We can use CC_MD4_Init/Update/Final without having to allocate memory |
| 7300 | directly. |
| 7301 | |
| 7302 | Closes #4979 |
| 7303 | |
| 7304 | Marc Hoersken (2 Mar 2020) |
| 7305 | - ci/tests: some MacOS builds randomly take longer than 20min |
| 7306 | |
| 7307 | Daniel Stenberg (2 Mar 2020) |
| 7308 | - multi_wait: stop loop when sread() returns zero |
| 7309 | |
| 7310 | It's unclear why it would ever return zero here, but this change fixes |
| 7311 | Robert's problem and it shouldn't loop forever... |
| 7312 | |
| 7313 | Reported-by: Robert Dunaj |
| 7314 | Bug: https://curl.haxx.se/mail/archive-2020-02/0011.html |
| 7315 | Closes #5019 |
| 7316 | |
| 7317 | - http: mark POSTs with no body as "upload done" from the start |
| 7318 | |
| 7319 | As we have logic that checks if we get a >= 400 reponse code back before |
| 7320 | the upload is done, which then got confused since it wasn't "done" but |
| 7321 | yet there was no data to send! |
| 7322 | |
| 7323 | Reported-by: IvanoG on github |
| 7324 | Fixes #4996 |
| 7325 | Closes #5002 |
| 7326 | |
| 7327 | - tests: disable 962, 963 and 964 on Windows |
| 7328 | |
| 7329 | These tests are also doing UTF-8 SMTP. |
| 7330 | |
| 7331 | Follow-up to df207d2dd93b9e73 |
| 7332 | |
| 7333 | Marc Hoersken (2 Mar 2020) |
| 7334 | - ci/tests: fine-tune Azure Pipeline timeouts with a small puffer |
| 7335 | |
| 7336 | Daniel Stenberg (2 Mar 2020) |
| 7337 | - configure: bump the AC_COPYRIGHT year range |
| 7338 | |
| 7339 | - [Steve Holme brought this change] |
| 7340 | |
| 7341 | tests: disable SMTP UTF-8 tests on Windows |
| 7342 | |
| 7343 | Fixes #4988 |
| 7344 | Closes #4992 |
| 7345 | |
| 7346 | - formdata/mime: copyright year range update |
| 7347 | |
| 7348 | Due to the merge/revert cycle |
| 7349 | |
| 7350 | - Revert "mime: latch last read callback status." |
| 7351 | |
| 7352 | This reverts commit 87869e38d7afdec3ef1bb4965711458b088e254f. |
| 7353 | |
| 7354 | Fixes #5014 |
| 7355 | Closes #5015 |
| 7356 | Reopens #4833 |
| 7357 | |
| 7358 | - Revert "mime: do not perform more than one read in a row" |
| 7359 | |
| 7360 | This reverts commit ed0f357f7d25566110d4302f33759f4ffb5a6f83. |
| 7361 | |
| 7362 | - Revert "mime: fix the binary encoder to handle large data properly" |
| 7363 | |
| 7364 | This reverts commit b2caaa0681f329eed317ffb6ae6927f4a539f0c1. |
| 7365 | |
| 7366 | - altsvc: both h3 backends now speak h3-27 |
| 7367 | |
| 7368 | ... also updated the HTTP3 build description for ngtcp2 accordingly. |
| 7369 | |
| 7370 | - [Patrick Monnerat brought this change] |
| 7371 | |
| 7372 | mime: fix the binary encoder to handle large data properly |
| 7373 | |
| 7374 | New test 666 checks this is effective. |
| 7375 | As upload buffer size is significant in this kind of tests, shorten it |
| 7376 | in similar test 652. |
| 7377 | |
| 7378 | Fixes #4860 |
| 7379 | Reported-by: RuurdBeerstra on github |
| 7380 | |
| 7381 | - [Patrick Monnerat brought this change] |
| 7382 | |
| 7383 | mime: do not perform more than one read in a row |
| 7384 | |
| 7385 | Input buffer filling may delay the data sending if data reads are slow. |
| 7386 | To overcome this problem, file and callback data reads do not accumulate |
| 7387 | in buffer anymore. All other data (memory data and mime framing) are |
| 7388 | considered as fast and still concatenated in buffer. |
| 7389 | As this may highly impact performance in terms of data overhead, an early |
| 7390 | end of part data check is added to spare a read call. |
| 7391 | When encoding a part's data, an encoder may require more bytes than made |
| 7392 | available by a single read. In this case, the above rule does not apply |
| 7393 | and reads are performed until the encoder is able to deliver some data. |
| 7394 | |
| 7395 | Tests 643, 644, 645, 650 and 654 have been adapted to the output data |
| 7396 | changes, with test data size reduced to avoid the boredom of long lists of |
| 7397 | 1-byte chunks in verification data. |
| 7398 | New test 664 checks mimepost using single-byte read callback with encoder. |
| 7399 | New test 665 checks the end of part data early detection. |
| 7400 | |
| 7401 | Fixes #4826 |
| 7402 | Reported-by: MrdUkk on github |
| 7403 | |
| 7404 | - [Patrick Monnerat brought this change] |
| 7405 | |
| 7406 | mime: latch last read callback status. |
| 7407 | |
| 7408 | In case a read callback returns a status (pause, abort, eof, |
| 7409 | error) instead of a byte count, drain the bytes read so far but |
| 7410 | remember this status for further processing. |
| 7411 | Takes care of not losing data when pausing, and properly resume a |
| 7412 | paused mime structure when requested. |
| 7413 | New tests 670-673 check unpausing cases, with easy or multi |
| 7414 | interface and mime or form api. |
| 7415 | |
| 7416 | Fixes #4813 |
| 7417 | Reported-by: MrdUkk on github |
| 7418 | Closes #4833 |
| 7419 | |
| 7420 | Steve Holme (1 Mar 2020) |
| 7421 | - unit1651: Fixed conversion compilation warning |
| 7422 | |
| 7423 | 371:17: warning: conversion to 'unsigned char' from 'int' may alter its |
| 7424 | value [-Wconversion] |
| 7425 | |
| 7426 | Closes #5008 |
| 7427 | |
| 7428 | - configure.ac: Disable metalink support if an incompatible SSL/TLS specified |
| 7429 | |
| 7430 | tool_metalink only supports cryptography from OpenSSL, GnuTLS, NSS, |
| 7431 | The Win32 Crypto library and Apple's Common Crypto library. |
| 7432 | |
| 7433 | If an TLS backend such as mbedTLS or WolfSSL is specified then the |
| 7434 | following error is given during compilation along, with a load of |
| 7435 | unresolved extern errors: |
| 7436 | |
| 7437 | Can't compile METALINK support without a crypto library. |
| 7438 | |
| 7439 | Reviewed-by: Daniel Stenberg |
| 7440 | Closes #5006 |
| 7441 | |
| 7442 | Marc Hoersken (1 Mar 2020) |
| 7443 | - ci/tests: Update Azure DevOps pipeline job display names |
| 7444 | |
| 7445 | Make the configure step more descriptive and align others. |
| 7446 | |
| 7447 | - ci/tests: Fix typo in previous commit 597cf2 |
| 7448 | |
| 7449 | - ci/tests: Make sure that the AZURE_ACCESS_TOKEN is available |
| 7450 | |
| 7451 | For security reasons the access token is not available to PR builds. |
| 7452 | Therefore we should not try to use the DevOps API with an empty token. |