Daniel Stenberg [Mon, 11 Dec 2023 18:36:27 +0000 (19:36 +0100)]
readwrite_data: loop less
This function is made to loop in order to drain incoming data
faster. Completely removing the loop has a measerably negative impact on
transfer speeds.
Downsides with the looping include
- it might call the progress callback much more seldom. Especially if
the write callback is slow.
- rate limiting becomes less exact
- a single transfer might "starve out" other parallel transfers
- QUIC timers for other connections can't be maintained correctly
The long term fix should be to remove the loop and optimize coming back
to avoid the transfer speed penalty.
This fix lower the max loop count to reduce the starvation problem, and
avoids the loop completely for when rate-limiting is in progress.
Stefan Eissing [Wed, 13 Dec 2023 10:25:20 +0000 (11:25 +0100)]
lib: eliminate `conn->cselect_bits`
- use `data->state.dselect_bits` everywhere instead
- remove `bool *comeback` parameter as non-zero
`data->state.dselect_bits` will indicate that IO is
incomplete.
Q: "The cmake build selected to run gcc with -fPIC on my box while the
plain configure script did not."
A: With CMake, since 2ebc74c36a19a1700af394c16855ce144d9878e3 #11546
and fc9bfb14520712672b4784e8b48256fb29204011 #11627, we explicitly
enable PIC for libcurl shared lib. Or when building libcurl for
shared and static lib in a single pass. We do this by default for
Windows or when enabled by the user via `SHARE_LIB_OBJECT`.
Otherwise we don't touch this setting. Meaning the default set by
CMake (if any) or the toolchain is used. On Debian Bookworm, this
means that PIC is disabled for static libs by default. Some platforms
(like macOS), has PIC enabled by default.
autotools supports the double-pass mode only, and in that case
CMake seems to match PIC behaviour now (as tested on Linux with gcc.)
Viktor Szakats [Sun, 10 Dec 2023 03:33:17 +0000 (03:33 +0000)]
cmake: delete obsolete TODO items [ci skip]
There is always room for improvement, but CMake is up to par now with
autotools, so there is no longer a good reason to keep around these
inline TODO items.
Answering one of questions:
Q: "The gcc command line use neither -g nor any -O options. As a
developer, I also treasure our configure scripts's --enable-debug
option that sets a long range of "picky" compiler options."
A: CMake offers the `CMAKE_BUILD_TYPE` variable to control debug info
and optimization level. E.g.:
- `Release` = `-O3` + no debug info
- `MinSizeRel` = `-Os` + no debug info
- `Debug` = `-O0` + debug info
Daniel Stenberg [Sat, 9 Dec 2023 23:05:04 +0000 (00:05 +0100)]
cmdline/gen: fix the sorting of the man page options
They were previously sorted based on the file names, which use a .d
extension, making "data" get placed after "data-binary" etc. Making the
sort ignore the extention fixes the ordering.
Reported-by: Boris Verkhovskiy
Bug: https://curl.se/mail/archive-2023-12/0014.html
Closes #12494
Daniel Stenberg [Fri, 8 Dec 2023 08:34:05 +0000 (09:34 +0100)]
tests: rename tests scripts to the test number
It is hard to name the scripts sensibly. Lots of them are similarly
named and the name did not tell which test that used them.
The new approach is rather to name them based on the test number that
runs them. Also helps us see which scripts are for individual tests
rather than for general test infra.
Daniel Stenberg [Tue, 5 Dec 2023 14:55:35 +0000 (15:55 +0100)]
lib: strndup/memdup instead of malloc, memcpy and null-terminate
- bufref: use strndup
- cookie: use strndup
- formdata: use strndup
- ftp: use strndup
- gtls: use aprintf instead of malloc + strcpy * 2
- http: use strndup
- mbedtls: use strndup
- md4: use memdup
- ntlm: use memdup
- ntlm_sspi: use strndup
- pingpong: use memdup
- rtsp: use strndup instead of malloc, memcpy and null-terminate
- sectransp: use strndup
- socks_gssapi.c: use memdup
- vtls: use dynbuf instead of malloc, snprintf and memcpy
- vtls: use strdup instead of malloc + memcpy
- wolfssh: use strndup
Daniel Stenberg [Wed, 6 Dec 2023 13:13:17 +0000 (14:13 +0100)]
url: for disabled protocols, mention if found in redirect
To help users better understand where the URL (and denied scheme) comes
from. Also removed "in libcurl" from the message, since the disabling
can be done by the application.
The error message now says "not supported" or "disabled" depending on
why it was denied:
Protocol "hej" not supported
Protocol "http" disabled
And in redirects:
Protocol "hej" not supported (in redirect)
Protocol "http" disabled (in redirect)
Stefan Eissing [Mon, 4 Dec 2023 11:13:25 +0000 (12:13 +0100)]
ngtcp2: ignore errors on unknown streams
- expecially in is_alive checks on connections, we might
see incoming packets on streams already forgotten and closed,
leading to errors reported by nghttp3. Ignore those.
Same with details:
https://ci.appveyor.com/project/curlorg/curl/builds/48662893/job/ol8a78q9gmilb6wt#L1263
```
tests/libtest/lib1545.c:38:3: error: 'curl_formadd' is deprecated: since 7.56.0. Use curl_mime_init() [-Werror=deprecated-declarations]
38 | curl_formadd(&m_formpost, &lastptr, CURLFORM_COPYNAME, "file",
| ^~~~~~~~~~~~
[...]
```
David Benjamin [Fri, 1 Dec 2023 19:40:32 +0000 (14:40 -0500)]
vtls: consistently use typedef names for OpenSSL structs
The foo_st names don't appear in OpenSSL public API documentation. The
FOO typedefs are more common. This header was already referencing
SSL_CTX via <openssl/ssl.h>. There is a comment about avoiding
<openssl/x509v3.h>, but OpenSSL actually declares all the typedefs in
<openssl/ossl_typ.h>, which is already included by <openssl/ssl.h> (and
every other OpenSSL header), so just use that. Though I've included it
just to be explicit.
(I'm also fairly sure including <openssl/ssl.h> already triggers the
Schannel conflicts anyway. The comment was probably just out of date.)
Stefan Eissing [Fri, 1 Dec 2023 13:00:15 +0000 (14:00 +0100)]
ngtcp2: fix races in stream handling
- fix cases where ngtcp2 invokes callbacks on streams that
nghttp3 has already forgotten. Ignore the NGHTTP3_ERR_STREAM_NOT_FOUND
in these cases as it is normal behaviour.
Emanuele Torre [Fri, 1 Dec 2023 00:51:47 +0000 (01:51 +0100)]
tool_writeout_json: fix JSON encoding of non-ascii bytes
char variables if unspecified can be either signed or unsigned depending
on the platform according to the C standard; in most platforms, they are
signed.
This meant that the *i<32 waas always true for bytes with the top bit
set. So they were always getting encoded as \uXXXX, and then since they
were also signed negative, they were getting extended with 1s causing
'\xe2' to be expanded to \uffffffe2, for example:
Daniel Stenberg [Thu, 30 Nov 2023 09:32:16 +0000 (10:32 +0100)]
multi: during ratelimit multi_getsock should return no sockets
... as there is nothing to wait for then, it just waits. Otherwise, this
causes much more CPU work and updates than necessary during ratelimit
periods.
Dmitry Karpov [Wed, 29 Nov 2023 20:49:06 +0000 (12:49 -0800)]
transfer: abort pause send when connection is marked for closing
This handles cases of some bi-directional "upgrade" scenarios
(i.e. WebSockets) where sending is paused until some "upgrade" handshake
is completed, but server rejects the handshake and closes the
connection.
Daniel Stenberg [Tue, 28 Nov 2023 16:07:44 +0000 (17:07 +0100)]
test1477: verify that libcurl-errors.3 and public headers are synced
The script errorcodes.pl extracts all error codes from all headers and
checks that they are all documented, then checks that all documented
error codes are also specified in a header file.
Daniel Stenberg [Mon, 27 Nov 2023 14:31:00 +0000 (15:31 +0100)]
tool_cb_prg: make the carriage return fit for wide progress bars
When the progress bar was made max width (256 columns), the fly()
function attempted to generate its output buffer too long so that the
trailing carriage return would not fit and then the output would show
wrongly. The fly function is called when the expected total transfer is
unknown, which could be one or more progress calls before the actual
progress meter get shown when the expected transfer size is provided.
This new take also replaces the msnprintf() call with a much simpler
memset() for speed.
Daniel Stenberg [Sat, 25 Nov 2023 23:10:17 +0000 (00:10 +0100)]
conncache: use the closure handle when disconnecting surplus connections
Use the closure handle for disconnecting connection cache entries so
that anything that happens during the disconnect is not stored and
associated with the 'data' handle which already just finished a transfer
and it is important that details from the unrelated disconnect does not
taint meta-data in the data handle.
Like storing the response code.
This also adjust test 1506. Unfortunately it also removes a key part of
the test that verifies that a connection is closed since when this
output vanishes (because the closure handle is used), we don't know
exactly that the connection actually gets closed in this test...
Reported-by: ohyeaah on github
Fixes #12367
Closes #12405
Stefan Eissing [Fri, 24 Nov 2023 12:45:55 +0000 (13:45 +0100)]
quic: make eyeballers connect retries stop at weird replies
- when a connect immediately goes into DRAINING state, do
not attempt retries in the QUIC connection filter. Instead,
return CURLE_WEIRD_SERVER_REPLY
- When eyeballing, interpret CURLE_WEIRD_SERVER_REPLY as an
inconclusive answer. When all addresses have been attempted,
rewind the address list once on an inconclusive answer.
- refs #11832 where connects were retried indefinitely until
the overall timeout fired
Stefan Eissing [Tue, 21 Nov 2023 10:24:18 +0000 (11:24 +0100)]
transfer: cleanup done+excess handling
- add `SingleRequest->download_done` as indicator that
all download bytes have been received
- remove `stop_reading` bool from readwrite functions
- move excess body handling into client download writer
Daniel Stenberg [Thu, 23 Nov 2023 14:52:57 +0000 (15:52 +0100)]
fopen: create new file using old file's mode
Because the function renames the temp file to the target name as a last
step, if the file was previously owned by a different user, not ORing
the old mode could otherwise end up creating a file that was no longer
readable by the original owner after save.
Viktor Szakats [Wed, 22 Nov 2023 15:08:09 +0000 (15:08 +0000)]
openssl: fix building with v3 `no-deprecated` + add CI test
- build quictls with `no-deprecated` in CI to have test coverage for
this OpenSSL 3 configuration.
- don't call `OpenSSL_add_all_algorithms()`, `OpenSSL_add_all_digests()`.
The caller code is meant for OpenSSL 3, while these two functions were
only necessary before OpenSSL 1.1.0. They are missing from OpenSSL 3
if built with option `no-deprecated`, causing build errors:
```
vtls/openssl.c:4097:3: error: call to undeclared function 'OpenSSL_add_all_algorithms'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
vtls/openssl.c:4098:3: error: call to undeclared function 'OpenSSL_add_all_digests'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
```
Ref: https://ci.appveyor.com/project/curlorg/curl-for-win/builds/48587418?fullLog=true#L7667
Regression from b6e6d4ff8f253c8b8055bab9d4d6a10f9be109f3 #12030
Bug: https://github.com/curl/curl/issues/12380#issuecomment-1822944669 Reviewed-by: Alex Bozarth
- vquic/curl_ngtcp2: fix using `SSL_get_peer_certificate` with
`no-deprecated` quictls 3 builds.
Do it by moving an existing solution for this from `vtls/openssl.c`
to `vtls/openssl.h` and adjusting caller code.
```
vquic/curl_ngtcp2.c:1950:19: error: implicit declaration of function 'SSL_get_peer_certificate'; did you mean 'SSL_get1_peer_certificate'? [-Wimplicit-function-declaration]
```
Ref: https://github.com/curl/curl/actions/runs/6960723097/job/18940818625#step:24:1178
- curl_ntlm_core: fix `-Wunused-parameter`, `-Wunused-variable` and
`-Wunused-function` when trying to build curl with NTLM enabled but
without the necessary TLS backend (with DES) support.
Viktor Szakats [Wed, 22 Nov 2023 01:08:33 +0000 (01:08 +0000)]
curl.h: delete Symbian OS references
curl deprecated Symbian OS in 3d64031fa7a80ac4ae3fd09a5939196268b92f81
via #5989. Delete references to it from public headers, because there
is no fresh release to use those headers with.
Reviewed-by: Dan Fandrich Reviewed-by: Jay Satiro
Closes #12378
Viktor Szakats [Tue, 21 Nov 2023 16:54:49 +0000 (16:54 +0000)]
windows: use built-in `_WIN32` macro to detect Windows
Windows compilers define `_WIN32` automatically. Windows SDK headers
or build env defines `WIN32`, or we have to take care of it. The
agreement seems to be that `_WIN32` is the preferred practice here.
Make the source code rely on that to detect we're building for Windows.
Public `curl.h` was using `WIN32`, `__WIN32__` and `CURL_WIN32` for
Windows detection, next to the official `_WIN32`. After this patch it
only uses `_WIN32` for this. Also, make it stop defining `CURL_WIN32`.
There is a slight chance these break compatibility with Windows
compilers that fail to define `_WIN32`. I'm not aware of any obsolete
or modern compiler affected, but in case there is one, one possible
solution is to define this macro manually.
grepping for `WIN32` remains useful to discover Windows-specific code.
Also:
- extend `checksrc` to ensure we're not using `WIN32` anymore.
- apply minor formatting here and there.
- delete unnecessary checks for `!MSDOS` when `_WIN32` is present.
Co-authored-by: Jay Satiro Reviewed-by: Daniel Stenberg
Closes #12376
Stefan Eissing [Tue, 21 Nov 2023 14:46:41 +0000 (15:46 +0100)]
url: ConnectionExists revisited
- have common pattern of `if not match, continue`
- revert pages long if()s to return early
- move dead connection check to later since it may
be relatively expensive
- check multiuse also when NOT building with NGHTTP2
- for MULTIUSE bundles, verify that the inspected
connection indeed supports multiplexing when in use
(bundles may contain a mix of connection, afaict)
Viktor Szakats [Wed, 15 Nov 2023 14:43:36 +0000 (14:43 +0000)]
build: add more picky warnings and fix them
Enable more picky compiler warnings. I've found these options in the
nghttp3 project when implementing the CMake quick picky warning
functionality for it [1].
`-Wunused-macros` was too noisy to keep around, but fixed a few issues
it revealed while testing.
- autotools: reflect the more precisely-versioned clang warnings.
Follow-up to 033f8e2a08eb1d3102f08c4d8c8e85470f8b460e #12324
- autotools: sync between clang and gcc the way we set `no-multichar`.
- autotools: avoid setting `-Wstrict-aliasing=3` twice.
- autotools: disable `-Wmissing-noreturn` for MSYS gcc targets [2].
It triggers in libtool-generated stub code.
- lib/timeval: delete a redundant `!MSDOS` guard from a `WIN32` branch.
- lib/curl_setup.h: delete duplicate declaration for `fileno`.
Added in initial commit ae1912cb0d494b48d514d937826c9fe83ec96c4d
(1999-12-29). This suggests this may not be needed anymore, but if
it does, we may restore this for those specific (non-Windows) systems.
- lib: delete unused macro `FTP_BUFFER_ALLOCSIZE` since c1d6fe2aaa5a26e49a69a4f2495b3cc7a24d9394.
- lib: delete unused macro `isxdigit_ascii` since f65f750742068f579f4ee6d8539ed9d5f0afcb85.
- lib/mqtt: delete unused macro `MQTT_HEADER_LEN`.
- lib/multi: delete unused macro `SH_READ`/`SH_WRITE`.
- lib/hostip: add `noreturn` function attribute via new `CURL_NORETURN`
macro.
- lib/mprintf: delete duplicate declaration for `Curl_dyn_vprintf`.
- lib/rand: fix `-Wunreachable-code` and related fallouts [3].
- lib/setopt: fix `-Wunreachable-code-break`.
- lib/system_win32 and lib/timeval: fix double declarations for
`Curl_freq` and `Curl_isVistaOrGreater` in CMake UNITY mode [4].
- lib/warnless: fix double declarations in CMake UNITY mode [5].
This was due to force-disabling the header guard of `warnless.h` to
to reapply it to source code coming after `warnless.c` in UNITY
builds. This reapplied declarations too, causing the warnings.
Solved by adding a header guard for the lines that actually need
to be reapplied.
- lib/vauth/digest: fix `-Wunreachable-code-break` [6].
- lib/vssh/libssh2: fix `-Wunreachable-code-break` and delete redundant
block.
- lib/vtls/sectransp: fix `-Wunreachable-code-break` [7].
- lib/vtls/sectransp: suppress `-Wunreachable-code`.
Detected in `else` branches of dynamic feature checks, with results
known at compile-time, e.g.
```c
if(SecCertificateCopySubjectSummary) /* -> true */
```
Likely fixable as a separate micro-project, but given SecureTransport
is deprecated anyway, let's just silence these locally.
- src/tool_help: delete duplicate declaration for `helptext`.
- src/tool_xattr: fix `-Wunreachable-code`.
- tests: delete duplicate declaration for `unitfail` [8].
- tests: delete duplicate declaration for `strncasecompare`.
- tests/libtest: delete duplicate declaration for `gethostname`.
Originally added in 687df5c8c39c370a59999b9afc0917d808d978b7
(2010-08-02).
Got complicated later: c49e9683b85ba9d12cbb6eebc4ab2c8dba68fbdc
If there are still systems around with warnings, we may restore the
prototype, but limited for those systems.
- tests/lib2305: delete duplicate declaration for
`libtest_debug_config`.
- tests/h2-download: fix `-Wunreachable-code-break`.
projects / thirdparty / curl.git / log
