Thomas Petazzoni [Fri, 11 Aug 2017 16:52:37 +0000 (18:52 +0200)]
curl/system.h: support more architectures
The long list of architectures in include/curl/system.h is annoying to
maintain, and needs to be extended for each and every architecture to
support.
Instead, let's rely on the __SIZEOF_LONG__ define of the gcc compiler
(we are in the GNUC condition anyway), which tells us if long is 4
bytes or 8 bytes.
This fixes the build of libcurl 7.55.0 on architectures such as
OpenRISC or ARC.
Closes #1766
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Even Rouault [Fri, 11 Aug 2017 09:29:09 +0000 (11:29 +0200)]
parse_proxy(): fix memory leak in case of invalid proxy server name
Fixes the below leak:
$ valgrind --leak-check=full ~/install-curl-git/bin/curl --proxy "http://a:b@/x" http://127.0.0.1
curl: (5) Couldn't resolve proxy name
==5048==
==5048== HEAP SUMMARY:
==5048== in use at exit: 532 bytes in 12 blocks
==5048== total heap usage: 5,288 allocs, 5,276 frees, 445,271 bytes allocated
==5048==
==5048== 2 bytes in 1 blocks are definitely lost in loss record 1 of 12
==5048== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==5048== by 0x4E6CB79: parse_login_details (url.c:5614)
==5048== by 0x4E6BA82: parse_proxy (url.c:5091)
==5048== by 0x4E6C46D: create_conn_helper_init_proxy (url.c:5346)
==5048== by 0x4E6EA18: create_conn (url.c:6498)
==5048== by 0x4E6F9B4: Curl_connect (url.c:6967)
==5048== by 0x4E86D05: multi_runsingle (multi.c:1436)
==5048== by 0x4E88432: curl_multi_perform (multi.c:2160)
==5048== by 0x4E7C515: easy_transfer (easy.c:708)
==5048== by 0x4E7C74A: easy_perform (easy.c:794)
==5048== by 0x4E7C7B1: curl_easy_perform (easy.c:813)
==5048== by 0x414025: operate_do (tool_operate.c:1563)
==5048==
==5048== 2 bytes in 1 blocks are definitely lost in loss record 2 of 12
==5048== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==5048== by 0x4E6CBB6: parse_login_details (url.c:5621)
==5048== by 0x4E6BA82: parse_proxy (url.c:5091)
==5048== by 0x4E6C46D: create_conn_helper_init_proxy (url.c:5346)
==5048== by 0x4E6EA18: create_conn (url.c:6498)
==5048== by 0x4E6F9B4: Curl_connect (url.c:6967)
==5048== by 0x4E86D05: multi_runsingle (multi.c:1436)
==5048== by 0x4E88432: curl_multi_perform (multi.c:2160)
==5048== by 0x4E7C515: easy_transfer (easy.c:708)
==5048== by 0x4E7C74A: easy_perform (easy.c:794)
==5048== by 0x4E7C7B1: curl_easy_perform (easy.c:813)
==5048== by 0x414025: operate_do (tool_operate.c:1563)
Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=2984
Credit to OSS Fuzz for discovery
Marcel Raad [Wed, 9 Aug 2017 12:11:27 +0000 (14:11 +0200)]
build: check out *.sln files with Windows line endings
Visual Studio doesn't like LF line endings in solution files and always
converts them to CRLF when doing changes to the solution. Notably, this
affects the solutions in the release archive.
Adam Sampson [Wed, 9 Aug 2017 13:11:17 +0000 (14:11 +0100)]
curl: do bounds check using a double comparison
The fix for this in 8661a0aacc01492e0436275ff36a21734f2541bb wasn't
complete: if the parsed number in num is larger than will fit in a long,
the conversion is undefined behaviour (causing test1427 to fail for me
on IA32 with GCC 7.1, although it passes on AMD64 and ARMv7). Getting
rid of the cast means the comparison will be done using doubles.
It might make more sense for the max argument to also be a double...
Daniel Stenberg [Sun, 6 Aug 2017 18:10:40 +0000 (20:10 +0200)]
curl: detect and bail out early on parameter integer overflows
Make the number parser aware of the maximum limit curl accepts for a
value and return an error immediately if larger, instead of running an
integer overflow later.
Daniel Stenberg [Sun, 6 Aug 2017 21:42:50 +0000 (23:42 +0200)]
easy_events: make event data static
First: this function is only used in debug-builds and not in
release/real builds. It is used to drive tests using the event-based
API.
A pointer to the local struct is passed to CURLMOPT_TIMERDATA, but the
CURLMOPT_TIMERFUNCTION calback can in fact be called even after this
funtion returns, namely when curl_multi_remove_handle() is called.
Isaac Boukris [Fri, 21 Jul 2017 23:00:46 +0000 (02:00 +0300)]
gssapi: fix memory leak of output token in multi round context
When multiple rounds are needed to establish a security context
(usually ntlm), we overwrite old token with a new one without free.
Found by proposed gss tests using stub a gss implementation (by
valgrind error), though I have confirmed the leak with a real
gssapi implementation as well.
Marcel Raad [Thu, 3 Aug 2017 09:01:25 +0000 (11:01 +0200)]
travis: explicitly specify dist
This makes the builds more reproducible as travis is currently rolling
out trusty as default dist [1]. Specifically, this avoids coverage
check failures when trusty is used as seen in [2] until we figure out
what's wrong.
The MSVC warning level defaults to 3 in CMake. Change it to 4, which is
consistent with the Visual Studio and NMake builds. Disable level 4
warning C4127 for the library and additionally C4306 for the test
servers to get a clean CURL_WERROR build as that warning is raised in
some macros in older Visual Studio versions.
Marcel Raad [Tue, 1 Aug 2017 09:56:41 +0000 (11:56 +0200)]
curl_threads: fix MSVC compiler warning
Use LongToHandle to convert from long to HANDLE in the Win32
implementation.
This should fix the following warning when compiling with
MSVC 11 (2012) in 64-bit mode:
lib\curl_threads.c(113): warning C4306:
'type cast' : conversion from 'long' to 'HANDLE' of greater size
There are some bugs in how timers are managed for a single easy handle
that causes the wrong "next timeout" value to be reported to the
application when a new minimum needs to be recomputed and that new
minimum should be an existing timer that isn't currently set for the
easy handle. When the application drives a set of easy handles via the
`curl_multi_socket_action()` API (for example), it gets told to wait the
wrong amount of time before the next call, which causes requests to
linger for a long time (or, it is my guess, possibly forever).
Daniel Stenberg [Fri, 28 Jul 2017 13:49:36 +0000 (15:49 +0200)]
timeval: struct curltime is a struct timeval replacement
... to make all libcurl internals able to use the same data types for
the struct members. The timeval struct differs subtly on several
platforms so it makes it cumbersome to use everywhere.
MSVC warns that gethostbyname is deprecated. Always use getaddrinfo
instead to fix this when IPv6 is enabled, also for IPv4 resolves. This
is also consistent with what libcurl does.
With warning level 4, MSVC warns about assignments within conditional
expressions. Change the while loop to a do-while loop to fix this. This
change is also consistent with CODE_STYLE.md.
This is consistent with 7bc64561a2e63ca93e4b0b31d350773ba80955c2, which
changed the warning level from 3 to 4 for the Visual Studio project
files. But disable the level 4 warning C4127 "conditional expression is
constant", as that one is issued by older versions of the Windows SDK
as well as curl itself under some circumstances.
build: remove WIN32_LEAN_AND_MEAN from individual build systems
It's defined for all build systems in curl_setup.h since commit beb08481d01a07a8b10938b1078a5e298b1c2912. This caused macro
redefinition warnings in the configure builds.
curl_setup: always define WIN32_LEAN_AND_MEAN on Windows
Make sure to always define WIN32_LEAN_AND_MEAN before including any
Windows headers to avoid pulling in unnecessary headers. This avoids
unnecessary macro clashes and compiler warnings.
Jay Satiro [Tue, 11 Jul 2017 05:53:23 +0000 (01:53 -0400)]
strerror: Preserve Windows error code in some functions
This is a follow-up to af02162 which removed (SET_)ERRNO macros. That
commit was an earlier draft that I committed by mistake, which was then
remedied by a5834e5 and e909de6, and now this commit. With this commit
there is now no difference between the current code and the changes that
were approved in the final draft.
Thanks-to: Max Dymond, Marcel Raad, Daniel Stenberg, Gisle Vanem
Ref: https://github.com/curl/curl/pull/1589