This patch-set implements a new QSslSocket backend based on OpenSSL 1.1.
1. General.
The code in this patch was organized to achieve these (somewhat contradicting)
objectives:
- keep the new code free of #if-ery, as far as possible;
- make it easy to clean away dead code when we're eventually able to retire
out-dated OpenSSL versions;
- reduce the amount of code duplication.
If changes in some file/component were insignificant (~5 one-liners per file),
we still use pp-checks like: #if QT_CONFIG(opensslv11) ... #else ... #endif -
the logic is simple and it's still easy to clean the code if we remove the legacy
back-end. Where it saved #if-ery, we also introduced 'forward-compatible'
macros implementing equivalents of 1.1 functions using older OpenSSL.
In case some class contains a lot of version-specific ifdefs (particularly where
nested #if-ery was complex) we choose to split code into: "pre11" h/cpp files,
"shared" h/cpp files (they preserve their original names, e.g qsslsocket_openssl.cpp)
and "11" h/cpp files. If in future we remove the legacy back-end, "pre11" should be
removed; "shared" and "11" parts - merged.
2. Configuration.
We introduced a new feature 'opensslv11' which complements the pre-existing
'openssl' and 'openssl-linked' features. The 'opensslv11' feature is enabled
by a simple test which either compiles successfully or ends in a compilation
error, depending on a value of the OPENSSL_VERSION_NUMBER constant. If the
feature was enabled, we also append an additional compilation flag
-DOPENSSL_API_COMPAT=0x10100000L to make sure our new code does not contain
deprecated structures, function calls, macro-invocations from OpenSSL < 1.1.
Change-Id: I2064efbe9685def5d2bb2233a66f7581954fb74a
Reviewed-by: André Klitzing <aklitzing@gmail.com>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The getentropy function, first found in OpenBSD, is present in glibc
since version 2.25 and Bionic since Android 6.0 and NDK r11. It uses the
Linux 3.17 getrandom system call. Unlike glibc's getrandom() wrapper,
the glibc implementation of getentropy() function is not a POSIX thread
cancellation point, so we prefer to use that even though we have to
break the reading into 256-byte blocks.
The big advantage is that these functions work even in the absence of a
/dev/urandom device node, in addition to a few cycles shaved off by not
having to open a file descriptor and close it at exit. What's more, the
glibc implementation blocks until entropy is available on early boot, so
we don't have to worry about a failure mode. The Bionic implementation
will fall back by itself to /dev/urandom and, failing that, gathering
entropy from elsewhere in the system in a way it cannot fail either.
uClibc has a wrapper to getrandom(2) but no getentropy(3). MUSL has
neither.
Change-Id: Ia53158e207a94bf49489fffd14c8cee1b968a619
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Follow the usual pattern:
Add a config test and automatic include of GLES3/gl32.h if there
is a GLES 3.2 capable header+lib at build time.
Then, regardless of this being enabled, expose all new 3.2 API
functions in QOpenGLExtraFunctions and resolve them dynamically
at run time.
This way 3.2 functions will be available when deployed to a 3.2
capable system (or OpenGL 3/4.x with the functions in question
available) regardless of what was present in the sysroot at build
time.
Change-Id: Ia52551f3178591e1e56ceac8e45d89c6b13f4927
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
GCC didn't support it until version 5 or 6, so add configure tests for
both <random> and <sys/auxv.h>. Normally I'd say "upgrade", but this is
too low-level and important a feature.
There's a good chance that all our supported compilers have <random>
anyway. As for <sys/auxv.h>, it's present on Glibc, Bionic and MUSL, but
I don't see it in uClibc (AT_RANDOM is a Linux-specific feature).
Change-Id: Ia3e896da908f42939148fffd14c5b2af491f7a77
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
This was originally enabled in the mkspecs for 64-bit QNX 7.0.0
but that broke when the qtConfig change was made. It looks like
qtConfig shouldn't be used in the platform mkspecs. I suspect
the stack-protector changes were left out of the 32-bit mkspecs
so that 6.6.0 builds wouldn't be affected.
Ignore the stack-protector/stack-protector-all possibility since
it isn't possible to access it without a command line option.
Specifying both options doesn't even make sense since
stack-protector-all encompasses stack-protector.
For now, leave out command line control of this feature.
Task-number: QTBUG-59644
Change-Id: I99323216be5b592dd2c3bef6d22da195764a6e65
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
A typo caused the test to never detect the system wide PCRE.
Task-number: QTBUG-59226
Change-Id: I42ada99aac240455d11b53d2ab59d712d8f811ff
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The patch fixes a number of bugs in code, and removes dead logic
clarifying that MIPS DSP, like ARM NEON, has no runtime detecton.
Change-Id: If2f4eea68da5b2eaa80b8e9c8258206d8c1b7173
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
PCRE1 is going towards EOL. PCRE2 is the way forward in terms
of new features, performance, and security improvements. The
APIs that QRegularExpression uses are similar so the required
modifications aren't extensive.
The biggest difference comes to JIT-compiling of the pattern.
In PCRE1, JIT-compiling did not modify the processed PCRE pattern,
but returned a new chunk of data.
This allowed multiple threads to keep matching using the same
processed data and NULL for the JIT data, until a thread
JIT-compiled and atomically set the shared JIT data to the results
of the compilation.
In PCRE2, JIT-compiling _modifies_ the processed PCRE pattern in a
way that it's thread unsafe [1]; the results of JIT-compilation
are stored somewhere inside the processed pattern.
This means the above approach cannot work -- a thread may be
matching while another one JIT-compiles, causing a data race.
While waiting for better workarounds from upstream, employ a
read/write mutex to protect the matching from JIT-compilation.
[1] https://lists.exim.org/lurker/message/20160104.105831.3cb25b39.en.html
[ChangeLog][General] QRegularExpression now requires the PCRE2
library, at least version 10.20. Support for the PCRE1 library
has been dropped. A copy of PCRE2 is shipped with Qt and will
automatically be used on those platforms which lack it.
Change-Id: I9fe11104230a096796df2d0bdcea861acf769f57
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This commit re-enables support for OpenVG in Qt, but not in the
same way as in Qt 4.8. The first part is about adding a test
and using the new configure system to enable OpenVG.
There is still support code in Qt for setting up EGL to provide a
surface and context for rendering with the OpenVG API, this commit
enables a path to do so.
Normally to get access to an EGLContext from a QWindow you do so via
QOpenGLContext, but in setups without OpenGL but with EGL and OpenVG
this doesn't make sense (there would be no QOpenGLContext). So the
intended way is to use a QWindow to get an EGLSurface, then create
an EGLContext directly (without going through QPA).
Change-Id: I0f75aadbaa3cd006deb7e6fd12cfbb574870fba4
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
Recent Khronos headers decided to break the world by guarding all
function prototypes with GL_GLEXT_PROTOTYPES which has traditionally
been used for extension headers only.
Until this gets corrected - see
https://lists.freedesktop.org/archives/mesa-dev/2016-September/128654.html
- add the define to the config tests and qopengl.h.
While 5.7 already has some of the qopengl.h fixes due to an upgraded
ANGLE shipping with newer headers, this is a cross-platform issue that
will surface everywhere eventually. Therefore we target the full set
of fixes to 5.6.
This time we also make sure the forced define of GL_GLEXT_PROTOTYPES
is removed before including the ext header, thus apps get the ext
protos only if they actually requested them.
Task-number: QTBUG-56764
Change-Id: Ib2c6d2e7b71b8fb8683424f43e6289e64e4ee46c
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
There may not be a version when libinput is not picked up via
pkg-config. Add a config test instead.
Task-number: QTBUG-56017
Change-Id: I421af4cef1b896413a4ebda561809a8b2a3386b3
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
convert the ugly config.tests/[...]/freetype.pri file into a custom
callback in configure.pri, and reinstate pkg-config use for freetype.
subsequently, use QMAKE_USE for the actual library references.
this fixes in particular cross-builds, as the new configure was not
passing the necessary information to the test any more, so the old .pri
file misbehaved.
Task-number: QTBUG-54911
Change-Id: I5fc9c254334a2675f7db4d54df4c77637e8e2487
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
5971b88e is not needed in new configure.
This merge also reverts "fix QMAKE_DEFAULT_*DIRS resolution with
apple SDK", 2c9d15d7, because it breaks iOS build with new
configure system.
Conflicts:
mkspecs/features/default_pre.prf
mkspecs/features/mac/toolchain.prf
mkspecs/features/toolchain.prf
src/dbus/qdbusconnection.cpp
src/plugins/sqldrivers/mysql/qsql_mysql.cpp
src/sql/drivers/mysql/qsql_mysql.cpp
src/widgets/widgets/qmenubar.cpp
src/widgets/widgets/qmenubar_p.h
tools/configure/configureapp.cpp
tools/configure/environment.cpp
tools/configure/environment.h
Change-Id: I995533dd334211ebd25912db05b639d6f908aaec
Yes, yes, this is just a configure test, but why do something stupid
in the code and then have to shut up Coverity manually?
Fix by making it a global, which means it will be zero-initialized
(I didn't want to do the obvious = 0, as that could protentially
create a "0 used as nullptr" warning at some point in the future.
Coverity-Id: 59485
Change-Id: I49ecd28be983a0e42b420d20da0db34a872c6f44
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This allows setting of variables, like FOO=bar.
Yes, it's intentional that there are no quotes.
Change-Id: Ib306f8f647014b399b87ffff13f1d9e6a10fa2f8
(cherry picked from commit e79200bf7f)
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Mike Krus <mike.krus@kdab.com>
so far, each library was distributed over a test and (optionally) a
'library' output of a feature. this was conceptually messy and limiting.
so instead, turn libraries into a category of their own.
libraries now support multiple properly separated sources, which makes
overriding them a lot saner. sources can be conditional to accommodate
platform differences.
as an immediate consequence, move (almost) all library references from
the config test projects to the json file.
a few tests were excluded, because they are doing somewhat magic things
that should not be handled in this bulk change:
- freetype: .pri file shared with actual source code
- clock-gettime: -lrt is conditional, and there is a .pri file which is
shared with actual source code
- ipc_posix: -lrt & -lpthread conditional
- iconv: -liconv conditional
the multi-source mechanism is used to make a variety of tests work on
windows, where the library name differs from unix (and sometimes between
build configurations). some tests still needed minor adjustments to
actually work.
on the way, fix up disagreements between manually specified libraries
and pkg-config lines (affecting several xcb-related tests).
Change-Id: Ic8c58556fa0cf8f981d386b13ea34b4431b127c5
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
it's bound to the bourne shell, which is not readily available on
windows hosts.
on the way, the pch, fvisibility, and bsymbolic_functions tests were
rewritten as regular compile tests. they now just verify that qmake's
built-in support for the tested features actually works.
Change-Id: Ibac246f21b5ececa40da3f576dc789982eaf9fdf
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
mitshm, xcursor, xfixes, xrandr, xshape, and xsync were dead for a long
time (see also 4cb795cbdb).
glxfbconfig was also dead (see also d54b77d55).
x11/notype and x11/xkb became dead in 4535913c4f.
javascriptcore-jit became dead in 24f1025663.
stdint was another webkit vestige (see also 1b716724f7).
Change-Id: I04f408cb917c767951645c6445f15f24378fa43a
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Conflicts:
configure
5.7 now supports clang on android; but dev re-worked configure
src/gui/kernel/qevent.h
One side renamed a parameter of a constructor; the other added an
alternate constructor on the next line. Applied the rename to both
for consistency.
tests/auto/tools/moc/tst_moc.cpp
Each side added a new test at the end.
.qmake.conf
Ignored 5.7's change to MODULE_VERSION.
configure.json
No conflict noticed by git; but changes in 5.7 were needed for the
re-worked configure to accommodate 5.7's stricter handling of C++11.
Change-Id: I9cda53836a32d7bf83828212c7ea00b1de3e09d2
Conflicts:
src/corelib/io/qtemporarydir.cpp
One side encapsulated a repeated piece of #if-ery in a local define;
the other added to the #if-ery. Made its addition to the other's.
src/corelib/kernel/qeventdispatcher_unix_p.h
One side moved some members into a struct; this collided with a #undef
check that neither side now has. Discarded the #undef part.
src/gui/opengl/qopengltexturehelper_p.h
5.7 deleted a bunch of methods; not clear why merge got confused.
src/tools/moc/moc.cpp
One added a name to the copyright header; another changed its URL.
Change-Id: I9e9032b819f030d67f1915445acf2793e98713fa
On macOS, the test script is passed the full path to the compiler, like
/usr/local/bin/icpc. That doesn't match "icpc".
Change-Id: I149e0540c00745fe8119fffd1463c87b8f6a89b2
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Detect host and target architecture from within qmake, and set
QT_ARCH accordingly.
Change-Id: I30255f88c7645d197bd07355a1dff02b377cbbe8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reference an SQLite3 function to verify that the library is being
linked. Discovered that the test didn't do this when I switched the
Windows configure to use a compile test to determine whether the
system has SQLite3. The test passed even though the initial
configure changes failed to provide the test with information about
the SQLite3 libraries.
Change-Id: I3114cfc2dec3a42a60c8e1e432eb8375b440d7e0
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Where accept4() is used, NetBSD offers paccept() as a replacement function.
Modify check for using accept4() and use paccept() on NetBSD.
See http://netbsd.gw.com/cgi-bin/man-cgi?paccept++NetBSD-current
and http://reviews.llvm.org/D12485
Change-Id: I9b3ecba5f3afad6c357d3f7b8f89589bf313e273
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Add pkg-config to the libpng usages as not all systems have the
symlink libpng.so -> libpng<version>.so (affected: NetBSD) that changes
with the version of the lib. If no-pkg-config is used, use -lpng as
before. Tested with FreeBSD 10.3, NetBSD 7.0.1 using png 1.6.21
Change-Id: I5c87f380c84da3d5c56c94da53adb900791c8caa
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Command line arguments, configure tests and features are now
defined in a json file and a configure.pri containing some
custom functions.
qmake uses the json file to determine command line arguments,
tests and features to be executed at configuration time.
A new qt_configure.prf contains all the infrastructure to parse
the command line, run the configure tests, determine the set of
available features and create a report about it.
Change-Id: If30ac089dd3e9f8c67ebd642bde10cfa960893bb
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
The use of clock_gettime() is limited to systems having _POSIX_TIMERS
defined, however OpenBSD implements clock_gettime() but does not have the
posix define. Enable using clock_gettime() on OpenBSD as well.
Change-Id: I785954fe61b42b15755ca625a766c9a95179ae8d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Qt does not support libressl which reports itself as openssl but uses
a high version number. This means that we expect it to have features
that it doesn't and will fail to build. Instead detect this situation
at configure time and disable the ssl support.
Change-Id: I73cca4d1544df2aaca0723c38ba63758c287ec41
Reviewed-by: Ralf Nolden <nolden@kde.org>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
