It is only used by default on Windows, but can be enabled by setting
QMAKE_LINK_OBJECT_MAX.
The current file name is .project_name.
Use a sensible file name on this case.
Change-Id: Ic250d024e5feca739027c79a6e285d616234c4ea
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
This option changes the order of include paths, which can cause problems
of various kinds. See https://bugs.debian.org/958479 for an example.
The benefit of that option is minimal for what it was intended.
Pick-to: 5.15 5.12
Change-Id: I80eeabd09764df290b60bc59aeb2f90d07723608
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Ubuntu 20.04 enables -fcf-protection by default.
PCRE2 10.35 sees this but complains that -mshstk is also necessary
to build its JIT. Detect whether the compiler is enabling Intel CET
automatically, and if so, build PCRE2 with the right options.
Pick-to: 5.15
Change-Id: I3440e689b81f4f07055f211a4fa7331a43eb410d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Cascade Lake (CLX) added AVX512VNNI, Ice Lake has added that plus the
integer multiplication (IFMA), the vector bit manipulation instructions
(VBMI and VBMI2), bit algorithms (BITALG), the extended vector AES
instructions (VAES), among others.
I haven't turned on the configure check for those yet. I believe the
VBMI and IFMA instructions will be interesting for Qt image algorithms,
but we'll need some Ice Lakes to do benchmarking on.
Change-Id: Ib5d667bf77a740c28d2efffd15cb43dd821d4b2d
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
The Intel whitepaer[1] recommends using the RDSEED over RDRAND whenever
present. libstdc++ from GCC 10 will also use it in std::random_device.
[ChangeLog][QtCore][QRandomGenerator] The system() random generator will
now use the RDSEED instruction on x86 processors whenever available as
the first source of random data. It will fall back to RDRAND and then to
the system functions, in that order.
[1] https://software.intel.com/en-us/articles/intel-digital-random-number-generator-drng-software-implementation-guide
Change-Id: I907a43cd9a714da288a2fffd15bab176e54e1975
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
As the comment says, Haswell is a nice divider and is a good
optimization target.
I'm using -march=core-avx2 instead of -march=haswell because the latter
form was only added to GCC 4.9 but we still support 4.7 and that has
support for AVX2.
This commit changes the AVX2-optimized code in QtGui to Haswell-
optimized instead. That means, for example, that qdrawhelper_avx2.cpp
can now use the FMA instructions.
Change-Id: If025d476890745368955fffd153129c1716ba006
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Add qmake feature and configure option, which optimze the size of static
exectuable. Use for static build.
Enabled via configure --gc-binaries, or CONFIG += gc-binaries in 3rd party
projects.
Change-Id: I3c25b02caaef6a4afc6019afc9c67122dd11696d
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Fixes the default C version used with gcc < 5
Change-Id: I948dece961caed8e6b181e1c6e6b9dc43c46583e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
It is considered slightly faster than the default mode[1],
but on Windows it causes trouble when aborting the build,
it leaves behind zero-sized object files which cause link
error. See discussion in the bug-make mailing list[2].
[1] https://stackoverflow.com/a/1512947/764870
[2] http://lists.gnu.org/archive/html/bug-make/2017-06/msg00066.html
Change-Id: I7aa0b328a8c743fdfe9b0aece02b329066515076
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
The instruction is "RDRAND", but the feature name, according to GCC, is
RDRND, so I had to change some macros in qsimd_p.h.
Change-Id: Icd0e0d4b27cb4e5eb892fffd14b5166779137e63
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Adds default off configure flag to use compiler optimizations
for size instead of the default speed/size trade-off.
Change-Id: I36702064ef2cc743d2d03a386adf5cefd5371b6e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
The AES instructions were first introduced with the Westmere shrink
(22nm) of the Nehalem architecture. The SHA instructions are still
pending on Intel architecture, but is available on AMD family 17h (gcc
argument -march=znver1).
Both features operate on SSE registers, so that's why the MSVC command-
line argument is the SSE2 one and the configure-time tests depend on
features.sse2.
The qmake feature names end in "ni" because "aes" and "sha" are too
simple and could clash with other uses. The QT_COMPILER_SUPPORTS_ macro
doesn't have the "NI" suffix because it has to match the GCC/Clang
predefined macro.
Change-Id: I445bb15619f6401494e8fffd149dbd1f862ff51c
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Use F16C or ARM FP16 if available at compile time.
Configure check added because older clang compilers have F16C defines
and flags but not all the intrinsics.
Change-Id: I71f358b8fd003e70ab8fcf35097414591e485112
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>
Enables optimizing with -Og if GCC has the option available, this
should produce faster debug binaries without compromising debugability.
Is a privateConfig to limit it to the default Qt build.
Includes two fixes for false positives of maybe_uninitialized triggered
by -Og on gcc 4.9.
Change-Id: I466d7a4070295714189024369312e6cbd36cfacf
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
GCC 5 combined with a recent binutils have a new optimization that
allows them to generate copy relocations even in -fPIE code. Clang has
the same functionality when compiling an executable with -flto. We need
to let the compilers know that they cannot use copy relocations, so they
need to use really position-independent code.
Position independent code throughout is not really required. We just
need the compilers to use position-independent access to symbols coming
from the Qt libraries, but there's currently no other way of doing that.
Task-number: QTBUG-45755
Change-Id: I0d4913955e3745b69672ffff13db5df7377398c5
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
The flags are the same across all GCC flavors so put the definition
in the common gcc-base.conf instead of duplicating it in g++-base.conf
and clang.conf
Change-Id: I1ba2c3c314d3a02b559c384aecef74240f69f659
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Due to common subexpression elimination instruction set extensions may
leak from the objects where they were enabled when doing link-time
optimizations.
To avoid that this patch disables LTCG/LTO on files built with extra
instruction set extensions.
Change-Id: Ie34ad900be7fb04a0dc4d3562187ee170c183333
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
By using the special "ar" and "ranlib" tools, the symbol table is made
visible, so we don't need fat LTO binaries. Since we need to store the
new tool names, we may as well clean up ltcg.prf with variable names for
the fat mode too.
Change-Id: I7e53af0c74a3d069313f38500b72538af1d61128
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
GCC and Clang support compiler intrinsic error detections tools:
address, memory, thread, undefined
Let users conveniently enable it in qmake, for instance with
CONFIG += sanitizer sanitize_address
Also add a -sanitize [...] option to configure to use it by default
for both the Qt libraries, and user applications.
[ChangeLog][configure] Added support for GCC/Clang -fsanitize= options
Change-Id: Ie5418abcdf41842566df510d7707e41739e66f87
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This code hasn't been tested for at least 4 years. It's not maintained
and probably doesn't work.
Change-Id: I4b9a5179e34111b400914f91caa6b741b69771bb
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
GCC currently requires fat object files for static libraries, since the
linker would otherwise not load the .o file from the archive at all and
the linking would fail with a lot of undefined references. Clang on
Linux also needs this, but it has no equivalent flag, so enabling LTCG
for Clang on static libraries will result in linker error.
This commit does not add support for enabling it in configure. It can be
enabled on a per-project basis by doing CONFIG += ltcg or by passing
-config ltcg to qmake's command-line.
Change-Id: I52cf99f1ed9f1701e23a3b457ba3502fd28126ce
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Move it from bootstrap.pro into qt_module.prf so it will apply to any
other bootstrapped libraries, like libQmlDevTools.
Variable called "SPLIT_SECTIONS" because -fdata-sections could be added
in the future, if it proves to be a benefit.
Change-Id: I3fbb004f111620a84e58e9112e9bce3afd95631e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This patch adds a new config option to qmake to enable full optimization
where it makes sense. This currently is supported on all gcc like
compilers by exchanging -O2 for -O3.
In qtbase it is used to enable full optimizations on qtcore and qtgui
and in a later patch can be used to replace similar existing logic in
QtWebKit's WTF and JavaScriptCore modules.
This fixes a performance regression from gcc 4.7 to 4.8 in the software
renderer.
An aliasing error in qregion.cpp which was exposed by more aggresive
optimization has been solved as well.
Change-Id: Ic2c6c41b79cb3846212b40e7bcc11ff492beb27f
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
This commit will make qmake use -isystem automatically for any
compilers that declare support for it for any paths that are listed in
QMAKE_DEFAULT_INCDIRS.
Change-Id: I36fefc6d5bba61671f65669f0ea42704b3c3cf31
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Replace all tabs with proper space characters and consistently align
the '=' characters. The default alignment for the '=' of 25 characters
has been left as is to get a minimal diff. Lines with the '=' further
to the right and those belonging to 'proper code (TM)' have not been
touched.
The work was mostly done using the following python script (might
come in handy again...):
import sys, re
indent_eq = 25 + 0*4 # 25 characters was the most widely used indentation for the '=' character
p = re.compile(r'(\w+)[ \t]*([\-\+]?)(=$|= )[ \t]*(.*$)')
for fn in sys.argv[1:]:
with open(fn, 'r+') as f:
lines = []
nl_count = 0
continuity_indent = None
for l in f:
m = p.match(l)
nl = l
if m:
n_spaces = max(m.start(3), indent_eq - 1) - len(m.group(2)) - len(m.group(1))
if m.group(2) and m.start(2) >= indent_eq-1 and m.start(2) % 4 == 0:
n_spaces -= 1 # left-shift '+=' by one if the '+' is aligned to a multiple of 4
n_spaces = max(1, n_spaces) # we want at least one space before '='/'+='
nl = m.group(1) + ' '*n_spaces + ''.join(m.group(2,3,4)) + '\n'
continuity_indent = nl.find('= ') + 2 if l[-2] == '\\' else None # remember indent on '\\$'
elif continuity_indent:
nl = ' '*continuity_indent + l.lstrip()
if l[-2] != '\\': # check when to stop the continuation
continuity_indent = None
elif l.startswith('#'):
nl = l.expandtabs(2)
if l != nl:
nl_count += 1
lines.append(nl)
if nl_count > 0:
print fn, nl_count, len(lines)
f.seek(0)
f.writelines(lines)
f.truncate()
Change-Id: I1d2870d0a2fe2e30d398c140fe523e69dd20c81b
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Remove all trailing whitespace from the following list of files:
*.cpp *.h *.conf *.qdoc *.pro *.pri *.mm *.rc *.pl *.qps *.xpm *.txt *README
excluding 3rdparty, test-data and auto generated code.
Note A): the only non 3rdparty c++-files that still
have trailing whitespace after this change are:
* src/corelib/codecs/cp949codetbl_p.h
* src/corelib/codecs/qjpunicode.cpp
* src/corelib/codecs/qbig5codec.cpp
* src/corelib/xml/qxmlstream_p.h
* src/tools/qdoc/qmlparser/qqmljsgrammar.cpp
* src/tools/uic/ui4.cpp
* tests/auto/other/qtokenautomaton/tokenizers/*
* tests/benchmarks/corelib/tools/qstring/data.cpp
* util/lexgen/tokenizer.cpp
Note B): in about 30 files some overlapping 'leading tab' and
'TAB character in non-leading whitespace' issues have been fixed
to make the sanity bot happy. Plus some general ws-fixes here
and there as asked for during review.
Change-Id: Ia713113c34d82442d6ce4d93d8b1cf545075d11d
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
The only difference between the two is that iOS append @executable_path/
to QMAKE_LFLAGS_SONAME, but since shared libraries are not supported on
iOS anyways, this is not really something we have to care about.
Change-Id: I4797a4dfb94d9b3af03af22618351b98b48f8255
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Richard Moe Gustavsen <richard.gustavsen@digia.com>
This allows us to have different flags for the compilers for
supporting the same feature. For example, the official flag in GCC to
support AVX2 is -mavx2, but ICC does not support it (yet), requiring
-march=core-avx2 or -xCORE-AVX2. That flag, instead, enables support
for all the features that the "Core-AVX2" processor (codename Haswell)
will support. And clearly, the MSVC flags are different.
Change-Id: I33b6d8617520925e807747180a8dbaafd79b7a9a
Reviewed-by: Sean Harmer <sean.harmer@kdab.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
This significantly reduces the size of the generated code
in places where we don't need exceptions.
The -(no-)exceptions configure flag has been removed in the
process, as there is now a fine grained way to control this
on a per module level, and Qt is being compiled without
exceptions in most places.
Change-Id: I99a15c5d03339db1fbffd4987935d0d671cdbc32
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Modified common MacOS gcc config for iOS compatibility. Added iOS
mkspecs for iOS devices and simulator using G++, LLVM-G++ or CLANG
compilers.
Change-Id: If0a0b1a3ce4e0dc4db319a315e227a9d31549399
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Put in qconfig.h whether qt is compiled with reduced relocations.
When using -Bsymbolic-functions (enabled by default on Qt)
but not -fPIE, the comparison of the function pointers fail
because the addresses are different in Qt, and in the executable.
Hence we now enable -fPIE by default on qmake, and force a compilation
error when it is not enabled and built with reduced relocations.
Done-with: Sune Vuorela <sune@vuorela.dk>
Change-Id: Ib3fdba06fab6e8a93b75b4c6cf16cc973ab335db
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This is the beginning of revision history for this module. If you
want to look at revision history older than this, please refer to the
Qt Git wiki for how to use Git history grafting. At the time of
writing, this wiki is located here:
http://qt.gitorious.org/qt/pages/GitIntroductionWithQt
If you have already performed the grafting and you don't see any
history beyond this commit, try running "git log" with the "--follow"
argument.
Branched from the monolithic repo, Qt master branch, at commit
896db169ea224deb96c59ce8af800d019de63f12
