While testing for {rB40dcf686f04f}, compiler flags got mixed up and
non-working ASan configuration was committed.
Platform file, which is `include`d after the `CMAKE_C_FLAGS_DEBUG` etc.,
are set, overwrites those flags instead of appending to them. To fix this,
`PLATFORM_CFLAGS` is used to pass the `-fsanitize=*` flags to the C/C++
compiler.
Tested on fresh build using both Xcode and Ninja, with & without ccache.
Also silence a clang warning for multi-config generators:
the object size sanitizer has no effect at -O0, but is explicitly
enabled: -fsanitize=object-size [-Winvalid-command-line-argument]
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8879
For work the GP team plans to land soon (T79877) potrace was taken
on as an additional optional dependency.
This diff adds building the library to the deps builder and takes
care of the integration into the build-system with the `WITH_POTRACE`
cmake switch.
Differential Revision: https://developer.blender.org/D8662
Reviewed by: brecht, sergey
* Don't link against Mesa OpenGL software emulation libraries from the
lib folder, they are not intended to be used for building, only install.
* Don't use static OpenMP library by default, it's usually not built to
work in shared libraries.
* Disable jemalloc on all platforms, the reason it fails is not specific
to Linux.
Differential Revision: https://developer.blender.org/D8827
Every time CMake is re-run, Xcode shows a popup asking if
user wants to manage schemes automatically or manually.
Building Blender wiki page recommends managing schemes automatically.
This change sets the default behavior to "automatically" and generates
the .xcscheme files while CMake is running, instead of hogging Xcode
later on. With tests enabled, the number of schemes is 203.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8820
CMake caches the location of the CRT runtime in the
MSVC_REDIST_DIR variable, and uses it to copy the required
dll's during the install phase.
This variable is only initialized when it does not exist.
Leading to issues when compiler updates are installed and
the compiler version slightly changes, cmake still looks
in the old location for the runtime and warns about the
files not existing.
This change fixes the issue by checking if the redist dir
exists and if not unsets it so InstallRequiredSystemLibraries
can have another go at figuring out where they live.
This is for design task T67744, Boolean Redesign.
It adds a choice of solver to the Boolean modifier and the
Intersect (Boolean) and Intersect (Knife) tools.
The 'Fast' choice is the current Bmesh boolean.
The new 'Exact' choice is a more advanced algorithm that supports
overlapping geometry and uses more robust calculations, but is
slower than the Fast choice.
The default with this commit is set to 'Exact'. We can decide before
the 2.91 release whether or not this is the right choice, but this
choice now will get us more testing and feedback on the new code.
It is now possible to build against a shared embree library.
Before it was only possible to build against static Embree libraries.
Reviewed By: Brecht
Differential Revision: http://developer.blender.org/D8702
OSL requires RTTI to be off, this is done with the /GR- flag for
MSVC, however /GR is in the default CXX flags leading to warning
D9025 : overriding '/GR' with '/GR-'
which cannot be suppressed.
/GR is on by default and this flag is not required, so removing
it from the default CXX flags makes it possible later use /GR-
without generating warnings.
Add `-Wno-maybe-uninitialized` option to `CMAKE_CXX_FLAGS_RELEASE` and
`CMAKE_CXX_FLAGS_RELWITHDEBINFO` variables on GCC/Linux.
In Release builds GCC's `-Wmaybe-uninitialized` warning is unreliable,
and thus causes noise that can drown out other warnings. These warnings
are now silenced in release mode builds.a
Debug builds seem fine, so flags for debug builds are not touched by
this commit.
No functional changes.
Reviewed By: Sergey
Differential Revision: https://developer.blender.org/D8615
This patch changes openvdb from a static to a dynamic library.
this is in preparation for enabling pyopenvdb at some point
in the future.
Differential Revision: https://developer.blender.org/D8282
Reviewed by: brecht
Fix package name missmatch in a few module files. IE "ALEMBIC" was
defined in the file but the find_package commands used "Alembic"
Some modules state that they set and use the _LIBRARY variable but the
do in fact not do this. Removed these comments from those files.
Required for the new boolean code, disabled by default
until all platforms have landed the libs and the boolean
code actually lands in master.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8384
Add our own copy of the gtest discovery scripts from CMake a few reasons:
* Use the very latest version which supports PRE_TEST for Windows
* Fix usage of [] symbols in file paths that fail with the zsh shell
* Disable asan leak checker when discovering tests
This means Windows also no longer requires the very latest CMake 3.18.
This bumps the minimum requirement for cmake from 3.10 to 3.18 on windows
if `WITH_GTESTS` is enabled.
Reviewed By: sergey brecht sybren campbellbarton
Differential Revision: https://developer.blender.org/D8405
Description of `USER_HEADER_SEARCH_PATHS` build setting:
"
This is a list of paths to folders to be searched by the compiler
for included or imported user header files (those headers listed
in quotes) when compiling C, Objective-C, C++, or Objective-C++.
Paths are delimited by whitespace, so any paths with spaces in
them need to be properly quoted. See Always Search User Paths
(Deprecated) (ALWAYS_SEARCH_USER_PATHS) for more details
on how this setting is used. If the compiler doesn't support the
concept of user headers, then the search paths are prepended to
the any existing header search paths defined in Header Search
Paths (HEADER_SEARCH_PATHS).
"
http://help.apple.com/xcode/mac/current/#/itcaec37c2a6
Xcode doesn't use `HEADER_SEARCH_PATHS` for auto-complete. Only the
header files in the same directory as the current file are suggested.
CMake as of now correctly sets `SYSTEM_HEADER_SEARCH_PATHS` and lumps the
rest in `HEADER_SEARCH_PATHS`. The standard way is to use
`USER_HEADER_SEARCH_PATHS` & `SYSTEM_HEADER_SEARCH_PATHS` and let
`HEADER_SEARCH_PATHS` be used as a fallback for compilers which do not
distinguish between `<*.h>` and `"*.h"` syntax.
So set `USER_HEADER_SEARCH_PATHS` to the include paths specified
in the `CMakeLists.txt` files of all targets.
In the situation where the precompiled libraries are used on Linux +
GCC, a version of GCC older than 9.3 is guaranteed to cause problems.
This just implents a fatal error message when we know it doesn't make
sense to continue. We could do more checks and add some warnings, but
it's very likely that these will be ignored amongst the other noise.
Reviewed By: sergey, brecht
Differential Revision: https://developer.blender.org/D8396
This patch changes the discovery of pre-compiled kernels, to look for any PTX, even if
it does not match the current architecture version exactly. It works because the driver can
JIT-compile PTX generated for architectures less than or equal to the current one.
This e.g. makes it possible to render on a new GPU architecture even if no pre-compiled
binary kernel was distributed for it as part of the Blender installation.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8332
This commit introduces a new way to build unit tests. It is now possible
for each module to generate its own test library. The tests in these
libraries are then bundled into a single executable.
The test executable can be run with `ctest`. Even though the tests
reside in a single executable, they are still exposed as individual
tests to `ctest`, and thus can be selected via its `-R` argument.
Not yet ported tests still build & run as before.
The following rules apply:
- Test code should reside in the same directory as the code under test.
- Tests that target functionality in `somefile.{c,cc}` should reside in
`somefile_test.cc`.
- The namespace for tests is the `tests` sub-namespace of the code under
test. For example, tests for `blender::bke` should be in
`blender::bke:tests`.
- The test files should be listed in the module's `CMakeLists.txt` in a
`blender_add_test_lib()` call. See the `blenkernel` module for an
example.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7649
Enabling all `make deps` dependencies with the exception of Embree and OIDN.
After that, Blender can be compiled on an Apple Silicon Mac just like on any
Intel based Mac. There are still compiler warnings that need to be
investigated and there are probably a couple of bug still to be discovered
and to be fixed.
Most patches to the dependencies are simple and are about disabling SSE and
setting the proper architecture to compiile for. Notable exception is Python,
where I back ported a yet to be accepted PR for upstream Python:
https://github.com/python/cpython/pull/21249
Cross compiling or buliding a Universal Binary is not supported yet.
The minimum macOS target version for x86_64 remains at 10.13, the target
for arm64 is 11.00.
Differential Revision: https://developer.blender.org/D8236
C++17 does not work on 10.12, and Apple extended support ended for 10.12 in
October 2019.
Maniphest Tasks: T76783, T76184
Differential Revision: https://developer.blender.org/D8179
The spelling and capitalization of package name passed to find_package()
and find_package_handle_standard_args() needs to match.
Silences CMake warning about mismatch.
Differential Revision: https://developer.blender.org/D8247
Clang Tidy is a Clang based "linter" tool which goal is to help
fixing typical programming errors.
It is run as a separate compile step of every file, which slows
compilation down but allows to fully analyze the file the same
way as compiler does and catch non-trivial bugprone cases.
This change includes:
- CMake option called `WITH_CLANG_TIDY` which enables Clang Tidy
linter tool on all source in the `source/` directory.
This option is only available on Linux, as it is currently the
easiest platform to get the Clang Tidy toolchain to work.
- CMake module which is aimed to find latest available Clang Tidy.
- Set of rules which allows to have Blender fully compiled without
extra issues.
The goal of this change is to provide a base ground so that solving
all the warnings can happen later on, as a team effort.
It should be possible to use Clang Tidy side-by-side with both GCC
and Clang, but there seems to be some tweaks to be done in CMake to
make it really work for Blender. For now use Clang toolchain if
there are issues with GCC+Clang Tidy.
It will be worked on in the nearest future to bring seamless
experience for all configurations.
Currently there is no official way of getting Clang Tidy on macOS,
and on Windows there are some difficulties of hooking up Clang Tidy
from LLVM package to the MSVC compiler toolchain.
The actual warnings in the code will be addressed as a part of the
Code Quality Days, task T78535.
Differential Revision: https://developer.blender.org/D7937
This requires ISPC for building OpenImageDenoise, so that is now added as
a dependency as well. Blender itself does not need ISPC for building so it
is not included as part of the precompiled libraries.
Differential Revision: https://developer.blender.org/D7641
This is the cluster of OIIO and friends , since they are all kinda tangled best to deal with this as a single unit
OIIO 2.1.15.0
png 1.6.37
jpeg 2.0.4
opencolorio 1.1.1
tiff 4.1.0
OSL 1.10.10
pugixml 1.10
openjpeg 2.3.1
Differential Revision: https://developer.blender.org/D7727
Reviewed by: brecht
The file subversion is no longer used in the Python API or user interface,
and is now internal to Blender.
User interface, Python API and file I/O metadata now use more consistent
formatting for version numbers. Official releases use "2.83.0", "2.83.1",
and releases under development use "2.90.0 Alpha", "2.90.0 Beta".
Some Python add-ons may need to lower the Blender version in bl_info to
(2, 83, 0) or (2, 90, 0) if they used a subversion number higher than 0.
https://wiki.blender.org/wiki/Reference/Release_Notes/2.83/Python_API#Compatibility
This change is in preparation of LTS releases, and also brings us more
in line with semantic versioning.
Fixes T76058.
Differential Revision: https://developer.blender.org/D7748
For a more detailed description of the issue see the commit
message for rB497cd3d7dd6e497be484eb78a8ddb23f53b20343
This change moves fftw to a shared library and reverts the bandaid
we did for 2.83.
This is not as much a fix as a work around, but given the real
involves replacing how we build fftw, it is not eligible for 2.83
which is in BCON3 already.
The root of the issue lies with (how we build) fftw3
The first issue is: fftw does not build with MSVC, there are other
dependencies that are not compatible with MSVC and for those we
build the libraries required with mingw64, same for fftw
The second issue is: for reasons unknown we really really really
liked all deps to link statically so wherever possible we did so.
Now during the building of the fftw it linked a few symbols from
libgcc (which we do not ship) like __chkstk_ms, for which we passed
some flags to stop generating calls to it. Problem solved! There
is no way this could possibly turn around and bite us in the rear.
fast forward to today mystery crashes that look like a race condition.
What is happening is, we tell the linker that each thread will require
a 2-megabyte stack, now if every thread immediately allocated 2 megs,
that be 'rough' on the memory usage. So, what happens is (for all apps
not just blender), 2 megs are reserved but not backed by any real memory
and the first page is allocated for use by the stack, now as the stack
grows, it will eventually grow out of that first page, and end up in
an area that has not been allocated yet, to deal with that the allocated
page is followed by a guard page, someone touches the guard page it's
time to grow the stack!
Meanwhile in FFTW is it's doing substantial allocation using alloca
(up to 64 kb) on the stack, jumping over the guard page, and ending
up in reserved but not yet committed memory, causing an access violation.
Now if you think, that doesn't sound right! something should have
protected us from that! You are correct! That thing was __chkstk_ms
which we disabled.
Given we do not want a dependency on libgcc while building with MSVC
the proper solution is to build fftw as a shared library which will
statically link any bits and pieces it needs, however that change
is a little bit too big to be doing in BCON3.
So as a work around, we change the size the stack grows from 8k to
68k which gives fftw a little bit more wiggle room to keep it out
of trouble most of the time.
Note this only sidesteps the issue, this may come up again if the
conditions are just right, and a proper solution will need to be
implemented for 2.90.
Search for all potential library names in each directory, otherwise e.g.
libImath-2_2.a from a system directory will be preferred over libImath.a even
if we specified a directory.
It was disabled in D7520 to keep the pdb's from growing out
of control however the increased link time is just not worth
it.
I'll keep an eye on the dailies and see if we have to come up
with a different solution.
