Issue: #15441

Florian Apolloner authored 8 years ago and Brad King committed 8 years ago

Leave it undocumented for now because we intend to use it internally and
it cannot be made available everywhere.

Parse the `-gencode=`, `-arch`, and `-code` flags and generate a
`CodeGeneration` field in the project file.

Parse the `-cudart=` option and add a corresponding `CudaRuntime`
field to the generated project file.  Also add a matching `.lib`
to the list of libraries linked.

The CUDA Toolkit's VS integration provides abstractions for host
compiler options for `nvcc` to pass through `-Xcompiler` to the host
MSVC.  Populate our secondary flag table and use it to remove flags from
the `AdditionalCompilerOptions` in favor of their abstractions.

Unfortunately a bug in the CUDA 8.0 VS integration prevents us from
passing anything in `AdditionalCompilerOptions` reliably.  After taking
out the flags that have dedicated abstractions, drop the rest.

The CUDA Toolkit's VS integration does not properly compute the
intermediate files directory location when the path to the source
file is absolute.

Generate the `CudaCompile` elements in `.vcxproj` files.

The CUDA Toolkit's VS integration defines abstractions for both options
to `nvcc` and options to pass through `-Xcompiler` to the host MSVC.
We need a separate flag table to parse each set of flags into the
corresponding abstractions.  Add empty placeholders for these tables.

Remove a flag from the map and return its value.

This will allow a client to parse flags, replace the flag tables, and
then re-parse a field in which flags for a secondary tool were
collected.

If `CMAKE_GENERATOR_TOOLSET` does not have a `cuda=...` field then
find available CUDA toolsets and choose the highest version.

The NVIDIA CUDA Toolkit provides MSBuild toolset files for integration
with Visual Studio.  Multiple versions may be installed so we need a way
to tell our VS generators which CUDA toolset to use.  Extend the
`CMAKE_GENERATOR_TOOLSET` specification to provide a `cuda=...` field
specifying the version number.

Run MSBuild on a simple `.vcxproj` file to extract the location of the
toolset definitions.  This will later be useful for looking at available
BuildCustomizations.

Use `cmGeneratorTarget::ModuleDefinitionInfo` to combine the
implementation of `WINDOWS_EXPORT_ALL_SYMBOLS` with that of using a
`.def` file as a source.  Only one of these could be used within a
single target before anyway.

Create a `ModuleDefinitionInfo` structure for each configuration of a
target to hold corresponding information about the selected module
definition file (`.def` source).

Add a `cmGeneratorTarget::GetModuleDefinitionSources` method.

Steven Newbury authored 8 years ago and Brad King committed 8 years ago

Detect x32-abi through CMakeCompilerABI infrastruture and use this
information at runtime to determine the correct library paths with
`FIND_LIBRARY_USE_LIBX32_PATHS`.

Fixes: #15994

We need this to make Large File Support definitions consistent with
other translation units.

Michael Stürmer authored 8 years ago and Brad King committed 8 years ago

Fixes: #16697

The ordering is assumed in many places throughout the code and
modification of the enumeration should be done carefully.