[CIR] Fix tests with LLVM_LINK_LLVM_DYLIB #1222
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This option creates and links all tools against a single libLLVM shared
library (and the corresponding CLANG_LINK_CLANG_DYLIB option also gets
turned on by default). In order for this to work, we need to use
LINK_COMPONENTS instead of LINK_LIBS for all LLVM dependencies, and
clang_target_link_libraries for all Clang dependencies, so that they get
rewritten to use the dylib. Remove llvm_update_compile_flags while I'm
here, since the build macros handle that internally. Before this change,
we'd link against certain LLVM libraries both statically and
dynamically, leading to test failures from duplicate singletons.
The way this works for MLIR is fragile right now: MLIR can create its
own dylib as well but doesn't have build system support for linking
against that dylib. We end up folding the MLIR libraries into
libclang-cpp.so (because all Clang dependencies get pulled into it), but
MLIR tools still link the MLIR libraries statically. It'll still work,
but BUILD_SHARED_LIBS is possibly a better alternative for development.
Distributions like Fedora build their LLVM packages with
LLVM_LINK_LLVM_DYLIB, so we'll want to eventually have good MLIR support
for that setup too.