projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 2507e0a) | patch
Expose ScalarizerPass options to C++ (not just commandline)
authorBenoit Jacob <benoitjacob@google.com>
Mon, 14 Mar 2022 10:07:33 +0000 (11:07 +0100)
committerBjorn Pettersson <bjorn.a.pettersson@ericsson.com>
Mon, 14 Mar 2022 11:00:35 +0000 (12:00 +0100)
commit9879c555f21097aee15e73dd25bd89f652dba8ea
tree5f6f87f621e20a108949aa3efccc58c54502f09atree | snapshot
parent2507e0a257991fd46bd847c8fa0964bc70891addcommit | diff
Expose ScalarizerPass options to C++ (not just commandline)

Context: I needed this for https://github.com/google/iree/pull/8474 .
I found that TSan instrumentation expects vector sizes to be <= 16,
and in my project (IREE) we have tests with higher vector sizes.
That left some test functions uninstrumented, resulting in crashes as
instrumented code called into them.

Differential Revision: https://reviews.llvm.org/D121182
llvm/include/llvm/Transforms/Scalar/Scalarizer.h diff | blob | history
llvm/lib/Transforms/Scalar/Scalarizer.cpp diff | blob | history
Domain: System / Toolchain;