en Copyright 2015 Java http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp#2962f9df stop llvm-reduce from introducing undefsDifferential Revision: https://reviews.llvm.org/D128317 List of files: /llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp Tue, 21 Jun 2022 23:30:52 +0000 John Regehr <regehr@cs.utah.edu> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp#a5bbc6ef [NFC] Remove unnecessary "#include"s from header files List of files: /llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp Wed, 23 Feb 2022 09:20:48 +0000 Bill Wendling <isanbard@gmail.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp#6f288bd7 [llvm-reduce] Count chunks by running a preliminary reductionHaving a separate counting method runs the risk of a mismatch betweenthe actual reduction method and the counting method.Instead, create an Oracle that always returns true for shouldKeep(), runthe reduction, and count how many times shouldKeep() was called. Themodule should not be modified if shouldKeep() always returns true.Reviewed By: MeinersburDifferential Revision: https://reviews.llvm.org/D113537 List of files: /llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp Fri, 12 Nov 2021 02:45:59 +0000 Arthur Eubanks <aeubanks@google.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp#be0b47d5 [llvm-reduce] Skip replacing metadata and callee operandsMetadata operands tend to require special conditions, especially on dbgintrinsics. We also don't have a zero value for metadata.Replacing callee operands is a little weird, since calling undef/nulldoesn't make sense. It also causes tons of invalid reductions whenreducing calls to intrinsics since only arguments to intrinsics can beof the metadata type.Reviewed By: MeinersburDifferential Revision: https://reviews.llvm.org/D113532 List of files: /llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp Wed, 10 Nov 2021 01:47:43 +0000 Arthur Eubanks <aeubanks@google.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp#177a7037 [llvm-reduce] Actually skip invalid candidates in operands-to-argsThis was checked while counting but not actually when doing the reduction, resulting in crashes.Reviewed By: MeinersburDifferential Revision: https://reviews.llvm.org/D112766 List of files: /llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp Thu, 28 Oct 2021 22:50:34 +0000 Arthur Eubanks <aeubanks@google.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp#dd71b65c [llvm-reduce] Introduce operands-to-args pass.Instead of setting operands to undef as the "operands" pass does,convert the operands to a function argument. This avoids having tointroduce undef values into the IR which have some unpredictabilityduring optimizations.For instance, define void @func() { entry: %val = add i32 32, 21 store i32 %val, i32* null ret void }is reduced to define void @func(i32 %val) { entry: %val1 = add i32 32, 21 store i32 %val, i32* null ret void }(note that the instruction %val is renamed to %val1 when printingthe IR to avoid ambiguity; ideally %val1 would be removed by dce or theinstruction reduction pass)Any call to @func is replaced with a call to the function with thenew signature and filled with undef. This is not ideal for IPA passes,but those out-of-scope for now.Reviewed By: aeubanksDifferential Revision: https://reviews.llvm.org/D111503 List of files: /llvm-project-15.0.7/llvm/tools/llvm-reduce/deltas/ReduceOperandsToArgs.cpp Wed, 13 Oct 2021 14:05:54 +0000 Michael Kruse <llvm-project@meinersbur.de>