projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 9a35f19) | patch
Revert "[RISCV] Avoid changing etype for splat of 0 or -1"
authorPhilip Reames <preames@rivosinc.com>
Wed, 29 Jun 2022 17:24:38 +0000 (10:24 -0700)
committerPhilip Reames <listmail@philipreames.com>
Wed, 29 Jun 2022 17:27:02 +0000 (10:27 -0700)
commitdd48d3ad0e5fff78e6e3a7fd8ca830624ccb8e33
tree2ab1c8d8bb50d447c47cf1de932c24bac652d5catree | snapshot
parent9a35f19e3e851e5a62499416e276ff9ae13b7b07commit | diff
Revert "[RISCV] Avoid changing etype for splat of 0 or -1"

This reverts commit 755c84c62cda80b0acf51ccc5653fc6d64536f7e.  A bug was reported on the original review thread (https://reviews.llvm.org/D128006), and on inspection this patch is simply wrong.  It needs to be checking for VLInBytes, not MaxVL.  These happen to be the same when using AVL=VLMAX (which is quite common), but this does not fold when AVL != VLMAX.
llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp diff | blob | history
llvm/test/CodeGen/RISCV/rvv/constant-folding-crash.ll diff | blob | history
llvm/test/CodeGen/RISCV/rvv/fixed-vectors-calling-conv.ll diff | blob | history
llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-setcc.ll diff | blob | history
llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp2i.ll diff | blob | history
Domain: System / Toolchain;