OSDN Git Service

[MSAN] Handle x86 {round,min,max}sd intrinsics
authorGui Andrade <guiand@google.com>
Sat, 27 Jun 2020 06:46:04 +0000 (06:46 +0000)
committerGui Andrade <guiand@google.com>
Sat, 27 Jun 2020 06:46:04 +0000 (06:46 +0000)
commiteae84b41fe8e138bf0425ff4d009dee81574620a
treefab73f72694e5b694c231a5fd56243aa7772ebab
parent9aa9855a9c253274d277a7ef0c9897a52cebf1f0
[MSAN] Handle x86 {round,min,max}sd intrinsics

These need special handling over the simple vector intrinsics as they
behave more like a shuffle operation: taking the top half of the vector
from one input, and the bottom half separately. Previously, these were
being handled as though all bits of all operands were combined.

Differential Revision: https://reviews.llvm.org/D82398
llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp