[PATCH] D51231: [X86] Make Feature64Bit useful

Anna Thomas via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Sep 24 11:28:57 PDT 2018


anna added a comment.

In https://reviews.llvm.org/D51231#1243547, @craig.topper wrote:

> @anna, the "lm" in the flags from the cpuinfo should be the 64bit flag. Is this failing with clang, or llc, or some other program that uses llvm libraries?


Interesting. It's failing with our Azul JIT (Falcon) which uses the llvm libraries.
We populate the features using llvm's `sys::getHostCPUFeatures` and set the mcpu and mattr in the `EngineBuilder`.
Note that if I manually add "+64bit" in the feature population (after all features retrieved by `getHostCPUFeatures ` have been added into the mattr), there is no code gen failure on the KVM.


Repository:
  rL LLVM

https://reviews.llvm.org/D51231





More information about the llvm-commits mailing list