[PATCH] D59780: Support Intel Control-flow Enforcement Technology

Peter Smith via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Mar 26 09:16:22 PDT 2019


peter.smith added a comment.

> Is --force-cet the best option name? H. J. Lu pointed out that gold has -z ibtplt and other options, but I couldn't find these options in the binutils' repository. Is the option really implemneted to gold?

I think the name doesn't quite match the concept. To me --force means "do it anyway", for example "git force push". As it has been implemented, which is error if object hasn't got CET then I'd say that would be better described with "enforce", "mandate", or more simply "require". So I'd say --require-cet would be a better name for that concept.

BTI the AArch64 equivalent of CET, has just landed in binutils. It uses the command line option --force-bti to warn if there is an object with a missing property but always (force) the creation of the BTI enabled PLT and output .note.gnu.property. I would prefer if --force-cet and --force-bti didn't have opposite meanings. I'd like to implement --force-bti the way it has been done in binutils for compatibility, I'd be happy to implement something like --require-bti as well to match the current behaviour of --force-cet.   
binutils patch: https://www.sourceware.org/ml/binutils/2019-03/msg00021.html


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D59780/new/

https://reviews.llvm.org/D59780





More information about the llvm-commits mailing list