[PATCH] D53736: [BTF] Add BTF DebugInfo

Yonghong Song via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Nov 26 09:16:36 PST 2018


yonghong-song added a comment.

Thanks for the review! A few comments for some of your concerns.

> I find that there hasn't been actual discussion on the mailing list https://lists.llvm.org/pipermail/llvm-dev/2018-November/127953.html What about making a follow-up to remind people of that RFC? http://llvmweekly.org/ is also a great place to attract more attention.

This is regarding to the format (using auto vs. using explicit type). I do not have a strong preference on this and I just follow some general guideline e.g., using it in iterators, using it if it would just repeat the same already-appeared type name, using a particular "auto" usage pattern just because somebody else already used it that way during my code study.... Let me know what I need to change in this regard.

> I'm still learning these stuff to understand better about the whole picture.

The whole BPF and BTF discussion happens in netdev mailing list https://www.spinics.net/lists/netdev/. You are more than welcome to participate in the discussion. The commit message of this patch provided some limited information about what is BTF and why we want to implement in LLVM.

> Note that dwarf sections are still generated as dwarf is used by user space tools like llvm-objdump etc. for BPF target.

The size benefit is not our goal. As stated in commit message, dwarf is used for user space application and BTF is used for kernel.
The size benefit you mentioned in  https://facebookmicrosites.github.io/bpf/blog/2018/11/14/btf-enhancement.html  is part of work
for compile-once run-everywhere bpf program, presented at linux plumbers conference 2018.
(http://vger.kernel.org/lpc-bpf.html, see the slide under "compile-once run-everywhere BPF program).


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53736/new/

https://reviews.llvm.org/D53736





More information about the llvm-commits mailing list