[PATCH] D115007: Dump swift5 reflection section data into dsym bundle generated binary with dsymutil

Jonas Devlieghere via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Jan 11 09:13:28 PST 2022


JDevlieghere added inline comments.


================
Comment at: llvm/include/llvm/MC/MCObjectFileInfo.h:448
+  void initMachOMCObjectFileInfo(const Triple &T,
+                                 StringRef Swift5ReflectionSegmentName = {});
   void initELFMCObjectFileInfo(const Triple &T, bool Large);
----------------
rastogishubham wrote:
> aprantl wrote:
> > rastogishubham wrote:
> > > JDevlieghere wrote:
> > > > Similar as above, this seems really out of place.
> > > I don't think this makes sense, I do not think the swift5reflectionsegmentname should be part of the Triple, so it has to be it's own argument.
> > > Similar as above, this seems really out of place.
> > 
> > @JDevlieghere can you clarify what you mean? We need to make the reflection segment name configurable. What is it that you are suggesting?
> @JDevlieghere Hi Jonas, can you please clarify what you mean? As Adrian said we need to make the segment name configurable, and I don't think it is valid to make that part of the Triple.
> 
> Thanks!
I meant that I think it's weird that generic MC interfaces need to know about Swift section names. I understand the need to pass the information, I just don't think it should be part of the interface like this. If I follow the code correctly, the Swift 5 reflection segment name is stored in the MCContext. Instead of passing this the `Swift5ReflectionSegmentName` in here, can it query the `*Ctx` for the name? That would address my concern. 


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115007/new/

https://reviews.llvm.org/D115007



More information about the llvm-commits mailing list