Latest clang in LLVM 17.0.2 may produce invalid code signatures on Apple Silicon macs #4842
Replies: 2 comments 4 replies
-
Hard to really say from this. Do you have a minimal reproducible example? Go also has its own linker so it's difficult to deduce from this at which operation the code signature is breaking. For reference: in a normal C-only workflow, the "ad-hoc" code signature is produced automatically by Apple's |
Beta Was this translation helpful? Give feedback.
-
I've not had a chance to revisit this, and I don't think I will any time soon. It sounds like a pretty niche problem, and I've worked around it to my satisfaction. |
Beta Was this translation helpful? Give feedback.
-
Output of
brew config
Output of
brew doctor
Description of issue
Really weird one, but I've found that if I use the brew-provided llvm, binaries can (but don't necessarily) have malformed code signatures on my M1 mac! This isn't a problem with LLVM 16, and once I figured out what was happening it was easy enough to work around, but I don't have the first clue on where to look for more information, or to file an issue (if one is even warranted); any advice?
This presented itself because we have started using more Golang at work, and the final linker step invoked by
go build
in the toolchains I had installed used whateverclang
was on the path. I'd forgotten that I'd previously put whatever the latest LLVM was into my path, so that I'd have the latest stable clang for other reasons. Some of my builds just stopped working, invalid signatures, and I didn't twig it happened after the upgrade to LLVM 17, but I've just spent the day digging through it all to figure this out 😂Note: I do absolutely realise there is a warning about adding LLVM to the path in the formula, and I have simply removed it for now, but this is the first time it's caused me a problem.
Beta Was this translation helpful? Give feedback.
All reactions