This is mostly for debugging for now. With the upcoming parallel ghashes, I need a way to disable iterating in parallel.
Diff Detail
Diff Detail
- Repository
- rLLD LLVM Linker
Event Timeline
Comment Actions
lgtm
At first I was thinking, maybe we should extend this to /threads:N, so some build systems could reduce the parallelism when links run concurrently, but I think for Windows it's best to just let concrt/the system do its thing.
The other big source of expensive concurrency is ThinLTO, which logically we should wire up to /cgthreads.