Repository navigation
Fix missing GC factories when linked as profiling - #111
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
There is a particular bug that can happen when all of these conditions are met:
--build=profileWhen a D application is built with profiling enabled, a small profiling wrapper is put around nearly every function. The very first time such a wrapper is executed, it calls
rt.trace.trace_init(). This method can (and on my machine does) trigger GC.If the linking order is unfortunate, this GC call can happen before D's own initializers configure the GC factories. Then, no GC factory can be found, and the following error is printed before the program exits:
By flipping the
version-clause, we prevent this one extra wrapper being put this early.To trigger this bug, use a modern Mac, install LDC via homebrew, and then run in an empty directory:
This should ideally also be fixed in DMD/LDC. It probably should not be placing these wrappers on
pragma(crt_constructor)calls. I hope to get that done at some point, but this will unblock me faster.