Fix opaque closure codegen ABI #40001
Merged
Conversation
This file contains 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
vtjnash
reviewed
Mar 12, 2021
src/codegen.cpp
Outdated
Comment on lines
6477
to
6478
| argType = jl_unwrap_unionall( | ||
| (jl_value_t*)jl_opaque_closure_type); |
There was a problem hiding this comment.
seems strange to intentionally introduce an invalid type object here as the argument type with jl_unwrap_unionall
There was a problem hiding this comment.
Yeah, this is to make emit_getfield_knownidx below work since the layout is independent of the type parameters. I guess I could just do OpaqueClosure{Tuple{Any}, Any} to make something valid or we could just manually compute the offset. Maybe the latter makes the most sense.
There was a problem hiding this comment.
ah, yeah, I'm surprised we don't end up tagging that load with some nonsensical type too and causing confusion
I guess the latter possibly does make the most sense, as later the intent is to inline the data, rather than making it a separate Tuple{Any} allocation
There was a problem hiding this comment.
Yeah, alright, let me just do that.
At the moment our ABI for specfun opaque closures is that the opaque closure itself remains a GC tracked pointer out of which the function itself implicitly loads the environment and the world age. In the future we may want to improve this ABI, but it seems fine for now. However, this ABI wasn't quite implemented correctly after we've started turning on inference of OpaqueClosure where the first argument of the MethodInstance's linfo actually stores the type of the closure environment rather than the closure itself. Fix this by manually forcing the signature determination code to just treat that argument as `Any`, though of course in the future we may want to do something fancier.
vtjnash
approved these changes
Mar 15, 2021
ElOceanografo
pushed a commit
to ElOceanografo/julia
that referenced
this pull request
May 4, 2021
At the moment our ABI for specfun opaque closures is that the opaque closure itself remains a GC tracked pointer out of which the function itself implicitly loads the environment and the world age. In the future we may want to improve this ABI, but it seems fine for now. However, this ABI wasn't quite implemented correctly after we've started turning on inference of OpaqueClosure where the first argument of the MethodInstance's linfo actually stores the type of the closure environment rather than the closure itself. Fix this by manually forcing the signature determination code to just treat that argument as `Any`, though of course in the future we may want to do something fancier.
antoine-levitt
pushed a commit
to antoine-levitt/julia
that referenced
this pull request
May 9, 2021
At the moment our ABI for specfun opaque closures is that the opaque closure itself remains a GC tracked pointer out of which the function itself implicitly loads the environment and the world age. In the future we may want to improve this ABI, but it seems fine for now. However, this ABI wasn't quite implemented correctly after we've started turning on inference of OpaqueClosure where the first argument of the MethodInstance's linfo actually stores the type of the closure environment rather than the closure itself. Fix this by manually forcing the signature determination code to just treat that argument as `Any`, though of course in the future we may want to do something fancier.
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.
At the moment our ABI for specfun opaque closures is that the
opaque closure itself remains a GC tracked pointer out of which
the function itself implicitly loads the environment and the world
age. In the future we may want to improve this ABI, but it seems
fine for now. However, this ABI wasn't quite implemented correctly
after we've started turning on inference of OpaqueClosure where
the first argument of the MethodInstance's linfo actually stores
the type of the closure environment rather than the closure itself.
Fix this by manually forcing the signature determination code to
just treat that argument as
Any, though of course in the futurewe may want to do something fancier.