1.11: allow external abstract interpreter compilation #53488
+214
−179
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
aviatesk
commented
Feb 27, 2024
| @@ -321,7 +321,7 @@ function CodeInstance(interp::AbstractInterpreter, result::InferenceResult, | |||
| t = @_gc_preserve_begin inferred_result | |||
| relocatability = unsafe_load(unsafe_convert(Ptr{UInt8}, inferred_result), Core.sizeof(inferred_result)) | |||
| @_gc_preserve_end t | |||
| elseif inferred_result === nothing | |||
| elseif !(inferred_result isa CodeInfo) | |||
There was a problem hiding this comment.
This is the actual fix. The other changes are the refinements on the test cases (which were done in #53478).
There was a problem hiding this comment.
Hm but on master we still have
julia/base/compiler/typeinfer.jl
Line 331 in b3b2736
So I am not sure how it would be fixed there. IIRC even on master I needed a abs int overload that set relocatability?
There was a problem hiding this comment.
The relocatability indicates that we know the inferred field is safe to duplicate, which doesn't seem would be the case if this is not one of the expected objects? (i.e. String or Nothing)
|
@nanosoldier Edit: I made this on the wrong PR. Oh well. |
KristofferC
changed the base branch from
release-1.11
to
backports-release-1.11
As mentioned in #53478, the precompilation support for external abstract interpreters in v1.11 isn't perfect, and directly cherry-picking the refined test cases from #53478 into the v1.11 backport branch leads to a test failure (note that this particular problem has likely been fixed in the master branch, probably thanks to #53300). To address this, this commit does more than just cherry-pick the test case, and it also modifies the `CodeInstance(::AbstractInterpreter, ::InferenceResult)` constructor to allow precompilation for external abstract interpreters in v1.11.
aviatesk
force-pushed
the
avi/1.11-precompile-ext-absint
branch
from
e1c6f55
to
bd7c92e
Compare
vchuravy
requested changes
Apr 19, 2024
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.
As mentioned in #53478, the precompilation support for external abstract interpreters in v1.11 isn't perfect, and directly cherry-picking the refined test cases from #53478 into the v1.11 backport branch leads to a test failure (note that this particular problem has likely been fixed in the master branch, probably thanks to #53300). To address this, this commit does more than just cherry-pick the test case, and it also modifies the
CodeInstance(::AbstractInterpreter, ::InferenceResult)constructor to allow precompilation for external abstract interpreters in v1.11.