MolotovCherry requested wasmtime-compiler-reviewers for a review on PR #8718.
MolotovCherry requested elliottt for a review on PR #8718.
MolotovCherry opened PR #8718 from MolotovCherry:main to bytecodealliance:main:
JITModulecurrently is!Send, which makes it less useful. According to bjorn, it makes sense to make itSend.I'm fairly new to this, but please let me know any issues/concerns you have and I'll do my best to address them. Thanks!
cc @bjorn3
MolotovCherry edited PR #8718:
JITModulecurrently is!Send, which makes it less useful. According to bjorn, it makes sense to make itSend.I'm fairly new to this, but please let me know any issues/concerns you have with this and I'll do my best to address them. Thanks!
cc @bjorn3
MolotovCherry submitted PR review.
MolotovCherry created PR review comment:
Seeing that
libcall_namesalready originally requiredSend + SyncinJITBuilder, I wonder if it would be better to also addSyncto this.
MolotovCherry edited PR review comment.
MolotovCherry edited PR review comment.
bjorn3 submitted PR review.
bjorn3 created PR review comment:
Instead of implementing
SendonJITModule, which may cause issues if we add a new non-Sendfield, maybe it would be possible to implementSendon each type that doesn't currently likeCompiledBlobandPtrLen. For thesymbolsfield you can for example wrap the*const u8in a newtype.
bjorn3 submitted PR review.
bjorn3 created PR review comment:
I don't think there is much point in making it
Synctoo asJITModulecan't beSyncanyway. Forlibcall_namesit makes sense for it to beSyncto be able to passcranelift_module::default_libcall_names().
MolotovCherry updated PR #8718.
MolotovCherry submitted PR review.
MolotovCherry created PR review comment:
Should be mostly good now I think. I made a
SendWrapperfor fields which can be wrapped in it inJITModule, and impl'dSendon any types that didn't likeCompiledBlobandPtrLen. I also had to wrap all theNonNullfields as they were also offenders
MolotovCherry edited PR review comment.
bjorn3 submitted PR review.
bjorn3 created PR review comment:
Deriving Copy and Clone should be fine.
MolotovCherry updated PR #8718.
alexcrichton submitted PR review.
alexcrichton merged PR #8718.
Last updated: Dec 06 2025 at 06:05 UTC