a631d20a - cranelift: stack-switching support (#11003)

Thu, 04 Sep 2025 21:58:01 +0000

cranelift: stack-switching support (#11003)* cranelift: stack-switching supportThis initial commit represents the "pr2" base commit withminimal merge conflicts resolved. Due to OOB conflicts, thiscommit is not functional as-is, but using it as a base inorder to allow for easier reviewing of the delta fromthis commit to what will be used for the PR against upstream.Co-authored-by: Daniel Hillerström Co-authored-by: Paul Osborne * cranelift: stack-switching updates pass 1This first set of changes updates the base pr in order tocompiled and pass basic checks (compile, clippy, fmt) withthe biggest part of the change being to eliminate injectionof tracing/assertions in JIT'ed code.* cranelift: stack-switching: restore original visibility for a few func_environ members* cranelift: stack-switching conditional compilationAt this point, the only bit we really branch on is what wedo in order to avoid problems tying into wasmtime_environ.This is basd on the approach and macro used by the gc code forconverting presence/absence of the cranelift feature flag tocranelift compile time. This is a bit of a half-measure for nowas we still compile most stack-switching code in cranelift, butthis does enough to avoid causing problems with missing definitionsin wasmtime_environ.* cranelift: avoid "as" casts in stack-switchingReplace either with infallible From or fallible, panicingTryFrom alternatives where required.* cranelift: cleanup stack-switching control_effect signaturesAfter removing emission of runtime trace logging and assertions,there were several unused parameters. Remove those from theControlEffect signatures completely.* cranelift: rename stack-switching VMArray to VMHostArrayThis matches a change to the mirrored runtime type inthe upstream changes.* stack-switching: fix typoCo-authored-by: Daniel Hillerström * stack-switching: used Index impl for get_stack_slot_data* stack-switching: use smallvec over vec in several cases* stack-switching: avoid resumetable naming confusion* stack-switching: cleanup unused params from unchecked_get_continuationThe extra parameters here used to be used for emitting runtimeassertions, but with those gone we just had unused paramsand lifetimes, clean those out.* stack_switching: simplify store_data_entries assertion* stack-switching: simplify translate_table_{grow,fill} control flow* stack-switching: remove translate_resume_throw stubThere's already a stub elsewhere and this is not called, whenexceptions are added and it is time to revisit, this methodcan be restored.* stack-switching: compute control_context_size based on target triple* stack-switching: VMHostArrayRef updatesRename VMHostArray -> VMHostArrayRefChange impl to compute address with offset upfront rather thanon each load.* stack-switching: move cranelift code to live under func_environThis matches the directory structure for gc and aids in visibilityfor a few members required by stack-switching code in cranelift.* stack-switching: formatting fix* stack-switching: reduce visibility on a few additional items* stack-switching: simplify contobj fatptr con/de-struction* stack-switching: add disas tests to cover new instructions* stack-switching: fix layout of VMContObjIn the course of the various runtime updates, the layout of theruntime VMContObj got switched around. This resulted in failureswhen doing certain table operations on continuations.This change fixes that layout problem and adds some tests withoffsets to avoid the problem. Due to the way that we interactwith the VMContObj in cranelift, we don't use these offsets outsideof the tests.* Fix formatting of merge conflict resolution* cranelift: remove ir::function::get_stack_slot_dataThis method isn't required as sized_stack_slots is already pub.* stack-switching: reduce visibility of a couple func_environ methods* stack-switching: define VMContObj as two wordsThis change migrates VMContObj and its usages in cranelift and runtimeto work with the VMContObj fat pointer as two words in order to bettertarget different architectures (still gated to x86_64 for now).To support this, a size type was plumbed into the builtins functionsignature types (as is done for component types) that maps tousize.* fixup! stack-switching: define VMContObj as two words* stack-switching: add stub Val::ContRefThis type is not fully complete until continuation/gc integrationis revisited (#10248) but without these changes, test cases arenow failing on panics as we need some representation ofcontinuation references in the runtime Val enumeration.Runtime errors with TODO notes are added for the stubbedcode paths to revisit later.* fixup! stack-switching: add stub Val::ContRef* fixup! stack-switching: add stub Val::ContRef* fixup! stack-switching: define VMContObj as two wordsprtest:full* stack-switching: don't conflate host and target pointer sizesDisas tests were failing on i686 targeting x86_64 as the size of thehost pointer was leaking into what we were using to do codegenin a few paths. This patch is a bit of a hack as it seems likeusing a generic for T: *mut u8 (as an example) is a bitquestionable. To keep things small, I do a hacky typecheck to mappointers to the target pointer size here.* stack-switching: VMHostArray entry sizes based off env PtrSizeRevisiting the previous commit with an approach that should beless brittle.---------Co-authored-by: Frank Emrich Co-authored-by: Daniel Hillerström List of files: /wasmtime-44.0.1/crates/cranelift/src/func_environ/stack_switching/control_effect.rs

Changes in control_effect.rs

a631d20a - cranelift: stack-switching support (#11003)