In jit/VMFunctions.* we have BindVar/DefVar/DefLexical functions that behave like the interpreter versions, but there's no good reason to have the logic duplicated there. I have a stack of patches to clean this up and unify as BindVarOperation, DefVarOperation, DefLexicalOperation that we call from interpreter, Baseline, Ion.
This also adds a GetVariablesObject helper so we don't have to duplicate the logic there.
No change in behavior. Depends on D13698
Depends on D13700
Depends on D13701
Depends on D13702
Depends on D13703
Depends on D13704
Depends on D13705
I think this is the right thing to do but I'm not sure. Depends on D13706
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/autoland/rev/582477c043dd part 1 - Add BindVarOperation and use it for JSOP_BINDVAR in interpreter and JITs. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/3523e71282fd part 2 - Move DefVarOperation from Interpreter-inl.h to Interpreter.cpp. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/452034cea923 part 3 - Refactor DefVarOperation so interpreter and JITs can call it directly. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/8fccd1861a22 part 4 - Move DefLexicalOperation from Interpreter-inl.h to Interpreter.cpp. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/c6e9f5fd0ee7 part 5 - Refactor DefLexicalOperation to make it easier to call directly from JIT code in the next patch. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/745cbd9ae616 part 6 - Merge two DefLexicalOperation functions into one and have the JITs call it directly. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/0df13a6c64be part 7 - Change jit::CheckGlobalOrEvalDeclarationConflicts signature to make it less Baseline-specific. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/4e1a2cc33577 part 8 - Move CheckGlobalOrEvalDeclarationConflicts to EnvironmentObject.cpp and call it also in the interpreter. r=tcampbell https://hg.mozilla.org/integration/autoland/rev/d7c3229eb734 part 9 - Call recordreplay::AdvanceExecutionProgressCounter also for eval frames in InterpreterFrame::prologue. r=bhackett
You need to log in before you can comment on or make changes to this bug.