Updated exception grouping algorithm: now it simply gather all the BBs
dominated by an EH pad, rather than stopping at catchret/cleanupret/unreachable.
CFGStackify seems to work either way, and this makes things simpler:

1. We don't need to store frontier BBs ourselves and be able to just use existing MachineDominanceFrontier
2. We can replace catchret/cleanupret earlier in ExceptionPrepare, so we need a single pass rather than two passes.

• Rebased onto D44090
• Not this pass deletes the following unreachable instruction after adding a rethrow instruction.

Deleted ReplaceFuncletReturns pass and move the routine to ExceptionPrepare.

• One more use of DeleteContainerPointers

• Use unique_ptr for containers and delete destructors
• Make wasm.get.exception/ehselector intrinsic take a token argument

• Make wasm.get.exception/ehselector intrinsic take a token argument

• Make wasm.get.exception/ehselector intrinsic take a token argument

@majnemer Continuing the discussion from D44134:

• Use DeleteContainerPointers

• clang-format

• clang-format

• clang-format

• clang-format

• Make wasm.get.exception/selector intrinsics take a token argument

Test case fix was missing in the last commit

• Make wasm.get.exception/selector intrinsics take a token argument

comment fix

• getScopeMembership -> getEHScopeMembership

• collectScopeMembers -> collectEHScopeMembers

• funclets -> scopes in comments / strings

• Address comments

• Address comments

• Move stuff out of doInitialization again

Thanks a a lot for all your help!

Thank you for the reviews!

CatchStartBlock -> WasmCatchStartBlock

I split MachineBasicBlock::isEHFuncletEntry and MachineBasicBlock::setIsEHFuncletEntry into two functions in D47005, as in the same vein with D45559, and rebased this CL onto it. PTAL.

fix sorry

rebased onto D47005

I don't really know about SEH, but it looks like cleanup pads are treated as funclet entries in all personalities, whereas catchpads should not be treated as funclet entries in SEH, right? So I tried to preserve that semantics, but one thing I don't understand is why we have to set isEHFuncletEntry again in findUnwindDestinations again if we set them in visitCatchPad and visitCleanupPad. Anyway that's what the current code is doing and I tried to preserve the same behavior. I ran all existing tests including SEH ones and all of them passed (except XFAILs).

more

Address comments

rebase

CXX_Wasm -> Wasm_CXX

More switch-case order change; NFC

changed switch-case order. NFC

I renamed usesWindowsEHInstructions to isScopedEHPersonality as @majnemer suggested. But there are still a lot of uses of the term 'funclet' to denote scopes in other function names, local variable names, and comments. I guess we should update all of them together sometime, but that can be a different CL.

• usesWindowsEHInstructions -> isScopedEHPersonality

• debug location fix

test fixes

• Group two terminate pads (catch and catch_all) together

• removed an unnecessary call from a test case

• Deleted successors of catch-all terminatepads

removed setIsEHFuncletEntry

rebase

Removed a blank line

fixes

• Early exit if a function does not have a personality function

• tweak test case