[PATCH] D48345: [WebAssembly] Fix unwind destination mismatches in CFG stackify

Mon Mar 25 10:22:18 PDT 2019

aheejin marked 2 inline comments as done.
aheejin added inline comments.

================
Comment at: test/CodeGen/WebAssembly/cfg-stackify-eh-mismatch.ll:41
+
+define void @test0() personality i8* bitcast (i32 (...)* @__gxx_wasm_personality_v0 to i8*) {
+bb0:
----------------
aheejin wrote:
> dschuff wrote:
> > Any weird corner cases it would be worthwhile to test for? Mixing both kinds of mismatches? A case with uwind-to-caller and block signature needed? Would multiple nesting of mismatches affect what happens?
> Add some more tests.
> - Mixing both kinds of mismatches: Done (test8)
> - A case with uwind-to-caller and block signature needed: it's not possible because of newly added `rethrow`. Added a test that shows why it's not possible. (test7)
> - Multiple nesting of mismatches: test8 kind of covers this too.
> 
> Note that `-wasm-disable-ehpad-sort` is used to generate maximum # of mismatches for those mismatch tests.
Oh and please also note that I deleted this file and moved all tests to cfg-stackify-eh.ll.

Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D48345/new/

https://reviews.llvm.org/D48345