aheejin marked an inline comment as done.
aheejin added a comment.
In D94039#2480002 <https://reviews.llvm.org/D94039#2480002>, @tlively wrote:
> In the description I think "but LLVM does not have a way of that kind of
> behavior" is missing the word "modeling" => "but LLVM does not have a way of
> modeling that kind of behavior"
Thanks. Fixed.
================
Comment at: llvm/lib/Target/WebAssembly/WebAssemblyInstrControl.td:166
let isCodeGenOnly = 1, hasSideEffects = 1 in
-defm EXTRACT_EXCEPTION_I32 : NRI<(outs I32:$dst), (ins),
- [(set I32:$dst,
(int_wasm_extract_exception))],
+defm EXTRACT_EXCEPTION_I32 : NRI<(outs I32:$dst), (ins), [],
"extract_exception\t$dst">;
----------------
tlively wrote:
> Why do we need to keep this instruction definition? Is it removed in a later
> diff in this stack?
It is removed in D94041. I tried to keep removal of `exnref` and related
instructions (`br_on_exn`, and also this, because this pseudoinstruction is
used to support `br_on_exn`) in a separate CL because it was all mechanical
stuff but touched a broad range of files, so mixing it with CLs that actually
change stuff could look confusing.
================
Comment at: llvm/test/CodeGen/WebAssembly/wasmehprepare.ll:611
declare void @__clang_call_terminate(i8*)
+declare void @_ZSt9terminatev()
----------------
tlively wrote:
> What is this addition for?
I think I added this in relation to the terminate pad handling (D94050) but
while slicing up the whole thing into several CLs this was somehow put into
this CL. But come to think of it I think this is not really necessary even in
D94050 because we don't run `llc` on this file anyway. Will delete this.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D94039/new/
https://reviews.llvm.org/D94039
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
