[PATCH] D11041: New EH representation for MSVC compatibility

[David Majnemer]

majnemer marked 13 inline comments as done.

================
Comment at: docs/LangRef.rst:5275-5276
@@ +5274,4 @@
+
+      cleanupret <type> <value> unwind label <continue>
+      cleanupret <type> <value> unwind to caller
+
----------------
JosephTremoulet wrote:
> My impression from the RFC discussion was that cleanupret would also have a 'from' label indicating its cleanupblock (though I didn't see a reply to John's questions about that).  What happened to that?
We ended up believing we didn't need it.

================
Comment at: docs/LangRef.rst:5289-5290
@@ +5288,4 @@
+The '``cleanupret``' instruction requires one argument, which must have the
+same type as the result of any '``cleanupblock``' instruction in the same
+function.  It also has an optional successor,  ``continue``.
+
----------------
JosephTremoulet wrote:
> This implies that all cleanupblocks in a given function have the same result type; should that be mentioned as a restriction on cleanupblock and/or checked in the verifier?
This is implicit in the cleanupblock semantics section because the personality function chooses the result type and there can only be one personality function for a function.

And yes, the verifier should check this.


http://reviews.llvm.org/D11041