[PATCH] D52067: [inline Cost] Don't mark function accessing varargs as non-inlinable

Florian Hahn via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Fri Sep 14 02:08:11 PDT 2018


fhahn added inline comments.


================
Comment at: test/Transforms/Inline/inline-func-vaend.ll:1
+; RUN: opt < %s -inline -S | FileCheck %s
+
----------------
Would it be possible to simplify the test case? I don't think we need most of the loads, stores and the `%struct.__va_list` type. Also I think it would be good to move the test case to the other vararg related test  (inline-varargs.ll IIRC)


================
Comment at: test/Transforms/Inline/inline-func-vaend.ll:35
+
+; Function Attrs: alwaysinline nounwind
+define void @always_inlined(i32 %n, [1 x i32] %a.coerce) {
----------------
It looks like the alwaysinline attribute got dropped? I think it would be good to test both with and without the attribute.


================
Comment at: test/Transforms/Inline/inline-func-vaend.ll:64
+
+declare dso_local void @foo([1 x i32], i32)
----------------
do we need foo?


https://reviews.llvm.org/D52067





More information about the llvm-commits mailing list