[llvm] r303000 - Renable test that was disabled due to cost analysis

[Jessica Paquette via llvm-commits]

Do these tests make any assumptions about the inliner’s behaviour? If so, do you think it would be possible to include inlining thresholds in these tests?

For example, some of the inliner tests specify -inline-threshold=SomeIntegerValue.

My concern is that changes to the inliner’s cost model might break some of these tests. Threshold values might improve the stability of the tests. They would also make it easier to fix the tests/inliner in the event of cost model changes.

- Jessica

> On May 13, 2017, at 7:58 PM, Xinliang David Li via llvm-commits <llvm-commits at lists.llvm.org> wrote:
> 
> Author: davidxl
> Date: Sat May 13 21:58:39 2017
> New Revision: 303000
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=303000&view=rev
> Log:
> Renable test that was disabled due to cost analysis
> 
> Modified:
>    llvm/trunk/test/Transforms/Inline/partial-inline-act.ll
> 
> Modified: llvm/trunk/test/Transforms/Inline/partial-inline-act.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/Inline/partial-inline-act.ll?rev=303000&r1=302999&r2=303000&view=diff
> ==============================================================================
> --- llvm/trunk/test/Transforms/Inline/partial-inline-act.ll (original)
> +++ llvm/trunk/test/Transforms/Inline/partial-inline-act.ll Sat May 13 21:58:39 2017
> @@ -1,4 +1,4 @@
> -; RUN: opt < %s -partial-inliner -disable-output
> +; RUN: opt < %s -partial-inliner -skip-partial-inlining-cost-analysis -disable-output
> ; This testcase tests the assumption cache
> 
> define internal i32 @inlinedFunc(i1 %cond, i32* align 4 %align.val) {
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits