[llvm] r178970 - PPC FMR instruction does not have side effects

[Hal Finkel]

----- Original Message -----
> From: "David Blaikie" <dblaikie at gmail.com>
> To: "Hal Finkel" <hfinkel at anl.gov>
> Cc: llvm-commits at cs.uiuc.edu
> Sent: Sunday, April 28, 2013 9:56:20 AM
> Subject: Re: [llvm] r178970 - PPC FMR instruction does not have side effects
> 
> 
> 
> Can this (& subsequent similar tests) be tested?

I'm not sure; these were all 'by inspection' changes (by looking at the TableGen output files). And while they likely caused some changes somewhere in the generated code for real test cases, I don't want to just pick an arbitrary change and test for it. I'm certainly open to suggestions.

 -Hal

> On Apr 7, 2013 3:01 PM, "Hal Finkel" < hfinkel at anl.gov > wrote:
> 
> 
> Author: hfinkel
> Date: Sat Apr 6 23:56:16 2013
> New Revision: 178970
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=178970&view=rev
> Log:
> PPC FMR instruction does not have side effects
> 
> Modified:
> llvm/trunk/lib/Target/PowerPC/PPCInstrInfo.td
> 
> Modified: llvm/trunk/lib/Target/PowerPC/PPCInstrInfo.td
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCInstrInfo.td?rev=178970&r1=178969&r2=178970&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/PowerPC/PPCInstrInfo.td (original)
> +++ llvm/trunk/lib/Target/PowerPC/PPCInstrInfo.td Sat Apr 6 23:56:16
> 2013
> @@ -1201,6 +1201,7 @@ let Uses = [RM] in {
> /// often coalesced away and we don't want the dispatch group builder
> to think
> /// that they will fill slots (which could cause the load of a LSU
> reject to
> /// sneak into a d-group with a store).
> +let neverHasSideEffects = 1 in
> def FMR : XForm_26<63, 72, (outs F4RC:$frD), (ins F4RC:$frB),
> "fmr $frD, $frB", FPGeneral,
> []>, // (set f32:$frD, f32:$frB)
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>