[llvm] r224009 - Bugfix in InlineSpiller::traceSiblingValue().

[David Blaikie]

Any chance of a test case? (I assume this was brought up in code review -
but it's helpful to have the justification included in the commit message,
ideally)

On Thu, Dec 11, 2014 at 2:40 AM, Patrik Hagglund <
patrik.h.hagglund at ericsson.com> wrote:

> Author: patha
> Date: Thu Dec 11 04:40:17 2014
> New Revision: 224009
>
> URL: http://llvm.org/viewvc/llvm-project?rev=224009&view=rev
> Log:
> Bugfix in InlineSpiller::traceSiblingValue().
>
> Properly determine whether or not a phi was added by splitting.
> Check against the current VNInfo of OrigLI instead of against the
> OrigVNI argument.
>
> Patch provided by Jonas Paulsson. Reviewed by Quentin Colombet.
>
> Modified:
>     llvm/trunk/lib/CodeGen/InlineSpiller.cpp
>
> Modified: llvm/trunk/lib/CodeGen/InlineSpiller.cpp
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/InlineSpiller.cpp?rev=224009&r1=224008&r2=224009&view=diff
>
> ==============================================================================
> --- llvm/trunk/lib/CodeGen/InlineSpiller.cpp (original)
> +++ llvm/trunk/lib/CodeGen/InlineSpiller.cpp Thu Dec 11 04:40:17 2014
> @@ -508,6 +508,7 @@ MachineInstr *InlineSpiller::traceSiblin
>    SmallVector<std::pair<unsigned, VNInfo*>, 8> WorkList;
>    WorkList.push_back(std::make_pair(UseReg, UseVNI));
>
> +  LiveInterval &OrigLI = LIS.getInterval(Original);
>    do {
>      unsigned Reg;
>      VNInfo *VNI;
> @@ -521,8 +522,11 @@ MachineInstr *InlineSpiller::traceSiblin
>
>      // Trace through PHI-defs created by live range splitting.
>      if (VNI->isPHIDef()) {
> -      // Stop at original PHIs.  We don't know the value at the
> predecessors.
> -      if (VNI->def == OrigVNI->def) {
> +      // Stop at original PHIs.  We don't know the value at the
> +      // predecessors. Look up the VNInfo for the current definition
> +      // in OrigLI, to properly determine whether or not this phi was
> +      // added by splitting.
> +      if (VNI->def == OrigLI.getVNInfoAt(VNI->def)->def) {
>          DEBUG(dbgs() << "orig phi value\n");
>          SVI->second.DefByOrigPHI = true;
>          SVI->second.AllDefsAreReloads = false;
> @@ -542,7 +546,6 @@ MachineInstr *InlineSpiller::traceSiblin
>        // Separate all values dominated by OrigVNI into PHIs and non-PHIs.
>        SmallVector<VNInfo*, 8> PHIs, NonPHIs;
>        LiveInterval &LI = LIS.getInterval(Reg);
> -      LiveInterval &OrigLI = LIS.getInterval(Original);
>
>        for (LiveInterval::vni_iterator VI = LI.vni_begin(), VE =
> LI.vni_end();
>             VI != VE; ++VI) {
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141211/a2f2d183/attachment.html>