[llvm-commits] [llvm] r165877 - in /llvm/trunk: lib/CodeGen/RegisterCoalescer.cpp test/CodeGen/ARM/coalesce-subregs.ll

[Jakob Stoklund Olesen]

Author: stoklund
Date: Sat Oct 13 12:26:47 2012
New Revision: 165877

URL: http://llvm.org/viewvc/llvm-project?rev=165877&view=rev
Log:
Drop <def,dead> flags when merging into an unused lane.

The new coalescer can merge a dead def into an unused lane of an
otherwise live vector register.

Clear the <dead> flag when that happens since the flag refers to the
full virtual register which is still live after the partial dead def.

This fixes PR14079.

Modified:
    llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp
    llvm/trunk/test/CodeGen/ARM/coalesce-subregs.ll

Modified: llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp?rev=165877&r1=165876&r2=165877&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp (original)
+++ llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp Sat Oct 13 12:26:47 2012
@@ -1739,15 +1739,20 @@
       // has been replaced.
       Val &OtherV = Other.Vals[Vals[i].OtherVNI->id];
       bool EraseImpDef = OtherV.IsImplicitDef && OtherV.Resolution == CR_Keep;
-      if (!EraseImpDef && !Def.isBlock()) {
+      if (!Def.isBlock()) {
         // Remove <def,read-undef> flags. This def is now a partial redef.
+        // Also remove <def,dead> flags since the joined live range will
+        // continue past this instruction.
         for (MIOperands MO(Indexes->getInstructionFromIndex(Def));
              MO.isValid(); ++MO)
-          if (MO->isReg() && MO->isDef() && MO->getReg() == LI.reg)
-            MO->setIsUndef(false);
+          if (MO->isReg() && MO->isDef() && MO->getReg() == LI.reg) {
+            MO->setIsUndef(EraseImpDef);
+            MO->setIsDead(false);
+          }
         // This value will reach instructions below, but we need to make sure
         // the live range also reaches the instruction at Def.
-        EndPoints.push_back(Def);
+        if (!EraseImpDef)
+          EndPoints.push_back(Def);
       }
       DEBUG(dbgs() << "\t\tpruned " << PrintReg(Other.LI.reg) << " at " << Def
                    << ": " << Other.LI << '\n');

Modified: llvm/trunk/test/CodeGen/ARM/coalesce-subregs.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/coalesce-subregs.ll?rev=165877&r1=165876&r2=165877&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/coalesce-subregs.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/coalesce-subregs.ll Sat Oct 13 12:26:47 2012
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mcpu=cortex-a9 -verify-coalescing | FileCheck %s
+; RUN: llc < %s -mcpu=cortex-a9 -verify-coalescing -verify-machineinstrs | FileCheck %s
 target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32-S32"
 target triple = "thumbv7-apple-ios0.0.0"
 
@@ -263,3 +263,29 @@
 declare <2 x float> @baz(<2 x float>, <2 x float>, <2 x float>) nounwind readnone
 
 declare <2 x float> @baz67(<2 x float>, <2 x float>) nounwind readnone
+
+%struct.wombat.5 = type { %struct.quux, %struct.quux, %struct.quux, %struct.quux }
+%struct.quux = type { <4 x float> }
+
+; CHECK: pr14079
+define linkonce_odr arm_aapcs_vfpcc %struct.wombat.5 @pr14079(i8* nocapture %arg, i8* nocapture %arg1, i8* nocapture %arg2) nounwind uwtable inlinehint {
+bb:
+  %tmp = shufflevector <2 x i64> zeroinitializer, <2 x i64> undef, <1 x i32> zeroinitializer
+  %tmp3 = bitcast <1 x i64> %tmp to <2 x float>
+  %tmp4 = shufflevector <2 x float> %tmp3, <2 x float> zeroinitializer, <2 x i32> <i32 1, i32 3>
+  %tmp5 = shufflevector <2 x float> %tmp4, <2 x float> undef, <2 x i32> <i32 1, i32 3>
+  %tmp6 = bitcast <2 x float> %tmp5 to <1 x i64>
+  %tmp7 = shufflevector <1 x i64> undef, <1 x i64> %tmp6, <2 x i32> <i32 0, i32 1>
+  %tmp8 = bitcast <2 x i64> %tmp7 to <4 x float>
+  %tmp9 = shufflevector <2 x i64> zeroinitializer, <2 x i64> undef, <1 x i32> <i32 1>
+  %tmp10 = bitcast <1 x i64> %tmp9 to <2 x float>
+  %tmp11 = shufflevector <2 x float> %tmp10, <2 x float> undef, <2 x i32> <i32 0, i32 2>
+  %tmp12 = shufflevector <2 x float> %tmp11, <2 x float> undef, <2 x i32> <i32 0, i32 2>
+  %tmp13 = bitcast <2 x float> %tmp12 to <1 x i64>
+  %tmp14 = shufflevector <1 x i64> %tmp13, <1 x i64> undef, <2 x i32> <i32 0, i32 1>
+  %tmp15 = bitcast <2 x i64> %tmp14 to <4 x float>
+  %tmp16 = insertvalue %struct.wombat.5 undef, <4 x float> %tmp8, 1, 0
+  %tmp17 = insertvalue %struct.wombat.5 %tmp16, <4 x float> %tmp15, 2, 0
+  %tmp18 = insertvalue %struct.wombat.5 %tmp17, <4 x float> undef, 3, 0
+  ret %struct.wombat.5 %tmp18
+}