[llvm-commits] [llvm] r155724 - in /llvm/trunk: lib/Target/ARM/ARMInstrNEON.td lib/Target/ARM/ARMInstrVFP.td test/CodeGen/ARM/fusedMAC.ll

Lang Hames lhames at gmail.com
Fri Apr 27 11:51:24 PDT 2012 

Author: lhames
Date: Fri Apr 27 13:51:24 2012
New Revision: 155724

URL: http://llvm.org/viewvc/llvm-project?rev=155724&view=rev
Log:
Fix the order of the operands in the llvm.fma intrinsic patterns for ARM,
<rdar://problem/11325085>.


Modified:
    llvm/trunk/lib/Target/ARM/ARMInstrNEON.td
    llvm/trunk/lib/Target/ARM/ARMInstrVFP.td
    llvm/trunk/test/CodeGen/ARM/fusedMAC.ll

Modified: llvm/trunk/lib/Target/ARM/ARMInstrNEON.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrNEON.td?rev=155724&r1=155723&r2=155724&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrNEON.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrNEON.td Fri Apr 27 13:51:24 2012
@@ -4136,16 +4136,16 @@
                 Requires<[HasVFP4,UseFusedMAC]>;
 
 // Match @llvm.fma.* intrinsics
-def : Pat<(v2f32 (fma DPR:$src1, DPR:$Vn, DPR:$Vm)),
+def : Pat<(v2f32 (fma DPR:$Vn, DPR:$Vm, DPR:$src1)),
           (VFMAfd DPR:$src1, DPR:$Vn, DPR:$Vm)>,
           Requires<[HasVFP4]>;
-def : Pat<(v4f32 (fma QPR:$src1, QPR:$Vn, QPR:$Vm)),
+def : Pat<(v4f32 (fma QPR:$Vn, QPR:$Vm, QPR:$src1)),
           (VFMAfq QPR:$src1, QPR:$Vn, QPR:$Vm)>,
           Requires<[HasVFP4]>;
-def : Pat<(v2f32 (fma (fneg DPR:$src1), DPR:$Vn, DPR:$Vm)),
+def : Pat<(v2f32 (fma (fneg DPR:$Vn), DPR:$Vm, DPR:$src1)),
           (VFMSfd DPR:$src1, DPR:$Vn, DPR:$Vm)>,
       Requires<[HasVFP4]>;
-def : Pat<(v4f32 (fma (fneg QPR:$src1), QPR:$Vn, QPR:$Vm)),
+def : Pat<(v4f32 (fma (fneg QPR:$Vn), QPR:$Vm, QPR:$src1)),
           (VFMSfq QPR:$src1, QPR:$Vn, QPR:$Vm)>,
       Requires<[HasVFP4]>;
 

Modified: llvm/trunk/lib/Target/ARM/ARMInstrVFP.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrVFP.td?rev=155724&r1=155723&r2=155724&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrVFP.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrVFP.td Fri Apr 27 13:51:24 2012
@@ -1089,10 +1089,11 @@
           Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]>;
 
 // Match @llvm.fma.* intrinsics
-def : Pat<(f64 (fma DPR:$Ddin, DPR:$Dn, DPR:$Dm)),
+// (fma x, y, z) -> (vfms z, x, y)
+def : Pat<(f64 (fma DPR:$Dn, DPR:$Dm, DPR:$Ddin)),
           (VFMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(f32 (fma SPR:$Sdin, SPR:$Sn, SPR:$Sm)),
+def : Pat<(f32 (fma SPR:$Sn, SPR:$Sm, SPR:$Sdin)),
           (VFMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
 
@@ -1123,18 +1124,18 @@
           Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]>;
 
 // Match @llvm.fma.* intrinsics
-// (fma (fneg x), y, z) -> (vfms x, y, z)
-def : Pat<(f64 (fma (fneg DPR:$Ddin), DPR:$Dn, DPR:$Dm)),
+// (fma (fneg x), y, z) -> (vfms z, x, y)
+def : Pat<(f64 (fma (fneg DPR:$Dn), DPR:$Dm, DPR:$Ddin)),
           (VFMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(f32 (fma (fneg SPR:$Sdin), SPR:$Sn, SPR:$Sm)),
+def : Pat<(f32 (fma (fneg SPR:$Sn), SPR:$Sm, SPR:$Sdin)),
           (VFMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
-// (fneg (fma x, (fneg y), z) -> (vfms x, y, z)
-def : Pat<(fneg (f64 (fma DPR:$Ddin, (fneg DPR:$Dn), DPR:$Dm))),
+// (fma x, (fneg y), z) -> (vfms z, x, y)
+def : Pat<(f64 (fma DPR:$Dn, (fneg DPR:$Dm), DPR:$Ddin)),
           (VFMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(fneg (f32 (fma SPR:$Sdin, (fneg SPR:$Sn), SPR:$Sm))),
+def : Pat<(f32 (fma SPR:$Sn, (fneg SPR:$Sm), SPR:$Sdin)),
           (VFMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
 
@@ -1165,18 +1166,18 @@
           Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]>;
 
 // Match @llvm.fma.* intrinsics
-// (fneg (fma x, y, z)) -> (vfnma x, y, z)
-def : Pat<(fneg (fma (f64 DPR:$Ddin), (f64 DPR:$Dn), (f64 DPR:$Dm))),
+// (fneg (fma x, y, z)) -> (vfnma z, x, y)
+def : Pat<(fneg (fma (f64 DPR:$Dn), (f64 DPR:$Dm), (f64 DPR:$Ddin))),
           (VFNMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(fneg (fma (f32 SPR:$Sdin), (f32 SPR:$Sn), (f32 SPR:$Sm))),
+def : Pat<(fneg (fma (f32 SPR:$Sn), (f32 SPR:$Sm), (f32 SPR:$Sdin))),
           (VFNMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
-// (fma (fneg x), y, (fneg z)) -> (vfnma x, y, z)
-def : Pat<(f64 (fma (fneg DPR:$Ddin), DPR:$Dn, (fneg DPR:$Dm))),
+// (fma (fneg x), y, (fneg z)) -> (vfnma z, x, y)
+def : Pat<(f64 (fma (fneg DPR:$Dn), DPR:$Dm, (fneg DPR:$Ddin))),
           (VFNMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(f32 (fma (fneg SPR:$Sdin), SPR:$Sn, (fneg SPR:$Sm))),
+def : Pat<(f32 (fma (fneg SPR:$Sn), SPR:$Sm, (fneg SPR:$Sdin))),
           (VFNMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
 
@@ -1206,18 +1207,18 @@
           Requires<[HasVFP4,DontUseNEONForFP,UseFusedMAC]>;
 
 // Match @llvm.fma.* intrinsics
-// (fneg (fma (fneg x), y, z)) -> (vnfms x, y, z)
-def : Pat<(fneg (f64 (fma (fneg DPR:$Ddin), DPR:$Dn, DPR:$Dm))),
+// (fneg (fma (fneg x), y, z)) -> (vfnms z, x, y)
+def : Pat<(fneg (f64 (fma (fneg DPR:$Dn), DPR:$Dm, DPR:$Ddin))),
           (VFNMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(fneg (f32 (fma (fneg SPR:$Sdin), SPR:$Sn, SPR:$Sm))),
+def : Pat<(fneg (f32 (fma (fneg SPR:$Sn), SPR:$Sm, SPR:$Sdin))),
           (VFNMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
-// (fma x, (fneg y), z) -> (vnfms x, y, z)
-def : Pat<(f64 (fma DPR:$Ddin, (fneg DPR:$Dn), DPR:$Dm)),
+// (fneg (fma x, (fneg y), z) -> (vfnms z, x, y)
+def : Pat<(fneg (f64 (fma DPR:$Dn, (fneg DPR:$Dm), DPR:$Ddin))),
           (VFNMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
       Requires<[HasVFP4]>;
-def : Pat<(f32 (fma SPR:$Sdin, (fneg SPR:$Sn), SPR:$Sm)),
+def : Pat<(fneg (f32 (fma SPR:$Sn, (fneg SPR:$Sm), SPR:$Sdin))),
           (VFNMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
       Requires<[HasVFP4]>;
 

Modified: llvm/trunk/test/CodeGen/ARM/fusedMAC.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/fusedMAC.ll?rev=155724&r1=155723&r2=155724&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/fusedMAC.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/fusedMAC.ll Fri Apr 27 13:51:24 2012
@@ -138,8 +138,7 @@
 ; CHECK: vfms.f64
   %tmp1 = fsub double -0.0, %b
   %tmp2 = tail call double @llvm.fma.f64(double %a, double %tmp1, double %c) nounwind readnone
-  %tmp3 = fsub double -0.0, %tmp2
-  ret double %tmp3
+  ret double %tmp2
 }
 
 define double @test_fnms_f64(double %a, double %b, double %c) nounwind readnone ssp {
@@ -158,7 +157,8 @@
 ; CHECK: vfnms.f64
   %tmp1 = fsub double -0.0, %b
   %tmp2 = tail call double @llvm.fma.f64(double %a, double %tmp1, double %c) nounwind readnone
-  ret double %tmp2
+  %tmp3 = fsub double -0.0, %tmp2
+  ret double %tmp3
 }
 
 define double @test_fnma_f64(double %a, double %b, double %c) nounwind readnone ssp {

More information about the llvm-commits mailing list