[llvm] r199248 - Revert "[AArch64] Added vselect patterns with float and double types"

[Ana Pazos]

Hi Jiangning,

 

Thanks for letting me know.

I have merged the patch again and I see the test passing.

 

Ana.

 

From: Jiangning Liu [mailto:liujiangning1 at gmail.com] 
Sent: Thursday, January 16, 2014 5:41 PM
To: Ana Pazos
Cc: llvm-commits at cs.uiuc.edu for LLVM; Rafael Espindola
Subject: Re: [llvm] r199248 - Revert "[AArch64] Added vselect patterns with
float and double types"

 

Hi Ana,

 

Do you want to upstream this fix again?

 

Thanks,

-Jiangning

 

 

2014/1/15 Rafael Espindola <rafael.espindola at gmail.com>

Author: rafael
Date: Tue Jan 14 13:24:08 2014
New Revision: 199248

URL: http://llvm.org/viewvc/llvm-project?rev=199248
<http://llvm.org/viewvc/llvm-project?rev=199248&view=rev> &view=rev
Log:
Revert "[AArch64] Added vselect patterns with float and double types"

This reverts commit r199242.

It is causing CodeGen/AArch64/neon-bsl.ll to fail.

Modified:
    llvm/trunk/lib/Target/AArch64/AArch64InstrNEON.td
    llvm/trunk/test/CodeGen/AArch64/neon-bsl.ll

Modified: llvm/trunk/lib/Target/AArch64/AArch64InstrNEON.td
URL:
http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AArch64/AArch64Ins
trNEON.td?rev=199248
<http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AArch64/AArch64In
strNEON.td?rev=199248&r1=199247&r2=199248&view=diff>
&r1=199247&r2=199248&view=diff
============================================================================
==
--- llvm/trunk/lib/Target/AArch64/AArch64InstrNEON.td (original)
+++ llvm/trunk/lib/Target/AArch64/AArch64InstrNEON.td Tue Jan 14 13:24:08
2014
@@ -473,10 +473,6 @@ multiclass Neon_bitwise3V_patterns<SDPat
             (INST16B VPR128:$src, VPR128:$Rn, VPR128:$Rm)>;
   def : Pat<(v2i64 (opnode (v2i64 VPR128:$src), VPR128:$Rn, VPR128:$Rm)),
             (INST16B VPR128:$src, VPR128:$Rn, VPR128:$Rm)>;
-  def : Pat<(v2f64 (opnode (v2i64 VPR128:$src), VPR128:$Rn, VPR128:$Rm)),
-            (INST16B VPR128:$src, VPR128:$Rn, VPR128:$Rm)>;
-  def : Pat<(v4f32 (opnode (v4i32 VPR128:$src), VPR128:$Rn, VPR128:$Rm)),
-            (INST16B VPR128:$src, VPR128:$Rn, VPR128:$Rm)>;

   // Allow to match BSL instruction pattern with non-constant operand
   def : Pat<(v8i8 (or (and VPR64:$Rn, VPR64:$Rd),

Modified: llvm/trunk/test/CodeGen/AArch64/neon-bsl.ll
URL:
http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AArch64/neon-bsl
.ll?rev=199248
<http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AArch64/neon-bs
l.ll?rev=199248&r1=199247&r2=199248&view=diff>
&r1=199247&r2=199248&view=diff
============================================================================
==
--- llvm/trunk/test/CodeGen/AArch64/neon-bsl.ll (original)
+++ llvm/trunk/test/CodeGen/AArch64/neon-bsl.ll Tue Jan 14 13:24:08 2014
@@ -220,14 +220,3 @@ entry:
   ret <2 x double> %vbsl3.i
 }

-define <2 x double> @bsl2xf64(<2 x i1> %v1, <2 x double> %v2, <2 x double>
%v3) {
-;CHECK: bsl {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{v[0-9]+}}.16b
-  %1 = select <2 x i1> %v1, <2 x double> %v2, <2 x double> %v3
-  ret <2 x double> %1
-}
-
-define <4 x float> @bsl4xf32(<4 x i1> %v1, <4 x float> %v2, <4 x float>
%v3) {
-;CHECK: bsl {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{v[0-9]+}}.16b
-  %1 = select <4 x i1> %v1, <4 x float> %v2, <4 x float> %v3
-  ret <4 x float> %1
-}


_______________________________________________
llvm-commits mailing list
llvm-commits at cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits





 

-- 

Thanks,

-Jiangning

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140123/8038cab2/attachment.html>