From: Dan Gohman <gohman@apple.com>
Date: Thu, 16 Jun 2011 15:55:48 +0000 (+0000)
Subject: Add a comment describing why transforming (shl x, 1) to (add x, x) is to be
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=a0697a7ef5a980727d8361acd1fc429a5b1b8e2b;p=oota-llvm.git

Add a comment describing why transforming (shl x, 1) to (add x, x) is to be
considered safe enough in this context.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133159 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Target/X86/X86InstrCompiler.td b/lib/Target/X86/X86InstrCompiler.td
index 33534cd82bd..adcc747eb4b 100644
--- a/lib/Target/X86/X86InstrCompiler.td
+++ b/lib/Target/X86/X86InstrCompiler.td
@@ -1368,6 +1368,11 @@ def : Pat<(store (i8 (trunc_su (srl_su GR16:$src, (i8 8)))), addr:$dst),
 
 
 // (shl x, 1) ==> (add x, x)
+// Note that if x is undef (immediate or otherwise), we could theoretically
+// end up with the two uses of x getting different values, producing a result
+// where the least significant bit is not 0. However, the probability of this
+// happening is considered low enough that this is officially not a
+// "real problem".
 def : Pat<(shl GR8 :$src1, (i8 1)), (ADD8rr  GR8 :$src1, GR8 :$src1)>;
 def : Pat<(shl GR16:$src1, (i8 1)), (ADD16rr GR16:$src1, GR16:$src1)>;
 def : Pat<(shl GR32:$src1, (i8 1)), (ADD32rr GR32:$src1, GR32:$src1)>;