[llvm-commits] [llvm] r42579 - /llvm/trunk/lib/Target/X86/README.txt

[Chris Lattner]

Author: lattner
Date: Wed Oct  3 12:10:03 2007
New Revision: 42579

URL: http://llvm.org/viewvc/llvm-project?rev=42579&view=rev
Log:
add a note


Modified:
    llvm/trunk/lib/Target/X86/README.txt

Modified: llvm/trunk/lib/Target/X86/README.txt
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/README.txt?rev=42579&r1=42578&r2=42579&view=diff

==============================================================================
--- llvm/trunk/lib/Target/X86/README.txt (original)
+++ llvm/trunk/lib/Target/X86/README.txt Wed Oct  3 12:10:03 2007
@@ -1330,3 +1330,35 @@
 suggests using the 32-bit register (which is what ICC uses).
 
 //===---------------------------------------------------------------------===//
+
+rdar://5506677 - We compile this:
+
+define i32 @foo(double %x) {
+        %x14 = bitcast double %x to i64         ; <i64> [#uses=1]
+        %tmp713 = trunc i64 %x14 to i32         ; <i32> [#uses=1]
+        %tmp8 = and i32 %tmp713, 2147483647             ; <i32> [#uses=1]
+        ret i32 %tmp8
+}
+
+to:
+
+_foo:
+        subl    $12, %esp
+        fldl    16(%esp)
+        fstpl   (%esp)
+        movl    $2147483647, %eax
+        andl    (%esp), %eax
+        addl    $12, %esp
+        #FP_REG_KILL
+        ret
+
+It would be much better to eliminate the fldl/fstpl by folding the bitcast 
+into the load SDNode.  That would give us:
+
+_foo:
+        movl    $2147483647, %eax
+        andl    4(%esp), %eax
+        ret
+
+//===---------------------------------------------------------------------===//
+