From 2b066988909948dc3d53d01760bc2d71d32f3feb Mon Sep 17 00:00:00 2001
From: dim <dim@FreeBSD.org>
Date: Mon, 2 May 2011 19:34:44 +0000
Subject: Vendor import of llvm trunk r130700:
 http://llvm.org/svn/llvm-project/llvm/trunk@130700

---
 lib/Target/X86/Utils/X86ShuffleDecode.cpp | 66 +++++++++++++++++++++++++------
 1 file changed, 54 insertions(+), 12 deletions(-)

(limited to 'lib/Target/X86/Utils/X86ShuffleDecode.cpp')

diff --git a/lib/Target/X86/Utils/X86ShuffleDecode.cpp b/lib/Target/X86/Utils/X86ShuffleDecode.cpp
index 1287977..cd06060 100644
--- a/lib/Target/X86/Utils/X86ShuffleDecode.cpp
+++ b/lib/Target/X86/Utils/X86ShuffleDecode.cpp
@@ -1,4 +1,4 @@
-//===-- X86ShuffleDecode.h - X86 shuffle decode logic ---------------------===//
+//===-- X86ShuffleDecode.cpp - X86 shuffle decode logic -------------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -95,12 +95,29 @@ void DecodePSHUFLWMask(unsigned Imm,
   ShuffleMask.push_back(7);
 }
 
-void DecodePUNPCKLMask(unsigned NElts,
+void DecodePUNPCKLBWMask(unsigned NElts,
+                         SmallVectorImpl<unsigned> &ShuffleMask) {
+  DecodeUNPCKLPMask(MVT::getVectorVT(MVT::i8, NElts), ShuffleMask);
+}
+
+void DecodePUNPCKLWDMask(unsigned NElts,
+                         SmallVectorImpl<unsigned> &ShuffleMask) {
+  DecodeUNPCKLPMask(MVT::getVectorVT(MVT::i16, NElts), ShuffleMask);
+}
+
+void DecodePUNPCKLDQMask(unsigned NElts,
+                         SmallVectorImpl<unsigned> &ShuffleMask) {
+  DecodeUNPCKLPMask(MVT::getVectorVT(MVT::i32, NElts), ShuffleMask);
+}
+
+void DecodePUNPCKLQDQMask(unsigned NElts,
+                          SmallVectorImpl<unsigned> &ShuffleMask) {
+  DecodeUNPCKLPMask(MVT::getVectorVT(MVT::i64, NElts), ShuffleMask);
+}
+
+void DecodePUNPCKLMask(EVT VT,
                        SmallVectorImpl<unsigned> &ShuffleMask) {
-  for (unsigned i = 0; i != NElts/2; ++i) {
-    ShuffleMask.push_back(i);
-    ShuffleMask.push_back(i+NElts);
-  }
+  DecodeUNPCKLPMask(VT, ShuffleMask);
 }
 
 void DecodePUNPCKHMask(unsigned NElts,
@@ -133,15 +150,40 @@ void DecodeUNPCKHPMask(unsigned NElts,
   }
 }
 
+void DecodeUNPCKLPSMask(unsigned NElts,
+                        SmallVectorImpl<unsigned> &ShuffleMask) {
+  DecodeUNPCKLPMask(MVT::getVectorVT(MVT::i32, NElts), ShuffleMask);
+}
+
+void DecodeUNPCKLPDMask(unsigned NElts,
+                        SmallVectorImpl<unsigned> &ShuffleMask) {
+  DecodeUNPCKLPMask(MVT::getVectorVT(MVT::i64, NElts), ShuffleMask);
+}
 
 /// DecodeUNPCKLPMask - This decodes the shuffle masks for unpcklps/unpcklpd
-/// etc.  NElts indicates the number of elements in the vector allowing it to
-/// handle different datatypes and vector widths.
-void DecodeUNPCKLPMask(unsigned NElts,
+/// etc.  VT indicates the type of the vector allowing it to handle different
+/// datatypes and vector widths.
+void DecodeUNPCKLPMask(EVT VT,
                        SmallVectorImpl<unsigned> &ShuffleMask) {
-  for (unsigned i = 0; i != NElts/2; ++i) {
-    ShuffleMask.push_back(i);        // Reads from dest
-    ShuffleMask.push_back(i+NElts);  // Reads from src
+  unsigned NumElts = VT.getVectorNumElements();
+
+  // Handle vector lengths > 128 bits.  Define a "section" as a set of
+  // 128 bits.  AVX defines UNPCK* to operate independently on 128-bit
+  // sections.
+  unsigned NumSections = VT.getSizeInBits() / 128;
+  if (NumSections == 0 ) NumSections = 1;  // Handle MMX
+  unsigned NumSectionElts = NumElts / NumSections;
+
+  unsigned Start = 0;
+  unsigned End = NumSectionElts / 2;
+  for (unsigned s = 0; s < NumSections; ++s) {
+    for (unsigned i = Start; i != End; ++i) {
+      ShuffleMask.push_back(i);                 // Reads from dest/src1
+      ShuffleMask.push_back(i+NumSectionElts);  // Reads from src/src2
+    }
+    // Process the next 128 bits.
+    Start += NumSectionElts;
+    End += NumSectionElts;
   }
 }
 
-- 
cgit v1.1