[X86] Describe wbnoinvd instruction

Similar to the wbinvd instruction, except this
one does not invalidate caches. Ring 0 only.
The encoding matches a wbinvd instruction with
an F3 prefix.

Reviewers: craig.topper, zvi, ashlykov

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D43816

llvm-svn: 329847

include/llvm/IR/IntrinsicsX86.td

@@ -6401,3 +6401,12 @@ let TargetPrefix = "x86" in {
   def int_x86_clzero : GCCBuiltin<"__builtin_ia32_clzero">,
       Intrinsic<[], [llvm_ptr_ty], []>;
 }
+
+//===----------------------------------------------------------------------===//
+// Cache line write back intrinsics
+
+// Write back no-invalidate
+let TargetPrefix = "x86" in {
+  def int_x86_wbnoinvd : GCCBuiltin<"__builtin_ia32_wbnoinvd">,
+      Intrinsic<[], [], []>;
+}

lib/Support/Host.cpp

@@ -1213,9 +1213,12 @@ bool sys::getHostCPUFeatures(StringMap<bool> &Features) {
   Features["tbm"]    = HasExtLeaf1 && ((ECX >> 21) & 1);
   Features["mwaitx"] = HasExtLeaf1 && ((ECX >> 29) & 1);
 
+  // Miscellaneous memory related features, detected by
+  // using the 0x80000008 leaf of the CPUID instruction
   bool HasExtLeaf8 = MaxExtLevel >= 0x80000008 &&
                      !getX86CpuIDAndInfo(0x80000008, &EAX, &EBX, &ECX, &EDX);
-  Features["clzero"] = HasExtLeaf8 && ((EBX >> 0) & 1);
+  Features["clzero"]   = HasExtLeaf8 && ((EBX >> 0) & 1);
+  Features["wbnoinvd"] = HasExtLeaf8 && ((EBX >> 9) & 1);
 
   bool HasLeaf7 =
       MaxLevel >= 7 && !getX86CpuIDAndInfoEx(0x7, 0x0, &EAX, &EBX, &ECX, &EDX);

lib/Target/X86/X86.td

@@ -249,6 +249,8 @@ def FeatureCLFLUSHOPT : SubtargetFeature<"clflushopt", "HasCLFLUSHOPT", "true",
                                       "Flush A Cache Line Optimized">;
 def FeatureCLWB    : SubtargetFeature<"clwb", "HasCLWB", "true",
                                       "Cache Line Write Back">;
+def FeatureWBNOINVD    : SubtargetFeature<"wbnoinvd", "HasWBNOINVD", "true",
+                                      "Write Back No Invalidate">;
 def FeatureRDPID : SubtargetFeature<"rdpid", "HasRDPID", "true",
                                     "Support RDPID instructions">;
 // On some processors, instructions that implicitly take two memory operands are
@@ -825,6 +827,7 @@ def : IcelakeClientProc<"icelake-client">;
 class IcelakeServerProc<string Name> : ProcModel<Name, SkylakeServerModel,
                                                  ICLFeatures.Value, [
   ProcIntelICX,
+  FeatureWBNOINVD,
   FeatureHasFastGather
 ]>;
 def : IcelakeServerProc<"icelake-server">;

lib/Target/X86/X86InstrInfo.td

@@ -888,6 +888,7 @@ def HasSHSTK     : Predicate<"Subtarget->hasSHSTK()">;
 def HasIBT       : Predicate<"Subtarget->hasIBT()">;
 def HasCLFLUSHOPT : Predicate<"Subtarget->hasCLFLUSHOPT()">;
 def HasCLWB      : Predicate<"Subtarget->hasCLWB()">;
+def HasWBNOINVD  : Predicate<"Subtarget->hasWBNOINVD()">;
 def HasRDPID     : Predicate<"Subtarget->hasRDPID()">;
 def HasCmpxchg16b: Predicate<"Subtarget->hasCmpxchg16b()">;
 def Not64BitMode : Predicate<"!Subtarget->is64Bit()">,

lib/Target/X86/X86InstrSystem.td

@@ -482,6 +482,12 @@ let Defs = [EAX, EBX, ECX, EDX], Uses = [EAX, ECX] in
 let SchedRW = [WriteSystem] in {
 def INVD : I<0x08, RawFrm, (outs), (ins), "invd", [], IIC_INVD>, TB;
 def WBINVD : I<0x09, RawFrm, (outs), (ins), "wbinvd", [], IIC_INVD>, TB;
+
+// wbnoinvd is like wbinvd, except without invalidation
+// encoding: like wbinvd + an 0xF3 prefix
+def WBNOINVD : I<0x09, RawFrm, (outs), (ins), "wbnoinvd",
+                 [(int_x86_wbnoinvd)], IIC_INVD>, XS,
+                 Requires<[HasWBNOINVD]>;
 } // SchedRW
 
 //===----------------------------------------------------------------------===//

lib/Target/X86/X86Subtarget.cpp

@@ -324,6 +324,7 @@ void X86Subtarget::initializeEnvironment() {
   HasSGX = false;
   HasCLFLUSHOPT = false;
   HasCLWB = false;
+  HasWBNOINVD = false;
   HasRDPID = false;
   UseRetpoline = false;
   UseRetpolineExternalThunk = false;

lib/Target/X86/X86Subtarget.h

@@ -360,6 +360,9 @@ protected:
   /// Processor supports Cache Line Write Back instruction
   bool HasCLWB;
 
+  /// Processor supports Write Back No Invalidate instruction
+  bool HasWBNOINVD;
+
   /// Processor support RDPID instruction
   bool HasRDPID;
 
@@ -621,6 +624,7 @@ public:
   bool hasIBT() const { return HasIBT; }
   bool hasCLFLUSHOPT() const { return HasCLFLUSHOPT; }
   bool hasCLWB() const { return HasCLWB; }
+  bool hasWBNOINVD() const { return HasWBNOINVD; }
   bool hasRDPID() const { return HasRDPID; }
   bool useRetpoline() const { return UseRetpoline; }
   bool useRetpolineExternalThunk() const { return UseRetpolineExternalThunk; }

test/CodeGen/X86/wbnoinvd-intrinsic.ll

@@ -0,0 +1,18 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+wbnoinvd | FileCheck %s -check-prefix=CHECK32
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+wbnoinvd | FileCheck %s -check-prefix=CHECK64
+
+define void @wbnoinvd() nounwind {
+; CHECK32-LABEL: wbnoinvd:
+; CHECK32:       # %bb.0:
+; CHECK32-NEXT:    wbnoinvd
+; CHECK32-NEXT:    retl
+;
+; CHECK64-LABEL: wbnoinvd:
+; CHECK64:       # %bb.0:
+; CHECK64-NEXT:    wbnoinvd
+; CHECK64-NEXT:    retq
+  tail call void @llvm.x86.wbnoinvd()
+  ret void
+}
+declare void @llvm.x86.wbnoinvd() nounwind

test/MC/Disassembler/X86/x86-16.txt

@@ -791,3 +791,6 @@
 
 # CHECK: callw	-1
 0xe8 0xff 0xff
+
+# CHECK: wbnoinvd
+0xf3 0x0f 0x09

test/MC/Disassembler/X86/x86-32.txt

@@ -820,3 +820,6 @@
 
 # CHECK: ptwritel %eax
 0xf3 0x0f 0xae 0xe0
+
+# CHECK: wbnoinvd
+0xf3 0x0f 0x09

test/MC/Disassembler/X86/x86-64.txt

@@ -516,3 +516,6 @@
 
 # CHECK: ptwriteq %rax
 0xf3 0x48 0x0f 0xae 0xe0
+
+# CHECK: wbnoinvd
+0xf3 0x0f 0x09

test/MC/X86/x86-16.s

@@ -969,3 +969,7 @@ data32
 // CHECK: lgdtw 4(%eax)
 // CHECK:  encoding: [0x67,0x0f,0x01,0x50,0x04]
 data32 lgdt 4(%eax)
+
+// CHECK: wbnoinvd
+// CHECK:  encoding: [0xf3,0x0f,0x09]
+wbnoinvd

test/MC/X86/x86-32-coverage.s

@@ -2788,6 +2788,10 @@
 // CHECK:  encoding: [0x0f,0x09]
         	wbinvd
 
+// CHECK: wbnoinvd
+// CHECK:  encoding: [0xf3,0x0f,0x09]
+        	wbnoinvd
+
 // CHECK: cpuid
 // CHECK:  encoding: [0x0f,0xa2]
         	cpuid

test/MC/X86/x86-64.s

@@ -1559,6 +1559,10 @@ ptwriteq 0xdeadbeef(%rbx,%rcx,8)
 // CHECK:  encoding: [0xf3,0x48,0x0f,0xae,0xe0]
 ptwriteq %rax
 
+// CHECK: wbnoinvd
+// CHECK:  encoding: [0xf3,0x0f,0x09]
+wbnoinvd
+
 //  __asm __volatile(
 //    "pushf        \n\t"
 //    "popf       \n\t"