Merged main:157ba33007fe into amd-gfx:f5e3081cf03f

Local branch amd-gfx f5e3081 Merged main:89a2e7015968 into amd-gfx:acddebee9f46
Remote branch main 157ba33 [CSKY,test] Update switch.ll

Author: SC llvm team

compiler-rt/include/sanitizer/allocator_interface.h

@@ -83,8 +83,8 @@ void SANITIZER_CDECL __sanitizer_free_hook(const volatile void *ptr);
    other threads.
 */
 int SANITIZER_CDECL __sanitizer_install_malloc_and_free_hooks(
-    void (*SANITIZER_CDECL malloc_hook)(const volatile void *, size_t),
-    void (*SANITIZER_CDECL free_hook)(const volatile void *));
+    void(SANITIZER_CDECL *malloc_hook)(const volatile void *, size_t),
+    void(SANITIZER_CDECL *free_hook)(const volatile void *));
 
 /* Drains allocator quarantines (calling thread's and global ones), returns
    freed memory back to OS and releases other non-essential internal allocator

compiler-rt/include/sanitizer/dfsan_interface.h

@@ -25,16 +25,16 @@ typedef uint8_t dfsan_label;
 typedef uint32_t dfsan_origin;
 
 /// Signature of the callback argument to dfsan_set_write_callback().
-typedef void (*SANITIZER_CDECL dfsan_write_callback_t)(int fd, const void *buf,
-                                                       size_t count);
+typedef void(SANITIZER_CDECL *dfsan_write_callback_t)(int fd, const void *buf,
+                                                      size_t count);
 
 /// Signature of the callback argument to dfsan_set_conditional_callback().
-typedef void (*SANITIZER_CDECL dfsan_conditional_callback_t)(
+typedef void(SANITIZER_CDECL *dfsan_conditional_callback_t)(
     dfsan_label label, dfsan_origin origin);
 
 /// Signature of the callback argument to dfsan_set_reaches_function_callback().
 /// The description is intended to hold the name of the variable.
-typedef void (*SANITIZER_CDECL dfsan_reaches_function_callback_t)(
+typedef void(SANITIZER_CDECL *dfsan_reaches_function_callback_t)(
     dfsan_label label, dfsan_origin origin, const char *file, unsigned int line,
     const char *function);
 

compiler-rt/include/sanitizer/msan_interface.h

@@ -108,7 +108,7 @@ const char *SANITIZER_CDECL __msan_default_options(void);
 
 /* Deprecated. Call __sanitizer_set_death_callback instead. */
 void SANITIZER_CDECL
-__msan_set_death_callback(void (*SANITIZER_CDECL callback)(void));
+__msan_set_death_callback(void(SANITIZER_CDECL *callback)(void));
 
 /* Update shadow for the application copy of size bytes from src to dst.
    Src and dst are application addresses. This function does not copy the

lldb/source/Host/macosx/objcxx/HostInfoMacOSX.mm

@@ -417,6 +417,8 @@ static void ParseOSVersion(llvm::VersionTuple &version, NSString *Key) {
     // xcrun didn't find a matching SDK. Not an error, we'll try
     // different spellings.
     LLDB_LOG(log, "xcrun returned exit code {0}", status);
+    if (!output_str.empty())
+      LLDB_LOG(log, "xcrun output was:\n{0}", output_str);
     return "";
   }
   if (output_str.empty()) {

llvm/include/llvm/Config/llvm-config.h.cmake

@@ -16,7 +16,7 @@
 
 /* Indicate that this is LLVM compiled from the amd-gfx branch. */
 #define LLVM_HAVE_BRANCH_AMD_GFX
-#define LLVM_MAIN_REVISION 479279
+#define LLVM_MAIN_REVISION 479290
 
 /* Define if LLVM_ENABLE_DUMP is enabled */
 #cmakedefine LLVM_ENABLE_DUMP

llvm/lib/Analysis/VectorUtils.cpp

@@ -91,6 +91,8 @@ bool llvm::isTriviallyVectorizable(Intrinsic::ID ID) {
   case Intrinsic::canonicalize:
   case Intrinsic::fptosi_sat:
   case Intrinsic::fptoui_sat:
+  case Intrinsic::lrint:
+  case Intrinsic::llrint:
     return true;
   default:
     return false;
@@ -122,6 +124,8 @@ bool llvm::isVectorIntrinsicWithOverloadTypeAtArg(Intrinsic::ID ID,
   switch (ID) {
   case Intrinsic::fptosi_sat:
   case Intrinsic::fptoui_sat:
+  case Intrinsic::lrint:
+  case Intrinsic::llrint:
     return OpdIdx == -1 || OpdIdx == 0;
   case Intrinsic::is_fpclass:
     return OpdIdx == 0;

llvm/lib/CodeGen/TargetLoweringBase.cpp

@@ -2019,7 +2019,8 @@ void TargetLoweringBase::insertSSPDeclarations(Module &M) const {
     if (M.getDirectAccessExternalData() &&
         !TM.getTargetTriple().isWindowsGNUEnvironment() &&
         !TM.getTargetTriple().isOSFreeBSD() &&
-        !TM.getTargetTriple().isOSDarwin())
+        (!TM.getTargetTriple().isOSDarwin() ||
+         TM.getRelocationModel() == Reloc::Static))
       GV->setDSOLocal(true);
   }
 }

llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp

@@ -4978,7 +4978,7 @@ void ARMBaseInstrInfo::expandLoadStackGuardBase(MachineBasicBlock::iterator MI,
         TargetFlags |= ARMII::MO_DLLIMPORT;
       else if (IsIndirect)
         TargetFlags |= ARMII::MO_COFFSTUB;
-    } else if (Subtarget.isGVInGOT(GV)) {
+    } else if (IsIndirect) {
       TargetFlags |= ARMII::MO_GOT;
     }
 

llvm/lib/Target/ARM/ARMInstrInfo.cpp

@@ -104,8 +104,11 @@ void ARMInstrInfo::expandLoadStackGuard(MachineBasicBlock::iterator MI) const {
   const GlobalValue *GV =
       cast<GlobalValue>((*MI->memoperands_begin())->getValue());
 
-  if (!Subtarget.useMovt() || Subtarget.isGVInGOT(GV)) {
-    if (TM.isPositionIndependent())
+  bool ForceELFGOTPIC = Subtarget.isTargetELF() && !GV->isDSOLocal();
+  if (!Subtarget.useMovt() || ForceELFGOTPIC) {
+    // For ELF non-PIC, use GOT PIC code sequence as well because R_ARM_GOT_ABS
+    // does not have assembler support.
+    if (TM.isPositionIndependent() || ForceELFGOTPIC)
       expandLoadStackGuardBase(MI, ARM::LDRLIT_ga_pcrel, ARM::LDRi12);
     else
       expandLoadStackGuardBase(MI, ARM::LDRLIT_ga_abs, ARM::LDRi12);

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

@@ -978,6 +978,9 @@ struct OpenMPOpt {
       }
     }
 
+    if (OMPInfoCache.OpenMPPostLink)
+      Changed |= removeRuntimeSymbols();
+
     return Changed;
   }
 
@@ -1507,6 +1510,37 @@ struct OpenMPOpt {
     return Changed;
   }
 
+  /// Tries to remove known runtime symbols that are optional from the module.
+  bool removeRuntimeSymbols() {
+    // The RPC client symbol is defined in `libc` and indicates that something
+    // required an RPC server. If its users were all optimized out then we can
+    // safely remove it.
+    // TODO: This should be somewhere more common in the future.
+    if (GlobalVariable *GV = M.getNamedGlobal("__llvm_libc_rpc_client")) {
+      if (!GV->getType()->isPointerTy())
+        return false;
+
+      Constant *C = GV->getInitializer();
+      if (!C)
+        return false;
+
+      // Check to see if the only user of the RPC client is the external handle.
+      GlobalVariable *Client = dyn_cast<GlobalVariable>(C->stripPointerCasts());
+      if (!Client || Client->getNumUses() > 1 ||
+          Client->user_back() != GV->getInitializer())
+        return false;
+
+      Client->replaceAllUsesWith(PoisonValue::get(Client->getType()));
+      Client->eraseFromParent();
+
+      GV->replaceAllUsesWith(PoisonValue::get(GV->getType()));
+      GV->eraseFromParent();
+
+      return true;
+    }
+    return false;
+  }
+
   /// Tries to hide the latency of runtime calls that involve host to
   /// device memory transfers by splitting them into their "issue" and "wait"
   /// versions. The "issue" is moved upwards as much as possible. The "wait" is

llvm/lib/Transforms/Utils/SimplifyIndVar.cpp

@@ -539,7 +539,8 @@ bool SimplifyIndvar::eliminateTrunc(TruncInst *TI) {
   for (auto *ICI : ICmpUsers) {
     bool IsSwapped = L->isLoopInvariant(ICI->getOperand(0));
     auto *Op1 = IsSwapped ? ICI->getOperand(0) : ICI->getOperand(1);
-    Instruction *Ext = nullptr;
+    IRBuilder<> Builder(ICI);
+    Value *Ext = nullptr;
     // For signed/unsigned predicate, replace the old comparison with comparison
     // of immediate IV against sext/zext of the invariant argument. If we can
     // use either sext or zext (i.e. we are dealing with equality predicate),
@@ -550,18 +551,18 @@ bool SimplifyIndvar::eliminateTrunc(TruncInst *TI) {
     if (IsSwapped) Pred = ICmpInst::getSwappedPredicate(Pred);
     if (CanUseZExt(ICI)) {
       assert(DoesZExtCollapse && "Unprofitable zext?");
-      Ext = new ZExtInst(Op1, IVTy, "zext", ICI);
+      Ext = Builder.CreateZExt(Op1, IVTy, "zext");
       Pred = ICmpInst::getUnsignedPredicate(Pred);
     } else {
       assert(DoesSExtCollapse && "Unprofitable sext?");
-      Ext = new SExtInst(Op1, IVTy, "sext", ICI);
+      Ext = Builder.CreateSExt(Op1, IVTy, "sext");
       assert(Pred == ICmpInst::getSignedPredicate(Pred) && "Must be signed!");
     }
     bool Changed;
     L->makeLoopInvariant(Ext, Changed);
     (void)Changed;
-    ICmpInst *NewICI = new ICmpInst(ICI, Pred, IV, Ext);
-    ICI->replaceAllUsesWith(NewICI);
+    auto *NewCmp = Builder.CreateICmp(Pred, IV, Ext);
+    ICI->replaceAllUsesWith(NewCmp);
     DeadInsts.emplace_back(ICI);
   }
 

llvm/test/CodeGen/ARM/stack-guard-elf.ll

@@ -0,0 +1,55 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 3
+;; direct-access-external-data is false due to PIC Level, so __stack_chk_guard
+;; is dso_preemtable. Check that we use GOT PIC code sequence as well because
+;; R_ARM_GOT_ABS does not have assembler support.
+; RUN: llc -relocation-model=static < %s | FileCheck %s
+; RUN: llc -relocation-model=pic < %s | FileCheck %s
+
+target triple = "armv7a-linux-gnueabi"
+
+define i32 @test1() #0 {
+; CHECK-LABEL: test1:
+; CHECK:       @ %bb.0:
+; CHECK-NEXT:    push {r11, lr}
+; CHECK-NEXT:    sub sp, sp, #8
+; CHECK-NEXT:    sub sp, sp, #1024
+; CHECK-NEXT:    ldr r0, .LCPI0_0
+; CHECK-NEXT:  .LPC0_0:
+; CHECK-NEXT:    add r0, pc, r0
+; CHECK-NEXT:    ldr r0, [r0]
+; CHECK-NEXT:    ldr r0, [r0]
+; CHECK-NEXT:    str r0, [sp, #1028]
+; CHECK-NEXT:    add r0, sp, #4
+; CHECK-NEXT:    bl foo
+; CHECK-NEXT:    ldr r0, [sp, #1028]
+; CHECK-NEXT:    ldr r1, .LCPI0_1
+; CHECK-NEXT:  .LPC0_1:
+; CHECK-NEXT:    add r1, pc, r1
+; CHECK-NEXT:    ldr r1, [r1]
+; CHECK-NEXT:    ldr r1, [r1]
+; CHECK-NEXT:    cmp r1, r0
+; CHECK-NEXT:    moveq r0, #0
+; CHECK-NEXT:    addeq sp, sp, #8
+; CHECK-NEXT:    addeq sp, sp, #1024
+; CHECK-NEXT:    popeq {r11, pc}
+; CHECK-NEXT:  .LBB0_1:
+; CHECK-NEXT:    bl __stack_chk_fail
+; CHECK-NEXT:    .p2align 2
+; CHECK-NEXT:  @ %bb.2:
+; CHECK-NEXT:  .LCPI0_0:
+; CHECK-NEXT:  .Ltmp0:
+; CHECK-NEXT:    .long __stack_chk_guard(GOT_PREL)-((.LPC0_0+8)-.Ltmp0)
+; CHECK-NEXT:  .LCPI0_1:
+; CHECK-NEXT:  .Ltmp1:
+; CHECK-NEXT:    .long __stack_chk_guard(GOT_PREL)-((.LPC0_1+8)-.Ltmp1)
+  %a1 = alloca [256 x i32], align 4
+  call void @foo(ptr %a1) #3
+  ret i32 0
+}
+
+declare void @foo(ptr)
+
+attributes #0 = { nounwind sspstrong }
+
+!llvm.module.flags = !{!0}
+!0 = !{i32 8, !"PIC Level", i32 2}

llvm/test/CodeGen/ARM/stack_guard_remat.ll

@@ -1,9 +1,11 @@
-; RUN: llc < %s -mtriple=arm-apple-ios -relocation-model=pic -no-integrated-as | FileCheck %s -check-prefix=PIC
-; RUN: llc < %s -mtriple=arm-apple-ios -relocation-model=static -no-integrated-as | FileCheck %s -check-prefix=NO-PIC -check-prefix=STATIC
-; RUN: llc < %s -mtriple=arm-apple-ios -relocation-model=dynamic-no-pic -no-integrated-as | FileCheck %s  -check-prefix=NO-PIC -check-prefix=DYNAMIC-NO-PIC
-; RUN: llc < %s -mtriple=armv7-apple-ios -mcpu=cortex-a8 -relocation-model=pic -no-integrated-as | FileCheck %s -check-prefix=PIC-V7
-; RUN: llc < %s -mtriple=armv7-apple-ios -mcpu=cortex-a8 -relocation-model=static -no-integrated-as | FileCheck %s -check-prefix=STATIC-V7
-; RUN: llc < %s -mtriple=armv7-apple-ios -mcpu=cortex-a8 -relocation-model=dynamic-no-pic -no-integrated-as | FileCheck %s -check-prefix=DYNAMIC-NO-PIC-V7
+; RUN: rm -rf %t && split-file %s %t && cd %t
+; RUN: cat main.ll pic-flag.ll > pic.ll
+; RUN: llc < pic.ll -mtriple=arm-apple-ios -relocation-model=pic -no-integrated-as | FileCheck %s -check-prefix=PIC
+; RUN: llc < main.ll -mtriple=arm-apple-ios -relocation-model=static -no-integrated-as | FileCheck %s -check-prefix=NO-PIC -check-prefix=STATIC
+; RUN: llc < main.ll -mtriple=arm-apple-ios -relocation-model=dynamic-no-pic -no-integrated-as | FileCheck %s -check-prefixes=NO-PIC,DYNAMIC-NO-PIC
+; RUN: llc < pic.ll -mtriple=armv7-apple-ios -mcpu=cortex-a8 -relocation-model=pic -no-integrated-as | FileCheck %s -check-prefix=PIC-V7
+; RUN: llc < main.ll -mtriple=armv7-apple-ios -mcpu=cortex-a8 -relocation-model=static -no-integrated-as | FileCheck %s -check-prefix=STATIC-V7
+; RUN: llc < main.ll -mtriple=armv7-apple-ios -mcpu=cortex-a8 -relocation-model=dynamic-no-pic -no-integrated-as | FileCheck %s -check-prefix=DYNAMIC-NO-PIC-V7
 
 ;PIC:   foo2
 ;PIC:   ldr [[R0:r[0-9]+]], [[LABEL0:LCPI[0-9_]+]]
@@ -47,6 +49,7 @@
 ;DYNAMIC-NO-PIC-V7: L___stack_chk_guard$non_lazy_ptr:
 ;DYNAMIC-NO-PIC-V7:   .indirect_symbol        ___stack_chk_guard
 
+;--- main.ll
 ; Function Attrs: nounwind ssp
 define i32 @test_stack_guard_remat() #0 {
   %a1 = alloca [256 x i32], align 4
@@ -67,5 +70,6 @@ declare void @llvm.lifetime.end.p0(i64, ptr nocapture)
 
 attributes #0 = { nounwind ssp "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
 
+;--- pic-flag.ll
 !llvm.module.flags = !{!0}
 !0 = !{i32 7, !"PIC Level", i32 2}