diff --git a/llvm/lib/Target/AMDGPU/AMDGPULowerKernelArguments.cpp b/llvm/lib/Target/AMDGPU/AMDGPULowerKernelArguments.cpp
--- a/llvm/lib/Target/AMDGPU/AMDGPULowerKernelArguments.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPULowerKernelArguments.cpp
@@ -13,10 +13,12 @@
 
 #include "AMDGPU.h"
 #include "GCNSubtarget.h"
+#include "SIMachineFunctionInfo.h"
 #include "llvm/CodeGen/TargetPassConfig.h"
-#include "llvm/IR/IntrinsicsAMDGPU.h"
 #include "llvm/IR/IRBuilder.h"
+#include "llvm/IR/IntrinsicsAMDGPU.h"
 #include "llvm/IR/MDBuilder.h"
+#include "llvm/IR/Metadata.h"
 #include "llvm/Target/TargetMachine.h"
 #define DEBUG_TYPE "amdgpu-lower-kernel-arguments"
 
@@ -24,6 +26,86 @@
 
 namespace {
 
+class PreloadKernelArgInfo {
+private:
+  Function &F;
+
+  const GCNSubtarget &ST;
+
+  unsigned NumFreeUserSGPRs;
+
+public:
+  SmallVector<llvm::Metadata *, 8> KernelArgMetadata;
+
+  PreloadKernelArgInfo(Function &F, const GCNSubtarget &ST) : F(F), ST(ST) {
+    setInitialFreeUserSGPRsCount();
+  }
+
+  // Returns the maximum number of user SGPRs that we have available to preload
+  // arguments.
+  void setInitialFreeUserSGPRsCount() {
+    const Module *M = F.getParent();
+
+    const unsigned MaxUserSGRPs = 16;
+    unsigned NumUserSGPRs = 0;
+
+    const CallingConv::ID CC = F.getCallingConv();
+    const bool IsKernel =
+        CC == CallingConv::AMDGPU_KERNEL || CC == CallingConv::SPIR_KERNEL;
+    const bool HasCalls = F.hasFnAttribute("amdgpu-calls");
+    const bool HasStackObjects = F.hasFnAttribute("amdgpu-stack-objects");
+
+    // ImplicitBufferPtr
+    if (ST.isMesaGfxShader(F))
+      NumUserSGPRs += 2;
+
+    // PrivateSegmentBuffer
+    if (ST.isAmdHsaOrMesa(F) && !ST.enableFlatScratch())
+      NumUserSGPRs += 4;
+
+    // DispatchPtr
+    if (!F.hasFnAttribute("amdgpu-no-dispatch-ptr"))
+      NumUserSGPRs += 2;
+
+    // QueuePtr
+    if (!F.hasFnAttribute("amdgpu-no-queue-ptr") &&
+        AMDGPU::getCodeObjectVersion(*M) < AMDGPU::AMDHSA_COV5) {
+      NumUserSGPRs += 2;
+    }
+
+    // KernargSegmentPtr
+    if (IsKernel && (!F.arg_empty() || ST.getImplicitArgNumBytes(F) != 0))
+      NumUserSGPRs += 2;
+
+    // DispatchID
+    if (!F.hasFnAttribute("amdgpu-no-dispatch-id"))
+      NumUserSGPRs += 2;
+
+    // FlatScratchInit
+    if (ST.hasFlatAddressSpace() && AMDGPU::isEntryFunctionCC(CC) &&
+        (ST.isAmdHsaOrMesa(F) || ST.enableFlatScratch()) &&
+        (HasCalls || HasStackObjects || ST.enableFlatScratch()) &&
+        !ST.flatScratchIsArchitected()) {
+      NumUserSGPRs += 2;
+    }
+
+    // LDSKernelId
+    if (!IsKernel && !F.hasFnAttribute("amdgpu-no-lds-kernel-id"))
+      NumUserSGPRs += 1;
+
+    NumFreeUserSGPRs = MaxUserSGRPs - NumUserSGPRs;
+  }
+
+  bool allocPreloadArg(unsigned AllocSize) {
+    unsigned AllocSizeDword = alignTo(AllocSize, 4) / 4;
+    if (AllocSizeDword > NumFreeUserSGPRs)
+      return false;
+
+    NumFreeUserSGPRs -= AllocSizeDword;
+    return true;
+  }
+};
+
 class AMDGPULowerKernelArguments : public FunctionPass{
 public:
   static char ID;
@@ -87,8 +169,14 @@
       Attribute::getWithDereferenceableBytes(Ctx, TotalKernArgSize));
 
   uint64_t ExplicitArgOffset = 0;
+  // Preloaded kernel arguments must be sequential.
+  bool InPreloadSequence = false;
+  bool HasPreloadArgs = false;
+  PreloadKernelArgInfo PreloadInfo(F, ST);
 
+  int Idx = -1;
   for (Argument &Arg : F.args()) {
+    Idx++;
     const bool IsByRef = Arg.hasByRefAttr();
     Type *ArgTy = IsByRef ? Arg.getParamByRefType() : Arg.getType();
     MaybeAlign ParamAlign = IsByRef ? Arg.getParamAlign() : std::nullopt;
@@ -103,6 +191,19 @@
     if (Arg.use_empty())
       continue;
 
+    if (Arg.hasInRegAttr() && (InPreloadSequence || !HasPreloadArgs) &&
+        PreloadInfo.allocPreloadArg(AllocSize)) {
+      // Preload this argument.
+      InPreloadSequence = true;
+      HasPreloadArgs = true;
+      MDBuilder MDB(Ctx);
+      PreloadInfo.KernelArgMetadata.push_back(
+          llvm::MDNode::get(Ctx, MDB.createConstant(llvm::ConstantInt::get(
+                                     Builder.getInt32Ty(), Idx))));
+    } else if (InPreloadSequence) {
+      InPreloadSequence = false;
+    }
+
     // If this is byval, the loads are already explicit in the function. We just
     // need to rewrite the pointer values.
     if (IsByRef) {
@@ -226,6 +327,11 @@
     }
   }
 
+  if (HasPreloadArgs) {
+    F.setMetadata("preload_kernel_args",
+                  llvm::MDNode::get(Ctx, PreloadInfo.KernelArgMetadata));
+  }
+
   KernArgSegment->addRetAttr(
       Attribute::getWithAlignment(Ctx, std::max(KernArgBaseAlign, MaxAlign)));
 
diff --git a/llvm/test/CodeGen/AMDGPU/preload-kernal-args-metadata.ll b/llvm/test/CodeGen/AMDGPU/preload-kernal-args-metadata.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/preload-kernal-args-metadata.ll
@@ -0,0 +1,451 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature
+; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -amdgpu-attributor -amdgpu-lower-kernel-arguments -S < %s | FileCheck -check-prefix=NO-PRELOAD %s
+; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -amdgpu-attributor -amdgpu-lower-kernel-arguments -amdgpu-kernarg-preload-count=1 -S < %s | FileCheck -check-prefix=PRELOAD-1 %s
+; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -amdgpu-attributor -amdgpu-lower-kernel-arguments -amdgpu-kernarg-preload-count=3 -S < %s | FileCheck -check-prefix=PRELOAD-3 %s
+; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -amdgpu-attributor -amdgpu-lower-kernel-arguments -amdgpu-kernarg-preload-count=8 -S < %s | FileCheck -check-prefix=PRELOAD-8 %s
+
+; Metadata for each runline is at the bottom of the file.
+
+define amdgpu_kernel void @test_preload_metadata_kernel_2(ptr addrspace(1) %in, ptr addrspace(1) %out) #0 {
+; NO-PRELOAD-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_2
+; NO-PRELOAD-SAME: (ptr addrspace(1) [[IN:%.*]], ptr addrspace(1) [[OUT:%.*]]) #[[ATTR0:[0-9]+]] {
+; NO-PRELOAD-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(16) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; NO-PRELOAD-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 0
+; NO-PRELOAD-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 8
+; NO-PRELOAD-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; NO-PRELOAD-NEXT:    ret void
+;
+; PRELOAD-1-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_2
+; PRELOAD-1-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) [[OUT:%.*]]) #[[ATTR0:[0-9]+]] !preload_kernel_args !0 {
+; PRELOAD-1-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(16) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-1-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 0
+; PRELOAD-1-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 8
+; PRELOAD-1-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-1-NEXT:    ret void
+;
+; PRELOAD-3-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_2
+; PRELOAD-3-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[OUT:%.*]]) #[[ATTR0:[0-9]+]] !preload_kernel_args !0 {
+; PRELOAD-3-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(16) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-3-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 0
+; PRELOAD-3-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 8
+; PRELOAD-3-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-3-NEXT:    ret void
+;
+; PRELOAD-8-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_2
+; PRELOAD-8-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[OUT:%.*]]) #[[ATTR0:[0-9]+]] !preload_kernel_args !0 {
+; PRELOAD-8-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(16) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-8-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 0
+; PRELOAD-8-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_2_KERNARG_SEGMENT]], i64 8
+; PRELOAD-8-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-8-NEXT:    ret void
+;
+  %load = load i32, ptr addrspace(1) %in
+  store i32 %load, ptr addrspace(1) %out
+  ret void
+}
+
+define amdgpu_kernel void @test_preload_metadata_kernel_4(ptr addrspace(1) %in, ptr addrspace(1) %in1, ptr addrspace(1) %out, ptr addrspace(1) %out1) #0 {
+; NO-PRELOAD-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4
+; NO-PRELOAD-SAME: (ptr addrspace(1) [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) [[OUT1:%.*]]) #[[ATTR0]] {
+; NO-PRELOAD-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; NO-PRELOAD-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 0
+; NO-PRELOAD-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 8
+; NO-PRELOAD-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 16
+; NO-PRELOAD-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 24
+; NO-PRELOAD-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; NO-PRELOAD-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    ret void
+;
+; PRELOAD-1-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4
+; PRELOAD-1-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !0 {
+; PRELOAD-1-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-1-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 0
+; PRELOAD-1-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 8
+; PRELOAD-1-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 16
+; PRELOAD-1-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 24
+; PRELOAD-1-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-1-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-1-NEXT:    ret void
+;
+; PRELOAD-3-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4
+; PRELOAD-3-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !4 {
+; PRELOAD-3-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-3-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 0
+; PRELOAD-3-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 8
+; PRELOAD-3-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 16
+; PRELOAD-3-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 24
+; PRELOAD-3-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-3-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-3-NEXT:    ret void
+;
+; PRELOAD-8-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4
+; PRELOAD-8-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !4 {
+; PRELOAD-8-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-8-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 0
+; PRELOAD-8-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 8
+; PRELOAD-8-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 16
+; PRELOAD-8-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_KERNARG_SEGMENT]], i64 24
+; PRELOAD-8-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-8-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-8-NEXT:    ret void
+;
+  %load = load i32, ptr addrspace(1) %in
+  %load1 = load i32, ptr addrspace(1) %in1
+  store i32 %load, ptr addrspace(1) %out
+  store i32 %load1, ptr addrspace(1) %out1
+  ret void
+}
+
+define amdgpu_kernel void @test_preload_metadata_kernel_8(ptr addrspace(1) %in, ptr addrspace(1) %in1, ptr addrspace(1) %in2, ptr addrspace(1) %in3, ptr addrspace(1) %out, ptr addrspace(1) %out1, ptr addrspace(1) %out2, ptr addrspace(1) %out3) #0 {
+; NO-PRELOAD-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_8
+; NO-PRELOAD-SAME: (ptr addrspace(1) [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) [[IN2:%.*]], ptr addrspace(1) [[IN3:%.*]], ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) [[OUT1:%.*]], ptr addrspace(1) [[OUT2:%.*]], ptr addrspace(1) [[OUT3:%.*]]) #[[ATTR0]] {
+; NO-PRELOAD-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; NO-PRELOAD-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 0
+; NO-PRELOAD-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 8
+; NO-PRELOAD-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[IN2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 16
+; NO-PRELOAD-NEXT:    [[IN2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN2_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[IN3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 24
+; NO-PRELOAD-NEXT:    [[IN3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN3_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 32
+; NO-PRELOAD-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 40
+; NO-PRELOAD-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 48
+; NO-PRELOAD-NEXT:    [[OUT2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT2_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 56
+; NO-PRELOAD-NEXT:    [[OUT3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT3_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; NO-PRELOAD-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    [[LOAD2:%.*]] = load i32, ptr addrspace(1) [[IN2_LOAD]], align 4
+; NO-PRELOAD-NEXT:    [[LOAD3:%.*]] = load i32, ptr addrspace(1) [[IN3_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD2]], ptr addrspace(1) [[OUT2_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD3]], ptr addrspace(1) [[OUT3_LOAD]], align 4
+; NO-PRELOAD-NEXT:    ret void
+;
+; PRELOAD-1-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_8
+; PRELOAD-1-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) [[IN2:%.*]], ptr addrspace(1) [[IN3:%.*]], ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) [[OUT1:%.*]], ptr addrspace(1) [[OUT2:%.*]], ptr addrspace(1) [[OUT3:%.*]]) #[[ATTR0]] !preload_kernel_args !0 {
+; PRELOAD-1-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-1-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 0
+; PRELOAD-1-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 8
+; PRELOAD-1-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[IN2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 16
+; PRELOAD-1-NEXT:    [[IN2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN2_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[IN3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 24
+; PRELOAD-1-NEXT:    [[IN3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN3_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 32
+; PRELOAD-1-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 40
+; PRELOAD-1-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 48
+; PRELOAD-1-NEXT:    [[OUT2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT2_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 56
+; PRELOAD-1-NEXT:    [[OUT3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT3_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-1-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-1-NEXT:    [[LOAD2:%.*]] = load i32, ptr addrspace(1) [[IN2_LOAD]], align 4
+; PRELOAD-1-NEXT:    [[LOAD3:%.*]] = load i32, ptr addrspace(1) [[IN3_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD2]], ptr addrspace(1) [[OUT2_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD3]], ptr addrspace(1) [[OUT3_LOAD]], align 4
+; PRELOAD-1-NEXT:    ret void
+;
+; PRELOAD-3-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_8
+; PRELOAD-3-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[IN2:%.*]], ptr addrspace(1) [[IN3:%.*]], ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) [[OUT1:%.*]], ptr addrspace(1) [[OUT2:%.*]], ptr addrspace(1) [[OUT3:%.*]]) #[[ATTR0]] !preload_kernel_args !4 {
+; PRELOAD-3-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-3-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 0
+; PRELOAD-3-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 8
+; PRELOAD-3-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[IN2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 16
+; PRELOAD-3-NEXT:    [[IN2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN2_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[IN3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 24
+; PRELOAD-3-NEXT:    [[IN3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN3_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 32
+; PRELOAD-3-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 40
+; PRELOAD-3-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 48
+; PRELOAD-3-NEXT:    [[OUT2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT2_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 56
+; PRELOAD-3-NEXT:    [[OUT3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT3_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-3-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-3-NEXT:    [[LOAD2:%.*]] = load i32, ptr addrspace(1) [[IN2_LOAD]], align 4
+; PRELOAD-3-NEXT:    [[LOAD3:%.*]] = load i32, ptr addrspace(1) [[IN3_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD2]], ptr addrspace(1) [[OUT2_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD3]], ptr addrspace(1) [[OUT3_LOAD]], align 4
+; PRELOAD-3-NEXT:    ret void
+;
+; PRELOAD-8-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_8
+; PRELOAD-8-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[IN2:%.*]], ptr addrspace(1) inreg [[IN3:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]], ptr addrspace(1) inreg [[OUT2:%.*]], ptr addrspace(1) inreg [[OUT3:%.*]]) #[[ATTR0]] !preload_kernel_args !7 {
+; PRELOAD-8-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-8-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 0
+; PRELOAD-8-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 8
+; PRELOAD-8-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[IN2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 16
+; PRELOAD-8-NEXT:    [[IN2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN2_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[IN3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 24
+; PRELOAD-8-NEXT:    [[IN3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN3_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 32
+; PRELOAD-8-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 40
+; PRELOAD-8-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT2_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 48
+; PRELOAD-8-NEXT:    [[OUT2_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT2_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT3_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_8_KERNARG_SEGMENT]], i64 56
+; PRELOAD-8-NEXT:    [[OUT3_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT3_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-8-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-8-NEXT:    [[LOAD2:%.*]] = load i32, ptr addrspace(1) [[IN2_LOAD]], align 4
+; PRELOAD-8-NEXT:    [[LOAD3:%.*]] = load i32, ptr addrspace(1) [[IN3_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD2]], ptr addrspace(1) [[OUT2_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD3]], ptr addrspace(1) [[OUT3_LOAD]], align 4
+; PRELOAD-8-NEXT:    ret void
+;
+  %load = load i32, ptr addrspace(1) %in
+  %load1 = load i32, ptr addrspace(1) %in1
+  %load2 = load i32, ptr addrspace(1) %in2
+  %load3 = load i32, ptr addrspace(1) %in3
+  store i32 %load, ptr addrspace(1) %out
+  store i32 %load1, ptr addrspace(1) %out1
+  store i32 %load2, ptr addrspace(1) %out2
+  store i32 %load3, ptr addrspace(1) %out3
+  ret void
+}
+
+; Preload args with inreg in the NO-PRELOAD case.
+
+define amdgpu_kernel void @test_preload_metadata_kernel_4_inreg_offset(ptr addrspace(1) %in, ptr addrspace(1) %in1, ptr addrspace(1) inreg %out, ptr addrspace(1) inreg %out1) #0 {
+; NO-PRELOAD-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset
+; NO-PRELOAD-SAME: (ptr addrspace(1) [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !1 {
+; NO-PRELOAD-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; NO-PRELOAD-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 0
+; NO-PRELOAD-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 8
+; NO-PRELOAD-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 16
+; NO-PRELOAD-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 24
+; NO-PRELOAD-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; NO-PRELOAD-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    ret void
+;
+; PRELOAD-1-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset
+; PRELOAD-1-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !0 {
+; PRELOAD-1-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-1-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 0
+; PRELOAD-1-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 8
+; PRELOAD-1-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 16
+; PRELOAD-1-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 24
+; PRELOAD-1-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-1-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-1-NEXT:    ret void
+;
+; PRELOAD-3-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset
+; PRELOAD-3-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !6 {
+; PRELOAD-3-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-3-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 0
+; PRELOAD-3-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 8
+; PRELOAD-3-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 16
+; PRELOAD-3-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 24
+; PRELOAD-3-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-3-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-3-NEXT:    ret void
+;
+; PRELOAD-8-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset
+; PRELOAD-8-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !4 {
+; PRELOAD-8-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-8-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 0
+; PRELOAD-8-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 8
+; PRELOAD-8-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 16
+; PRELOAD-8-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_KERNARG_SEGMENT]], i64 24
+; PRELOAD-8-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-8-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-8-NEXT:    ret void
+;
+  %load = load i32, ptr addrspace(1) %in
+  %load1 = load i32, ptr addrspace(1) %in1
+  store i32 %load, ptr addrspace(1) %out
+  store i32 %load1, ptr addrspace(1) %out1
+  ret void
+}
+
+; Only preload the first sequence of arguments with the inreg attribute. In the NO-PRELOAD case this is just the first argument.
+
+define amdgpu_kernel void @test_preload_metadata_kernel_4_inreg_offset_two_sequence(ptr addrspace(1) inreg %in, ptr addrspace(1) %in1, ptr addrspace(1) inreg %out, ptr addrspace(1) inreg %out1) #0 {
+; NO-PRELOAD-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset_two_sequence
+; NO-PRELOAD-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !4 {
+; NO-PRELOAD-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; NO-PRELOAD-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 0
+; NO-PRELOAD-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 8
+; NO-PRELOAD-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 16
+; NO-PRELOAD-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 24
+; NO-PRELOAD-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !0
+; NO-PRELOAD-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; NO-PRELOAD-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; NO-PRELOAD-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; NO-PRELOAD-NEXT:    ret void
+;
+; PRELOAD-1-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset_two_sequence
+; PRELOAD-1-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !0 {
+; PRELOAD-1-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-1-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 0
+; PRELOAD-1-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 8
+; PRELOAD-1-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 16
+; PRELOAD-1-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !2
+; PRELOAD-1-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 24
+; PRELOAD-1-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !2
+; PRELOAD-1-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-1-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-1-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-1-NEXT:    ret void
+;
+; PRELOAD-3-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset_two_sequence
+; PRELOAD-3-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !6 {
+; PRELOAD-3-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-3-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 0
+; PRELOAD-3-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 8
+; PRELOAD-3-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 16
+; PRELOAD-3-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-3-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 24
+; PRELOAD-3-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-3-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-3-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-3-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-3-NEXT:    ret void
+;
+; PRELOAD-8-LABEL: define {{[^@]+}}@test_preload_metadata_kernel_4_inreg_offset_two_sequence
+; PRELOAD-8-SAME: (ptr addrspace(1) inreg [[IN:%.*]], ptr addrspace(1) inreg [[IN1:%.*]], ptr addrspace(1) inreg [[OUT:%.*]], ptr addrspace(1) inreg [[OUT1:%.*]]) #[[ATTR0]] !preload_kernel_args !4 {
+; PRELOAD-8-NEXT:    [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT:%.*]] = call nonnull align 16 dereferenceable(32) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
+; PRELOAD-8-NEXT:    [[IN_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 0
+; PRELOAD-8-NEXT:    [[IN_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[IN1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 8
+; PRELOAD-8-NEXT:    [[IN1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[IN1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 16
+; PRELOAD-8-NEXT:    [[OUT_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT_KERNARG_OFFSET]], align 16, !invariant.load !3
+; PRELOAD-8-NEXT:    [[OUT1_KERNARG_OFFSET:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[TEST_PRELOAD_METADATA_KERNEL_4_INREG_OFFSET_TWO_SEQUENCE_KERNARG_SEGMENT]], i64 24
+; PRELOAD-8-NEXT:    [[OUT1_LOAD:%.*]] = load ptr addrspace(1), ptr addrspace(4) [[OUT1_KERNARG_OFFSET]], align 8, !invariant.load !3
+; PRELOAD-8-NEXT:    [[LOAD:%.*]] = load i32, ptr addrspace(1) [[IN_LOAD]], align 4
+; PRELOAD-8-NEXT:    [[LOAD1:%.*]] = load i32, ptr addrspace(1) [[IN1_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD]], ptr addrspace(1) [[OUT_LOAD]], align 4
+; PRELOAD-8-NEXT:    store i32 [[LOAD1]], ptr addrspace(1) [[OUT1_LOAD]], align 4
+; PRELOAD-8-NEXT:    ret void
+;
+  %load = load i32, ptr addrspace(1) %in
+  %load1 = load i32, ptr addrspace(1) %in1
+  store i32 %load, ptr addrspace(1) %out
+  store i32 %load1, ptr addrspace(1) %out1
+  ret void
+}
+
+attributes #0 = { nounwind }
+
+; Metadata
+
+; NO-PRELOAD: !0 = !{}
+; NO-PRELOAD-NEXT: !1 = !{!2, !3}
+; NO-PRELOAD-NEXT: !2 = !{i32 2}
+; NO-PRELOAD-NEXT: !3 = !{i32 3}
+; NO-PRELOAD-NEXT: !4 = !{!5}
+; NO-PRELOAD-NEXT: !5 = !{i32 0}
+
+; PRELOAD-1: 0 = !{!1}
+; PRELOAD-1-NEXT: !1 = !{i32 0}
+; PRELOAD-1-NEXT: !2 = !{}
+
+; PRELOAD-3: !0 = !{!1, !2}
+; PRELOAD-3-NEXT: !1 = !{i32 0}
+; PRELOAD-3-NEXT: !2 = !{i32 1}
+; PRELOAD-3-NEXT: !3 = !{}
+; PRELOAD-3-NEXT: !4 = !{!1, !2, !5}
+; PRELOAD-3-NEXT: !5 = !{i32 2}
+; PRELOAD-3-NEXT: !6 = !{!1, !2, !5, !7}
+; PRELOAD-3-NEXT: !7 = !{i32 3}
+
+; PRELOAD-8: !0 = !{!1, !2}
+; PREALOD-8-NEXT: !1 = !{i32 0}
+; PREALOD-8-NEXT: !2 = !{i32 1}
+; PREALOD-8-NEXT: !3 = !{!1, !2, !4, !5}
+; PREALOD-8-NEXT: !4 = !{i32 2}
+; PREALOD-8-NEXT: !5 = !{i32 3}
+; PREALOD-8-NEXT: !6 = !{!1, !2, !4, !5, !7}
+; PREALOD-8-NEXT: !7 = !{i32 4}
+; PREALOD-8-NEXT: !8 = !{}