diff --git a/llvm/test/tools/llvm-readobj/ELF/note-amdgpu-invalid.s b/llvm/test/tools/llvm-readobj/ELF/note-amdgpu-invalid.s
new file mode 100644
--- /dev/null
+++ b/llvm/test/tools/llvm-readobj/ELF/note-amdgpu-invalid.s
@@ -0,0 +1,63 @@
+# RUN: yaml2obj %s -o %t.o
+# RUN: llvm-readobj --notes %t.o | FileCheck %s --check-prefix=LLVM
+# RUN: llvm-readelf --notes %t.o | FileCheck %s --check-prefix=GNU
+
+# GNU:      Displaying notes found in: .note.foo
+# GNU-NEXT:   Owner                Data size        Description
+# GNU-NEXT:   AMDGPU               0x000000d4       NT_AMDGPU_METADATA (AMDGPU Metadata)
+# GNU-NEXT:     AMDGPU Metadata:
+# GNU-NEXT:         Invalid AMDGPU Metadata
+# GNU-NEXT:         ---
+# GNU-NEXT: amdhsa.kernels:
+# GNU-NEXT:   - .group_segment_fixed_size: 2
+# GNU-NEXT:     .kernarg_segment_align: 4
+# GNU-NEXT:     .kernarg_segment_size: 1
+# GNU-NEXT:     .max_flat_workgroup_size: 8
+# GNU-NEXT:     .name:           foo
+# GNU-NEXT:     .private_segment_fixed_size: 3
+# GNU-NEXT:     .sgpr_count:     6
+# GNU-NEXT:     .symbol:         foo
+# GNU-NEXT:     .vgpr_count:     7
+# GNU-NEXT:     .wavefront_size: 5
+# GNU-NEXT: ...
+
+# LLVM:      Notes [
+# LLVM-NEXT:   NoteSection {
+# LLVM-NEXT:     Name: .note.foo
+# LLVM-NEXT:     Offset: 0x40
+# LLVM-NEXT:     Size: 0xE8
+# LLVM-NEXT:     Note {
+# LLVM-NEXT:       Owner: AMDGPU
+# LLVM-NEXT:       Data size: 0xD4
+# LLVM-NEXT:       Type: NT_AMDGPU_METADATA (AMDGPU Metadata)
+# LLVM-NEXT:       AMDGPU Metadata: Invalid AMDGPU Metadata
+# LLVM-NEXT: ---
+# LLVM-NEXT: amdhsa.kernels:
+# LLVM-NEXT:   - .group_segment_fixed_size: 2
+# LLVM-NEXT:     .kernarg_segment_align: 4
+# LLVM-NEXT:     .kernarg_segment_size: 1
+# LLVM-NEXT:     .max_flat_workgroup_size: 8
+# LLVM-NEXT:     .name:           foo
+# LLVM-NEXT:     .private_segment_fixed_size: 3
+# LLVM-NEXT:     .sgpr_count:     6
+# LLVM-NEXT:     .symbol:         foo
+# LLVM-NEXT:     .vgpr_count:     7
+# LLVM-NEXT:     .wavefront_size: 5
+# LLVM-NEXT: ...
+# LLVM-EMPTY:
+# LLVM-NEXT:     }
+# LLVM-NEXT:   }
+# LLVM-NEXT: ]
+
+## Use yaml2obj instead of llvm-mc for more test portability. This was
+## generated by grabbing section data from note-amdgpu.s and removing the amdhsa.version field.
+
+--- !ELF
+FileHeader:
+  Class: ELFCLASS64
+  Data:  ELFDATA2LSB
+  Type:  ET_REL
+Sections:
+  - Name:        .note.foo
+    Type:        SHT_NOTE
+    Content:     07000000D400000020000000414D44475055000081ae616d646873612e6b65726e656c73918ab92e67726f75705f7365676d656e745f66697865645f73697a6502b62e6b65726e6172675f7365676d656e745f616c69676e04b52e6b65726e6172675f7365676d656e745f73697a6501b82e6d61785f666c61745f776f726b67726f75705f73697a6508a52e6e616d65a3666f6fbb2e707269766174655f7365676d656e745f66697865645f73697a6503ab2e736770725f636f756e7406a72e73796d626f6ca3666f6fab2e766770725f636f756e7407af2e7761766566726f6e745f73697a6505