[llvm-branch-commits] [llvm] [AMDGPU] Add `.amdgpu.info` section for per-function metadata (PR #192384)

[Juan Manuel Martinez Caamaño via llvm-branch-commits]

================
@@ -0,0 +1,77 @@
+//===--- AMDGPUObjLinkingInfo.h ---------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+/// \file
+/// Enums shared between the AMDGPU backend (LLVM) and the ELF linker (LLD)
+/// for the `.amdgpu.info` object-linking metadata section.
+///
+/// Binary layout of each entry: [kind: u8] [len: u8] [payload: <len> bytes].
+/// Unknown kinds are forward-compatible: a consumer skips them by reading len.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_SUPPORT_AMDGPUOBJECTLINKINGINFO_H
+#define LLVM_SUPPORT_AMDGPUOBJECTLINKINGINFO_H
+
+#include <cstdint>
+
+namespace llvm {
+namespace AMDGPU {
+
+/// Entry kind values for the `.amdgpu.info` section.
+///
+/// Entries that appear between an INFO_FUNC and the next INFO_FUNC (or end of
+/// section) belong to the function scope opened by that INFO_FUNC.
+enum InfoKind : uint8_t {
+  /// Opens a new function scope.  Payload is an 8-byte symbol reference
+  /// (relocated) identifying the function.  All subsequent entries until the
+  /// next INFO_FUNC belong to this function.
+  INFO_FUNC = 1,
+  /// Number of architectural VGPRs used by the function.  [u32]
+  INFO_NUM_VGPR = 2,
+  /// Number of accumulator VGPRs (AGPRs) used by the function.  [u32]
+  INFO_NUM_AGPR = 3,
+  /// Number of SGPRs explicitly used by the function.  [u32]
+  INFO_NUM_SGPR = 4,
+  /// Private (scratch) memory size in bytes required by the function.  [u32]
+  INFO_PRIVATE_SEGMENT_SIZE = 5,
+  /// Target occupancy for the function, expressed as the maximum number of
+  /// waves per EU the compiler expects.  The linker uses this to guide
+  /// resource allocation decisions (e.g. LDS partitioning) so that the
+  /// resulting program can meet the occupancy target.  [u32]
+  INFO_OCCUPANCY = 6,
+  /// Bitfield of FuncInfoFlags properties for the function.  [u32]
+  INFO_FLAGS = 7,
+  /// Dependency edge: the function uses the resource identified by the
+  /// 8-byte relocated symbol (e.g. an LDS variable or named barrier).
+  INFO_USE = 8,
+  /// Direct call edge: the function calls the callee identified by the
+  /// 8-byte relocated symbol.
+  INFO_CALL = 9,
+  /// Indirect call edge: the function contains an indirect call whose
+  /// callee is expected to match the type-ID string at the given
+  /// `.amdgpu.strtab` offset.  [u32]
+  INFO_INDIRECT_CALL = 10,
+  /// Function type ID: tags an address-taken function with a type-ID
+  /// string (at the given `.amdgpu.strtab` offset) so the linker can match
+  /// it against INFO_INDIRECT_CALL entries.  [u32]
+  INFO_TYPEID = 11,
+};
+
+/// Per-function flags packed into INFO_FLAGS entries.
+enum FuncInfoFlags : uint32_t {
+  FUNC_IS_KERNEL = 0x1,
+  FUNC_USES_VCC = 0x2,
+  FUNC_USES_FLAT_SCRATCH = 0x4,
+  FUNC_HAS_DYN_STACK = 0x8,
----------------
jmmartinez wrote:

Silly remark, but for bitsets I'd rather use the less error prone:
```cpp
0x1 << 0
0x1 << 1
0x1 << 2
```

Because I've already seen 0x10 used instead of 0xa elsewhere.

https://github.com/llvm/llvm-project/pull/192384