[LLVM] Add file magic detection for SPIR-V files.

[jhuber6]

Summary:
More SPIR-V related patches are being upstreamed. We should add support
to detect when a binary file is SPIR-V. This will be used in the future
when support for SPIR-V is added to the offloading runtime or more
support for bundling.

The magic number is described in the official documentation:
https://registry.khronos.org/SPIR-V/specs/1.0/SPIRV.html#Magic. Notably,
SPIR-V files are streams of 32-bit words. This means that the magic
numbers differ depending on the endianness. Here we simply check the
strandard and byte-reversed versions.

[llvmbot]

@llvm/pr-subscribers-clang

@llvm/pr-subscribers-llvm-binary-utilities

Author: Joseph Huber (jhuber6)

Changes

Summary:
More SPIR-V related patches are being upstreamed. We should add support
to detect when a binary file is SPIR-V. This will be used in the future
when support for SPIR-V is added to the offloading runtime or more
support for bundling.

The magic number is described in the official documentation:
https://registry.khronos.org/SPIR-V/specs/1.0/SPIRV.html#Magic. Notably,
SPIR-V files are streams of 32-bit words. This means that the magic
numbers differ depending on the endianness. Here we simply check the
strandard and byte-reversed versions.

---

Full diff: https://github.com/llvm/llvm-project/pull/75363.diff

4 Files Affected:

• (modified) llvm/include/llvm/BinaryFormat/Magic.h (+1)
• (modified) llvm/lib/BinaryFormat/Magic.cpp (+9)
• (modified) llvm/lib/Object/Binary.cpp (+1)
• (modified) llvm/lib/Object/ObjectFile.cpp (+1)

diff --git a/llvm/include/llvm/BinaryFormat/Magic.h b/llvm/include/llvm/BinaryFormat/Magic.h
index a28710dcdfaf2c..c635a269576587 100644
--- a/llvm/include/llvm/BinaryFormat/Magic.h
+++ b/llvm/include/llvm/BinaryFormat/Magic.h
@@ -57,6 +57,7 @@ struct file_magic {
     dxcontainer_object,        ///< DirectX container file
     offload_bundle,            ///< Clang offload bundle file
     offload_bundle_compressed, ///< Compressed clang offload bundle file
+    spirv_object,              ///< A binary SPIR-V file
   };
 
   bool is_object() const { return V != unknown; }
diff --git a/llvm/lib/BinaryFormat/Magic.cpp b/llvm/lib/BinaryFormat/Magic.cpp
index 255937a5bdd04a..b0f0043f0e492a 100644
--- a/llvm/lib/BinaryFormat/Magic.cpp
+++ b/llvm/lib/BinaryFormat/Magic.cpp
@@ -72,6 +72,15 @@ file_magic llvm::identify_magic(StringRef Magic) {
   case 0x03:
     if (startswith(Magic, "\x03\xF0\x00"))
       return file_magic::goff_object;
+    // SPIR-V format in little-endian mode.
+    if (startswith(Magic, "\x03\x02\x23\x07"))
+      return file_magic::spirv_object;
+    break;
+
+  case 0x07:
+    // SPIR-V format in big-endian mode.
+    if (startswith(Magic, "\x07\x23\x02\x03"))
+      return file_magic::spirv_object;
     break;
 
   case 0x10:
diff --git a/llvm/lib/Object/Binary.cpp b/llvm/lib/Object/Binary.cpp
index 0ee9f7fac448a2..0b9d95485287dc 100644
--- a/llvm/lib/Object/Binary.cpp
+++ b/llvm/lib/Object/Binary.cpp
@@ -89,6 +89,7 @@ Expected<std::unique_ptr<Binary>> object::createBinary(MemoryBufferRef Buffer,
   case file_magic::dxcontainer_object:
   case file_magic::offload_bundle:
   case file_magic::offload_bundle_compressed:
+  case file_magic::spirv_object:
     // Unrecognized object file format.
     return errorCodeToError(object_error::invalid_file_type);
   case file_magic::offload_binary:
diff --git a/llvm/lib/Object/ObjectFile.cpp b/llvm/lib/Object/ObjectFile.cpp
index 428166f58070d0..ca921836b7f65a 100644
--- a/llvm/lib/Object/ObjectFile.cpp
+++ b/llvm/lib/Object/ObjectFile.cpp
@@ -160,6 +160,7 @@ ObjectFile::createObjectFile(MemoryBufferRef Object, file_magic Type,
   case file_magic::dxcontainer_object:
   case file_magic::offload_bundle:
   case file_magic::offload_bundle_compressed:
+  case file_magic::spirv_object:
     return errorCodeToError(object_error::invalid_file_type);
   case file_magic::tapi_file:
     return errorCodeToError(object_error::invalid_file_type);

[dwblaikie]

Perhaps llvm::identify_magic already has unit test coverage that could be extended to cover the new cases?

[jhuber6]

Added a test, for whatever reason I had to do a completely clean build to get the test to correctly pick up my changes to Magic.cpp, don't know why.

[github-actions]

:white_check_mark: With the latest revision this PR passed the C/C++ code formatter.

[dwblaikie]

SGTM