[BinaryFormat] Adjust OSABI functions and add unittests

[MaskRay]

Adjust #89280:

• ELFOSABI_LINUX is a historical alias that should not be used in new
  code. readelf -h displays "UNIX - GNU" instead of "Linux".
• "OS" is inappropriate. Some values are architecture-specific, e.g.
  ELFOSABI_ARM.
• Drop lowercase, which seems a job of the caller.

Add some unittests.

[llvmbot]

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

Author: Fangrui Song (MaskRay)

Changes

Adjust #89280:
ELFOSABI_LINUX is a historical alias that should not be used in new
code. readelf -h displays "UNIX - GNU" instead of "Linux". Add some
unittests.

---

Full diff: https://github.com/llvm/llvm-project/pull/90270.diff

3 Files Affected:

• (modified) llvm/lib/BinaryFormat/ELF.cpp (+3-3)
• (modified) llvm/unittests/BinaryFormat/CMakeLists.txt (+1)
• (added) llvm/unittests/BinaryFormat/ELFTest.cpp (+32)

diff --git a/llvm/lib/BinaryFormat/ELF.cpp b/llvm/lib/BinaryFormat/ELF.cpp
index 8c10ed1a980bbc..a07e5423a2ab66 100644
--- a/llvm/lib/BinaryFormat/ELF.cpp
+++ b/llvm/lib/BinaryFormat/ELF.cpp
@@ -573,7 +573,7 @@ uint8_t ELF::convertOSToOSAbi(StringRef OS) {
   return StringSwitch<uint16_t>(LowerOS)
       .StartsWith("hpux", ELFOSABI_HPUX)
       .StartsWith("netbsd", ELFOSABI_NETBSD)
-      .StartsWith("linux", ELFOSABI_LINUX)
+      .StartsWith("gnu", ELFOSABI_GNU)
       .StartsWith("hurd", ELFOSABI_HURD)
       .StartsWith("solaris", ELFOSABI_SOLARIS)
       .StartsWith("aix", ELFOSABI_AIX)
@@ -603,8 +603,8 @@ StringRef ELF::convertOSAbiToOS(uint8_t OSAbi) {
     return "hpux";
   case ELFOSABI_NETBSD:
     return "netbsd";
-  case ELFOSABI_LINUX:
-    return "linux";
+  case ELFOSABI_GNU:
+    return "gnu";
   case ELFOSABI_HURD:
     return "hurd";
   case ELFOSABI_SOLARIS:
diff --git a/llvm/unittests/BinaryFormat/CMakeLists.txt b/llvm/unittests/BinaryFormat/CMakeLists.txt
index f0c42a0dd02b8e..40d3bc4dca0b66 100644
--- a/llvm/unittests/BinaryFormat/CMakeLists.txt
+++ b/llvm/unittests/BinaryFormat/CMakeLists.txt
@@ -5,6 +5,7 @@ set(LLVM_LINK_COMPONENTS
 
 add_llvm_unittest(BinaryFormatTests
   DwarfTest.cpp
+  ELFTest.cpp
   MachOTest.cpp
   MsgPackDocumentTest.cpp
   MsgPackReaderTest.cpp
diff --git a/llvm/unittests/BinaryFormat/ELFTest.cpp b/llvm/unittests/BinaryFormat/ELFTest.cpp
new file mode 100644
index 00000000000000..29793bf99c6000
--- /dev/null
+++ b/llvm/unittests/BinaryFormat/ELFTest.cpp
@@ -0,0 +1,32 @@
+//===- ELFTest.cpp ----------------------------------------------*- 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
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/BinaryFormat/ELF.h"
+#include "gtest/gtest.h"
+
+using namespace llvm;
+using namespace llvm::ELF;
+
+namespace {
+TEST(ELFTest, OSAbi) {
+  EXPECT_EQ(ELFOSABI_GNU, convertOSToOSAbi("gnu"));
+  EXPECT_EQ(ELFOSABI_FREEBSD, convertOSToOSAbi("freebsd"));
+  EXPECT_EQ(ELFOSABI_STANDALONE, convertOSToOSAbi("standalone"));
+  EXPECT_EQ(ELFOSABI_NONE, convertOSToOSAbi("none"));
+  // Test unrecognized strings.
+  EXPECT_EQ(ELFOSABI_NONE, convertOSToOSAbi(""));
+  EXPECT_EQ(ELFOSABI_NONE, convertOSToOSAbi("linux"));
+
+  EXPECT_EQ("gnu", convertOSAbiToOS(ELFOSABI_GNU));
+  EXPECT_EQ("freebsd", convertOSAbiToOS(ELFOSABI_FREEBSD));
+  EXPECT_EQ("standalone", convertOSAbiToOS(ELFOSABI_STANDALONE));
+  EXPECT_EQ("none", convertOSAbiToOS(ELFOSABI_NONE));
+  // Test unrecognized values.
+  EXPECT_EQ("none", convertOSAbiToOS(0xfe));
+}
+} // namespace

[jh7370]

Couple of nits, but otherwise LGTM.

[jh7370]

Nit: whilst changing the signature, now might be a good time to fix the casing of Abi to ABI, since it's an acronym and therefore should be all uppercase. Same below, including the parameter.

[jh7370]

Nit: I believe the C++ marker only needs to be in headers.

[MaskRay]

-*- C++ -*- is indeed a rare exception, so I am removing the marker.

% rg 'C\+\+ \*-=' -l -g '*.cpp'
llvm/tools/llvm-pdbutil/PrettyFunctionDumper.cpp
llvm/tools/llvm-pdbutil/PrettyBuiltinDumper.cpp
llvm/tools/llvm-pdbutil/PrettyTypeDumper.cpp
llvm/tools/llvm-pdbutil/PrettyExternalSymbolDumper.cpp
llvm/tools/llvm-pdbutil/PrettyCompilandDumper.cpp
llvm/tools/llvm-pdbutil/PrettyTypedefDumper.cpp
llvm/lib/CodeGen/GlobalISel/GISelKnownBits.cpp
llvm/lib/MC/SPIRVObjectWriter.cpp