[ClangOffloadBundler] make hipv4 and hip compatible (#91637)

yxsamliu · web-flow · commit ca3917538de1 · 2024-05-09T17:14:43.000-04:00
The distinction between the hip and hipv4 offload kinds is historically
based. Originally, these designations might have indicated different
versions of the code object ABI (Application Binary Interface). However,
as the system has evolved, the ABI version is now embedded directly
within the code object itself, making these historical distinctions
irrelevant during the unbundling process. Consequently, hip and hipv4
are treated as compatible in current implementations, facilitating
interchangeable handling of code objects without differentiation based
on offload kind. This change streamlines code management within the
ecosystem.
diff --git a/clang/docs/ClangOffloadBundler.rst b/clang/docs/ClangOffloadBundler.rst
@@ -245,7 +245,7 @@ Where:
                     object as a data section with the name ``.hip_fatbin``.
 
       hipv4         Offload code object for the HIP language. Used for AMD GPU
-                    code objects with at least ABI version V4 when the
+                    code objects with at least ABI version V4 and above when the
                     ``clang-offload-bundler`` is used to create a *fat binary*
                     to be loaded by the HIP runtime. The fat binary can be
                     loaded directly from a file, or be embedded in the host code
@@ -254,6 +254,14 @@ Where:
       openmp        Offload code object for the OpenMP language extension.
       ============= ==============================================================
 
+Note: The distinction between the `hip` and `hipv4` offload kinds is historically based.
+Originally, these designations might have indicated different versions of the
+code object ABI. However, as the system has evolved, the ABI version is now embedded
+directly within the code object itself, making these historical distinctions irrelevant
+during the unbundling process. Consequently, `hip` and `hipv4` are treated as compatible
+in current implementations, facilitating interchangeable handling of code objects
+without differentiation based on offload kind.
+
 **target-triple**
     The target triple of the code object. See `Target Triple
     <https://clang.llvm.org/docs/CrossCompilation.html#target-triple>`_.
@@ -295,7 +303,7 @@ Compatibility Rules for Bundle Entry ID
   A code object, specified using its Bundle Entry ID, can be loaded and
   executed on a target processor, if:
 
-  * Their offload kinds are the same.
+  * Their offload kinds are the same or comptible.
   * Their target triples are compatible.
   * Their Target IDs are compatible as defined in :ref:`compatibility-target-id`.
 
diff --git a/clang/lib/Driver/OffloadBundler.cpp b/clang/lib/Driver/OffloadBundler.cpp
@@ -113,8 +113,11 @@ bool OffloadTargetInfo::isOffloadKindValid() const {
 
 bool OffloadTargetInfo::isOffloadKindCompatible(
     const StringRef TargetOffloadKind) const {
-  if (OffloadKind == TargetOffloadKind)
+  if ((OffloadKind == TargetOffloadKind) ||
+      (OffloadKind == "hip" && TargetOffloadKind == "hipv4") ||
+      (OffloadKind == "hipv4" && TargetOffloadKind == "hip"))
     return true;
+
   if (BundlerConfig.HipOpenmpCompatible) {
     bool HIPCompatibleWithOpenMP = OffloadKind.starts_with_insensitive("hip") &&
                                    TargetOffloadKind == "openmp";
diff --git a/clang/test/Driver/clang-offload-bundler.c b/clang/test/Driver/clang-offload-bundler.c
@@ -505,6 +505,17 @@
 // RUN:   -output=%t.res.tgt1 -input=%t.hip.bundle.bc -unbundle 2>&1 | FileCheck %s -check-prefix=NOGFX906
 // NOGFX906: error: Can't find bundles for hip-amdgcn-amd-amdhsa--gfx906
 
+//
+// Check hip and hipv4 are compatible as offload kind.
+//
+// RUN: clang-offload-bundler -type=o -targets=hip-amdgcn-amd-amdhsa--gfx90a -input=%t.tgt1 -output=%t.bundle3.o
+// RUN: clang-offload-bundler -type=o -targets=hipv4-amdgcn-amd-amdhsa--gfx90a:sramecc-:xnack+ -output=%t.res.tgt1 -input=%t.bundle3.o -unbundle
+// RUN: diff %t.tgt1 %t.res.tgt1
+
+// RUN: clang-offload-bundler -type=o -targets=hipv4-amdgcn-amd-amdhsa--gfx90a -input=%t.tgt1 -output=%t.bundle3.o
+// RUN: clang-offload-bundler -type=o -targets=hip-amdgcn-amd-amdhsa--gfx90a:sramecc-:xnack+ -output=%t.res.tgt1 -input=%t.bundle3.o -unbundle
+// RUN: diff %t.tgt1 %t.res.tgt1
+
 //
 // Check archive unbundling
 //
diff --git a/clang/test/Driver/linker-wrapper.c b/clang/test/Driver/linker-wrapper.c
@@ -120,7 +120,7 @@ __attribute__((visibility("protected"), used)) int x;
 
 // HIP: clang{{.*}} -o [[IMG_GFX908:.+]] --target=amdgcn-amd-amdhsa -mcpu=gfx908
 // HIP: clang{{.*}} -o [[IMG_GFX90A:.+]] --target=amdgcn-amd-amdhsa -mcpu=gfx90a
-// HIP: clang-offload-bundler{{.*}}-type=o -bundle-align=4096 -compress -compression-level=6 -targets=host-x86_64-unknown-linux,hipv4-amdgcn-amd-amdhsa--gfx90a,hipv4-amdgcn-amd-amdhsa--gfx908 -input=/dev/null -input=[[IMG_GFX90A]] -input=[[IMG_GFX908]] -output={{.*}}.hipfb
+// HIP: clang-offload-bundler{{.*}}-type=o -bundle-align=4096 -compress -compression-level=6 -targets=host-x86_64-unknown-linux,hip-amdgcn-amd-amdhsa--gfx90a,hip-amdgcn-amd-amdhsa--gfx908 -input=/dev/null -input=[[IMG_GFX90A]] -input=[[IMG_GFX908]] -output={{.*}}.hipfb
 
 // RUN: clang-offload-packager -o %t.out \
 // RUN:   --image=file=%t.elf.o,kind=openmp,triple=amdgcn-amd-amdhsa,arch=gfx908 \
@@ -210,7 +210,7 @@ __attribute__((visibility("protected"), used)) int x;
 // RUN:   %t.o -o a.out 2>&1 | FileCheck %s --check-prefix=RELOCATABLE-LINK-HIP
 
 // RELOCATABLE-LINK-HIP: clang{{.*}} -o {{.*}}.img --target=amdgcn-amd-amdhsa
-// RELOCATABLE-LINK-HIP: clang-offload-bundler{{.*}} -type=o -bundle-align=4096 -targets=host-x86_64-unknown-linux,hipv4-amdgcn-amd-amdhsa--gfx90a -input=/dev/null -input={{.*}} -output={{.*}}
+// RELOCATABLE-LINK-HIP: clang-offload-bundler{{.*}} -type=o -bundle-align=4096 -targets=host-x86_64-unknown-linux,hip-amdgcn-amd-amdhsa--gfx90a -input=/dev/null -input={{.*}} -output={{.*}}
 // RELOCATABLE-LINK-HIP: /usr/bin/ld.lld{{.*}}-r
 // RELOCATABLE-LINK-HIP: llvm-objcopy{{.*}}a.out --remove-section .llvm.offloading
 
diff --git a/clang/tools/clang-linker-wrapper/ClangLinkerWrapper.cpp b/clang/tools/clang-linker-wrapper/ClangLinkerWrapper.cpp
@@ -413,7 +413,7 @@ fatbinary(ArrayRef<std::pair<StringRef, StringRef>> InputFiles,
 
   SmallVector<StringRef> Targets = {"-targets=host-x86_64-unknown-linux"};
   for (const auto &[File, Arch] : InputFiles)
-    Targets.push_back(Saver.save("hipv4-amdgcn-amd-amdhsa--" + Arch));
+    Targets.push_back(Saver.save("hip-amdgcn-amd-amdhsa--" + Arch));
   CmdArgs.push_back(Saver.save(llvm::join(Targets, ",")));
 
 #ifdef _WIN32
