[PATCH] D16998: [AMDGPU] llvm-objdump: disassembling amdgcn object file

[Michael Spencer via llvm-commits]

Bigcheese added a subscriber: Bigcheese.
Bigcheese requested changes to this revision.
Bigcheese added a reviewer: Bigcheese.
This revision now requires changes to proceed.

================
Comment at: tools/llvm-objdump/llvm-objdump.cpp:395
@@ +394,3 @@
+    OS << left_justify(IS.str(), 60) << format("// %012X: ", Address);
+    for (auto D : makeArrayRef(reinterpret_cast<const uint32_t*>(Bytes.data()),
+                               Bytes.size() / sizeof(uint32_t)))
----------------
This should be using a endian specific type. Either llvm::support::ulittle32_t or ubig32_t.

================
Comment at: tools/llvm-objdump/llvm-objdump.cpp:1077
@@ +1076,3 @@
+        while (End > Limit &&
+               *reinterpret_cast<const uint32_t*>(&Bytes[End - 4]) == 0)
+          End -= 4;
----------------
Same with the endianness.


http://reviews.llvm.org/D16998