[llvm] 8d3f112 - [M68k] Update pointer data layout

[Ricky Taylor via llvm-commits]

Author: Ricky Taylor
Date: 2021-08-27T11:47:27+01:00
New Revision: 8d3f112f0cdbed2311aead86bcd72e763ad55255

URL: https://github.com/llvm/llvm-project/commit/8d3f112f0cdbed2311aead86bcd72e763ad55255
DIFF: https://github.com/llvm/llvm-project/commit/8d3f112f0cdbed2311aead86bcd72e763ad55255.diff

LOG: [M68k] Update pointer data layout

Fixes PR51626.

The M68k requires that all instruction, word and long word reads are
aligned to word boundaries. From the 68020 onwards, there is a
performance benefit from aligning long words to long word boundaries.

The M68k uses the same data layout for pointers and integers.

In line with this, this commit updates the pointer data layout to
match the layout already set for 32-bit integers: 32:16:32.

Differential Revision: https://reviews.llvm.org/D108792

Added: 
    

Modified: 
    clang/lib/Basic/Targets/M68k.cpp
    llvm/lib/Target/M68k/M68kTargetMachine.cpp

Removed: 
    


################################################################################
diff  --git a/clang/lib/Basic/Targets/M68k.cpp b/clang/lib/Basic/Targets/M68k.cpp
index 31cb36d37636c..c0cd8fa90ed6b 100644
--- a/clang/lib/Basic/Targets/M68k.cpp
+++ b/clang/lib/Basic/Targets/M68k.cpp
@@ -37,8 +37,8 @@ M68kTargetInfo::M68kTargetInfo(const llvm::Triple &Triple,
   // FIXME how to wire it with the used object format?
   Layout += "-m:e";
 
-  // M68k pointers are always 32 bit wide even for 16 bit cpus
-  Layout += "-p:32:32";
+  // M68k pointers are always 32 bit wide even for 16-bit CPUs
+  Layout += "-p:32:16:32";
 
   // M68k integer data types
   Layout += "-i8:8:8-i16:16:16-i32:16:32";

diff  --git a/llvm/lib/Target/M68k/M68kTargetMachine.cpp b/llvm/lib/Target/M68k/M68kTargetMachine.cpp
index 5b8fd3d41b147..cb7d8f8b25e39 100644
--- a/llvm/lib/Target/M68k/M68kTargetMachine.cpp
+++ b/llvm/lib/Target/M68k/M68kTargetMachine.cpp
@@ -49,10 +49,14 @@ std::string computeDataLayout(const Triple &TT, StringRef CPU,
   // FIXME how to wire it with the used object format?
   Ret += "-m:e";
 
-  // M68k pointers are always 32 bit wide even for 16 bit cpus
-  Ret += "-p:32:32";
-
-  // M68k requires i8 to align on 2 byte boundry
+  // M68k pointers are always 32 bit wide even for 16-bit CPUs.
+  // The ABI only specifies 16-bit alignment.
+  // On at least the 68020+ with a 32-bit bus, there is a performance benefit
+  // to having 32-bit alignment.
+  Ret += "-p:32:16:32";
+
+  // Bytes do not require special alignment, words are word aligned and
+  // long words are word aligned at minimum.
   Ret += "-i8:8:8-i16:16:16-i32:16:32";
 
   // FIXME no floats at the moment