Revert "[AIX] Avoid structor alias; die before bad alias codegen"

[Jake-Egan]

Summary

Avoiding structor alias is no longer needed because AIX now has an alias implementation here: https://reviews.llvm.org/D83252.

This reverts commit b116ded57da3530e661f871f4191c59cd9e091cd.

[jasonliu]

clang/lib/Driver/ToolChains/Clang.cpp

5031–5032

Format as clang-format suggest. You could run clang-format or git-clang-format just for your commits by the way.

clang/test/Driver/aix-constructor-alias.c

3

Instead of removing this file, we want to check if -mconstructor-aliases is passed in for AIX.
Check if there are other existing files doing it for other target, then you just need to add the check for AIX target in other files.

llvm/test/CodeGen/PowerPC/aix-alias.ll

4 ↗

(On Diff #346270)

I don't think we want to remove the testing of our alias implementation.

[stevewan]

Might be a dumb question, but I see this is a revert to rGb116ded, do we also want to revert the changes added to llvm/lib/Target/PowerPC/PPCAsmPrinter.cpp?

clang/lib/Driver/ToolChains/Clang.cpp

5031–5032

FWIW, if clang-format is installed on your host (you can do this via homebrew on mac IIRC), arc would use it as the linter when you do arc lint or arc diff.

[Jake-Egan]

In D102724#2768663, @stevewan wrote:

Might be a dumb question, but I see this is a revert to rGb116ded, do we also want to revert the changes added to llvm/lib/Target/PowerPC/PPCAsmPrinter.cpp?

The changes to that file seems to be already reverted by: https://reviews.llvm.org/D79127

[Jake-Egan]

Added the changes jasonliu asked for.

[jasonliu]

LGTM.

[jsji]

because AIX now has an alias implementation.

Can we please include the patches or commits that implement the alias in description?

[jasonliu]

Good idea, fyi, this is the patch: https://reviews.llvm.org/D83252

[jsji]

In D102724#2769756, @jasonliu wrote:

Good idea, fyi, this is the patch: https://reviews.llvm.org/D83252

Thanks.

clang/lib/Driver/ToolChains/Clang.cpp

#endif
   }
   if (Args.hasArg(options::OPT_fdebug_pass_arguments)) {
     CmdArgs.push_back("-mdebug-pass");
     CmdArgs.push_back("Arguments");
   }
 
   // Enable -mconstructor-aliases except on darwin, where we have to work around
   // a linker bug (see <rdar://problem/7651567>), and CUDA/AMDGPU device code,
-  // where aliases aren't supported. Similarly, aliases aren't yet supported
-  // for AIX.
-  if (!RawTriple.isOSDarwin() && !RawTriple.isNVPTX() &&
-      !RawTriple.isAMDGPU() && !RawTriple.isOSAIX())
+  // where aliases aren't supported.
+  if (!RawTriple.isOSDarwin() && !RawTriple.isNVPTX() && !RawTriple.isAMDGPU())
     CmdArgs.push_back("-mconstructor-aliases");
 
   // Darwin's kernel doesn't support guard variables; just die if we
   // try to use them.
   if (KernelOrKext && RawTriple.isOSDarwin())
     CmdArgs.push_back("-fforbid-guard-variables");
 
   if (Args.hasFlag(options::OPT_mms_bitfields, options::OPT_mno_ms_bitfields,

clang/test/Driver/aix-constructor-alias.c

-// Check that we don't pass -mconstructor-aliases when compiling for AIX.
+// Check that we pass -mconstructor-aliases when compiling for AIX.
 
 // RUN: %clang -### -target powerpc-ibm-aix7.1.0.0 %s -c -o %t.o 2>&1 \
 // RUN:   | FileCheck %s
 // RUN: %clang -### -target powerpc64-ibm-aix7.1.0.0 %s -c -o %t.o 2>&1 \
 // RUN:   | FileCheck %s
-// CHECK-NOT: "-mconstructor-aliases"
+// CHECK: "-mconstructor-aliases"