[Pass Builder] Allow Module Inliner for full lto

[IBricchi]

Summary

Currently there is no way to enable the module inliner
when linking with full lto. This patch enables that
option.

[aeubanks]

llvm/lib/Passes/PassBuilderPipelines.cpp

1713

formatting is wrong, please run git clang-format

[IBricchi]

Sorry about that, should be correct now

[aeubanks]

taking a closer look, this doesn't make sense. the LTO pipeline doesn't run the function simplification pipeline in ModuleInlinerWrapperPass, but buildModuleInlinerPipeline will run the function simplification pipeline. maybe you just want ModuleInlinerPass?

[IBricchi]

Oh I see. I was trying to copy over the changes from how it was done in the thinLTO pipeline, but you're right, I only really want the ModuleInlinerPass. I've updated the patch to reflect that.

[IBricchi]

ping

[IBricchi]

Thanks!

I don't have commit access so would you mind committing it for me, my username is ibricchi and email is [e-mail].

llvm/lib/Passes/PassBuilderPipelines.cpp

PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
   MPM.addPass(createModuleToFunctionPassAdaptor(std::move(PeepholeFPM),
                                                 PTO.EagerlyInvalidateAnalyses));
 
   // Note: historically, the PruneEH pass was run first to deduce nounwind and
   // generally clean up exception handling overhead. It isn't clear this is
   // valuable as the inliner doesn't currently care whether it is inlining an
   // invoke or a call.
   // Run the inliner now.
+  if (EnableModuleInliner) {
+    MPM.addPass(ModuleInlinerPass(getInlineParamsFromOptLevel(Level),
+                                  UseInlineAdvisor,
+                                  ThinOrFullLTOPhase::FullLTOPostLink));
+  } else {
     MPM.addPass(ModuleInlinerWrapperPass(
         getInlineParamsFromOptLevel(Level),
         /* MandatoryFirst */ true,
         InlineContext{ThinOrFullLTOPhase::FullLTOPostLink,
                       InlinePass::CGSCCInliner}));
+  }
 
   // Perform context disambiguation after inlining, since that would reduce the
   // amount of additional cloning required to distinguish the allocation
   // contexts.
   if (EnableMemProfContextDisambiguation)
     MPM.addPass(MemProfContextDisambiguation());
 
   // Optimize globals again after we ran the inliner.