[NewPM/Inliner] Customize threshold based on optlevel and sizelevel

[eraman]

Summary

(Note: It will be cleaner to expose the OptimizationLevel enum to InlineCost and Inliner instead of mapping back from the enum to SizeLevel/OptLevel, but I don't know what is the right header file to move the enum into)

[eraman]

Ping.

[chandlerc]

lib/Passes/PassBuilder.cpp

558 ↗

(On Diff #102915)

As discussed in person, I would just call 'getInlineParams' here and use the existing constructor of the pass object.

[eraman]

Call getInlineParams in PassBuilder and pass the Params object to the existing Inliner constructor.

[chandlerc]

Minor nitpick below. Feel free to submit.

lib/Passes/PassBuilder.cpp

872–873 ↗

(On Diff #104333)

Noting that this is duplicated, maybe a static helper function:

static InlineParams getInlineParamsFromOptLevel(...)

And have it do the call to getInlineParams?

[eraman]

Address Chandler's comments.

llvm/trunk/lib/Passes/PassBuilder.cpp

if (RunProfileGen) {
     Options.DoCounterPromotion = true;
     MPM.addPass(InstrProfiling(Options));
   }
 
   if (!ProfileUseFile.empty())
     MPM.addPass(PGOInstrumentationUse(ProfileUseFile));
 }
 
+static InlineParams
+getInlineParamsFromOptLevel(PassBuilder::OptimizationLevel Level) {
+  auto O3 = PassBuilder::O3;
+  unsigned OptLevel = Level > O3 ? 2 : Level;
+  unsigned SizeLevel = Level > O3 ? Level - O3 : 0;
+  return getInlineParams(OptLevel, SizeLevel);
+}
+
 ModulePassManager
 PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
                                                bool DebugLogging) {
   ModulePassManager MPM(DebugLogging);
 
   // Do basic inference of function attributes from known properties of system
   // libraries and other oracles.
   MPM.addPass(InferFunctionAttrsPass());
PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
   // 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 first. The theory is that we are walking bottom-up and so
   // the callees have already been fully optimized, and we want to inline them
   // into the callers so that our optimizations can reflect that.
-  // FIXME; Customize the threshold based on optimization level.
-  MainCGPipeline.addPass(InlinerPass());
+  MainCGPipeline.addPass(InlinerPass(getInlineParamsFromOptLevel(Level)));
 
   // Now deduce any function attributes based in the current code.
   MainCGPipeline.addPass(PostOrderFunctionAttrsPass());
 
   // When at O3 add argument promotion to the pass pipeline.
   // FIXME: It isn't at all clear why this should be limited to O3.
   if (Level == O3)
     MainCGPipeline.addPass(ArgumentPromotionPass());
ModulePassManager PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
   // FIXME: add peephole extensions here as the legacy PM does.
   MPM.addPass(createModuleToFunctionPassAdaptor(InstCombinePass()));
 
   // 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.
-  MPM.addPass(createModuleToPostOrderCGSCCPassAdaptor(InlinerPass()));
+  MPM.addPass(createModuleToPostOrderCGSCCPassAdaptor(
+      InlinerPass(getInlineParamsFromOptLevel(Level))));
 
   // Optimize globals again after we ran the inliner.
   MPM.addPass(GlobalOptPass());
 
   // Garbage collect dead functions.
   // FIXME: Add ArgumentPromotion pass after once it's ported.
   MPM.addPass(GlobalDCEPass());