[CSSPGO] Turn off PseudoProbeUpdatePass for non-FDO builds.

[hoy]

Summary

PseudoProbeUpdatePass is used to distribute sample counts among dulplicated probes. It doesn't make sense for it to run without a sample profile. The pass takes 1% of the build time.

[wenlei]

lgtm.

llvm/lib/Passes/PassBuilderPipelines.cpp

1262–1263

linter might complain the length of the line.

[hoy]

llvm/lib/Passes/PassBuilderPipelines.cpp

1262–1263

Interestingly, linter is actually fine with this. The lint build job just passed. My local linting doesn't do anything about it. I guess the linting rule has changed?

[hoy]

Fixing linter issue.

[wlei]

LGTM, thanks!

llvm/lib/Passes/PassBuilderPipelines.cpp

PassBuilder::buildPerModuleDefaultPipeline(OptimizationLevel Level,
   // Add the core simplification pipeline.
   MPM.addPass(buildModuleSimplificationPipeline(
       Level, LTOPreLink ? ThinOrFullLTOPhase::FullLTOPreLink
                         : ThinOrFullLTOPhase::None));
 
   // Now add the optimization pipeline.
   MPM.addPass(buildModuleOptimizationPipeline(Level, LTOPreLink));
 
-  if (PGOOpt && PGOOpt->PseudoProbeForProfiling)
+  if (PGOOpt && PGOOpt->PseudoProbeForProfiling &&
+      PGOOpt->Action == PGOOptions::SampleUse)
     MPM.addPass(PseudoProbeUpdatePass());
 
   // Emit annotation remarks.
   addAnnotationRemarksPass(MPM);
 
   if (LTOPreLink)
     addRequiredLTOPreLinkPasses(MPM);
 
PassBuilder::buildThinLTOPreLinkDefaultPipeline(OptimizationLevel Level) {
   // Reduce the size of the IR as much as possible.
   MPM.addPass(GlobalOptPass());
 
   // Module simplification splits coroutines, but does not fully clean up
   // coroutine intrinsics. To ensure ThinLTO optimization passes don't trip up
   // on these, we schedule the cleanup here.
   MPM.addPass(createModuleToFunctionPassAdaptor(CoroCleanupPass()));
 
-  if (PGOOpt && PGOOpt->PseudoProbeForProfiling)
+  if (PGOOpt && PGOOpt->PseudoProbeForProfiling &&
+      PGOOpt->Action == PGOOptions::SampleUse)
     MPM.addPass(PseudoProbeUpdatePass());
 
   // Handle OptimizerLastEPCallbacks added by clang on PreLink. Actual
   // optimization is going to be done in PostLink stage, but clang can't
   // add callbacks there in case of in-process ThinLTO called by linker.
   for (auto &C : OptimizerLastEPCallbacks)
     C(MPM, Level);