[llvm][NFC] Add an explicit 'ComputeFullInlineCost' API

[mtrofin]

Summary

Added getInliningCostEstimate, which is essentially what getInlineCost
computes if passed default inlining params, and non-null ORE or
InlineParams::ComputeFullInlineCost.

[davidxl]

Why not passing the 'computeFullCost' as an additional parameter to InlineCostAnalyzer?

llvm/lib/Analysis/InlineCost.cpp

719

should it define 'shouldStop()' to return false?

[mtrofin]

feedback

[mtrofin]

In D78730#2002357, @davidxl wrote:

Why not passing the 'computeFullCost' as an additional parameter to InlineCostAnalyzer?

Because InlineCostAnalyzer's finalizeAnalysis looks at threshold (see lines 633-635).

Clarified documentation of the new API.

[mtrofin]

comments

[davidxl]

Perhaps add a 'ignoreThreshold' state in InlineCostANalyzer then? IgnoreThreshold implies 'ComputeFullCost' and will be honored by the finalize Method.

[mtrofin]

simplified

[mtrofin]

In D78730#2002590, @davidxl wrote:

Perhaps add a 'ignoreThreshold' state in InlineCostANalyzer then? IgnoreThreshold implies 'ComputeFullCost' and will be honored by the finalize Method.

That's simpler, indeed. Done.

[davidxl]

looks good but I think the shouldStop method should check IgnoreThreshold as well.

[mtrofin]

feedback

[davidxl]

lgtm

llvm/include/llvm/Analysis/InlineCost.h

 /// InlineResult::failure() if the inlining may never happen because of user
 /// directives or incompatibilities detectable without needing callee traversal.
 /// Otherwise returns None, meaning that inlining should be decided based on
 /// other criteria (e.g. cost modeling).
 Optional<InlineResult> getAttributeBasedInliningDecision(
     CallBase &Call, Function *Callee, TargetTransformInfo &CalleeTTI,
     function_ref<const TargetLibraryInfo &(Function &)> GetTLI);
 
+/// Get the cost estimate ignoring thresholds. This is similar to getInlineCost
+/// when passed InlineParams::ComputeFullInlineCost, or a non-null ORE. It
+/// uses default InlineParams otherwise.
+/// Contrary to getInlineCost, which makes a threshold-based final evaluation of
+/// should/shouldn't inline, captured in InlineResult, getInliningCostEstimate
+/// returns:
+/// - None, if the inlining cannot happen (is illegal)
+/// - an integer, representing the cost.
+Optional<int> getInliningCostEstimate(
+    CallBase &Call, TargetTransformInfo &CalleeTTI,
+    std::function<AssumptionCache &(Function &)> &GetAssumptionCache,
+    Optional<function_ref<BlockFrequencyInfo &(Function &)>> GetBFI,
+    ProfileSummaryInfo *PSI, OptimizationRemarkEmitter *ORE);
+
 /// Minimal filter to detect invalid constructs for inlining.
 InlineResult isInlineViable(Function &Callee);
 } // namespace llvm
 
 #endif

llvm/lib/Analysis/InlineCost.cpp

class InlineCostCallAnalyzer final : public CallAnalyzer {
 
   /// Upper bound for the inlining cost. Bonuses are being applied to account
   /// for speculative "expected profit" of the inlining decision.
   int Threshold = 0;
 
   /// Attempt to evaluate indirect calls to boost its inline cost.
   const bool BoostIndirectCalls;
 
+  /// Ignore the threshold when finalizing analysis.
+  const bool IgnoreThreshold;
+
   /// Inlining cost measured in abstract units, accounts for all the
   /// instructions expected to be executed for a given function invocation.
   /// Instructions that are statically proven to be dead based on call-site
   /// arguments are not counted here.
   int Cost = 0;
 
   bool SingleBB = true;
 
InlineResult finalizeAnalysis() override {
     // We applied the maximum possible vector bonus at the beginning. Now,
     // subtract the excess bonus, if any, from the Threshold before
     // comparing against Cost.
     if (NumVectorInstructions <= NumInstructions / 10)
       Threshold -= VectorBonus;
     else if (NumVectorInstructions <= NumInstructions / 2)
       Threshold -= VectorBonus / 2;
 
-    if (Cost < std::max(1, Threshold))
+    if (IgnoreThreshold || Cost < std::max(1, Threshold))
       return InlineResult::success();
     return InlineResult::failure("Cost over threshold.");
   }
   bool shouldStop() override {
     // Bail out the moment we cross the threshold. This means we'll under-count
     // the cost, but only when undercounting doesn't matter.
-    return Cost >= Threshold && !ComputeFullInlineCost;
+    return !IgnoreThreshold && Cost >= Threshold && !ComputeFullInlineCost;
   }
 
   void onLoadEliminationOpportunity() override {
     LoadEliminationCost += InlineConstants::InstrCost;
   }
 
   InlineResult onAnalysisStart() override {
     // Perform some tweaks to the cost and threshold based on the direct
class InlineCostCallAnalyzer final : public CallAnalyzer {
   }
 
 public:
   InlineCostCallAnalyzer(
       const TargetTransformInfo &TTI,
       std::function<AssumptionCache &(Function &)> &GetAssumptionCache,
       Optional<function_ref<BlockFrequencyInfo &(Function &)>> &GetBFI,
       ProfileSummaryInfo *PSI, OptimizationRemarkEmitter *ORE, Function &Callee,
-      CallBase &Call, const InlineParams &Params, bool BoostIndirect = true)
+      CallBase &Call, const InlineParams &Params, bool BoostIndirect = true,
+      bool IgnoreThreshold = false)
       : CallAnalyzer(TTI, GetAssumptionCache, GetBFI, PSI, ORE, Callee, Call),
         ComputeFullInlineCost(OptComputeFullInlineCost ||
                               Params.ComputeFullInlineCost || ORE),
         Params(Params), Threshold(Params.DefaultThreshold),
-        BoostIndirectCalls(BoostIndirect) {}
+        BoostIndirectCalls(BoostIndirect), IgnoreThreshold(IgnoreThreshold) {}
 
   /// Annotation Writer for cost annotation
   CostAnnotationWriter Writer;
 
   void dump();
 
   virtual ~InlineCostCallAnalyzer() {}
   int getThreshold() { return Threshold; }
   int getCost() { return Cost; }
 };
 } // namespace
 
 /// Test whether the given value is an Alloca-derived function argument.
 bool CallAnalyzer::isAllocaDerivedArg(Value *V) {
   return SROAArgValues.count(V);
 }
 
 void CallAnalyzer::disableSROAForArg(AllocaInst *SROAArg) {
   onDisableSROA(SROAArg);
   EnabledSROAAllocas.erase(SROAArg);
   disableLoadElimination();
InlineCost llvm::getInlineCost(
     std::function<AssumptionCache &(Function &)> &GetAssumptionCache,
     Optional<function_ref<BlockFrequencyInfo &(Function &)>> GetBFI,
     function_ref<const TargetLibraryInfo &(Function &)> GetTLI,
     ProfileSummaryInfo *PSI, OptimizationRemarkEmitter *ORE) {
   return getInlineCost(Call, Call.getCalledFunction(), Params, CalleeTTI,
                        GetAssumptionCache, GetBFI, GetTLI, PSI, ORE);
 }
 
+Optional<int> getInliningCostEstimate(
+    CallBase &Call, TargetTransformInfo &CalleeTTI,
+    std::function<AssumptionCache &(Function &)> &GetAssumptionCache,
+    Optional<function_ref<BlockFrequencyInfo &(Function &)>> GetBFI,
+    ProfileSummaryInfo *PSI, OptimizationRemarkEmitter *ORE) {
+  const InlineParams Params = {/* DefaultThreshold*/ 0,
+                               /*HintThreshold*/ {},
+                               /*ColdThreshold*/ {},
+                               /*OptSizeThreshold*/ {},
+                               /*OptMinSizeThreshold*/ {},
+                               /*HotCallSiteThreshold*/ {},
+                               /*LocallyHotCallSiteThreshold*/ {},
+                               /*ColdCallSiteThreshold*/ {},
+                               /* ComputeFullInlineCost*/ true};
+
+  InlineCostCallAnalyzer CA(CalleeTTI, GetAssumptionCache, GetBFI, PSI, ORE,
+                            *Call.getCalledFunction(), Call, Params, true,
+                            /*IgnoreThreshold*/ true);
+  auto R = CA.analyze();
+  if (!R.isSuccess())
+    return None;
+  return CA.getCost();
+}
+
 Optional<InlineResult> llvm::getAttributeBasedInliningDecision(
     CallBase &Call, Function *Callee, TargetTransformInfo &CalleeTTI,
     function_ref<const TargetLibraryInfo &(Function &)> GetTLI) {
 
   // Cannot inline indirect calls.
   if (!Callee)
     return InlineResult::failure("indirect call");