Cache attribute checks

[nickwasmer]

Summary

As a followup to D99362 cache the checks that attributes belong to the same context. Looking for presence in a FoldingSet is slow.

[dexonsmith]

I triggered a run in the compile-time tracker for this patch:
http://llvm-compile-time-tracker.com/compare.php?from=a7712091ea7a2b4a7b5c4764af7545a7574b651b&to=f01acba0cdc4625738be504449a71704595cd12f&stat=instructions

Looks like an improvement, but I'm not how link time is exposed in that interface... or maybe it isn't? @nikic, you mentioned there was a 1% regression in link time: can you confirm if this fixes it?

Note I also triggered a run where attributes themselves are also cached:
http://llvm-compile-time-tracker.com/compare.php?from=f01acba0cdc4625738be504449a71704595cd12f&to=68b7232d1d68abb75486de8588094742c3146ba9&stat=instructions
Seems more noisy.

[dexonsmith]

In D100738#2699767, @dexonsmith wrote:

I triggered a run in the compile-time tracker for this patch:
http://llvm-compile-time-tracker.com/compare.php?from=a7712091ea7a2b4a7b5c4764af7545a7574b651b&to=f01acba0cdc4625738be504449a71704595cd12f&stat=instructions

Looks like an improvement, but I'm not how link time is exposed in that interface... or maybe it isn't? @nikic, you mentioned there was a 1% regression in link time: can you confirm if this fixes it?

Ah, I found it; the "per-file details" checkbox says it's .29% faster:

tramp3d-v4	213713M	213394M (-0.15%)
    CMakeFiles/tramp3d-v4.dir/tramp3d-v4.cpp.o	100598M	100613M (+0.01%)
    tramp3d-v4.link	113114M	112781M (-0.29%)

for LegacyPM-ReleaseLTO-g, with my follow-up that adds attribute caching totally in the noise:

tramp3d-v4	213394M	213318M (-0.04%)
    CMakeFiles/tramp3d-v4.dir/tramp3d-v4.cpp.o	100613M	100575M (-0.04%)
    tramp3d-v4.link	112781M	112744M (-0.03%)

This patch LGTM (certainly an improvement), but maybe this isn't quite enough. @nikic, thoughts?

[nikic]

I think this is good enough. Doesn't recover fully, but most of the way.

llvm/lib/IR/Verifier.cpp

class Verifier : public InstVisitor<Verifier>, VerifierSupport {
   MapVector<Instruction *, Instruction *> SiblingFuncletInfo;
 
   /// Cache of constants visited in search of ConstantExprs.
   SmallPtrSet<const Constant *, 32> ConstantExprVisited;
 
   /// Cache of declarations of the llvm.experimental.deoptimize.<ty> intrinsic.
   SmallVector<const Function *, 4> DeoptimizeDeclarations;
 
+  /// Cache of attribute lists verified.
+  SmallPtrSet<const void *, 32> AttributeListsVisited;
+
   // Verify that this GlobalValue is only used in this module.
   // This map is used to avoid visiting uses twice. We can arrive at a user
   // twice, if they have multiple operands. In particular for very large
   // constant expressions, we can arrive at a particular user many times.
   SmallPtrSet<const Value *, 32> GlobalValueVisited;
 
   // Keeps track of duplicate function argument debug info.
   SmallVector<const DILocalVariable *, 16> DebugFnArgs;

 
 // Check parameter attributes against a function type.
 // The value V is printed in error messages.
 void Verifier::verifyFunctionAttrs(FunctionType *FT, AttributeList Attrs,
                                    const Value *V, bool IsIntrinsic) {
   if (Attrs.isEmpty())
     return;
 
+  if (AttributeListsVisited.insert(Attrs.getRawPointer()).second) {
     Assert(Attrs.hasParentContext(Context),
            "Attribute list does not match Module context!", &Attrs);
     for (const auto &AttrSet : Attrs) {
       Assert(!AttrSet.hasAttributes() || AttrSet.hasParentContext(Context),
              "Attribute set does not match Module context!", &AttrSet);
       for (const auto &A : AttrSet) {
         Assert(A.hasParentContext(Context),
                "Attribute does not match Module context!", &A);
       }
     }
+  }
 
   bool SawNest = false;
   bool SawReturned = false;
   bool SawSRet = false;
   bool SawSwiftSelf = false;
   bool SawSwiftError = false;
 
   // Verify return value attributes.