Move accurate-sample-profile into the function attribute.

[danielcdh]

Summary

We need to have accurate-sample-profile in function attribute so that it works with LTO.

[davidxl]

lib/Analysis/ProfileSummaryInfo.cpp

42

I don't see the harm of keeping the internal option too?

[davidxl]

How can we test it works for ThinLTO? some end-to-end test is also desired (given that this becomes an external feature).

[danielcdh]

In D37113#851728, @davidxl wrote:

How can we test it works for ThinLTO? some end-to-end test is also desired (given that this becomes an external feature).

Added an end-to-end test in https://reviews.llvm.org/D37091

[danielcdh]

update

[davidxl]

lgtm

include/llvm/IR/Attributes.td

 def ZExt : EnumAttr<"zeroext">;
 
 /// Target-independent string attributes.
 def LessPreciseFPMAD : StrBoolAttr<"less-precise-fpmad">;
 def NoInfsFPMath : StrBoolAttr<"no-infs-fp-math">;
 def NoNansFPMath : StrBoolAttr<"no-nans-fp-math">;
 def UnsafeFPMath : StrBoolAttr<"unsafe-fp-math">;
 def NoJumpTables : StrBoolAttr<"no-jump-tables">;
+def ProfileSampleAccurate : StrBoolAttr<"profile-sample-accurate">;
 
 class CompatRule<string F> {
   // The name of the function called to check the attribute of the caller and
   // callee and decide whether inlining should be allowed. The function's
   // signature must match "bool(const Function&, const Function &)", where the
   // first parameter is the reference to the caller and the second parameter is
   // the reference to the callee. It must return false if the attributes of the
   // caller and callee are incompatible, and true otherwise.

 }
 
 def : MergeRule<"setAND<LessPreciseFPMADAttr>">;
 def : MergeRule<"setAND<NoInfsFPMathAttr>">;
 def : MergeRule<"setAND<NoNansFPMathAttr>">;
 def : MergeRule<"setAND<UnsafeFPMathAttr>">;
 def : MergeRule<"setOR<NoImplicitFloatAttr>">;
 def : MergeRule<"setOR<NoJumpTablesAttr>">;
+def : MergeRule<"setOR<ProfileSampleAccurateAttr>">;
 def : MergeRule<"adjustCallerSSPLevel">;
 def : MergeRule<"adjustCallerStackProbes">;
 def : MergeRule<"adjustCallerStackProbeSize">;

lib/Analysis/ProfileSummaryInfo.cpp

static cl::opt<int> ProfileSummaryCutoffHot(
     cl::desc("A count is hot if it exceeds the minimum count to"
              " reach this percentile of total counts."));
 
 static cl::opt<int> ProfileSummaryCutoffCold(
     "profile-summary-cutoff-cold", cl::Hidden, cl::init(999999), cl::ZeroOrMore,
     cl::desc("A count is cold if it is below the minimum count"
              " to reach this percentile of total counts."));
 
-static cl::opt<bool> AccurateSampleProfile(
-    "accurate-sample-profile", cl::Hidden, cl::init(false),
+static cl::opt<bool> ProfileSampleAccurate(
+    "profile-sample-accurate", cl::Hidden, cl::init(false),
     cl::desc("If the sample profile is accurate, we will mark all un-sampled "
              "callsite as cold. Otherwise, treat un-sampled callsites as if "
              "we have no profile."));
 static cl::opt<unsigned> ProfileSummaryHugeWorkingSetSizeThreshold(
     "profile-summary-huge-working-set-size-threshold", cl::Hidden,
     cl::init(15000), cl::ZeroOrMore,
     cl::desc("The code working set size is considered huge if the number of"
              " blocks required to reach the -profile-summary-cutoff-hot"
bool ProfileSummaryInfo::isColdCallSite(const CallSite &CS,
   if (C)
     return isColdCount(*C);
 
   // In SamplePGO, if the caller has been sampled, and there is no profile
   // annotatedon the callsite, we consider the callsite as cold.
   // If there is no profile for the caller, and we know the profile is
   // accurate, we consider the callsite as cold.
   return (hasSampleProfile() &&
-          (CS.getCaller()->getEntryCount() || AccurateSampleProfile));
+          (CS.getCaller()->getEntryCount() || ProfileSampleAccurate ||
+           CS.getCaller()->hasFnAttribute("profile-sample-accurate")));
 }
 
 INITIALIZE_PASS(ProfileSummaryInfoWrapperPass, "profile-summary-info",
                 "Profile summary info", false, true)
 
 ProfileSummaryInfoWrapperPass::ProfileSummaryInfoWrapperPass()
     : ImmutablePass(ID) {
   initializeProfileSummaryInfoWrapperPassPass(*PassRegistry::getPassRegistry());

test/Transforms/Inline/inline-cold-callsite-samplepgo.ll

-; For SamplePGO, if -accurate-sample-profile is specified, cold callsite
+; For SamplePGO, if -profile-sample-accurate is specified, cold callsite
 ; heuristics should be honored if the caller has no profile.
 
 ; RUN: opt < %s -inline -S -inline-cold-callsite-threshold=0 | FileCheck %s
-; RUN: opt < %s -inline -S -inline-cold-callsite-threshold=0 -accurate-sample-profile | FileCheck %s --check-prefix=ACCURATE
 
 define i32 @callee(i32 %x) {
   %x1 = add i32 %x, 1
   %x2 = add i32 %x1, 1
   %x3 = add i32 %x2, 1
   call void @extern()
   call void @extern()
   ret i32 %x3
 }
 
 define i32 @caller(i32 %y1) {
+; CHECK-LABEL: @caller
 ; CHECK-NOT: call i32 @callee
-; ACCURATE: call i32 @callee
+  %y2 = call i32 @callee(i32 %y1)
+  ret i32 %y2
+}
+
+define i32 @caller_accurate(i32 %y1) #0 {
+; CHECK-LABEL: @caller_accurate
+; CHECK: call i32 @callee
   %y2 = call i32 @callee(i32 %y1)
   ret i32 %y2
 }
 
 declare void @extern()
 
+attributes #0 = { "profile-sample-accurate" }
+
 !llvm.module.flags = !{!1}
 !1 = !{i32 1, !"ProfileSummary", !2}
 !2 = !{!3, !4, !5, !6, !7, !8, !9, !10}
 !3 = !{!"ProfileFormat", !"SampleProfile"}
 !4 = !{!"TotalCount", i64 10000}
 !5 = !{!"MaxCount", i64 1000}
 !6 = !{!"MaxInternalCount", i64 1}
 !7 = !{!"MaxFunctionCount", i64 1000}
 !8 = !{!"NumCounts", i64 3}
 !9 = !{!"NumFunctions", i64 3}
 !10 = !{!"DetailedSummary", !11}
 !11 = !{!12, !13, !14}
 !12 = !{i32 10000, i64 100, i32 1}
 !13 = !{i32 999000, i64 100, i32 1}
 !14 = !{i32 999999, i64 1, i32 2}