[Inline] prevent inlining on noprofile mismatch

[nickdesaulniers]

Summary

Similar to
commit bc044a88ee3c ("[Inline] prevent inlining on stack protector mismatch")

The noprofile function attribute is meant to prevent compiler
instrumentation from being inserted into a function. Inlining may defeat
the developer's intent. If the caller and callee don't either BOTH have
the attribute or BOTH lack the attribute, suppress inline substitution.

This matches behavior being proposed in GCC:
https://gcc.gnu.org/pipermail/gcc-patches/2021-June/573511.html
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80223

Add LangRef entry for noprofile fn attr, similar to text added in D93422
and D104944.

[MaskRay]

LG.

The contract is that the annotated code region, even if inlined or cloned, does not get instrumentation.
I think suppressing inlining a NoProfileAttr callee into a caller without the attr is the only practical approach here,
because a fine-grained per-instruction attribute is obviously cumbersome and likely over-engineering.
(
For example, a smart inliner could still inline a noprofile callee, just filtering out the instrumentation instructions.
But this is definitely over-engineering.
)

Users (mainly the Linux kernel developers) should be aware that noprofile does not imply noinline.
It just behaves like noinline in many use cases.

Add @wmi for test coverage suggestions from the experience of D79959 "use-sample-profile".

[nickdesaulniers]

Similar to bc044a88ee3c, I should probably add a test for alwaysinline.

[nickdesaulniers]

Similar to D93422, I should update the LangRef for noprofile.

[nickdesaulniers]

• langref addition for noprofile, test alwaysinline callsite

[MaskRay]

llvm/docs/LangRef.rst

1662

My only problem is with (too strong) guarantee. I think we can only say the instructions will not instrumentation even inlined.

[nickdesaulniers]

• relax specifics regarding inlining noprofile in LangRef

[MaskRay]

llvm/docs/LangRef.rst

1662

For alwaysinline, I think the effect may be hazy. Dropping instrumentation is ok. To make alwaysinline take preference (the behavior of many other binary attributes), retaining instrumentation also looks ok.

[phosek]

LGTM

llvm/docs/LangRef.rst

``nomerge``
     Tail merging normally reduces the precision of source location information,
     making stack traces less useful for debugging. This attribute gives the
     user control over the tradeoff between code size and debug information
     precision.
 ``nonlazybind``
     This attribute suppresses lazy symbol binding for the function. This
     may make calls to the function faster, at the cost of extra program
     startup time if the function is not called during program startup.
+``noprofile``
+    This function attribute prevents instrumentation based profiling, used for
+    coverage or profile based optimization, from being added to a function,
+    even when inlined.
 ``noredzone``
     This attribute indicates that the code generator should not use a
     red zone, even if the target-specific ABI normally permits it.
 ``indirect-tls-seg-refs``
     This attribute indicates that the code generator should not use
     direct TLS access through segment registers, even if the
     target-specific ABI normally permits it.
 ``noreturn``

llvm/include/llvm/IR/Attributes.td

 def : CompatRule<"isEqual<SanitizeAddressAttr>">;
 def : CompatRule<"isEqual<SanitizeThreadAttr>">;
 def : CompatRule<"isEqual<SanitizeMemoryAttr>">;
 def : CompatRule<"isEqual<SanitizeHWAddressAttr>">;
 def : CompatRule<"isEqual<SanitizeMemTagAttr>">;
 def : CompatRule<"isEqual<SafeStackAttr>">;
 def : CompatRule<"isEqual<ShadowCallStackAttr>">;
 def : CompatRule<"isEqual<UseSampleProfileAttr>">;
+def : CompatRule<"isEqual<NoProfileAttr>">;
 
 class MergeRule<string F> {
   // The name of the function called to merge the attributes of the caller and
   // callee. The function's signature must match
   // "void(Function&, const Function &)", where the first parameter is the
   // reference to the caller and the second parameter is the reference to the
   // callee.
   string MergeFunc = F;

llvm/test/Transforms/Inline/inline_noprofile.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=inline %s -S -pass-remarks-missed=inline 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-INLINE
+; RUN: opt -passes=always-inline %s -S | FileCheck %s
+
+; Test that we don't inline when caller and callee don't have matching
+; noprofile fn attrs.
+
+; CHECK-INLINE: profile not inlined into profile_caller because it should never be inlined (cost=never): conflicting attributes
+; CHECK-INLINE: noprofile not inlined into noprofile_caller because it should never be inlined (cost=never): conflicting attributes
+
+define i32 @profile() { ret i32 42 }
+define i32 @noprofile() noprofile { ret i32 43 }
+define i32 @profile_aa() alwaysinline { ret i32 44 }
+define i32 @noprofile_aa() noprofile alwaysinline { ret i32 45 }
+
+define i32 @profile_caller() noprofile {
+; CHECK-LABEL: @profile_caller(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @profile()
+; CHECK-NEXT:    ret i32 44
+;
+  call i32 @profile()
+  %2 = call i32 @profile_aa()
+  ret i32 %2
+}
+
+define i32 @noprofile_caller() {
+; CHECK-LABEL: @noprofile_caller(
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @noprofile()
+; CHECK-NEXT:    ret i32 45
+;
+  call i32 @noprofile()
+  %2 = call i32 @noprofile_aa()
+  ret i32 %2
+}
+
+; Test that we do inline when caller and callee don't have matching
+; noprofile fn attrs, when CallInst is alwaysinline.
+define i32 @aa_callsite() {
+; CHECK-INLINE-LABEL: @aa_callsite(
+; CHECK-INLINE-NEXT:    ret i32 43
+;
+  %1 = call i32 @noprofile() alwaysinline
+  ret i32 %1
+}