b034905c82
When we collect a contextual profile, we sample the threads entering its root and only collect on one at a time (see `ContextRoot::Taken`). If we want to compare profiles between contextual profiles, and/or flat profiles, we have a problem: we don't know how to compare the counter values relative to each other. To that end, we add `ContextRoot::TotalEntries`, which is incremented every time a root is entered and serves as multiplier for the counter values collected under that root. We expose this in the profile and leave the normalization to the user of the profile, for a few reasons: * it's only needed if reasoning about all profiles in aggregate. * the goal, in compiler_rt, is to flush out the profile as quickly as possible, and performing multiplications adds an overhead that may not even be necessary if the consumer of the profile doesn't care about combining profiles * the information itself may be interesting as an indication of relative sampling of various contexts.
59 lines
1.5 KiB
LLVM
59 lines
1.5 KiB
LLVM
; Check that flattened profile lowering handles cold subgraphs that end in "unreachable"
|
|
; RUN: split-file %s %t
|
|
; RUN: llvm-ctxprof-util fromYAML --input=%t/profile.yaml --output=%t/profile.ctxprofdata
|
|
; RUN: opt -passes=ctx-prof-flatten %t/example.ll -use-ctx-profile=%t/profile.ctxprofdata -S -o - | FileCheck %s
|
|
|
|
; CHECK-LABEL: entry:
|
|
; CHECK: br i1 %t, label %yes, label %no, !prof ![[C1:[0-9]+]]
|
|
; CHECK-LABEL: no:
|
|
; CHECK-NOT: !prof
|
|
; CHECK-LABEL: no1:
|
|
; CHECK-NOT: !prof
|
|
; CHECK-LABEL: no2:
|
|
; CHECK-NOT: !prof
|
|
; CHECK-LABEL: yes:
|
|
; CHECK: br i1 %t3, label %yes1, label %yes2, !prof ![[C1]]
|
|
; CHECK-NOT: !prof
|
|
; CHECK: ![[C1]] = !{!"branch_weights", i32 6, i32 0}
|
|
|
|
;--- example.ll
|
|
define void @f1(i32 %cond) !guid !0 {
|
|
entry:
|
|
call void @llvm.instrprof.increment(ptr @f1, i64 42, i32 42, i32 0)
|
|
%t = icmp eq i32 %cond, 1
|
|
br i1 %t, label %yes, label %no
|
|
|
|
no:
|
|
%t2 = icmp eq i32 %cond, 2
|
|
br i1 %t2, label %no1, label %no2
|
|
no1:
|
|
unreachable
|
|
no2:
|
|
call void @llvm.instrprof.increment(ptr @f1, i64 42, i32 42, i32 1)
|
|
unreachable
|
|
yes:
|
|
%t3 = icmp eq i32 %cond, 3
|
|
br i1 %t3, label %yes1, label %yes2
|
|
yes1:
|
|
br label %exit
|
|
yes2:
|
|
call void @llvm.instrprof.increment(ptr @f1, i64 42, i32 42, i32 2)
|
|
%t4 = icmp eq i32 %cond, 4
|
|
br i1 %t4, label %yes3, label %yes4
|
|
yes3:
|
|
br label %exit
|
|
yes4:
|
|
call void @llvm.instrprof.increment(ptr @f1, i64 42, i32 42, i32 3)
|
|
unreachable
|
|
exit:
|
|
ret void
|
|
}
|
|
|
|
!0 = !{i64 1234}
|
|
|
|
;--- profile.yaml
|
|
Contexts:
|
|
- Guid: 1234
|
|
TotalRootEntryCount: 12
|
|
Counters: [6,0,0,0]
|