diff options
author | Kazu Hirata <kazu@google.com> | 2024-05-28 07:31:29 -0700 |
---|---|---|
committer | GitHub <noreply@github.com> | 2024-05-28 07:31:29 -0700 |
commit | d2a103e682d65c3bfdff1d6a6f7b114e6cf4ff76 (patch) | |
tree | b0fc7f8c8a7f81ee5d9ed63ba0e2e839127ff715 /llvm/lib/ProfileData | |
parent | 3864bfd2e0ce7e32fd623c550660885599383e6a (diff) | |
download | llvm-d2a103e682d65c3bfdff1d6a6f7b114e6cf4ff76.zip llvm-d2a103e682d65c3bfdff1d6a6f7b114e6cf4ff76.tar.gz llvm-d2a103e682d65c3bfdff1d6a6f7b114e6cf4ff76.tar.bz2 |
[memprof] Remove const from the return type of toMemProfRecord (#93415)
"const" being removed in this patch prevents the move semantics from
being used in:
AI.CallStack = Callback(IndexedAI.CSId);
With this patch on an indexed MemProf Version 2 profile, the cycle
count and instruction count go down by 13.3% and 26.3%, respectively,
with "llvm-profdata show" modified to deserialize all MemProfRecords.
Diffstat (limited to 'llvm/lib/ProfileData')
-rw-r--r-- | llvm/lib/ProfileData/MemProf.cpp | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/llvm/lib/ProfileData/MemProf.cpp b/llvm/lib/ProfileData/MemProf.cpp index e560864..89afe7c 100644 --- a/llvm/lib/ProfileData/MemProf.cpp +++ b/llvm/lib/ProfileData/MemProf.cpp @@ -245,8 +245,8 @@ IndexedMemProfRecord::deserialize(const MemProfSchema &Schema, } MemProfRecord IndexedMemProfRecord::toMemProfRecord( - llvm::function_ref<const llvm::SmallVector<Frame>(const CallStackId)> - Callback) const { + llvm::function_ref<llvm::SmallVector<Frame>(const CallStackId)> Callback) + const { MemProfRecord Record; Record.AllocSites.reserve(AllocSites.size()); |