blob: 12d3d9d827d26a98c90d3a699cd0b113eca7d971 [file] [log] [blame]
commit 13f83365cdb5bb752066ee8f4149ae24dfe46cf1
Author: haoyuintel <haoyu.zhang@intel.com>
Date: Tue Nov 8 14:51:14 2022 +0800
[Driver] Add -fsample-profile-use-profi
This patch enable `-sample-profile-use-profi` in Clang frontend as user-facing
feature. By using this patch, we can use the cflag of `-fsample-profile-use-profi`
instead of `-mllvm -sample-profile-use-profi`.
Reviewed By: hans, MaskRay
Differential Revision: https://reviews.llvm.org/D136846
diff --git a/clang/docs/UsersManual.rst b/clang/docs/UsersManual.rst
index 11bc5c906611..9b03db9e0f74 100644
--- a/clang/docs/UsersManual.rst
+++ b/clang/docs/UsersManual.rst
@@ -2241,6 +2241,15 @@ usual build cycle when using sample profilers for optimization:
$ clang++ -O2 -gline-tables-only -fprofile-sample-use=code.prof code.cc -o code
+ [OPTIONAL] Sampling-based profiles can have inaccuracies or missing block/
+ edge counters. The profile inference algorithm (profi) can be used to infer
+ missing blocks and edge counts, and improve the quality of profile data.
+ Enable it with ``-fsample-profile-use-profi``.
+
+ .. code-block:: console
+
+ $ clang++ -O2 -gline-tables-only -fprofile-sample-use=code.prof \
+ -fsample-profile-use-profi code.cc -o code
Sample Profile Formats
""""""""""""""""""""""
diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
index 608840b2d369..ca16bd976559 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -1251,6 +1251,13 @@ def fprofile_sample_accurate : Flag<["-"], "fprofile-sample-accurate">,
as cold. Otherwise, treat callsites without profile samples as if
we have no profile}]>,
MarshallingInfoFlag<CodeGenOpts<"ProfileSampleAccurate">>;
+def fsample_profile_use_profi : Flag<["-"], "fsample-profile-use-profi">,
+ Flags<[NoXarchOption, CC1Option]>, Group<f_Group>,
+ HelpText<"Use profi to infer block and edge counts">,
+ DocBrief<[{Infer block and edge counts. If the profiles have errors or missing
+ blocks caused by sampling, profile inference (profi) can convert
+ basic block counts to branch probabilites to fix them by extended
+ and re-engineered classic MCMF (min-cost max-flow) approach.}]>;
def fno_profile_sample_accurate : Flag<["-"], "fno-profile-sample-accurate">,
Group<f_Group>, Flags<[NoXarchOption]>;
def fauto_profile : Flag<["-"], "fauto-profile">, Group<f_Group>,
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp b/clang/lib/Driver/ToolChains/Clang.cpp
index bffc8dc61160..4e404c579a57 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -5729,6 +5729,12 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
Args.AddLastArg(CmdArgs, options::OPT_fclang_abi_compat_EQ);
+ if (getLastProfileSampleUseArg(Args) &&
+ Args.hasArg(options::OPT_fsample_profile_use_profi)) {
+ CmdArgs.push_back("-mllvm");
+ CmdArgs.push_back("-sample-profile-use-profi");
+ }
+
// Add runtime flag for PS4/PS5 when PGO, coverage, or sanitizers are enabled.
if (RawTriple.isPS() &&
!Args.hasArg(options::OPT_nostdlib, options::OPT_nodefaultlibs)) {
diff --git a/clang/test/Driver/pgo-sample-use-profi.c b/clang/test/Driver/pgo-sample-use-profi.c
new file mode 100644
index 000000000000..454a511a0628
--- /dev/null
+++ b/clang/test/Driver/pgo-sample-use-profi.c
@@ -0,0 +1,4 @@
+/// Test if profi flat is enabled in frontend as user-facing feature.
+// RUN: %clang -c -fsample-profile-use-profi -fprofile-sample-use=/dev/null -### %s 2>&1 | FileCheck %s
+
+// CHECK: "-mllvm" "-sample-profile-use-profi"