[memprof] Add MemProf format Version 3 #93608
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This patch adds Version 3 for development purposes. For now, this patch adds V3 as a copy of V2. For the most part, this patch adds "case Version3:" wherever "case Version2:" appears. One exception is writeMemProfV3, which is copied from writeMemProfV2 but updated to write out memprof::Version3 to the MemProf header. We'll incrementally modify writeMemProfV3 in subsequent patches.
|
@llvm/pr-subscribers-pgo Author: Kazu Hirata (kazutakahirata) ChangesThis patch adds Version 3 for development purposes. For now, this For the most part, this patch adds "case Version3:" wherever "case Full diff: https://github.com/llvm/llvm-project/pull/93608.diff 6 Files Affected:
diff --git a/llvm/include/llvm/ProfileData/MemProf.h b/llvm/include/llvm/ProfileData/MemProf.h
index 17cef15344285..d44a2d1e2fb11 100644
--- a/llvm/include/llvm/ProfileData/MemProf.h
+++ b/llvm/include/llvm/ProfileData/MemProf.h
@@ -28,10 +28,12 @@ enum IndexedVersion : uint64_t {
Version1 = 1,
// Version 2: Added a call stack table.
Version2 = 2,
+ // Version 3: Under development.
+ Version3 = 3,
};
constexpr uint64_t MinimumSupportedVersion = Version0;
-constexpr uint64_t MaximumSupportedVersion = Version2;
+constexpr uint64_t MaximumSupportedVersion = Version3;
// Verify that the minimum and maximum satisfy the obvious constraint.
static_assert(MinimumSupportedVersion <= MaximumSupportedVersion);
diff --git a/llvm/lib/ProfileData/InstrProfReader.cpp b/llvm/lib/ProfileData/InstrProfReader.cpp
index 836206a4fd86e..798236c295194 100644
--- a/llvm/lib/ProfileData/InstrProfReader.cpp
+++ b/llvm/lib/ProfileData/InstrProfReader.cpp
@@ -1212,7 +1212,8 @@ Error IndexedMemProfReader::deserialize(const unsigned char *Start,
const uint64_t FirstWord =
support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
- if (FirstWord == memprof::Version1 || FirstWord == memprof::Version2) {
+ if (FirstWord == memprof::Version1 || FirstWord == memprof::Version2 ||
+ FirstWord == memprof::Version3) {
// Everything is good. We can proceed to deserialize the rest.
Version = static_cast<memprof::IndexedVersion>(FirstWord);
} else if (FirstWord >= 24) {
@@ -1559,6 +1560,7 @@ IndexedMemProfReader::getMemProfRecord(const uint64_t FuncNameHash) const {
"MemProfCallStackTable must not be available");
return getMemProfRecordV0(IndexedRecord, *MemProfFrameTable);
case memprof::Version2:
+ case memprof::Version3:
assert(MemProfFrameTable && "MemProfFrameTable must be available");
assert(MemProfCallStackTable && "MemProfCallStackTable must be available");
return getMemProfRecordV2(IndexedRecord, *MemProfFrameTable,
diff --git a/llvm/lib/ProfileData/InstrProfWriter.cpp b/llvm/lib/ProfileData/InstrProfWriter.cpp
index b67a9700b680a..b16714ae8b9a2 100644
--- a/llvm/lib/ProfileData/InstrProfWriter.cpp
+++ b/llvm/lib/ProfileData/InstrProfWriter.cpp
@@ -617,6 +617,56 @@ static Error writeMemProfV2(ProfOStream &OS,
return Error::success();
}
+// Write out MemProf Version3 as follows:
+// uint64_t Version
+// uint64_t RecordTableOffset = RecordTableGenerator.Emit
+// uint64_t FramePayloadOffset = Offset for the frame payload
+// uint64_t FrameTableOffset = FrameTableGenerator.Emit
+// uint64_t CallStackPayloadOffset = Offset for the call stack payload
+// uint64_t CallStackTableOffset = CallStackTableGenerator.Emit
+// uint64_t Num schema entries
+// uint64_t Schema entry 0
+// uint64_t Schema entry 1
+// ....
+// uint64_t Schema entry N - 1
+// OnDiskChainedHashTable MemProfRecordData
+// OnDiskChainedHashTable MemProfFrameData
+// OnDiskChainedHashTable MemProfCallStackData
+static Error writeMemProfV3(ProfOStream &OS,
+ memprof::IndexedMemProfData &MemProfData,
+ bool MemProfFullSchema) {
+ OS.write(memprof::Version3);
+ uint64_t HeaderUpdatePos = OS.tell();
+ OS.write(0ULL); // Reserve space for the memprof record table offset.
+ OS.write(0ULL); // Reserve space for the memprof frame payload offset.
+ OS.write(0ULL); // Reserve space for the memprof frame table offset.
+ OS.write(0ULL); // Reserve space for the memprof call stack payload offset.
+ OS.write(0ULL); // Reserve space for the memprof call stack table offset.
+
+ auto Schema = memprof::getHotColdSchema();
+ if (MemProfFullSchema)
+ Schema = memprof::getFullSchema();
+ writeMemProfSchema(OS, Schema);
+
+ uint64_t RecordTableOffset = writeMemProfRecords(OS, MemProfData.RecordData,
+ &Schema, memprof::Version3);
+
+ uint64_t FramePayloadOffset = OS.tell();
+ uint64_t FrameTableOffset = writeMemProfFrames(OS, MemProfData.FrameData);
+
+ uint64_t CallStackPayloadOffset = OS.tell();
+ uint64_t CallStackTableOffset =
+ writeMemProfCallStacks(OS, MemProfData.CallStackData);
+
+ uint64_t Header[] = {
+ RecordTableOffset, FramePayloadOffset, FrameTableOffset,
+ CallStackPayloadOffset, CallStackTableOffset,
+ };
+ OS.patch({{HeaderUpdatePos, Header, std::size(Header)}});
+
+ return Error::success();
+}
+
// Write out the MemProf data in a requested version.
static Error writeMemProf(ProfOStream &OS,
memprof::IndexedMemProfData &MemProfData,
@@ -629,6 +679,8 @@ static Error writeMemProf(ProfOStream &OS,
return writeMemProfV1(OS, MemProfData);
case memprof::Version2:
return writeMemProfV2(OS, MemProfData, MemProfFullSchema);
+ case memprof::Version3:
+ return writeMemProfV3(OS, MemProfData, MemProfFullSchema);
}
return make_error<InstrProfError>(
diff --git a/llvm/lib/ProfileData/MemProf.cpp b/llvm/lib/ProfileData/MemProf.cpp
index 89afe7c39027c..2f0e53736c82e 100644
--- a/llvm/lib/ProfileData/MemProf.cpp
+++ b/llvm/lib/ProfileData/MemProf.cpp
@@ -52,6 +52,7 @@ size_t IndexedAllocationInfo::serializedSize(const MemProfSchema &Schema,
case Version1:
return serializedSizeV0(*this, Schema);
case Version2:
+ case Version3:
return serializedSizeV2(*this, Schema);
}
llvm_unreachable("unsupported MemProf version");
@@ -95,6 +96,7 @@ size_t IndexedMemProfRecord::serializedSize(const MemProfSchema &Schema,
case Version1:
return serializedSizeV0(*this, Schema);
case Version2:
+ case Version3:
return serializedSizeV2(*this, Schema);
}
llvm_unreachable("unsupported MemProf version");
@@ -149,6 +151,7 @@ void IndexedMemProfRecord::serialize(const MemProfSchema &Schema,
serializeV0(*this, Schema, OS);
return;
case Version2:
+ case Version3:
serializeV2(*this, Schema, OS);
return;
}
@@ -239,6 +242,7 @@ IndexedMemProfRecord::deserialize(const MemProfSchema &Schema,
case Version1:
return deserializeV0(Schema, Ptr);
case Version2:
+ case Version3:
return deserializeV2(Schema, Ptr);
}
llvm_unreachable("unsupported MemProf version");
diff --git a/llvm/test/tools/llvm-profdata/memprof-merge-versions.test b/llvm/test/tools/llvm-profdata/memprof-merge-versions.test
index 28f65e0781bc6..aa7d0329425dc 100644
--- a/llvm/test/tools/llvm-profdata/memprof-merge-versions.test
+++ b/llvm/test/tools/llvm-profdata/memprof-merge-versions.test
@@ -19,6 +19,12 @@ RUN: llvm-profdata show %t.prof.v2 | FileCheck %s
RUN: llvm-profdata merge %t.proftext %p/Inputs/basic.memprofraw --memprof-version=2 --memprof-full-schema --profiled-binary %p/Inputs/basic.memprofexe -o %t.prof.v2
RUN: llvm-profdata show %t.prof.v2 | FileCheck %s
+RUN: llvm-profdata merge %t.proftext %p/Inputs/basic.memprofraw --memprof-version=3 --profiled-binary %p/Inputs/basic.memprofexe -o %t.prof.v3
+RUN: llvm-profdata show %t.prof.v3 | FileCheck %s
+
+RUN: llvm-profdata merge %t.proftext %p/Inputs/basic.memprofraw --memprof-version=3 --memprof-full-schema --profiled-binary %p/Inputs/basic.memprofexe -o %t.prof.v3
+RUN: llvm-profdata show %t.prof.v3 | FileCheck %s
+
For now we only check the validity of the instrumented profile since we don't
have a way to display the contents of the memprof indexed format yet.
diff --git a/llvm/tools/llvm-profdata/llvm-profdata.cpp b/llvm/tools/llvm-profdata/llvm-profdata.cpp
index 28c3afa101647..fae6d1e989ab5 100644
--- a/llvm/tools/llvm-profdata/llvm-profdata.cpp
+++ b/llvm/tools/llvm-profdata/llvm-profdata.cpp
@@ -306,7 +306,8 @@ cl::opt<memprof::IndexedVersion> MemProfVersionRequested(
cl::init(memprof::Version0),
cl::values(clEnumValN(memprof::Version0, "0", "version 0"),
clEnumValN(memprof::Version1, "1", "version 1"),
- clEnumValN(memprof::Version2, "2", "version 2")));
+ clEnumValN(memprof::Version2, "2", "version 2"),
+ clEnumValN(memprof::Version3, "3", "version 3")));
cl::opt<bool> MemProfFullSchema(
"memprof-full-schema", cl::Hidden, cl::sub(MergeSubcommand),
|
teresajohnson
approved these changes
May 28, 2024
vg0204
pushed a commit
to vg0204/llvm-project
that referenced
this pull request
May 29, 2024
This patch adds Version 3 for development purposes. For now, this patch adds V3 as a copy of V2. For the most part, this patch adds "case Version3:" wherever "case Version2:" appears. One exception is writeMemProfV3, which is copied from writeMemProfV2 but updated to write out memprof::Version3 to the MemProf header. We'll incrementally modify writeMemProfV3 in subsequent patches.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This patch adds Version 3 for development purposes. For now, this
patch adds V3 as a copy of V2.
For the most part, this patch adds "case Version3:" wherever "case
Version2:" appears. One exception is writeMemProfV3, which is copied
from writeMemProfV2 but updated to write out memprof::Version3 to the
MemProf header. We'll incrementally modify writeMemProfV3 in
subsequent patches.