[lld][BP] Print total size of startup symbols #145106
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
Member
|
@llvm/pr-subscribers-lld-macho @llvm/pr-subscribers-lld-elf Author: Ellis Hoag (ellishg) ChangesA good proxy to estimate the number of page faults during startup is the total size of startup functions. Assuming profiles are up-to-date, we can measure this total size pretty easily. Note that if profile data is old, this number could be wrong. Full diff: https://github.com/llvm/llvm-project/pull/145106.diff 3 Files Affected:
diff --git a/lld/include/lld/Common/BPSectionOrdererBase.inc b/lld/include/lld/Common/BPSectionOrdererBase.inc
index bb2e55af1eb35..fbeca187aa76f 100644
--- a/lld/include/lld/Common/BPSectionOrdererBase.inc
+++ b/lld/include/lld/Common/BPSectionOrdererBase.inc
@@ -292,45 +292,54 @@ auto BPOrderer<D>::computeOrder(
bp.run(nodesForDataCompression);
}
- unsigned numStartupSections = 0;
- unsigned numCodeCompressionSections = 0;
- unsigned numDuplicateCodeSections = 0;
- unsigned numDataCompressionSections = 0;
- unsigned numDuplicateDataSections = 0;
+ unsigned numStartupSections = 0, startupSize = 0;
+ unsigned numCodeCompressionSections = 0, codeCompressionSize = 0;
+ unsigned numDuplicateCodeSections = 0, duplicateCodeSize = 0;
+ unsigned numDataCompressionSections = 0, dataCompressionSize = 0;
+ unsigned numDuplicateDataSections = 0, duplicateDataSize = 0;
SetVector<const Section *> orderedSections;
// Order startup functions,
for (auto &node : nodesForStartup) {
const auto *isec = sections[node.Id];
- if (orderedSections.insert(isec))
+ if (orderedSections.insert(isec)) {
+ startupSize += D::getSize(*isec);
++numStartupSections;
+ }
}
// then functions for compression,
for (auto &node : nodesForFunctionCompression) {
const auto *isec = sections[node.Id];
- if (orderedSections.insert(isec))
+ if (orderedSections.insert(isec)) {
+ codeCompressionSize += D::getSize(*isec);
++numCodeCompressionSections;
-
+ }
auto It = duplicateSectionIdxs.find(node.Id);
if (It == duplicateSectionIdxs.end())
continue;
for (auto dupSecIdx : It->getSecond()) {
const auto *dupIsec = sections[dupSecIdx];
- if (orderedSections.insert(dupIsec))
+ if (orderedSections.insert(dupIsec)) {
+ duplicateCodeSize += D::getSize(*isec);
++numDuplicateCodeSections;
+ }
}
}
// then data for compression.
for (auto &node : nodesForDataCompression) {
const auto *isec = sections[node.Id];
- if (orderedSections.insert(isec))
+ if (orderedSections.insert(isec)) {
+ dataCompressionSize += D::getSize(*isec);
++numDataCompressionSections;
+ }
auto It = duplicateSectionIdxs.find(node.Id);
if (It == duplicateSectionIdxs.end())
continue;
for (auto dupSecIdx : It->getSecond()) {
const auto *dupIsec = sections[dupSecIdx];
- if (orderedSections.insert(dupIsec))
+ if (orderedSections.insert(dupIsec)) {
+ duplicateDataSize += D::getSize(*isec);
++numDuplicateDataSections;
+ }
}
}
@@ -339,14 +348,21 @@ auto BPOrderer<D>::computeOrder(
numStartupSections + numCodeCompressionSections +
numDuplicateCodeSections + numDataCompressionSections +
numDuplicateDataSections;
- dbgs()
- << "Ordered " << numTotalOrderedSections
- << " sections using balanced partitioning:\n Functions for startup: "
- << numStartupSections
- << "\n Functions for compression: " << numCodeCompressionSections
- << "\n Duplicate functions: " << numDuplicateCodeSections
- << "\n Data for compression: " << numDataCompressionSections
- << "\n Duplicate data: " << numDuplicateDataSections << "\n";
+ unsigned totalOrderedSize = startupSize + codeCompressionSize +
+ duplicateCodeSize + dataCompressionSize +
+ duplicateDataSize;
+ dbgs() << "Ordered " << numTotalOrderedSections << " sections ("
+ << totalOrderedSize << " bytes) using balanced partitioning:\n";
+ dbgs() << " Functions for startup: " << numStartupSections << " ("
+ << startupSize << " bytes)\n";
+ dbgs() << " Functions for compression: " << numCodeCompressionSections
+ << " (" << codeCompressionSize << " bytes)\n";
+ dbgs() << " Duplicate functions: " << numDuplicateCodeSections << " ("
+ << duplicateCodeSize << " bytes)\n";
+ dbgs() << " Data for compression: " << numDataCompressionSections << " ("
+ << dataCompressionSize << " bytes)\n";
+ dbgs() << " Duplicate data: " << numDuplicateDataSections << " ("
+ << duplicateDataSize << " bytes)\n";
if (!profilePath.empty()) {
// Evaluate this function order for startup
diff --git a/lld/test/ELF/bp-section-orderer.s b/lld/test/ELF/bp-section-orderer.s
index 4889db63cd4d1..4a8374f7991fa 100644
--- a/lld/test/ELF/bp-section-orderer.s
+++ b/lld/test/ELF/bp-section-orderer.s
@@ -21,7 +21,7 @@
# RUN: llvm-profdata merge a.proftext -o a.profdata
# RUN: ld.lld a.o --irpgo-profile=a.profdata --bp-startup-sort=function --verbose-bp-section-orderer --icf=all --gc-sections 2>&1 | FileCheck %s --check-prefix=STARTUP-FUNC-ORDER
-# STARTUP-FUNC-ORDER: Ordered 3 sections using balanced partitioning
+# STARTUP-FUNC-ORDER: Ordered 3 sections ({{[0-9]+}} bytes) using balanced partitioning
# STARTUP-FUNC-ORDER: Total area under the page fault curve: 3.
# RUN: ld.lld -o out.s a.o --irpgo-profile=a.profdata --bp-startup-sort=function
@@ -49,10 +49,10 @@
# RUN: llvm-nm -jn out.cbs | tr '\n' , | FileCheck %s --check-prefix=CBOTH-STARTUP
# CBOTH-STARTUP: s5,s3,s4,s2,s1,A,B,C,F,E,D,merged1,merged2,_start,d4,d1,d3,d2,{{$}}
-# BP-COMPRESSION-FUNC: Ordered 9 sections using balanced partitioning
-# BP-COMPRESSION-ICF-FUNC: Ordered 8 sections using balanced partitioning
-# BP-COMPRESSION-DATA: Ordered 9 sections using balanced partitioning
-# BP-COMPRESSION-BOTH: Ordered 18 sections using balanced partitioning
+# BP-COMPRESSION-FUNC: Ordered 9 sections ({{[0-9]+}} bytes) using balanced partitioning
+# BP-COMPRESSION-ICF-FUNC: Ordered 8 sections ({{[0-9]+}} bytes) using balanced partitioning
+# BP-COMPRESSION-DATA: Ordered 9 sections ({{[0-9]+}} bytes) using balanced partitioning
+# BP-COMPRESSION-BOTH: Ordered 18 sections ({{[0-9]+}} bytes) using balanced partitioning
#--- a.proftext
:ir
diff --git a/lld/test/MachO/bp-section-orderer.s b/lld/test/MachO/bp-section-orderer.s
index 93d0b85731de7..d933f2a27851c 100644
--- a/lld/test/MachO/bp-section-orderer.s
+++ b/lld/test/MachO/bp-section-orderer.s
@@ -9,8 +9,8 @@
# RUN: %lld -arch arm64 -lSystem -e _main -o %t/a.out %t/a.o --irpgo-profile %t/a.profdata --bp-startup-sort=function --verbose-bp-section-orderer 2>&1 | FileCheck %s --check-prefix=STARTUP
# RUN: %lld -arch arm64 -lSystem -e _main -o %t/a.out %t/a.o --irpgo-profile=%t/a.profdata --bp-startup-sort=function --verbose-bp-section-orderer --icf=all --bp-compression-sort=none 2>&1 | FileCheck %s --check-prefix=STARTUP-ICF
-# STARTUP: Ordered 5 sections using balanced partitioning
-# STARTUP-ICF: Ordered 4 sections using balanced partitioning
+# STARTUP: Ordered 5 sections ({{[0-9]+}} bytes) using balanced partitioning
+# STARTUP-ICF: Ordered 4 sections ({{[0-9]+}} bytes) using balanced partitioning
# Check that orderfiles take precedence over BP
# RUN: %no-fatal-warnings-lld -arch arm64 -lSystem -e _main -o - %t/a.o -order_file %t/a.orderfile --irpgo-profile-sort=%t/a.profdata | llvm-nm --numeric-sort --format=just-symbols - | FileCheck %s --check-prefix=ORDERFILE
@@ -50,10 +50,10 @@
# RUN: %lld -arch arm64 -lSystem -e _main -o %t/a.out %t/a.o --verbose-bp-section-orderer --bp-compression-sort=both 2>&1 | FileCheck %s --check-prefix=COMPRESSION-BOTH
# RUN: %lld -arch arm64 -lSystem -e _main -o %t/a.out %t/a.o --verbose-bp-section-orderer --bp-compression-sort=both --irpgo-profile=%t/a.profdata --bp-startup-sort=function 2>&1 | FileCheck %s --check-prefix=COMPRESSION-BOTH
-# COMPRESSION-FUNC: Ordered 9 sections using balanced partitioning
-# COMPRESSION-ICF-FUNC: Ordered 7 sections using balanced partitioning
-# COMPRESSION-DATA: Ordered 7 sections using balanced partitioning
-# COMPRESSION-BOTH: Ordered 16 sections using balanced partitioning
+# COMPRESSION-FUNC: Ordered 9 sections ({{[0-9]+}} bytes) using balanced partitioning
+# COMPRESSION-ICF-FUNC: Ordered 7 sections ({{[0-9]+}} bytes) using balanced partitioning
+# COMPRESSION-DATA: Ordered 7 sections ({{[0-9]+}} bytes) using balanced partitioning
+# COMPRESSION-BOTH: Ordered 16 sections ({{[0-9]+}} bytes) using balanced partitioning
#--- a.s
.text
|
Contributor
Author
|
CC @Colibrow |
Contributor
|
LGTM. |
MaskRay
approved these changes
Jun 21, 2025
lld/test/ELF/bp-section-orderer.s
Outdated
| # BP-COMPRESSION-ICF-FUNC: Ordered 8 sections using balanced partitioning | ||
| # BP-COMPRESSION-DATA: Ordered 9 sections using balanced partitioning | ||
| # BP-COMPRESSION-BOTH: Ordered 18 sections using balanced partitioning | ||
| # BP-COMPRESSION-FUNC: Ordered 9 sections ({{[0-9]+}} bytes) using balanced partitioning |
lld/test/MachO/bp-section-orderer.s
Outdated
| # RUN: %lld -arch arm64 -lSystem -e _main -o %t/a.out %t/a.o --irpgo-profile=%t/a.profdata --bp-startup-sort=function --verbose-bp-section-orderer --icf=all --bp-compression-sort=none 2>&1 | FileCheck %s --check-prefix=STARTUP-ICF | ||
| # STARTUP: Ordered 5 sections using balanced partitioning | ||
| # STARTUP-ICF: Ordered 4 sections using balanced partitioning | ||
| # STARTUP: Ordered 5 sections ({{[0-9]+}} bytes) using balanced partitioning |
anthonyhatran
pushed a commit
to anthonyhatran/llvm-project
that referenced
this pull request
Jun 26, 2025
A good proxy to estimate the number of page faults during startup is the total size of startup functions. Assuming profiles are up-to-date, we can measure this total size pretty easily. Note that if profile data is old, this number could be wrong.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
A good proxy to estimate the number of page faults during startup is the total size of startup functions. Assuming profiles are up-to-date, we can measure this total size pretty easily. Note that if profile data is old, this number could be wrong.