[NFC]In codegen pipeline, turn static-data-splitter pass on/off with its own option #134752
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
|
@llvm/pr-subscribers-backend-aarch64 Author: Mingming Liu (mingmingl-llvm) ChangesPer discussion in #129781 (comment), we'd like to refactor out the requirement of MFS. Full diff: https://github.com/llvm/llvm-project/pull/134752.diff 5 Files Affected:
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index fa1bb84ec5319..45fae653e4181 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/llvm/lib/CodeGen/TargetPassConfig.cpp
@@ -1235,13 +1235,11 @@ void TargetPassConfig::addMachinePasses() {
addPass(createMIRAddFSDiscriminatorsPass(
sampleprof::FSDiscriminatorPass::PassLast));
- // Machine function splitter uses the basic block sections feature.
- // When used along with `-basic-block-sections=`, the basic-block-sections
- // feature takes precedence. This means functions eligible for
- // basic-block-sections optimizations (`=all`, or `=list=` with function
- // included in the list profile) will get that optimization instead.
- if (TM->Options.EnableMachineFunctionSplitter ||
- EnableMachineFunctionSplitter) {
+ bool MIRProfileLoaderPassEnabled = false;
+ auto EnableMIRProfileLoaderPass = [&]() {
+ if (MIRProfileLoaderPassEnabled)
+ return;
+
const std::string ProfileFile = getFSProfileFile(TM);
if (!ProfileFile.empty()) {
if (EnableFSDiscriminator) {
@@ -1256,14 +1254,25 @@ void TargetPassConfig::addMachinePasses() {
"performance.\n";
}
}
+ MIRProfileLoaderPassEnabled = true;
+ };
+ // Machine function splitter uses the basic block sections feature.
+ // When used along with `-basic-block-sections=`, the basic-block-sections
+ // feature takes precedence. This means functions eligible for
+ // basic-block-sections optimizations (`=all`, or `=list=` with function
+ // included in the list profile) will get that optimization instead.
+ if (TM->Options.EnableMachineFunctionSplitter ||
+ EnableMachineFunctionSplitter) {
+ EnableMIRProfileLoaderPass();
addPass(createMachineFunctionSplitterPass());
- if (SplitStaticData || TM->Options.EnableStaticDataPartitioning) {
- // The static data splitter pass is a machine function pass. and
- // static data annotator pass is a module-wide pass. See the file comment
- // in StaticDataAnnotator.cpp for the motivation.
- addPass(createStaticDataSplitterPass());
- addPass(createStaticDataAnnotatorPass());
- }
+ }
+ if (SplitStaticData || TM->Options.EnableStaticDataPartitioning) {
+ EnableMIRProfileLoaderPass();
+ // The static data splitter pass is a machine function pass. and
+ // static data annotator pass is a module-wide pass. See the file comment
+ // in StaticDataAnnotator.cpp for the motivation.
+ addPass(createStaticDataSplitterPass());
+ addPass(createStaticDataAnnotatorPass());
}
// We run the BasicBlockSections pass if either we need BB sections or BB
// address map (or both).
diff --git a/llvm/test/CodeGen/AArch64/constant-pool-partition.ll b/llvm/test/CodeGen/AArch64/constant-pool-partition.ll
index ab627b02a1bc7..d4447131e9de1 100644
--- a/llvm/test/CodeGen/AArch64/constant-pool-partition.ll
+++ b/llvm/test/CodeGen/AArch64/constant-pool-partition.ll
@@ -1,12 +1,10 @@
-; RUN: llc -mtriple=aarch64 -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -unique-section-names=false \
+; RUN: llc -mtriple=aarch64 -partition-static-data-sections \
+; RUN: -function-sections -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
; Repeat the RUN command above for big-endian systems.
-; RUN: llc -mtriple=aarch64_be -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -unique-section-names=false \
+; RUN: llc -mtriple=aarch64_be -partition-static-data-sections \
+; RUN: -function-sections -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
; Tests that constant pool hotness is aggregated across the module. The
diff --git a/llvm/test/CodeGen/X86/constant-pool-partition.ll b/llvm/test/CodeGen/X86/constant-pool-partition.ll
index d2c87b7b3fc14..515284fb2cf1a 100644
--- a/llvm/test/CodeGen/X86/constant-pool-partition.ll
+++ b/llvm/test/CodeGen/X86/constant-pool-partition.ll
@@ -10,18 +10,16 @@ target triple = "x86_64-grtev4-linux-gnu"
; 2. Similarly if a constant is accessed by both cold function and un-profiled
; function, constant pools for this constant should not have .unlikely suffix.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true -data-sections=true \
-; RUN: -unique-section-names=false \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -data-sections -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true -data-sections=true \
-; RUN: -unique-section-names=true \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -data-sections -unique-section-names \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=false -data-sections=false \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections=false -data-sections=false \
; RUN: -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
diff --git a/llvm/test/CodeGen/X86/global-variable-partition.ll b/llvm/test/CodeGen/X86/global-variable-partition.ll
index 91084d038cfe0..ce06d1712f840 100644
--- a/llvm/test/CodeGen/X86/global-variable-partition.ll
+++ b/llvm/test/CodeGen/X86/global-variable-partition.ll
@@ -11,22 +11,22 @@ target triple = "x86_64-unknown-linux-gnu"
; This RUN command sets `-data-sections=true -unique-section-names=true` so data
; sections are uniqufied by numbers.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -data-sections=true \
-; RUN: -unique-section-names=true -relocation-model=pic \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -relocation-model=pic \
+; RUN: -partition-static-data-sections=true \
+; RUN: -data-sections=true -unique-section-names=true \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=SYM,COMMON --dump-input=always
; This RUN command sets `-data-sections=true -unique-section-names=false` so
; data sections are uniqufied by variable names.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -data-sections=true \
-; RUN: -unique-section-names=false -relocation-model=pic \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -relocation-model=pic \
+; RUN: -partition-static-data-sections=true \
+; RUN: -data-sections=true -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=UNIQ,COMMON --dump-input=always
; This RUN command sets `-data-sections=false -unique-section-names=false`.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -data-sections=false \
-; RUN: -unique-section-names=false -relocation-model=pic \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -relocation-model=pic \
+; RUN: -partition-static-data-sections=true \
+; RUN: -data-sections=false -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=AGG,COMMON --dump-input=always
; For @.str and @.str.1
@@ -42,19 +42,19 @@ target triple = "x86_64-unknown-linux-gnu"
; For @hot_relro_array
; COMMON: .type hot_relro_array,@object
; SYM-NEXT: .section .data.rel.ro.hot.hot_relro_array
-; UNIQ-NEXT: .section .data.rel.ro.hot.,"aw",@progbits,unique,3
+; UNIQ-NEXT: .section .data.rel.ro.hot.,"aw",@progbits,unique,1
; AGG-NEXT: .section .data.rel.ro.hot.,"aw",@progbits
; For @hot_data, which is accessed by {cold_func, unprofiled_func, hot_func}.
; COMMON: .type hot_data,@object
; SYM-NEXT: .section .data.hot.hot_data,"aw",@progbits
-; UNIQ-NEXT: .section .data.hot.,"aw",@progbits,unique,4
+; UNIQ-NEXT: .section .data.hot.,"aw",@progbits,unique,2
; AGG-NEXT: .section .data.hot.,"aw",@progbits
; For @hot_bss, which is accessed by {unprofiled_func, hot_func}.
; COMMON: .type hot_bss,@object
; SYM-NEXT: .section .bss.hot.hot_bss,"aw",@nobits
-; UNIQ-NEXT: .section .bss.hot.,"aw",@nobits,unique,5
+; UNIQ-NEXT: .section .bss.hot.,"aw",@nobits,unique,3
; AGG-NEXT: .section .bss.hot.,"aw",@nobits
; For @.str.2
@@ -68,13 +68,13 @@ target triple = "x86_64-unknown-linux-gnu"
; For @cold_bss
; COMMON: .type cold_bss,@object
; SYM-NEXT: .section .bss.unlikely.cold_bss,"aw",@nobits
-; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,6
+; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,4
; AGG-NEXT: .section .bss.unlikely.,"aw",@nobits
; For @cold_data
; COMMON: .type cold_data,@object
; SYM-NEXT: .section .data.unlikely.cold_data,"aw",@progbits
-; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,7
+; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,5
; AGG-NEXT: .section .data.unlikely.,"aw",@progbits
; For @cold_data_custom_foo_section
@@ -87,7 +87,7 @@ target triple = "x86_64-unknown-linux-gnu"
; For @cold_relro_array
; COMMON: .type cold_relro_array,@object
; SYM-NEXT: .section .data.rel.ro.unlikely.cold_relro_array,"aw",@progbits
-; UNIQ-NEXT: .section .data.rel.ro.unlikely.,"aw",@progbits,unique,8
+; UNIQ-NEXT: .section .data.rel.ro.unlikely.,"aw",@progbits,unique,6
; AGG-NEXT: .section .data.rel.ro.unlikely.,"aw",@progbits
; Currently static-data-splitter only analyzes access from code.
@@ -97,19 +97,19 @@ target triple = "x86_64-unknown-linux-gnu"
; For @bss2
; COMMON: .type bss2,@object
; SYM-NEXT: .section .bss.unlikely.bss2,"aw",@nobits
-; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,9
+; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,7
; AGG-NEXT: .section .bss.unlikely.,"aw",@nobits
; For @data3
; COMMON: .type data3,@object
; SYM-NEXT: .section .data.unlikely.data3,"aw",@progbits
-; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,10
+; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,8
; AGG-NEXT: .section .data.unlikely.,"aw",@progbits
; For @data_with_unknown_hotness
; SYM: .type .Ldata_with_unknown_hotness,@object # @data_with_unknown_hotness
; SYM: .section .data..Ldata_with_unknown_hotness,"aw",@progbits
-; UNIQ: .section .data,"aw",@progbits,unique,11
+; UNIQ: .section .data,"aw",@progbits,unique,9
; The `.section` directive is omitted for .data with -unique-section-names=false.
; See MCSectionELF::shouldOmitSectionDirective for the implementation details.
; AGG: .data
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 0d76f8a5a91ed..40dbc8131e22b 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -13,19 +13,19 @@
; STAT: 2 static-data-splitter - Number of hot jump tables seen
; STAT: 2 static-data-splitter - Number of jump tables with unknown hotness
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -min-jump-table-entries=2 -unique-section-names=false \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -unique-section-names=false \
+; RUN: -min-jump-table-entries=2 \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=NUM,JT
; Section names will optionally have `.<func>` if -function-sections is enabled.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -min-jump-table-entries=2 \
+; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=false \
-; RUN: -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections=false -min-jump-table-entries=2 \
+; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
; In function @foo, the 2 switch instructions to jt0.* and jt1.* are placed in
; hot-prefixed sections, and the 2 switch instructions to jt2.* and jt3.* are
|
|
@llvm/pr-subscribers-backend-x86 Author: Mingming Liu (mingmingl-llvm) ChangesPer discussion in #129781 (comment), we'd like to refactor out the requirement of MFS. Full diff: https://github.com/llvm/llvm-project/pull/134752.diff 5 Files Affected:
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index fa1bb84ec5319..45fae653e4181 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/llvm/lib/CodeGen/TargetPassConfig.cpp
@@ -1235,13 +1235,11 @@ void TargetPassConfig::addMachinePasses() {
addPass(createMIRAddFSDiscriminatorsPass(
sampleprof::FSDiscriminatorPass::PassLast));
- // Machine function splitter uses the basic block sections feature.
- // When used along with `-basic-block-sections=`, the basic-block-sections
- // feature takes precedence. This means functions eligible for
- // basic-block-sections optimizations (`=all`, or `=list=` with function
- // included in the list profile) will get that optimization instead.
- if (TM->Options.EnableMachineFunctionSplitter ||
- EnableMachineFunctionSplitter) {
+ bool MIRProfileLoaderPassEnabled = false;
+ auto EnableMIRProfileLoaderPass = [&]() {
+ if (MIRProfileLoaderPassEnabled)
+ return;
+
const std::string ProfileFile = getFSProfileFile(TM);
if (!ProfileFile.empty()) {
if (EnableFSDiscriminator) {
@@ -1256,14 +1254,25 @@ void TargetPassConfig::addMachinePasses() {
"performance.\n";
}
}
+ MIRProfileLoaderPassEnabled = true;
+ };
+ // Machine function splitter uses the basic block sections feature.
+ // When used along with `-basic-block-sections=`, the basic-block-sections
+ // feature takes precedence. This means functions eligible for
+ // basic-block-sections optimizations (`=all`, or `=list=` with function
+ // included in the list profile) will get that optimization instead.
+ if (TM->Options.EnableMachineFunctionSplitter ||
+ EnableMachineFunctionSplitter) {
+ EnableMIRProfileLoaderPass();
addPass(createMachineFunctionSplitterPass());
- if (SplitStaticData || TM->Options.EnableStaticDataPartitioning) {
- // The static data splitter pass is a machine function pass. and
- // static data annotator pass is a module-wide pass. See the file comment
- // in StaticDataAnnotator.cpp for the motivation.
- addPass(createStaticDataSplitterPass());
- addPass(createStaticDataAnnotatorPass());
- }
+ }
+ if (SplitStaticData || TM->Options.EnableStaticDataPartitioning) {
+ EnableMIRProfileLoaderPass();
+ // The static data splitter pass is a machine function pass. and
+ // static data annotator pass is a module-wide pass. See the file comment
+ // in StaticDataAnnotator.cpp for the motivation.
+ addPass(createStaticDataSplitterPass());
+ addPass(createStaticDataAnnotatorPass());
}
// We run the BasicBlockSections pass if either we need BB sections or BB
// address map (or both).
diff --git a/llvm/test/CodeGen/AArch64/constant-pool-partition.ll b/llvm/test/CodeGen/AArch64/constant-pool-partition.ll
index ab627b02a1bc7..d4447131e9de1 100644
--- a/llvm/test/CodeGen/AArch64/constant-pool-partition.ll
+++ b/llvm/test/CodeGen/AArch64/constant-pool-partition.ll
@@ -1,12 +1,10 @@
-; RUN: llc -mtriple=aarch64 -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -unique-section-names=false \
+; RUN: llc -mtriple=aarch64 -partition-static-data-sections \
+; RUN: -function-sections -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
; Repeat the RUN command above for big-endian systems.
-; RUN: llc -mtriple=aarch64_be -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -unique-section-names=false \
+; RUN: llc -mtriple=aarch64_be -partition-static-data-sections \
+; RUN: -function-sections -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
; Tests that constant pool hotness is aggregated across the module. The
diff --git a/llvm/test/CodeGen/X86/constant-pool-partition.ll b/llvm/test/CodeGen/X86/constant-pool-partition.ll
index d2c87b7b3fc14..515284fb2cf1a 100644
--- a/llvm/test/CodeGen/X86/constant-pool-partition.ll
+++ b/llvm/test/CodeGen/X86/constant-pool-partition.ll
@@ -10,18 +10,16 @@ target triple = "x86_64-grtev4-linux-gnu"
; 2. Similarly if a constant is accessed by both cold function and un-profiled
; function, constant pools for this constant should not have .unlikely suffix.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true -data-sections=true \
-; RUN: -unique-section-names=false \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -data-sections -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true -data-sections=true \
-; RUN: -unique-section-names=true \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -data-sections -unique-section-names \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=false -data-sections=false \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections=false -data-sections=false \
; RUN: -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --dump-input=always
diff --git a/llvm/test/CodeGen/X86/global-variable-partition.ll b/llvm/test/CodeGen/X86/global-variable-partition.ll
index 91084d038cfe0..ce06d1712f840 100644
--- a/llvm/test/CodeGen/X86/global-variable-partition.ll
+++ b/llvm/test/CodeGen/X86/global-variable-partition.ll
@@ -11,22 +11,22 @@ target triple = "x86_64-unknown-linux-gnu"
; This RUN command sets `-data-sections=true -unique-section-names=true` so data
; sections are uniqufied by numbers.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -data-sections=true \
-; RUN: -unique-section-names=true -relocation-model=pic \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -relocation-model=pic \
+; RUN: -partition-static-data-sections=true \
+; RUN: -data-sections=true -unique-section-names=true \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=SYM,COMMON --dump-input=always
; This RUN command sets `-data-sections=true -unique-section-names=false` so
; data sections are uniqufied by variable names.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -data-sections=true \
-; RUN: -unique-section-names=false -relocation-model=pic \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -relocation-model=pic \
+; RUN: -partition-static-data-sections=true \
+; RUN: -data-sections=true -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=UNIQ,COMMON --dump-input=always
; This RUN command sets `-data-sections=false -unique-section-names=false`.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -data-sections=false \
-; RUN: -unique-section-names=false -relocation-model=pic \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -relocation-model=pic \
+; RUN: -partition-static-data-sections=true \
+; RUN: -data-sections=false -unique-section-names=false \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=AGG,COMMON --dump-input=always
; For @.str and @.str.1
@@ -42,19 +42,19 @@ target triple = "x86_64-unknown-linux-gnu"
; For @hot_relro_array
; COMMON: .type hot_relro_array,@object
; SYM-NEXT: .section .data.rel.ro.hot.hot_relro_array
-; UNIQ-NEXT: .section .data.rel.ro.hot.,"aw",@progbits,unique,3
+; UNIQ-NEXT: .section .data.rel.ro.hot.,"aw",@progbits,unique,1
; AGG-NEXT: .section .data.rel.ro.hot.,"aw",@progbits
; For @hot_data, which is accessed by {cold_func, unprofiled_func, hot_func}.
; COMMON: .type hot_data,@object
; SYM-NEXT: .section .data.hot.hot_data,"aw",@progbits
-; UNIQ-NEXT: .section .data.hot.,"aw",@progbits,unique,4
+; UNIQ-NEXT: .section .data.hot.,"aw",@progbits,unique,2
; AGG-NEXT: .section .data.hot.,"aw",@progbits
; For @hot_bss, which is accessed by {unprofiled_func, hot_func}.
; COMMON: .type hot_bss,@object
; SYM-NEXT: .section .bss.hot.hot_bss,"aw",@nobits
-; UNIQ-NEXT: .section .bss.hot.,"aw",@nobits,unique,5
+; UNIQ-NEXT: .section .bss.hot.,"aw",@nobits,unique,3
; AGG-NEXT: .section .bss.hot.,"aw",@nobits
; For @.str.2
@@ -68,13 +68,13 @@ target triple = "x86_64-unknown-linux-gnu"
; For @cold_bss
; COMMON: .type cold_bss,@object
; SYM-NEXT: .section .bss.unlikely.cold_bss,"aw",@nobits
-; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,6
+; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,4
; AGG-NEXT: .section .bss.unlikely.,"aw",@nobits
; For @cold_data
; COMMON: .type cold_data,@object
; SYM-NEXT: .section .data.unlikely.cold_data,"aw",@progbits
-; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,7
+; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,5
; AGG-NEXT: .section .data.unlikely.,"aw",@progbits
; For @cold_data_custom_foo_section
@@ -87,7 +87,7 @@ target triple = "x86_64-unknown-linux-gnu"
; For @cold_relro_array
; COMMON: .type cold_relro_array,@object
; SYM-NEXT: .section .data.rel.ro.unlikely.cold_relro_array,"aw",@progbits
-; UNIQ-NEXT: .section .data.rel.ro.unlikely.,"aw",@progbits,unique,8
+; UNIQ-NEXT: .section .data.rel.ro.unlikely.,"aw",@progbits,unique,6
; AGG-NEXT: .section .data.rel.ro.unlikely.,"aw",@progbits
; Currently static-data-splitter only analyzes access from code.
@@ -97,19 +97,19 @@ target triple = "x86_64-unknown-linux-gnu"
; For @bss2
; COMMON: .type bss2,@object
; SYM-NEXT: .section .bss.unlikely.bss2,"aw",@nobits
-; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,9
+; UNIQ-NEXT: .section .bss.unlikely.,"aw",@nobits,unique,7
; AGG-NEXT: .section .bss.unlikely.,"aw",@nobits
; For @data3
; COMMON: .type data3,@object
; SYM-NEXT: .section .data.unlikely.data3,"aw",@progbits
-; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,10
+; UNIQ-NEXT: .section .data.unlikely.,"aw",@progbits,unique,8
; AGG-NEXT: .section .data.unlikely.,"aw",@progbits
; For @data_with_unknown_hotness
; SYM: .type .Ldata_with_unknown_hotness,@object # @data_with_unknown_hotness
; SYM: .section .data..Ldata_with_unknown_hotness,"aw",@progbits
-; UNIQ: .section .data,"aw",@progbits,unique,11
+; UNIQ: .section .data,"aw",@progbits,unique,9
; The `.section` directive is omitted for .data with -unique-section-names=false.
; See MCSectionELF::shouldOmitSectionDirective for the implementation details.
; AGG: .data
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 0d76f8a5a91ed..40dbc8131e22b 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -13,19 +13,19 @@
; STAT: 2 static-data-splitter - Number of hot jump tables seen
; STAT: 2 static-data-splitter - Number of jump tables with unknown hotness
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -min-jump-table-entries=2 -unique-section-names=false \
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -unique-section-names=false \
+; RUN: -min-jump-table-entries=2 \
; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=NUM,JT
; Section names will optionally have `.<func>` if -function-sections is enabled.
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=true \
-; RUN: -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections -min-jump-table-entries=2 \
+; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT
-; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions \
-; RUN: -partition-static-data-sections=true -function-sections=false \
-; RUN: -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -partition-static-data-sections \
+; RUN: -function-sections=false -min-jump-table-entries=2 \
+; RUN: %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
; In function @foo, the 2 switch instructions to jt0.* and jt1.* are placed in
; hot-prefixed sections, and the 2 switch instructions to jt2.* and jt3.* are
|
snehasish
reviewed
Apr 8, 2025
There was a problem hiding this comment.
I think we can simplify this a bit if we move the staticdatapartitioning pass before the machine function splitter. Since no new CFG changes are introduced in the static data splitter pass we can only load the FS profile once before both of these passes.
So something like this --
if (TM->Options.EnableMachineFunctionSplitter ||
EnableMachineFunctionSplitter || SplitStaticData || TM->Options.EnableStaticDataPartitioning) {
const std::string ProfileFile = getFSProfileFile(TM);
...
}
if (TM->Options.EnableMachineFunctionSplitter ||
EnableMachineFunctionSplitter) {
addPass(createMachineFunctionSplitterPass());
}
if (SplitStaticData || TM->Options.EnableStaticDataPartitioning) {
addPass(createStaticDataSplitterPass());
addPass(createStaticDataAnnotatorPass());
}
thanks for the suggestion! Yeah it's indeed much simpler, done. |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/66/builds/12264 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/52/builds/7417 Here is the relevant piece of the build log for the reference |
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.
Per discussion in #129781 (comment), we'd like to refactor out the requirement of MFS.