[llvm-branch-commits] [llvm] [AsmPrinter][ELF] Support profile-guided section prefix for jump tables' (read-only) data sections (PR #122215)
Mingming Liu via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Thu Jan 16 09:21:51 PST 2025
https://github.com/mingmingl-llvm updated https://github.com/llvm/llvm-project/pull/122215
>From dd748277dff2b30ed02bfa466eeca7102aa93eb4 Mon Sep 17 00:00:00 2001
From: mingmingl <mingmingl at google.com>
Date: Fri, 10 Jan 2025 13:53:08 -0800
Subject: [PATCH 1/6] rely to upstream
---
llvm/include/llvm/CodeGen/MachineFunction.h | 2 +-
.../llvm/CodeGen/MachineJumpTableInfo.h | 9 +-
llvm/include/llvm/CodeGen/Passes.h | 2 +-
llvm/lib/CodeGen/MachineFunction.cpp | 12 +-
llvm/lib/CodeGen/StaticDataSplitter.cpp | 87 +++---
llvm/test/CodeGen/X86/jump-table-partition.ll | 251 +++++++++++-------
6 files changed, 223 insertions(+), 140 deletions(-)
diff --git a/llvm/include/llvm/CodeGen/MachineFunction.h b/llvm/include/llvm/CodeGen/MachineFunction.h
index c0f983d1c6787b..dcdbcaec168d22 100644
--- a/llvm/include/llvm/CodeGen/MachineFunction.h
+++ b/llvm/include/llvm/CodeGen/MachineFunction.h
@@ -91,7 +91,7 @@ template <> struct ilist_callback_traits<MachineBasicBlock> {
// The hotness of static data tracked by a MachineFunction and not represented
// as a global object in the module IR / MIR. Typical examples are
// MachineJumpTableInfo and MachineConstantPool.
-enum class DataHotness {
+enum class MachineFunctionDataHotness {
Unknown,
Cold,
Hot,
diff --git a/llvm/include/llvm/CodeGen/MachineJumpTableInfo.h b/llvm/include/llvm/CodeGen/MachineJumpTableInfo.h
index cc1f54a81b9bb4..e3675d6489b350 100644
--- a/llvm/include/llvm/CodeGen/MachineJumpTableInfo.h
+++ b/llvm/include/llvm/CodeGen/MachineJumpTableInfo.h
@@ -28,7 +28,7 @@ namespace llvm {
class MachineBasicBlock;
class DataLayout;
class raw_ostream;
-enum class DataHotness;
+enum class MachineFunctionDataHotness;
/// MachineJumpTableEntry - One jump table in the jump table info.
///
@@ -36,7 +36,7 @@ struct MachineJumpTableEntry {
/// MBBs - The vector of basic blocks from which to create the jump table.
std::vector<MachineBasicBlock*> MBBs;
- DataHotness Hotness;
+ MachineFunctionDataHotness Hotness;
explicit MachineJumpTableEntry(const std::vector<MachineBasicBlock *> &M);
};
@@ -109,7 +109,10 @@ class MachineJumpTableInfo {
return JumpTables;
}
- void updateJumpTableHotness(size_t JTI, DataHotness Hotness);
+ // Update machine jump table entry's hotness. Return true if the hotness is
+ // updated.
+ bool updateJumpTableEntryHotness(size_t JTI,
+ MachineFunctionDataHotness Hotness);
/// RemoveJumpTable - Mark the specific index as being dead. This will
/// prevent it from being emitted.
diff --git a/llvm/include/llvm/CodeGen/Passes.h b/llvm/include/llvm/CodeGen/Passes.h
index 16423d03ff7018..b5d2a7e6bf035b 100644
--- a/llvm/include/llvm/CodeGen/Passes.h
+++ b/llvm/include/llvm/CodeGen/Passes.h
@@ -71,7 +71,7 @@ namespace llvm {
/// using profile information.
MachineFunctionPass *createMachineFunctionSplitterPass();
- /// createStaticDataSplitterPass - This pass partions static data sections
+ /// createStaticDataSplitterPass - This pass partitions a static data section
/// into a hot and cold section using profile information.
MachineFunctionPass *createStaticDataSplitterPass();
diff --git a/llvm/lib/CodeGen/MachineFunction.cpp b/llvm/lib/CodeGen/MachineFunction.cpp
index b5a89f3bcf42f1..d09e93d79aae6c 100644
--- a/llvm/lib/CodeGen/MachineFunction.cpp
+++ b/llvm/lib/CodeGen/MachineFunction.cpp
@@ -1293,7 +1293,7 @@ const unsigned MachineFunction::DebugOperandMemNumber = 1000000;
MachineJumpTableEntry::MachineJumpTableEntry(
const std::vector<MachineBasicBlock *> &MBBs)
- : MBBs(MBBs), Hotness(DataHotness::Unknown) {}
+ : MBBs(MBBs), Hotness(MachineFunctionDataHotness::Unknown) {}
/// Return the size of each entry in the jump table.
unsigned MachineJumpTableInfo::getEntrySize(const DataLayout &TD) const {
@@ -1344,13 +1344,17 @@ unsigned MachineJumpTableInfo::createJumpTableIndex(
return JumpTables.size()-1;
}
-void MachineJumpTableInfo::updateJumpTableHotness(size_t JTI,
- DataHotness Hotness) {
+bool MachineJumpTableInfo::updateJumpTableEntryHotness(
+ size_t JTI, MachineFunctionDataHotness Hotness) {
assert(JTI < JumpTables.size() && "Invalid JTI!");
// Note record the largest hotness is important for mergable data (constant
// pools). Even if jump table instances are not merged, record the largest
// value seen fwiw.
- JumpTables[JTI].Hotness = std::max(JumpTables[JTI].Hotness, Hotness);
+ if (Hotness <= JumpTables[JTI].Hotness)
+ return false;
+
+ JumpTables[JTI].Hotness = Hotness;
+ return true;
}
/// If Old is the target of any jump tables, update the jump tables to branch
diff --git a/llvm/lib/CodeGen/StaticDataSplitter.cpp b/llvm/lib/CodeGen/StaticDataSplitter.cpp
index 482a61027cf985..9e2cfe18256e35 100644
--- a/llvm/lib/CodeGen/StaticDataSplitter.cpp
+++ b/llvm/lib/CodeGen/StaticDataSplitter.cpp
@@ -6,13 +6,16 @@
//
//===----------------------------------------------------------------------===//
//
-// This pass uses profile information to partition static data sections into
-// hot and cold ones. It begins to split jump tables based on profile, and
-// subsequent patches will handle constant pools and other module internal data.
+// The pass uses branch profile data to assign hotness based section qualifiers
+// for the following types of static data:
+// - Jump tables
+// - Constant pools (TODO)
+// - Other module-internal data (TODO)
//
// For the original RFC of this pass please see
-// https://discourse.llvm.org/t/rfc-profile-guided-static-data-partitioning/83744.
+// https://discourse.llvm.org/t/rfc-profile-guided-static-data-partitioning/83744
+#include "llvm/ADT/ScopeExit.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/ProfileSummaryInfo.h"
#include "llvm/CodeGen/MBFIWrapper.h"
@@ -35,8 +38,15 @@ using namespace llvm;
STATISTIC(NumHotJumpTables, "Number of hot jump tables seen");
STATISTIC(NumColdJumpTables, "Number of cold jump tables seen");
STATISTIC(NumUnknownJumpTables,
- "Number of jump tables with unknown hotness. Such jump tables will "
- "be placed in the hot-suffixed section by default.");
+ "Number of jump tables with unknown hotness. Option "
+ "-static-data-default-hotness specifies the hotness.");
+
+static cl::opt<MachineFunctionDataHotness> StaticDataDefaultHotness(
+ "static-data-default-hotness", cl::Hidden,
+ cl::desc("The hotness for static data with unknown hotness"),
+ cl::init(MachineFunctionDataHotness::Hot),
+ cl::values(clEnumValN(MachineFunctionDataHotness::Hot, "hot", "Hot"),
+ clEnumValN(MachineFunctionDataHotness::Cold, "cold", "Cold")));
class StaticDataSplitter : public MachineFunctionPass {
const MachineBranchProbabilityInfo *MBPI = nullptr;
@@ -82,12 +92,7 @@ bool StaticDataSplitter::runOnMachineFunction(MachineFunction &MF) {
bool StaticDataSplitter::splitJumpTablesWithProfiles(
MachineFunction &MF, MachineJumpTableInfo &MJTI) {
int NumChangedJumpTables = 0;
- // Regard a jump table as hot by default. If the source and all of destination
- // blocks are cold, regard the jump table as cold. While a destination block
- // does not read a jump table (unless it's also a source block), a hot
- // destination heuristically makes its jump table hot to accommodate for
- // potential profile data skews (from sampled profiles, for example).
- DataHotness Hotness = DataHotness::Hot;
+
for (const auto &MBB : MF) {
// IMPORTANT, `getJumpTableIndex` is a thin wrapper around per-target
// interface `TargetInstrInfo::getjumpTableIndex`, and only X86 implements
@@ -97,24 +102,14 @@ bool StaticDataSplitter::splitJumpTablesWithProfiles(
if (JTI == -1)
continue;
- bool AllBlocksCold = true;
-
- if (!PSI->isColdBlock(&MBB, MBFI))
- AllBlocksCold = false;
+ auto Hotness = MachineFunctionDataHotness::Hot;
- for (const MachineBasicBlock *MBB : MJTI.getJumpTables()[JTI].MBBs)
- if (!PSI->isColdBlock(MBB, MBFI))
- AllBlocksCold = false;
-
- if (AllBlocksCold) {
- Hotness = DataHotness::Cold;
- ++NumColdJumpTables;
- } else {
- ++NumHotJumpTables;
- }
+ // Hotness is based on source basic block hotness.
+ if (PSI->isColdBlock(&MBB, MBFI))
+ Hotness = MachineFunctionDataHotness::Cold;
- MF.getJumpTableInfo()->updateJumpTableHotness(JTI, Hotness);
- ++NumChangedJumpTables;
+ if (MF.getJumpTableInfo()->updateJumpTableEntryHotness(JTI, Hotness))
+ ++NumChangedJumpTables;
}
return NumChangedJumpTables > 0;
}
@@ -124,18 +119,40 @@ bool StaticDataSplitter::splitJumpTables(MachineFunction &MF) {
if (!MJTI || MJTI->getJumpTables().empty())
return false;
+ const bool ProfileAvailable = PSI && PSI->hasProfileSummary() && MBFI &&
+ MF.getFunction().hasProfileData();
+ auto statOnExit = llvm::make_scope_exit([&] {
+ if (!AreStatisticsEnabled())
+ return;
+
+ if (!ProfileAvailable) {
+ NumUnknownJumpTables += MJTI->getJumpTables().size();
+ return;
+ }
+
+ for (size_t JTI = 0; JTI < MJTI->getJumpTables().size(); JTI++) {
+ auto Hotness = MJTI->getJumpTables()[JTI].Hotness;
+ if (Hotness == MachineFunctionDataHotness::Hot)
+ NumHotJumpTables++;
+ else {
+ assert(Hotness == MachineFunctionDataHotness::Cold &&
+ "A jump table is hot or cold when profile information is "
+ "available.");
+ NumColdJumpTables++;
+ }
+ }
+ });
+
// Place jump tables according to block hotness if function has profile data.
- if (PSI && PSI->hasProfileSummary() && MBFI &&
- MF.getFunction().hasProfileData())
+ if (ProfileAvailable)
return splitJumpTablesWithProfiles(MF, *MJTI);
- // Conservatively place all jump tables in the hot-suffixed section if profile
- // information for the function is not available, or the target doesn't
- // implement `TargetInstrInfo::getJumpTableIndex` yet.
+ // If function profile is unavailable, -static-data-default-hotness specifies
+ // the hotness.
for (size_t JTI = 0; JTI < MJTI->getJumpTables().size(); JTI++)
- MF.getJumpTableInfo()->updateJumpTableHotness(JTI, DataHotness::Hot);
+ MF.getJumpTableInfo()->updateJumpTableEntryHotness(
+ JTI, StaticDataDefaultHotness);
- NumUnknownJumpTables += MJTI->getJumpTables().size();
return true;
}
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 3a8f1395f6b283..3b5af2c098553f 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -1,104 +1,172 @@
-;; -stats requires asserts
+; -stats requires asserts
; requires: asserts
-; RUN: llc -stop-after=block-placement %s -o - | llc --run-pass=static-data-splitter -stats -x mir -o - 2>&1 | FileCheck %s --check-prefix=STAT
+; Stop after 'finalize-isel' for simpler MIR, and lower the minimum number of
+; jump table entries so 'switch' needs fewer cases to generate a jump table.
+; RUN: llc -stop-after=finalize-isel -min-jump-table-entries=2 %s -o %t.mir
+; RUN: llc --run-pass=static-data-splitter -stats -x mir %t.mir -o - 2>&1 | FileCheck %s --check-prefix=STAT
-; `func_with_hot_jumptable` contains a hot jump table and `func_with_cold_jumptable` contains a cold one.
-; `func_without_entry_count` simulates the functions without profile information (e.g., not instrumented or not profiled),
-; it's jump table hotness is unknown and regarded as hot conservatively.
-;
; Tests stat messages are expected.
; TODO: Update test to verify section suffixes when target-lowering and assembler changes are implemented.
-;
-; STAT-DAG: 1 static-data-splitter - Number of cold jump tables seen
-; STAT-DAG: 1 static-data-splitter - Number of hot jump tables seen
+; TODO: Also run static-data-splitter pass with -static-data-default-hotness=cold and check data section suffix.
+
+; STAT-DAG: 2 static-data-splitter - Number of cold jump tables seen
+; STAT-DAG: 3 static-data-splitter - Number of hot jump tables seen
; STAT-DAG: 1 static-data-splitter - Number of jump tables with unknown hotness
+; @foo has four jump tables, jt0, jt1, jt2 and jt3 in the input basic block
+; order; jt0 and jt2 are hot, and jt1 and jt3 are cold.
+;
+; @func_with_hot_jt is a function with one entry count, and a hot loop using a
+; jump table.
+
+; @func_without_entry_count simulates the functions without profile information
+; (e.g., not instrumented or not profiled), it's jump table hotness is unknown
+; and regarded as hot conservatively.
+
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
- at .str.2 = private constant [7 x i8] c"case 3\00"
- at .str.3 = private constant [7 x i8] c"case 4\00"
- at .str.4 = private constant [7 x i8] c"case 5\00"
- at str.9 = private constant [7 x i8] c"case 2\00"
- at str.10 = private constant [7 x i8] c"case 1\00"
- at str.11 = private constant [8 x i8] c"default\00"
+ at str.9 = private constant [7 x i8] c".str.9\00"
+ at str.10 = private constant [8 x i8] c".str.10\00"
+ at str.11 = private constant [8 x i8] c".str.11\00"
-define i32 @func_with_hot_jumptable(i32 %num) !prof !13 {
+ at case2 = private constant [7 x i8] c"case 2\00"
+ at case1 = private constant [7 x i8] c"case 1\00"
+ at default = private constant [8 x i8] c"default\00"
+ at jt3 = private constant [3 x i8] c"jt\00"
+
+; jt0 and jt2 are hot. jt1 and jt3 are cold.
+define i32 @foo(i32 %num) !prof !13 {
entry:
- switch i32 %num, label %sw.default [
- i32 1, label %sw.bb
- i32 2, label %sw.bb1
- i32 3, label %sw.bb3
- i32 4, label %sw.bb5
- i32 5, label %sw.bb7
+ %mod3 = sdiv i32 %num, 3
+ switch i32 %mod3, label %jt0.default [
+ i32 1, label %jt0.bb1
+ i32 2, label %jt0.bb2
], !prof !14
-sw.bb: ; preds = %entry
- %puts11 = tail call i32 @puts(ptr @str.10)
- br label %sw.epilog
+jt0.bb1:
+ call i32 @puts(ptr @case1)
+ br label %jt0.epilog
-sw.bb1: ; preds = %entry
- %puts = tail call i32 @puts(ptr @str.9)
- br label %sw.epilog
+jt0.bb2:
+ call i32 @puts(ptr @case2)
+ br label %jt0.epilog
-sw.bb3: ; preds = %entry
- %call4 = tail call i32 (ptr, ...) @printf(ptr @.str.2)
- br label %sw.bb5
+jt0.default:
+ call i32 @puts(ptr @default)
+ br label %jt0.epilog
-sw.bb5: ; preds = %entry, %sw.bb3
- %call6 = tail call i32 (ptr, ...) @printf(ptr @.str.3)
- br label %sw.bb7
+jt0.epilog:
+ %zero = icmp eq i32 %num, 0
+ br i1 %zero, label %cold, label %hot, !prof !17
-sw.bb7: ; preds = %entry, %sw.bb5
- %call8 = tail call i32 (ptr, ...) @printf(ptr @.str.4)
- br label %sw.epilog
+cold:
+ %c1 = call i32 @compute(i32 %num)
+ switch i32 %c1, label %jt1.default [
+ i32 1, label %jt1.bb1
+ i32 2, label %jt1.bb2
+ ], !prof !14
-sw.default: ; preds = %entry
- %puts12 = tail call i32 @puts(ptr @str.11)
- br label %sw.epilog
+jt1.bb1:
+ call i32 @puts(ptr @case1)
+ br label %jt1.epilog
-sw.epilog: ; preds = %sw.default, %sw.bb7, %sw.bb1, %sw.bb
- %div = sdiv i32 %num, 3
- ret i32 %div
-}
+jt1.bb2:
+ call i32 @puts(ptr @case2)
+ br label %jt1.epilog
-define void @func_with_cold_jumptable(i32 %num) !prof !15 {
-entry:
- switch i32 %num, label %sw.default [
- i32 1, label %sw.bb
- i32 2, label %sw.bb1
- i32 3, label %sw.bb3
- i32 4, label %sw.bb5
- i32 5, label %sw.bb7
- ], !prof !16
+jt1.default:
+ call i32 @puts(ptr @default)
+ br label %jt1.epilog
-sw.bb: ; preds = %entry
- %puts10 = tail call i32 @puts(ptr @str.10)
- br label %sw.epilog
+jt1.epilog:
+ br label %return
-sw.bb1: ; preds = %entry
- %puts = tail call i32 @puts(ptr @str.9)
- br label %sw.epilog
+hot:
+ %c2 = call i32 @transform(i32 %num)
+ switch i32 %c2, label %jt2.default [
+ i32 1, label %jt2.bb1
+ i32 2, label %jt2.bb2
+ ], !prof !14
-sw.bb3: ; preds = %entry
- %call4 = tail call i32 (ptr, ...) @printf(ptr @.str.2)
- br label %sw.bb5
+jt2.bb1:
+ call i32 @puts(ptr @case1)
+ br label %jt1.epilog
-sw.bb5: ; preds = %entry, %sw.bb3
- %call6 = tail call i32 (ptr, ...) @printf(ptr @.str.3)
- br label %sw.bb7
+jt2.bb2:
+ call i32 @puts(ptr @case2)
+ br label %jt1.epilog
-sw.bb7: ; preds = %entry, %sw.bb5
- %call8 = tail call i32 (ptr, ...) @printf(ptr @.str.4)
- br label %sw.epilog
+jt2.default:
+ call i32 @puts(ptr @default)
+ br label %jt2.epilog
-sw.default: ; preds = %entry
- %puts11 = tail call i32 @puts(ptr @str.11)
- br label %sw.epilog
+jt2.epilog:
+ %c2cmp = icmp ne i32 %c2, 0
+ br i1 %c2cmp, label %return, label %jt3.prologue, !prof !18
-sw.epilog: ; preds = %sw.default, %sw.bb7, %sw.bb1, %sw.bb
- ret void
+jt3.prologue:
+ %c3 = call i32 @cleanup(i32 %num)
+ switch i32 %c3, label %jt3.default [
+ i32 1, label %jt3.bb1
+ i32 2, label %jt3.bb2
+ ], !prof !14
+
+jt3.bb1:
+ call i32 @puts(ptr @case1)
+ br label %jt3.epilog
+
+jt3.bb2:
+ call i32 @puts(ptr @case2)
+ br label %jt3.epilog
+
+jt3.default:
+ call i32 @puts(ptr @default)
+ br label %jt3.epilog
+
+jt3.epilog:
+ call i32 @puts(ptr @jt3)
+ br label %return
+
+return:
+ ret i32 %mod3
+}
+
+define i32 @func_with_hot_jt() !prof !15 {
+entry:
+ br label %for.body
+
+for.cond.cleanup:
+ ret i32 0
+
+for.body:
+ %lsr.iv = phi i32 [ 100000, %entry ], [ %lsr.iv.next, %loop.exit ]
+ %i.04 = phi i32 [ 0, %entry ], [ %inc, %loop.exit ]
+ %0 = urem i32 %i.04, 100
+ switch i32 %0, label %sw.default [
+ i32 1, label %loop.exit
+ i32 2, label %sw.bb1.i
+ i32 3, label %sw.bb3.i
+ ], !prof !19
+
+sw.bb1.i:
+ br label %loop.exit
+
+sw.bb3.i:
+ call i32 (ptr, ...) @printf(ptr @case1)
+ br label %sw.default
+
+sw.default:
+ br label %loop.exit
+
+loop.exit:
+ %str.5.sink.i = phi ptr [ @str.10, %sw.default ], [ @str.9, %sw.bb1.i ], [ @case2, %for.body ]
+ call i32 @puts(ptr %str.5.sink.i)
+ %inc = add i32 %i.04, 1
+ %lsr.iv.next = add i32 %lsr.iv, -1
+ %exitcond.not = icmp eq i32 %lsr.iv.next, 0
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body, !prof !17
}
define void @func_without_entry_count(i32 %num) {
@@ -106,41 +174,29 @@ entry:
switch i32 %num, label %sw.default [
i32 1, label %sw.bb
i32 2, label %sw.bb1
- i32 3, label %sw.bb3
- i32 4, label %sw.bb5
- i32 5, label %sw.bb7
]
-sw.bb: ; preds = %entry
- %puts10 = tail call i32 @puts(ptr @str.10)
+sw.bb:
+ call i32 @puts(ptr @str.10)
br label %sw.epilog
-sw.bb1: ; preds = %entry
- %puts = tail call i32 @puts(ptr @str.9)
- br label %sw.epilog
-
-sw.bb3: ; preds = %entry
- %call4 = tail call i32 (ptr, ...) @printf(ptr @.str.2)
- br label %sw.bb5
-
-sw.bb5: ; preds = %entry, %sw.bb3
- %call6 = tail call i32 (ptr, ...) @printf(ptr @.str.3)
- br label %sw.bb7
-
-sw.bb7: ; preds = %entry, %sw.bb5
- %call8 = tail call i32 (ptr, ...) @printf(ptr @.str.4)
+sw.bb1:
+ call i32 @puts(ptr @str.9)
br label %sw.epilog
-sw.default: ; preds = %entry
- %puts11 = tail call i32 @puts(ptr @str.11)
+sw.default:
+ call i32 @puts(ptr @str.11)
br label %sw.epilog
-sw.epilog: ; preds = %sw.default, %sw.bb7, %sw.bb1, %sw.bb
+sw.epilog:
ret void
}
declare i32 @puts(ptr)
declare i32 @printf(ptr, ...)
+declare i32 @compute(i32)
+declare i32 @transform(i32)
+declare i32 @cleanup(i32)
!llvm.module.flags = !{!0}
@@ -158,6 +214,9 @@ declare i32 @printf(ptr, ...)
!11 = !{i32 990000, i64 10000, i32 7}
!12 = !{i32 999999, i64 1, i32 9}
!13 = !{!"function_entry_count", i64 100000}
-!14 = !{!"branch_weights", i32 50000, i32 10000, i32 10000, i32 10000, i32 10000, i32 10000}
+!14 = !{!"branch_weights", i32 60000, i32 20000, i32 20000}
!15 = !{!"function_entry_count", i64 1}
!16 = !{!"branch_weights", i32 1, i32 0, i32 0, i32 0, i32 0, i32 0}
+!17 = !{!"branch_weights", i32 1, i32 99999}
+!18 = !{!"branch_weights", i32 99998, i32 1}
+!19 = !{!"branch_weights", i32 97000, i32 1000, i32 1000, i32 1000}
>From 8d3a985df083bf766d28e089ce3f7dcab2b53b00 Mon Sep 17 00:00:00 2001
From: mingmingl <mingmingl at google.com>
Date: Fri, 10 Jan 2025 16:31:42 -0800
Subject: [PATCH 2/6] Emit jump table with section suffix
---
llvm/include/llvm/CodeGen/AsmPrinter.h | 8 +-
.../CodeGen/TargetLoweringObjectFileImpl.h | 3 +
.../llvm/Target/TargetLoweringObjectFile.h | 5 +
llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp | 108 +++++++++++++-----
.../CodeGen/TargetLoweringObjectFileImpl.cpp | 29 +++--
llvm/lib/CodeGen/TargetPassConfig.cpp | 2 +-
llvm/lib/Target/TargetLoweringObjectFile.cpp | 6 +
llvm/test/CodeGen/X86/jump-table-partition.ll | 59 ++++++----
8 files changed, 159 insertions(+), 61 deletions(-)
diff --git a/llvm/include/llvm/CodeGen/AsmPrinter.h b/llvm/include/llvm/CodeGen/AsmPrinter.h
index c9a88d7b1c015c..4a7de22b844a65 100644
--- a/llvm/include/llvm/CodeGen/AsmPrinter.h
+++ b/llvm/include/llvm/CodeGen/AsmPrinter.h
@@ -453,6 +453,10 @@ class AsmPrinter : public MachineFunctionPass {
/// function to the current output stream.
virtual void emitJumpTableInfo();
+ virtual void emitJumpTables(ArrayRef<unsigned> JumpTableIndices,
+ MCSection *JumpTableSection, bool JTInDiffSection,
+ const MachineJumpTableInfo &MJTI);
+
/// Emit the specified global variable to the .s file.
virtual void emitGlobalVariable(const GlobalVariable *GV);
@@ -892,10 +896,10 @@ class AsmPrinter : public MachineFunctionPass {
// Internal Implementation Details
//===------------------------------------------------------------------===//
- void emitJumpTableEntry(const MachineJumpTableInfo *MJTI,
+ void emitJumpTableEntry(const MachineJumpTableInfo &MJTI,
const MachineBasicBlock *MBB, unsigned uid) const;
- void emitJumpTableSizesSection(const MachineJumpTableInfo *MJTI,
+ void emitJumpTableSizesSection(const MachineJumpTableInfo &MJTI,
const Function &F) const;
void emitLLVMUsedList(const ConstantArray *InitList);
diff --git a/llvm/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h b/llvm/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h
index a2a9e5d499e527..3d48d380fcb245 100644
--- a/llvm/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h
+++ b/llvm/include/llvm/CodeGen/TargetLoweringObjectFileImpl.h
@@ -74,6 +74,9 @@ class TargetLoweringObjectFileELF : public TargetLoweringObjectFile {
MCSection *getSectionForJumpTable(const Function &F,
const TargetMachine &TM) const override;
+ MCSection *
+ getSectionForJumpTable(const Function &F, const TargetMachine &TM,
+ const MachineJumpTableEntry *JTE) const override;
MCSection *getSectionForLSDA(const Function &F, const MCSymbol &FnSym,
const TargetMachine &TM) const override;
diff --git a/llvm/include/llvm/Target/TargetLoweringObjectFile.h b/llvm/include/llvm/Target/TargetLoweringObjectFile.h
index 4864ba843f4886..577adc458fcbf1 100644
--- a/llvm/include/llvm/Target/TargetLoweringObjectFile.h
+++ b/llvm/include/llvm/Target/TargetLoweringObjectFile.h
@@ -27,6 +27,7 @@ class Function;
class GlobalObject;
class GlobalValue;
class MachineBasicBlock;
+class MachineJumpTableEntry;
class MachineModuleInfo;
class Mangler;
class MCContext;
@@ -132,6 +133,10 @@ class TargetLoweringObjectFile : public MCObjectFileInfo {
virtual MCSection *getSectionForJumpTable(const Function &F,
const TargetMachine &TM) const;
+ virtual MCSection *
+ getSectionForJumpTable(const Function &F, const TargetMachine &TM,
+ const MachineJumpTableEntry *JTE) const;
+
virtual MCSection *getSectionForLSDA(const Function &, const MCSymbol &,
const TargetMachine &) const {
return LSDASection;
diff --git a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
index d34fe0e86c7495..208c812ac493e1 100644
--- a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
@@ -168,6 +168,11 @@ static cl::opt<bool> BBAddrMapSkipEmitBBEntries(
"unnecessary for some PGOAnalysisMap features."),
cl::Hidden, cl::init(false));
+static cl::opt<bool>
+ EmitStaticDataHotnessSuffix("emit-static-data-hotness-suffix", cl::Hidden,
+ cl::init(false), cl::ZeroOrMore,
+ cl::desc("Emit static data hotness suffix"));
+
static cl::opt<bool> EmitJumpTableSizesSection(
"emit-jump-table-sizes-section",
cl::desc("Emit a section containing jump table addresses and sizes"),
@@ -2861,7 +2866,6 @@ void AsmPrinter::emitConstantPool() {
// Print assembly representations of the jump tables used by the current
// function.
void AsmPrinter::emitJumpTableInfo() {
- const DataLayout &DL = MF->getDataLayout();
const MachineJumpTableInfo *MJTI = MF->getJumpTableInfo();
if (!MJTI) return;
if (MJTI->getEntryKind() == MachineJumpTableInfo::EK_Inline) return;
@@ -2876,42 +2880,94 @@ void AsmPrinter::emitJumpTableInfo() {
MJTI->getEntryKind() == MachineJumpTableInfo::EK_LabelDifference32 ||
MJTI->getEntryKind() == MachineJumpTableInfo::EK_LabelDifference64,
F);
+
+ std::vector<unsigned> JumpTableIndices;
+ if (!EmitStaticDataHotnessSuffix) {
+ for (unsigned JTI = 0, JTSize = JT.size(); JTI < JTSize; ++JTI)
+ JumpTableIndices.push_back(JTI);
+ emitJumpTables(JumpTableIndices, TLOF.getSectionForJumpTable(F, TM),
+ JTInDiffSection, *MJTI);
+ return;
+ }
+
+ // Iterate all jump tables, put hot jump table indices towards the beginning
+ // of the vector, and cold jump table indices towards the end.
+ int NextHotJumpTableIndex = 0, NextColdJumpTableIndex = JT.size() - 1;
+ JumpTableIndices.resize(JT.size());
+ for (unsigned JTI = 0, JTSize = JT.size(); JTI < JTSize; ++JTI) {
+ if (JT[JTI].Hotness == MachineFunctionDataHotness::Cold)
+ JumpTableIndices[NextColdJumpTableIndex--] = JTI;
+ else
+ JumpTableIndices[NextHotJumpTableIndex++] = JTI;
+ }
+
+ if (NextHotJumpTableIndex != 0) {
+ emitJumpTables(
+ ArrayRef<unsigned>(JumpTableIndices).take_front(NextHotJumpTableIndex),
+ TLOF.getSectionForJumpTable(F, TM, &JT[0]), JTInDiffSection, *MJTI);
+ }
+
+ if (NextHotJumpTableIndex != (int)JT.size()) {
+ // Retain the relative orders of original jump tables.
+ for (int L = NextHotJumpTableIndex, R = JT.size() - 1; L < R; ++L, --R)
+ std::swap(JumpTableIndices[L], JumpTableIndices[R]);
+
+ emitJumpTables(
+ ArrayRef<unsigned>(JumpTableIndices)
+ .take_back(JT.size() - NextHotJumpTableIndex),
+ TLOF.getSectionForJumpTable(F, TM, &JT[JumpTableIndices[NextHotJumpTableIndex]]),
+ JTInDiffSection, *MJTI);
+ }
+
+ return;
+}
+
+void AsmPrinter::emitJumpTables(ArrayRef<unsigned> JumpTableIndices,
+ MCSection *JumpTableSection,
+ bool JTInDiffSection,
+ const MachineJumpTableInfo &MJTI) {
+ if (JumpTableIndices.empty())
+ return;
+
+ const DataLayout &DL = MF->getDataLayout();
if (JTInDiffSection) {
- // Drop it in the readonly section.
- MCSection *ReadOnlySection = TLOF.getSectionForJumpTable(F, TM);
- OutStreamer->switchSection(ReadOnlySection);
+ OutStreamer->switchSection(JumpTableSection);
}
- emitAlignment(Align(MJTI->getEntryAlignment(DL)));
+ emitAlignment(Align(MJTI.getEntryAlignment(MF->getDataLayout())));
// Jump tables in code sections are marked with a data_region directive
// where that's supported.
if (!JTInDiffSection)
OutStreamer->emitDataRegion(MCDR_DataRegionJT32);
- for (unsigned JTI = 0, e = JT.size(); JTI != e; ++JTI) {
- const std::vector<MachineBasicBlock*> &JTBBs = JT[JTI].MBBs;
+ const auto &JT = MJTI.getJumpTables();
+ for (unsigned Index = 0, e = JumpTableIndices.size(); Index != e; ++Index) {
+ const std::vector<MachineBasicBlock *> &JTBBs =
+ JT[JumpTableIndices[Index]].MBBs;
// If this jump table was deleted, ignore it.
- if (JTBBs.empty()) continue;
+ if (JTBBs.empty())
+ continue;
// For the EK_LabelDifference32 entry, if using .set avoids a relocation,
/// emit a .set directive for each unique entry.
- if (MJTI->getEntryKind() == MachineJumpTableInfo::EK_LabelDifference32 &&
+ if (MJTI.getEntryKind() == MachineJumpTableInfo::EK_LabelDifference32 &&
MAI->doesSetDirectiveSuppressReloc()) {
- SmallPtrSet<const MachineBasicBlock*, 16> EmittedSets;
+ SmallPtrSet<const MachineBasicBlock *, 16> EmittedSets;
const TargetLowering *TLI = MF->getSubtarget().getTargetLowering();
- const MCExpr *Base = TLI->getPICJumpTableRelocBaseExpr(MF,JTI,OutContext);
+ const MCExpr *Base = TLI->getPICJumpTableRelocBaseExpr(
+ MF, JumpTableIndices[Index], OutContext);
for (const MachineBasicBlock *MBB : JTBBs) {
if (!EmittedSets.insert(MBB).second)
continue;
// .set LJTSet, LBB32-base
const MCExpr *LHS =
- MCSymbolRefExpr::create(MBB->getSymbol(), OutContext);
- OutStreamer->emitAssignment(GetJTSetSymbol(JTI, MBB->getNumber()),
- MCBinaryExpr::createSub(LHS, Base,
- OutContext));
+ MCSymbolRefExpr::create(MBB->getSymbol(), OutContext);
+ OutStreamer->emitAssignment(
+ GetJTSetSymbol(JumpTableIndices[Index], MBB->getNumber()),
+ MCBinaryExpr::createSub(LHS, Base, OutContext));
}
}
@@ -2923,27 +2979,27 @@ void AsmPrinter::emitJumpTableInfo() {
// FIXME: This doesn't have to have any specific name, just any randomly
// named and numbered local label started with 'l' would work. Simplify
// GetJTISymbol.
- OutStreamer->emitLabel(GetJTISymbol(JTI, true));
+ OutStreamer->emitLabel(GetJTISymbol(JumpTableIndices[Index], true));
- MCSymbol* JTISymbol = GetJTISymbol(JTI);
+ MCSymbol *JTISymbol = GetJTISymbol(JumpTableIndices[Index]);
OutStreamer->emitLabel(JTISymbol);
// Defer MCAssembler based constant folding due to a performance issue. The
// label differences will be evaluated at write time.
for (const MachineBasicBlock *MBB : JTBBs)
- emitJumpTableEntry(MJTI, MBB, JTI);
+ emitJumpTableEntry(MJTI, MBB, JumpTableIndices[Index]);
}
if (EmitJumpTableSizesSection)
- emitJumpTableSizesSection(MJTI, F);
+ emitJumpTableSizesSection(MJTI, MF->getFunction());
if (!JTInDiffSection)
OutStreamer->emitDataRegion(MCDR_DataRegionEnd);
}
-void AsmPrinter::emitJumpTableSizesSection(const MachineJumpTableInfo *MJTI,
+void AsmPrinter::emitJumpTableSizesSection(const MachineJumpTableInfo &MJTI,
const Function &F) const {
- const std::vector<MachineJumpTableEntry> &JT = MJTI->getJumpTables();
+ const std::vector<MachineJumpTableEntry> &JT = MJTI.getJumpTables();
if (JT.empty())
return;
@@ -2991,17 +3047,17 @@ void AsmPrinter::emitJumpTableSizesSection(const MachineJumpTableInfo *MJTI,
/// EmitJumpTableEntry - Emit a jump table entry for the specified MBB to the
/// current stream.
-void AsmPrinter::emitJumpTableEntry(const MachineJumpTableInfo *MJTI,
+void AsmPrinter::emitJumpTableEntry(const MachineJumpTableInfo &MJTI,
const MachineBasicBlock *MBB,
unsigned UID) const {
assert(MBB && MBB->getNumber() >= 0 && "Invalid basic block");
const MCExpr *Value = nullptr;
- switch (MJTI->getEntryKind()) {
+ switch (MJTI.getEntryKind()) {
case MachineJumpTableInfo::EK_Inline:
llvm_unreachable("Cannot emit EK_Inline jump table entry");
case MachineJumpTableInfo::EK_Custom32:
Value = MF->getSubtarget().getTargetLowering()->LowerCustomJumpTableEntry(
- MJTI, MBB, UID, OutContext);
+ &MJTI, MBB, UID, OutContext);
break;
case MachineJumpTableInfo::EK_BlockAddress:
// EK_BlockAddress - Each entry is a plain address of block, e.g.:
@@ -3035,7 +3091,7 @@ void AsmPrinter::emitJumpTableEntry(const MachineJumpTableInfo *MJTI,
// If the .set directive avoids relocations, this is emitted as:
// .set L4_5_set_123, LBB123 - LJTI1_2
// .word L4_5_set_123
- if (MJTI->getEntryKind() == MachineJumpTableInfo::EK_LabelDifference32 &&
+ if (MJTI.getEntryKind() == MachineJumpTableInfo::EK_LabelDifference32 &&
MAI->doesSetDirectiveSuppressReloc()) {
Value = MCSymbolRefExpr::create(GetJTSetSymbol(UID, MBB->getNumber()),
OutContext);
@@ -3051,7 +3107,7 @@ void AsmPrinter::emitJumpTableEntry(const MachineJumpTableInfo *MJTI,
assert(Value && "Unknown entry kind!");
- unsigned EntrySize = MJTI->getEntrySize(getDataLayout());
+ unsigned EntrySize = MJTI.getEntrySize(getDataLayout());
OutStreamer->emitValue(Value, EntrySize);
}
diff --git a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
index be243c0e74e9db..e54acec98864bd 100644
--- a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
@@ -24,6 +24,7 @@
#include "llvm/CodeGen/BasicBlockSectionUtils.h"
#include "llvm/CodeGen/MachineBasicBlock.h"
#include "llvm/CodeGen/MachineFunction.h"
+#include "llvm/CodeGen/MachineJumpTableInfo.h"
#include "llvm/CodeGen/MachineModuleInfo.h"
#include "llvm/CodeGen/MachineModuleInfoImpls.h"
#include "llvm/IR/Comdat.h"
@@ -642,9 +643,11 @@ static StringRef getSectionPrefixForGlobal(SectionKind Kind, bool IsLarge) {
static SmallString<128>
getELFSectionNameForGlobal(const GlobalObject *GO, SectionKind Kind,
Mangler &Mang, const TargetMachine &TM,
- unsigned EntrySize, bool UniqueSectionName) {
+ unsigned EntrySize, bool UniqueSectionName,
+ const MachineJumpTableEntry *JTE) {
SmallString<128> Name =
getSectionPrefixForGlobal(Kind, TM.isLargeGlobalValue(GO));
+
if (Kind.isMergeableCString()) {
// We also need alignment here.
// FIXME: this is getting the alignment of the character, not the
@@ -663,7 +666,12 @@ getELFSectionNameForGlobal(const GlobalObject *GO, SectionKind Kind,
bool HasPrefix = false;
if (const auto *F = dyn_cast<Function>(GO)) {
- if (std::optional<StringRef> Prefix = F->getSectionPrefix()) {
+ // Jump table hotness takes precedence over its enclosing function's hotness
+ // if both are available.
+ if (JTE) {
+ if (JTE->Hotness == MachineFunctionDataHotness::Hot)
+ raw_svector_ostream(Name) << ".hot";
+ } else if (std::optional<StringRef> Prefix = F->getSectionPrefix()) {
raw_svector_ostream(Name) << '.' << *Prefix;
HasPrefix = true;
}
@@ -761,8 +769,8 @@ calcUniqueIDUpdateFlagsAndSize(const GlobalObject *GO, StringRef SectionName,
// implicitly for this symbol e.g. .rodata.str1.1, then we don't need
// to unique the section as the entry size for this symbol will be
// compatible with implicitly created sections.
- SmallString<128> ImplicitSectionNameStem =
- getELFSectionNameForGlobal(GO, Kind, Mang, TM, EntrySize, false);
+ SmallString<128> ImplicitSectionNameStem = getELFSectionNameForGlobal(
+ GO, Kind, Mang, TM, EntrySize, false, /*MJTE=*/nullptr);
if (SymbolMergeable &&
Ctx.isELFImplicitMergeableSectionNamePrefix(SectionName) &&
SectionName.starts_with(ImplicitSectionNameStem))
@@ -862,7 +870,8 @@ MCSection *TargetLoweringObjectFileELF::getExplicitSectionGlobal(
static MCSectionELF *selectELFSectionForGlobal(
MCContext &Ctx, const GlobalObject *GO, SectionKind Kind, Mangler &Mang,
const TargetMachine &TM, bool EmitUniqueSection, unsigned Flags,
- unsigned *NextUniqueID, const MCSymbolELF *AssociatedSymbol) {
+ unsigned *NextUniqueID, const MCSymbolELF *AssociatedSymbol,
+ const MachineJumpTableEntry *MJTE = nullptr) {
auto [Group, IsComdat, ExtraFlags] = getGlobalObjectInfo(GO, TM);
Flags |= ExtraFlags;
@@ -881,7 +890,7 @@ static MCSectionELF *selectELFSectionForGlobal(
}
}
SmallString<128> Name = getELFSectionNameForGlobal(
- GO, Kind, Mang, TM, EntrySize, UniqueSectionName);
+ GO, Kind, Mang, TM, EntrySize, UniqueSectionName, MJTE);
// Use 0 as the unique ID for execute-only text.
if (Kind.isExecuteOnly())
@@ -955,6 +964,12 @@ MCSection *TargetLoweringObjectFileELF::getUniqueSectionForFunction(
MCSection *TargetLoweringObjectFileELF::getSectionForJumpTable(
const Function &F, const TargetMachine &TM) const {
+ return getSectionForJumpTable(F, TM, nullptr);
+}
+
+MCSection *TargetLoweringObjectFileELF::getSectionForJumpTable(
+ const Function &F, const TargetMachine &TM,
+ const MachineJumpTableEntry *JTE) const {
// If the function can be removed, produce a unique section so that
// the table doesn't prevent the removal.
const Comdat *C = F.getComdat();
@@ -965,7 +980,7 @@ MCSection *TargetLoweringObjectFileELF::getSectionForJumpTable(
return selectELFSectionForGlobal(getContext(), &F, SectionKind::getReadOnly(),
getMangler(), TM, EmitUniqueSection,
ELF::SHF_ALLOC, &NextUniqueID,
- /* AssociatedSymbol */ nullptr);
+ /* AssociatedSymbol */ nullptr, JTE);
}
MCSection *TargetLoweringObjectFileELF::getSectionForLSDA(
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index 6a964c0910fc61..ae8379f5b9b33d 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/llvm/lib/CodeGen/TargetPassConfig.cpp
@@ -1261,9 +1261,9 @@ void TargetPassConfig::addMachinePasses() {
"performance.\n";
}
}
- addPass(createMachineFunctionSplitterPass());
if (SplitStaticData)
addPass(createStaticDataSplitterPass());
+ addPass(createMachineFunctionSplitterPass());
}
// We run the BasicBlockSections pass if either we need BB sections or BB
// address map (or both).
diff --git a/llvm/lib/Target/TargetLoweringObjectFile.cpp b/llvm/lib/Target/TargetLoweringObjectFile.cpp
index 4fe9d13d062265..f21f72ad41e98e 100644
--- a/llvm/lib/Target/TargetLoweringObjectFile.cpp
+++ b/llvm/lib/Target/TargetLoweringObjectFile.cpp
@@ -348,6 +348,12 @@ TargetLoweringObjectFile::SectionForGlobal(const GlobalObject *GO,
MCSection *TargetLoweringObjectFile::getSectionForJumpTable(
const Function &F, const TargetMachine &TM) const {
+ return getSectionForJumpTable(F, TM, nullptr);
+}
+
+MCSection *TargetLoweringObjectFile::getSectionForJumpTable(
+ const Function &F, const TargetMachine &TM,
+ const MachineJumpTableEntry *JTE) const {
Align Alignment(1);
return getSectionForConstant(F.getDataLayout(),
SectionKind::getReadOnly(), /*C=*/nullptr,
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 3b5af2c098553f..12f9b3d42afd54 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -6,6 +6,8 @@
; RUN: llc -stop-after=finalize-isel -min-jump-table-entries=2 %s -o %t.mir
; RUN: llc --run-pass=static-data-splitter -stats -x mir %t.mir -o - 2>&1 | FileCheck %s --check-prefix=STAT
+; RUN: llc -enable-split-machine-functions -split-static-data -emit-static-data-hotness-suffix=true -function-sections -min-jump-table-entries=2 -disable-block-placement %s -o - 2>&1 | FileCheck %s --check-prefix=SECTION
+
; Tests stat messages are expected.
; TODO: Update test to verify section suffixes when target-lowering and assembler changes are implemented.
; TODO: Also run static-data-splitter pass with -static-data-default-hotness=cold and check data section suffix.
@@ -14,6 +16,13 @@
; STAT-DAG: 3 static-data-splitter - Number of hot jump tables seen
; STAT-DAG: 1 static-data-splitter - Number of jump tables with unknown hotness
+; SECTION: .section .rodata.hot.foo,"a", at progbits
+; SECTION: .LJTI0_0:
+; SECTION: .LJTI0_2:
+; SECTION: .section .rodata.foo,"a", at progbits
+; SECTION: .LJTI0_1:
+; SECTION: .LJTI0_3:
+
; @foo has four jump tables, jt0, jt1, jt2 and jt3 in the input basic block
; order; jt0 and jt2 are hot, and jt1 and jt3 are cold.
;
@@ -59,29 +68,7 @@ jt0.default:
jt0.epilog:
%zero = icmp eq i32 %num, 0
- br i1 %zero, label %cold, label %hot, !prof !17
-
-cold:
- %c1 = call i32 @compute(i32 %num)
- switch i32 %c1, label %jt1.default [
- i32 1, label %jt1.bb1
- i32 2, label %jt1.bb2
- ], !prof !14
-
-jt1.bb1:
- call i32 @puts(ptr @case1)
- br label %jt1.epilog
-
-jt1.bb2:
- call i32 @puts(ptr @case2)
- br label %jt1.epilog
-
-jt1.default:
- call i32 @puts(ptr @default)
- br label %jt1.epilog
-
-jt1.epilog:
- br label %return
+ br i1 %zero, label %hot, label %cold, !prof !17
hot:
%c2 = call i32 @transform(i32 %num)
@@ -106,6 +93,28 @@ jt2.epilog:
%c2cmp = icmp ne i32 %c2, 0
br i1 %c2cmp, label %return, label %jt3.prologue, !prof !18
+cold:
+ %c1 = call i32 @compute(i32 %num)
+ switch i32 %c1, label %jt1.default [
+ i32 1, label %jt1.bb1
+ i32 2, label %jt1.bb2
+ ], !prof !14
+
+jt1.bb1:
+ call i32 @puts(ptr @case1)
+ br label %jt1.epilog
+
+jt1.bb2:
+ call i32 @puts(ptr @case2)
+ br label %jt1.epilog
+
+jt1.default:
+ call i32 @puts(ptr @default)
+ br label %jt1.epilog
+
+jt1.epilog:
+ br label %return
+
jt3.prologue:
%c3 = call i32 @cleanup(i32 %num)
switch i32 %c3, label %jt3.default [
@@ -166,7 +175,7 @@ loop.exit:
%inc = add i32 %i.04, 1
%lsr.iv.next = add i32 %lsr.iv, -1
%exitcond.not = icmp eq i32 %lsr.iv.next, 0
- br i1 %exitcond.not, label %for.cond.cleanup, label %for.body, !prof !17
+ br i1 %exitcond.not, label %for.body, label %for.cond.cleanup, !prof !17
}
define void @func_without_entry_count(i32 %num) {
@@ -217,6 +226,6 @@ declare i32 @cleanup(i32)
!14 = !{!"branch_weights", i32 60000, i32 20000, i32 20000}
!15 = !{!"function_entry_count", i64 1}
!16 = !{!"branch_weights", i32 1, i32 0, i32 0, i32 0, i32 0, i32 0}
-!17 = !{!"branch_weights", i32 1, i32 99999}
+!17 = !{!"branch_weights", i32 99999, i32 1}
!18 = !{!"branch_weights", i32 99998, i32 1}
!19 = !{!"branch_weights", i32 97000, i32 1000, i32 1000, i32 1000}
>From 27ef86dec4a69cbc89f9856e317fbdf65bf9f8d6 Mon Sep 17 00:00:00 2001
From: mingmingl <mingmingl at google.com>
Date: Tue, 14 Jan 2025 13:30:19 -0800
Subject: [PATCH 3/6] Introduce a TargetMachine option , and the command line
flag option to set it
---
llvm/include/llvm/CodeGen/CommandFlags.h | 2 +
llvm/include/llvm/Target/TargetMachine.h | 4 +
llvm/include/llvm/Target/TargetOptions.h | 3 +
llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp | 7 +-
llvm/lib/CodeGen/CommandFlags.cpp | 8 ++
.../CodeGen/TargetLoweringObjectFileImpl.cpp | 2 +-
llvm/lib/CodeGen/TargetPassConfig.cpp | 2 +-
llvm/test/CodeGen/X86/jump-table-partition.ll | 74 +++++++------------
8 files changed, 46 insertions(+), 56 deletions(-)
diff --git a/llvm/include/llvm/CodeGen/CommandFlags.h b/llvm/include/llvm/CodeGen/CommandFlags.h
index d5448d781363d4..000aed782a8057 100644
--- a/llvm/include/llvm/CodeGen/CommandFlags.h
+++ b/llvm/include/llvm/CodeGen/CommandFlags.h
@@ -136,6 +136,8 @@ bool getEmitCallSiteInfo();
bool getEnableMachineFunctionSplitter();
+bool getEnableStaticDataPartitioning();
+
bool getEnableDebugEntryValues();
bool getValueTrackingVariableLocations();
diff --git a/llvm/include/llvm/Target/TargetMachine.h b/llvm/include/llvm/Target/TargetMachine.h
index 9bdb110bd36839..4a54c706c0cb6a 100644
--- a/llvm/include/llvm/Target/TargetMachine.h
+++ b/llvm/include/llvm/Target/TargetMachine.h
@@ -305,6 +305,10 @@ class TargetMachine {
return Options.FunctionSections;
}
+ bool getEnableStaticDataPartitioning() const {
+ return Options.EnableStaticDataPartitioning;
+ }
+
/// Return true if visibility attribute should not be emitted in XCOFF,
/// corresponding to -mignore-xcoff-visibility.
bool getIgnoreXCOFFVisibility() const {
diff --git a/llvm/include/llvm/Target/TargetOptions.h b/llvm/include/llvm/Target/TargetOptions.h
index 88f253805ca99c..1ddee265effa73 100644
--- a/llvm/include/llvm/Target/TargetOptions.h
+++ b/llvm/include/llvm/Target/TargetOptions.h
@@ -312,6 +312,9 @@ namespace llvm {
/// Enables the MachineFunctionSplitter pass.
unsigned EnableMachineFunctionSplitter : 1;
+ /// Enables the StaticDataSplitter pass.
+ unsigned EnableStaticDataPartitioning : 1;
+
/// Set if the target supports default outlining behaviour.
unsigned SupportsDefaultOutlining : 1;
diff --git a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
index 208c812ac493e1..c4629561a4267e 100644
--- a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
@@ -168,11 +168,6 @@ static cl::opt<bool> BBAddrMapSkipEmitBBEntries(
"unnecessary for some PGOAnalysisMap features."),
cl::Hidden, cl::init(false));
-static cl::opt<bool>
- EmitStaticDataHotnessSuffix("emit-static-data-hotness-suffix", cl::Hidden,
- cl::init(false), cl::ZeroOrMore,
- cl::desc("Emit static data hotness suffix"));
-
static cl::opt<bool> EmitJumpTableSizesSection(
"emit-jump-table-sizes-section",
cl::desc("Emit a section containing jump table addresses and sizes"),
@@ -2882,7 +2877,7 @@ void AsmPrinter::emitJumpTableInfo() {
F);
std::vector<unsigned> JumpTableIndices;
- if (!EmitStaticDataHotnessSuffix) {
+ if (!TM.Options.EnableStaticDataPartitioning) {
for (unsigned JTI = 0, JTSize = JT.size(); JTI < JTSize; ++JTI)
JumpTableIndices.push_back(JTI);
emitJumpTables(JumpTableIndices, TLOF.getSectionForJumpTable(F, TM),
diff --git a/llvm/lib/CodeGen/CommandFlags.cpp b/llvm/lib/CodeGen/CommandFlags.cpp
index d180cfcea658c2..023656cde0089e 100644
--- a/llvm/lib/CodeGen/CommandFlags.cpp
+++ b/llvm/lib/CodeGen/CommandFlags.cpp
@@ -103,6 +103,7 @@ CGOPT(bool, EnableStackSizeSection)
CGOPT(bool, EnableAddrsig)
CGOPT(bool, EmitCallSiteInfo)
CGOPT(bool, EnableMachineFunctionSplitter)
+CGOPT(bool, EnableStaticDataPartitioning)
CGOPT(bool, EnableDebugEntryValues)
CGOPT(bool, ForceDwarfFrameSection)
CGOPT(bool, XRayFunctionIndex)
@@ -480,6 +481,12 @@ codegen::RegisterCodeGenFlags::RegisterCodeGenFlags() {
cl::init(false));
CGBINDOPT(EnableMachineFunctionSplitter);
+ static cl::opt<bool> EnableStaticDataPartitioning(
+ "partition-static-data-sections",
+ cl::desc("Partition data sections using profile information."),
+ cl::init(false));
+ CGBINDOPT(EnableStaticDataPartitioning);
+
static cl::opt<bool> ForceDwarfFrameSection(
"force-dwarf-frame-section",
cl::desc("Always emit a debug frame section."), cl::init(false));
@@ -586,6 +593,7 @@ codegen::InitTargetOptionsFromCodeGenFlags(const Triple &TheTriple) {
Options.ExceptionModel = getExceptionModel();
Options.EmitStackSizeSection = getEnableStackSizeSection();
Options.EnableMachineFunctionSplitter = getEnableMachineFunctionSplitter();
+ Options.EnableStaticDataPartitioning = getEnableStaticDataPartitioning();
Options.EmitAddrsig = getEnableAddrsig();
Options.EmitCallSiteInfo = getEmitCallSiteInfo();
Options.EnableDebugEntryValues = getEnableDebugEntryValues();
diff --git a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
index e54acec98864bd..5e19df037290a4 100644
--- a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
@@ -974,7 +974,7 @@ MCSection *TargetLoweringObjectFileELF::getSectionForJumpTable(
// the table doesn't prevent the removal.
const Comdat *C = F.getComdat();
bool EmitUniqueSection = TM.getFunctionSections() || C;
- if (!EmitUniqueSection)
+ if (!EmitUniqueSection && !TM.getEnableStaticDataPartitioning())
return ReadOnlySection;
return selectELFSectionForGlobal(getContext(), &F, SectionKind::getReadOnly(),
diff --git a/llvm/lib/CodeGen/TargetPassConfig.cpp b/llvm/lib/CodeGen/TargetPassConfig.cpp
index ae8379f5b9b33d..3f7b2a1b52ac31 100644
--- a/llvm/lib/CodeGen/TargetPassConfig.cpp
+++ b/llvm/lib/CodeGen/TargetPassConfig.cpp
@@ -1261,7 +1261,7 @@ void TargetPassConfig::addMachinePasses() {
"performance.\n";
}
}
- if (SplitStaticData)
+ if (SplitStaticData || TM->Options.EnableStaticDataPartitioning)
addPass(createStaticDataSplitterPass());
addPass(createMachineFunctionSplitterPass());
}
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 12f9b3d42afd54..8c542beef57bb4 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -6,22 +6,36 @@
; RUN: llc -stop-after=finalize-isel -min-jump-table-entries=2 %s -o %t.mir
; RUN: llc --run-pass=static-data-splitter -stats -x mir %t.mir -o - 2>&1 | FileCheck %s --check-prefix=STAT
-; RUN: llc -enable-split-machine-functions -split-static-data -emit-static-data-hotness-suffix=true -function-sections -min-jump-table-entries=2 -disable-block-placement %s -o - 2>&1 | FileCheck %s --check-prefix=SECTION
-
-; Tests stat messages are expected.
-; TODO: Update test to verify section suffixes when target-lowering and assembler changes are implemented.
-; TODO: Also run static-data-splitter pass with -static-data-default-hotness=cold and check data section suffix.
+; When 'partition-static-data-sections' is enabled, static data splitter pass will
+; categorize jump tables and assembly printer will place hot jump tables in the
+; `.hot`-suffixed read only section, and cold ones in the `.rodata` sections.
+; Section names will optionally have `.<func>` if -function-sections is enabled.
+; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=true -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,HOT
+; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=false -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
+
+; Tests that jump tables with unknown hotness are categorized as cold if `-static-data-default-hotness` specifies so.
+; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -min-jump-table-entries=2 -static-data-default-hotness=cold -function-sections=true %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULT
+ ; Tests stat messages are expected.
; STAT-DAG: 2 static-data-splitter - Number of cold jump tables seen
-; STAT-DAG: 3 static-data-splitter - Number of hot jump tables seen
+; STAT-DAG: 2 static-data-splitter - Number of hot jump tables seen
; STAT-DAG: 1 static-data-splitter - Number of jump tables with unknown hotness
-; SECTION: .section .rodata.hot.foo,"a", at progbits
-; SECTION: .LJTI0_0:
-; SECTION: .LJTI0_2:
-; SECTION: .section .rodata.foo,"a", at progbits
-; SECTION: .LJTI0_1:
-; SECTION: .LJTI0_3:
+; Tests that the first and third jump table are placed in a hot-suffixed section,
+; and the second and fourth are placed in the original section.
+; FUNC: .section .rodata.hot.foo,"a", at progbits
+; FUNCLESS: .section .rodata.hot,"a", at progbits
+; JT: .LJTI0_0:
+; JT: .LJTI0_2:
+; FUNC: .section .rodata.foo,"a", at progbits
+; FUNCLESS: .section .rodata,"a", at progbits
+; JT: .LJTI0_1:
+; JT: .LJTI0_3:
+; HOT: .section .rodata.hot.func_without_entry_count,"a", at progbits
+; HOT: .LJTI1_0:
+
+; DEFAULT: .section .rodata.func_without_entry_count,"a", at progbits
+; DEFAULT: .LJTI1_0:
; @foo has four jump tables, jt0, jt1, jt2 and jt3 in the input basic block
; order; jt0 and jt2 are hot, and jt1 and jt3 are cold.
@@ -142,42 +156,6 @@ return:
ret i32 %mod3
}
-define i32 @func_with_hot_jt() !prof !15 {
-entry:
- br label %for.body
-
-for.cond.cleanup:
- ret i32 0
-
-for.body:
- %lsr.iv = phi i32 [ 100000, %entry ], [ %lsr.iv.next, %loop.exit ]
- %i.04 = phi i32 [ 0, %entry ], [ %inc, %loop.exit ]
- %0 = urem i32 %i.04, 100
- switch i32 %0, label %sw.default [
- i32 1, label %loop.exit
- i32 2, label %sw.bb1.i
- i32 3, label %sw.bb3.i
- ], !prof !19
-
-sw.bb1.i:
- br label %loop.exit
-
-sw.bb3.i:
- call i32 (ptr, ...) @printf(ptr @case1)
- br label %sw.default
-
-sw.default:
- br label %loop.exit
-
-loop.exit:
- %str.5.sink.i = phi ptr [ @str.10, %sw.default ], [ @str.9, %sw.bb1.i ], [ @case2, %for.body ]
- call i32 @puts(ptr %str.5.sink.i)
- %inc = add i32 %i.04, 1
- %lsr.iv.next = add i32 %lsr.iv, -1
- %exitcond.not = icmp eq i32 %lsr.iv.next, 0
- br i1 %exitcond.not, label %for.body, label %for.cond.cleanup, !prof !17
-}
-
define void @func_without_entry_count(i32 %num) {
entry:
switch i32 %num, label %sw.default [
>From e816defdec877b7c54e27fbe0e51660f67a8e074 Mon Sep 17 00:00:00 2001
From: mingmingl <mingmingl at google.com>
Date: Tue, 14 Jan 2025 17:07:46 -0800
Subject: [PATCH 4/6] Emit jump tables into .hot and .unlikely sections
respectively
---
llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp | 12 ++++++++---
.../CodeGen/TargetLoweringObjectFileImpl.cpp | 2 ++
llvm/test/CodeGen/X86/jump-table-partition.ll | 21 ++++++++++---------
3 files changed, 22 insertions(+), 13 deletions(-)
diff --git a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
index c4629561a4267e..7f80ef6e8a533f 100644
--- a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
@@ -2885,8 +2885,14 @@ void AsmPrinter::emitJumpTableInfo() {
return;
}
+ // When static data partitioning is enabled, collect jump table entries that
+ // go into the same section together to reduce the amount of section switch
+ // statements.
+ //
// Iterate all jump tables, put hot jump table indices towards the beginning
- // of the vector, and cold jump table indices towards the end.
+ // of the vector, and cold jump table indices towards the end. Meanwhile
+ // retain the relative orders of original jump tables within a hot or unlikely
+ // section by reversing the cold jump table indices.
int NextHotJumpTableIndex = 0, NextColdJumpTableIndex = JT.size() - 1;
JumpTableIndices.resize(JT.size());
for (unsigned JTI = 0, JTSize = JT.size(); JTI < JTSize; ++JTI) {
@@ -2902,8 +2908,8 @@ void AsmPrinter::emitJumpTableInfo() {
TLOF.getSectionForJumpTable(F, TM, &JT[0]), JTInDiffSection, *MJTI);
}
- if (NextHotJumpTableIndex != (int)JT.size()) {
- // Retain the relative orders of original jump tables.
+ if (NextHotJumpTableIndex < (int)JT.size()) {
+ // Reverse the order of cold jump tables indices.
for (int L = NextHotJumpTableIndex, R = JT.size() - 1; L < R; ++L, --R)
std::swap(JumpTableIndices[L], JumpTableIndices[R]);
diff --git a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
index 5e19df037290a4..2a00ecf80ac1e2 100644
--- a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
@@ -671,6 +671,8 @@ getELFSectionNameForGlobal(const GlobalObject *GO, SectionKind Kind,
if (JTE) {
if (JTE->Hotness == MachineFunctionDataHotness::Hot)
raw_svector_ostream(Name) << ".hot";
+ else if (JTE->Hotness == MachineFunctionDataHotness::Cold)
+ raw_svector_ostream(Name) << ".unlikely";
} else if (std::optional<StringRef> Prefix = F->getSectionPrefix()) {
raw_svector_ostream(Name) << '.' << *Prefix;
HasPrefix = true;
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 8c542beef57bb4..6e359c14659125 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -8,13 +8,12 @@
; When 'partition-static-data-sections' is enabled, static data splitter pass will
; categorize jump tables and assembly printer will place hot jump tables in the
-; `.hot`-suffixed read only section, and cold ones in the `.rodata` sections.
+; `.rodata.hot`-prefixed section, and cold ones in the `.rodata.unlikely`-prefixed section.
; Section names will optionally have `.<func>` if -function-sections is enabled.
-; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=true -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,HOT
+; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=true -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULTHOT
; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=false -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
-; Tests that jump tables with unknown hotness are categorized as cold if `-static-data-default-hotness` specifies so.
-; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -min-jump-table-entries=2 -static-data-default-hotness=cold -function-sections=true %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULT
+; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -min-jump-table-entries=2 -static-data-default-hotness=cold -function-sections=true %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULTCOLD
; Tests stat messages are expected.
; STAT-DAG: 2 static-data-splitter - Number of cold jump tables seen
@@ -27,15 +26,17 @@
; FUNCLESS: .section .rodata.hot,"a", at progbits
; JT: .LJTI0_0:
; JT: .LJTI0_2:
-; FUNC: .section .rodata.foo,"a", at progbits
-; FUNCLESS: .section .rodata,"a", at progbits
+; FUNC: .section .rodata.unlikely.foo,"a", at progbits
+; FUNCLESS: .section .rodata.unlikely,"a", at progbits
; JT: .LJTI0_1:
; JT: .LJTI0_3:
-; HOT: .section .rodata.hot.func_without_entry_count,"a", at progbits
-; HOT: .LJTI1_0:
+; DEFAULTHOT: .section .rodata.hot.func_without_entry_count,"a", at progbits
+; DEFAULTHOT: .LJTI1_0:
+; FUNCLESS: .section .rodata.hot,"a", at progbits
+; FUNCLESS: .LJTI1_0:
-; DEFAULT: .section .rodata.func_without_entry_count,"a", at progbits
-; DEFAULT: .LJTI1_0:
+; DEFAULTCOLD: .section .rodata.unlikely.func_without_entry_count,"a", at progbits
+; DEFAULTCOLD: .LJTI1_0:
; @foo has four jump tables, jt0, jt1, jt2 and jt3 in the input basic block
; order; jt0 and jt2 are hot, and jt1 and jt3 are cold.
>From 7f3e4731a6bcda9e6fbdcaeb9a39b1edb47bef48 Mon Sep 17 00:00:00 2001
From: mingmingl <mingmingl at google.com>
Date: Thu, 16 Jan 2025 09:09:40 -0800
Subject: [PATCH 5/6] Add -mtriple=x86_64-unknown-linux-gnu for test
---
llvm/test/CodeGen/X86/jump-table-partition.ll | 33 ++++++++++---------
1 file changed, 18 insertions(+), 15 deletions(-)
diff --git a/llvm/test/CodeGen/X86/jump-table-partition.ll b/llvm/test/CodeGen/X86/jump-table-partition.ll
index 6e359c14659125..1a03bd9cc7bb6d 100644
--- a/llvm/test/CodeGen/X86/jump-table-partition.ll
+++ b/llvm/test/CodeGen/X86/jump-table-partition.ll
@@ -3,31 +3,34 @@
; Stop after 'finalize-isel' for simpler MIR, and lower the minimum number of
; jump table entries so 'switch' needs fewer cases to generate a jump table.
-; RUN: llc -stop-after=finalize-isel -min-jump-table-entries=2 %s -o %t.mir
-; RUN: llc --run-pass=static-data-splitter -stats -x mir %t.mir -o - 2>&1 | FileCheck %s --check-prefix=STAT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -stop-after=finalize-isel -min-jump-table-entries=2 %s -o %t.mir
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu --run-pass=static-data-splitter -stats -x mir %t.mir -o - 2>&1 | FileCheck %s --check-prefix=STAT
+
+ ; Tests stat messages are expected.
+; STAT-DAG: 2 static-data-splitter - Number of cold jump tables seen
+; STAT-DAG: 2 static-data-splitter - Number of hot jump tables seen
+; STAT-DAG: 1 static-data-splitter - Number of jump tables with unknown hotness
; When 'partition-static-data-sections' is enabled, static data splitter pass will
; categorize jump tables and assembly printer will place hot jump tables in the
; `.rodata.hot`-prefixed section, and cold ones in the `.rodata.unlikely`-prefixed section.
; Section names will optionally have `.<func>` if -function-sections is enabled.
-; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=true -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULTHOT
-; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -function-sections=false -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions -partition-static-data-sections=true -function-sections=true -min-jump-table-entries=2 -unique-section-names=false %s -o - 2>&1 | FileCheck %s --check-prefixes=LINEAR,JT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions -partition-static-data-sections=true -function-sections=true -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULTHOT
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions -partition-static-data-sections=true -function-sections=false -min-jump-table-entries=2 %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNCLESS,JT --implicit-check-not=unique
-; RUN: llc -enable-split-machine-functions -partition-static-data-sections=true -min-jump-table-entries=2 -static-data-default-hotness=cold -function-sections=true %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULTCOLD
-
- ; Tests stat messages are expected.
-; STAT-DAG: 2 static-data-splitter - Number of cold jump tables seen
-; STAT-DAG: 2 static-data-splitter - Number of hot jump tables seen
-; STAT-DAG: 1 static-data-splitter - Number of jump tables with unknown hotness
+; Tests that `-static-data-default-hotness` can override hotness for data with
+; unknown hotness.
+; RUN: llc -mtriple=x86_64-unknown-linux-gnu -enable-split-machine-functions -partition-static-data-sections=true -min-jump-table-entries=2 -static-data-default-hotness=cold -function-sections=true %s -o - 2>&1 | FileCheck %s --check-prefixes=FUNC,JT,DEFAULTCOLD
-; Tests that the first and third jump table are placed in a hot-suffixed section,
-; and the second and fourth are placed in the original section.
-; FUNC: .section .rodata.hot.foo,"a", at progbits
+; LINEAR: .section .rodata.hot,"a", at progbits,unique,2
+; FUNC: .section .rodata.hot.foo,"a", at progbits
; FUNCLESS: .section .rodata.hot,"a", at progbits
; JT: .LJTI0_0:
; JT: .LJTI0_2:
-; FUNC: .section .rodata.unlikely.foo,"a", at progbits
-; FUNCLESS: .section .rodata.unlikely,"a", at progbits
+; LINEAR: .section .rodata.unlikely,"a", at progbits,unique,3
+; FUNC: .section .rodata.unlikely.foo,"a", at progbits
+; FUNCLESS: .section .rodata.unlikely,"a", at progbits
; JT: .LJTI0_1:
; JT: .LJTI0_3:
; DEFAULTHOT: .section .rodata.hot.func_without_entry_count,"a", at progbits
>From 1bef9b1b6ed5d94e159d8168dfe34e93d2064b73 Mon Sep 17 00:00:00 2001
From: mingmingl <mingmingl at google.com>
Date: Thu, 16 Jan 2025 09:21:11 -0800
Subject: [PATCH 6/6] run clang format
---
llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
index 7f80ef6e8a533f..4730c33d748832 100644
--- a/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
+++ b/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
@@ -2916,7 +2916,8 @@ void AsmPrinter::emitJumpTableInfo() {
emitJumpTables(
ArrayRef<unsigned>(JumpTableIndices)
.take_back(JT.size() - NextHotJumpTableIndex),
- TLOF.getSectionForJumpTable(F, TM, &JT[JumpTableIndices[NextHotJumpTableIndex]]),
+ TLOF.getSectionForJumpTable(
+ F, TM, &JT[JumpTableIndices[NextHotJumpTableIndex]]),
JTInDiffSection, *MJTI);
}
More information about the llvm-branch-commits
mailing list