Teach LiveDebugValues about lexical scopes.

This addresses PR26055 LiveDebugValues is very slow.

Contrary to the old LiveDebugVariables pass LiveDebugValues currently
doesn't look at the lexical scopes before inserting a DBG_VALUE
intrinsic. This means that we often propagate DBG_VALUEs much further
down than necessary. This is especially noticeable in large C++
functions with many inlined method calls that all use the same
"this"-pointer.

For example, in the following code it makes no sense to propagate the
inlined variable a from the first inlined call to f() into any of the
subsequent basic blocks, because the variable will always be out of
scope:

void sink(int a);
void __attribute((always_inline)) f(int a) { sink(a); }
void foo(int i) {
   f(i);
   if (i)
     f(i);
   f(i);
}

This patch reuses the LexicalScopes infrastructure we have for
LiveDebugVariables to take this into account.

The effect on compile time and memory consumption is quite noticeable:
I tested a benchmark that is a large C++ source with an enormous
amount of inlined "this"-pointers that would previously eat >24GiB
(most of them for DBG_VALUE intrinsics) and whose compile time was
dominated by LiveDebugValues. With this patch applied the memory
consumption is 1GiB and 1.7% of the time is spent in LiveDebugValues.

https://reviews.llvm.org/D24994
Thanks to Daniel Berlin and Keith Walker for reviewing!

llvm-svn: 282611
This commit is contained in:
Adrian Prantl 2016-09-28 17:51:14 +00:00
parent d4d9dc8252
commit 7f5866c227
5 changed files with 278 additions and 16 deletions

View File

@ -23,6 +23,7 @@
#include "llvm/ADT/SparseBitVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/ADT/UniqueVector.h"
#include "llvm/CodeGen/LexicalScopes.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
@ -60,6 +61,26 @@ class LiveDebugValues : public MachineFunctionPass {
private:
const TargetRegisterInfo *TRI;
const TargetInstrInfo *TII;
LexicalScopes LS;
/// Keeps track of lexical scopes associated with a user value's source
/// location.
class UserValueScopes {
DebugLoc DL;
LexicalScopes &LS;
SmallPtrSet<const MachineBasicBlock *, 4> LBlocks;
public:
UserValueScopes(DebugLoc D, LexicalScopes &L) : DL(std::move(D)), LS(L) {}
/// Return true if current scope dominates at least one machine
/// instruction in a given machine basic block.
bool dominates(MachineBasicBlock *MBB) {
if (LBlocks.empty())
LS.getMachineBasicBlocks(DL, LBlocks);
return LBlocks.count(MBB) != 0 || LS.dominates(DL, MBB);
}
};
/// Based on std::pair so it can be used as an index into a DenseMap.
typedef std::pair<const DILocalVariable *, const DILocation *>
@ -83,7 +104,7 @@ private:
struct VarLoc {
const DebugVariable Var;
const MachineInstr &MI; ///< Only used for cloning a new DBG_VALUE.
mutable UserValueScopes UVS;
enum { InvalidKind = 0, RegisterKind } Kind;
/// The value location. Stored separately to avoid repeatedly
@ -96,9 +117,9 @@ private:
uint64_t Hash;
} Loc;
VarLoc(const MachineInstr &MI)
VarLoc(const MachineInstr &MI, LexicalScopes &LS)
: Var(MI.getDebugVariable(), MI.getDebugLoc()->getInlinedAt()), MI(MI),
Kind(InvalidKind) {
UVS(MI.getDebugLoc(), LS), Kind(InvalidKind) {
static_assert((sizeof(Loc) == sizeof(uint64_t)),
"hash does not cover all members of Loc");
assert(MI.isDebugValue() && "not a DBG_VALUE");
@ -125,6 +146,10 @@ private:
return 0;
}
/// Determine whether the lexical scope of this value's debug location
/// dominates MBB.
bool dominates(MachineBasicBlock &MBB) const { return UVS.dominates(&MBB); }
void dump() const { MI.dump(); }
bool operator==(const VarLoc &Other) const {
@ -229,6 +254,7 @@ public:
/// Calculate the liveness information for the given machine function.
bool runOnMachineFunction(MachineFunction &MF) override;
};
} // namespace
//===----------------------------------------------------------------------===//
@ -295,7 +321,7 @@ void LiveDebugValues::transferDebugValue(const MachineInstr &MI,
// Add the VarLoc to OpenRanges from this DBG_VALUE.
// TODO: Currently handles DBG_VALUE which has only reg as location.
if (isDbgValueDescribedByReg(MI)) {
VarLoc VL(MI);
VarLoc VL(MI, LS);
unsigned ID = VarLocIDs.insert(VL);
OpenRanges.insert(ID, VL.Var);
}
@ -399,6 +425,13 @@ bool LiveDebugValues::join(MachineBasicBlock &MBB, VarLocInMBB &OutLocs,
NumVisited++;
}
// Filter out DBG_VALUES that are out of scope.
VarLocSet KillSet;
for (auto ID : InLocsT)
if (!VarLocIDs[ID].dominates(MBB))
KillSet.set(ID);
InLocsT.intersectWithComplement(KillSet);
// As we are processing blocks in reverse post-order we
// should have processed at least one predecessor, unless it
// is the entry block which has no predecessor.
@ -520,12 +553,14 @@ bool LiveDebugValues::ExtendRanges(MachineFunction &MF) {
}
bool LiveDebugValues::runOnMachineFunction(MachineFunction &MF) {
if (!MF.getFunction()->getSubprogram())
// LiveDebugValues will already have removed all DBG_VALUEs.
return false;
TRI = MF.getSubtarget().getRegisterInfo();
TII = MF.getSubtarget().getInstrInfo();
LS.initialize(MF);
bool Changed = false;
Changed |= ExtendRanges(MF);
bool Changed = ExtendRanges(MF);
return Changed;
}

View File

@ -47,7 +47,6 @@
# CHECK: DBG_VALUE debug-use [[REG_I]], debug-use _, [[VAR_I]]
# CHECK: bb.4.for.cond.cleanup
# CHECK: DBG_VALUE debug-use [[REG_I]], debug-use _, [[VAR_I]]
# CHECK: DBG_VALUE debug-use [[REG_C]], debug-use _, [[VAR_C]]
# CHECK: DBG_VALUE debug-use [[REG_B]], debug-use _, [[VAR_B]]
# CHECK: DBG_VALUE debug-use [[REG_A]], debug-use _, [[VAR_A]]

View File

@ -37,8 +37,8 @@
; ASM: #DEBUG_VALUE: c <- %EAX
; ASM: testl %esi, %esi
; ASM: je .LBB0_2
; ASM: [[after_je:\.Ltmp.*]]:
; ASM: # BB#1: # %if.then
; ASM-DAG: #DEBUG_VALUE: c <- %EAX
; ASM-DAG: #DEBUG_VALUE: inlineinc:a <- %EAX
; ASM-DAG: #DEBUG_VALUE: a <- %EAX
; ASM-DAG: #DEBUG_VALUE: f:p <- %ESI
@ -65,7 +65,7 @@
; ASM: .cv_def_range [[after_getint]] [[after_inc_eax]], "A\021\021\000\000\000"
; ASM: .short 4414 # Record kind: S_LOCAL
; ASM: .asciz "c"
; ASM: .cv_def_range [[after_getint]] [[after_inc_eax]], "A\021\021\000\000\000"
; ASM: .cv_def_range [[after_getint]] [[after_je]], "A\021\021\000\000\000"
; ASM: .short 4414 # Record kind: S_LOCAL
; ASM: .asciz "b"
; ASM: .cv_def_range [[after_inc_eax]] [[after_if]], "A\021\021\000\000\000"
@ -132,7 +132,7 @@
; OBJ: LocalVariableAddrRange {
; OBJ: OffsetStart: .text+0xC
; OBJ: ISectStart: 0x0
; OBJ: Range: 0x6
; OBJ: Range: 0x4
; OBJ: }
; OBJ: }
; OBJ: Local {
@ -143,7 +143,7 @@
; OBJ: }
; OBJ: DefRangeRegister {
; OBJ: Register: 17
; OBJ: LocalVariableAddrRange {
; OBJ: MayHaveNoName: 0
; OBJ: OffsetStart: .text+0x12
; OBJ: ISectStart: 0x0
; OBJ: Range: 0x6

View File

@ -0,0 +1,228 @@
--- |
; RUN: llc -run-pass=livedebugvalues -march=x86-64 -o - %s | FileCheck %s
; Created from:
; void sink(int a);
; void __attribute((always_inline)) f(int a) { sink(a); }
; void foo(int i) {
; f(i);
; if (i)
; f(i);
; f(i);
; }
;
; This test verifies that LiveDebugValues doesn't propagate DBG_VALUEs into
; basic blocks that are beyond the scope of the source variable.
;
; CHECK: bb.1.if.then:
; CHECK: DBG_VALUE debug-use %ebx, debug-use _, !19, !13, debug-location !20
; CHECK-NOT: DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !21
; CHECK: DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !27
; CHECK: bb.2.if.end:
; CHECK: DBG_VALUE debug-use %ebx, debug-use _, !19, !13, debug-location !20
; CHECK-NOT: DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !21
; CHECK: DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !31
;
; ModuleID = 'livedebugvalues-limit.ll'
source_filename = "livedebugvalues-limit.c"
target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-apple-macosx"
; Function Attrs: alwaysinline nounwind ssp uwtable
define void @f(i32 %a) local_unnamed_addr #0 !dbg !7 {
entry:
tail call void @llvm.dbg.value(metadata i32 %a, i64 0, metadata !12, metadata !13), !dbg !14
tail call void @sink(i32 %a) #4, !dbg !15
ret void, !dbg !16
}
declare void @sink(i32) local_unnamed_addr
; Function Attrs: nounwind ssp uwtable
define void @foo(i32 %i) local_unnamed_addr #2 !dbg !17 {
entry:
tail call void @llvm.dbg.value(metadata i32 %i, i64 0, metadata !19, metadata !13), !dbg !20
tail call void @llvm.dbg.value(metadata i32 %i, i64 0, metadata !12, metadata !13) #4, !dbg !21
tail call void @sink(i32 %i) #4, !dbg !23
%tobool = icmp eq i32 %i, 0, !dbg !24
br i1 %tobool, label %if.end, label %if.then, !dbg !26
if.then: ; preds = %entry
tail call void @llvm.dbg.value(metadata i32 %i, i64 0, metadata !12, metadata !13) #4, !dbg !27
tail call void @sink(i32 %i) #4, !dbg !29
br label %if.end, !dbg !30
if.end: ; preds = %if.then, %entry
tail call void @llvm.dbg.value(metadata i32 %i, i64 0, metadata !12, metadata !13) #4, !dbg !31
tail call void @sink(i32 %i) #4, !dbg !33
ret void, !dbg !34
}
; Function Attrs: nounwind readnone
declare void @llvm.dbg.value(metadata, i64, metadata, metadata) #3
; Function Attrs: nounwind
declare void @llvm.stackprotector(i8*, i8**) #4
attributes #0 = { alwaysinline nounwind ssp uwtable }
attributes #2 = { nounwind ssp uwtable }
attributes #3 = { nounwind readnone }
attributes #4 = { nounwind }
!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3, !4, !5}
!llvm.ident = !{!6}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 4.0.0 (trunk 281923) (llvm/trunk 281916)", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !2)
!1 = !DIFile(filename: "livedebugvalues-limit.c", directory: "/Volumes/Fusion/Data/llvm")
!2 = !{}
!3 = !{i32 2, !"Dwarf Version", i32 4}
!4 = !{i32 2, !"Debug Info Version", i32 3}
!5 = !{i32 1, !"PIC Level", i32 2}
!6 = !{!"clang version 4.0.0 (trunk 281923) (llvm/trunk 281916)"}
!7 = distinct !DISubprogram(name: "f", scope: !1, file: !1, line: 3, type: !8, isLocal: false, isDefinition: true, scopeLine: 3, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !11)
!8 = !DISubroutineType(types: !9)
!9 = !{null, !10}
!10 = !DIBasicType(name: "int", size: 32, align: 32, encoding: DW_ATE_signed)
!11 = !{!12}
!12 = !DILocalVariable(name: "a", arg: 1, scope: !7, file: !1, line: 3, type: !10)
!13 = !DIExpression()
!14 = !DILocation(line: 3, column: 41, scope: !7)
!15 = !DILocation(line: 3, column: 46, scope: !7)
!16 = !DILocation(line: 3, column: 55, scope: !7)
!17 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 4, type: !8, isLocal: false, isDefinition: true, scopeLine: 4, flags: DIFlagPrototyped, isOptimized: true, unit: !0, variables: !18)
!18 = !{!19}
!19 = !DILocalVariable(name: "i", arg: 1, scope: !17, file: !1, line: 4, type: !10)
!20 = !DILocation(line: 4, column: 14, scope: !17)
!21 = !DILocation(line: 3, column: 41, scope: !7, inlinedAt: !22)
!22 = distinct !DILocation(line: 5, column: 3, scope: !17)
!23 = !DILocation(line: 3, column: 46, scope: !7, inlinedAt: !22)
!24 = !DILocation(line: 6, column: 7, scope: !25)
!25 = distinct !DILexicalBlock(scope: !17, file: !1, line: 6, column: 7)
!26 = !DILocation(line: 6, column: 7, scope: !17)
!27 = !DILocation(line: 3, column: 41, scope: !7, inlinedAt: !28)
!28 = distinct !DILocation(line: 7, column: 5, scope: !25)
!29 = !DILocation(line: 3, column: 46, scope: !7, inlinedAt: !28)
!30 = !DILocation(line: 7, column: 5, scope: !25)
!31 = !DILocation(line: 3, column: 41, scope: !7, inlinedAt: !32)
!32 = distinct !DILocation(line: 8, column: 3, scope: !17)
!33 = !DILocation(line: 3, column: 46, scope: !7, inlinedAt: !32)
!34 = !DILocation(line: 9, column: 1, scope: !17)
...
---
name: f
alignment: 4
exposesReturnsTwice: false
legalized: false
regBankSelected: false
selected: false
tracksRegLiveness: true
liveins:
- { reg: '%edi' }
calleeSavedRegisters: [ '%bh', '%bl', '%bp', '%bpl', '%bx', '%ebp', '%ebx',
'%rbp', '%rbx', '%r12', '%r13', '%r14', '%r15',
'%r12b', '%r13b', '%r14b', '%r15b', '%r12d', '%r13d',
'%r14d', '%r15d', '%r12w', '%r13w', '%r14w', '%r15w' ]
frameInfo:
isFrameAddressTaken: false
isReturnAddressTaken: false
hasStackMap: false
hasPatchPoint: false
stackSize: 8
offsetAdjustment: 0
maxAlignment: 0
adjustsStack: false
hasCalls: false
maxCallFrameSize: 0
hasOpaqueSPAdjustment: false
hasVAStart: false
hasMustTailInVarArgFunc: false
fixedStack:
- { id: 0, type: spill-slot, offset: -16, size: 8, alignment: 16 }
body: |
bb.0.entry:
liveins: %edi, %rbp
frame-setup PUSH64r killed %rbp, implicit-def %rsp, implicit %rsp
CFI_INSTRUCTION def_cfa_offset 16
CFI_INSTRUCTION offset %rbp, -16
%rbp = frame-setup MOV64rr %rsp
CFI_INSTRUCTION def_cfa_register %rbp
DBG_VALUE debug-use %edi, debug-use _, !12, !13, debug-location !14
%rbp = POP64r implicit-def %rsp, implicit %rsp, debug-location !15
TAILJMPd64 @sink, csr_64, implicit %rsp, implicit %rsp, implicit %edi, debug-location !15
...
---
name: foo
alignment: 4
exposesReturnsTwice: false
legalized: false
regBankSelected: false
selected: false
tracksRegLiveness: true
liveins:
- { reg: '%edi' }
calleeSavedRegisters: [ '%bh', '%bl', '%bp', '%bpl', '%bx', '%ebp', '%ebx',
'%rbp', '%rbx', '%r12', '%r13', '%r14', '%r15',
'%r12b', '%r13b', '%r14b', '%r15b', '%r12d', '%r13d',
'%r14d', '%r15d', '%r12w', '%r13w', '%r14w', '%r15w' ]
frameInfo:
isFrameAddressTaken: false
isReturnAddressTaken: false
hasStackMap: false
hasPatchPoint: false
stackSize: 24
offsetAdjustment: -8
maxAlignment: 0
adjustsStack: true
hasCalls: true
maxCallFrameSize: 0
hasOpaqueSPAdjustment: false
hasVAStart: false
hasMustTailInVarArgFunc: false
fixedStack:
- { id: 0, type: spill-slot, offset: -24, size: 8, alignment: 8, callee-saved-register: '%rbx' }
- { id: 1, type: spill-slot, offset: -16, size: 8, alignment: 16 }
body: |
bb.0.entry:
successors: %bb.2.if.end, %bb.1.if.then
liveins: %edi, %rbx, %rbp
frame-setup PUSH64r killed %rbp, implicit-def %rsp, implicit %rsp
CFI_INSTRUCTION def_cfa_offset 16
CFI_INSTRUCTION offset %rbp, -16
%rbp = frame-setup MOV64rr %rsp
CFI_INSTRUCTION def_cfa_register %rbp
frame-setup PUSH64r killed %rbx, implicit-def %rsp, implicit %rsp
frame-setup PUSH64r undef %rax, implicit-def %rsp, implicit %rsp
CFI_INSTRUCTION offset %rbx, -24
DBG_VALUE debug-use %edi, debug-use _, !19, !13, debug-location !20
%ebx = MOV32rr %edi
DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !21
DBG_VALUE debug-use %ebx, debug-use _, !19, !13, debug-location !20
CALL64pcrel32 @sink, csr_64, implicit %rsp, implicit %edi, implicit-def %rsp, debug-location !23
TEST32rr %ebx, %ebx, implicit-def %eflags, debug-location !24
JE_1 %bb.2.if.end, implicit %eflags
bb.1.if.then:
successors: %bb.2.if.end
liveins: %ebx, %rbp
DBG_VALUE debug-use %ebx, debug-use _, !19, !13, debug-location !20
DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !27
%edi = MOV32rr %ebx, debug-location !29
CALL64pcrel32 @sink, csr_64, implicit %rsp, implicit %edi, implicit-def %rsp, debug-location !29
bb.2.if.end:
liveins: %ebx, %rbp
DBG_VALUE debug-use %ebx, debug-use _, !19, !13, debug-location !20
%edi = MOV32rr killed %ebx, debug-location !33
%rsp = ADD64ri8 %rsp, 8, implicit-def dead %eflags, debug-location !33
DBG_VALUE debug-use %ebx, debug-use _, !12, !13, debug-location !31
%rbx = POP64r implicit-def %rsp, implicit %rsp, debug-location !33
%rbp = POP64r implicit-def %rsp, implicit %rsp, debug-location !33
TAILJMPd64 @sink, csr_64, implicit %rsp, implicit %rsp, implicit %edi, debug-location !33
...

View File

@ -32,13 +32,13 @@
; CHECK-NEXT: Length: 25
; CHECK-NEXT: Location description: 50 93 04
; CHECK: [[E]]: Beginning address index: 4
; CHECK-NEXT: Length: 23
; CHECK-NEXT: Length: 19
; CHECK-NEXT: Location description: 50 93 04
; CHECK: [[B]]: Beginning address index: 5
; CHECK-NEXT: Length: 21
; CHECK-NEXT: Length: 17
; CHECK-NEXT: Location description: 50 93 04
; CHECK: [[D]]: Beginning address index: 6
; CHECK-NEXT: Length: 21
; CHECK-NEXT: Length: 17
; CHECK-NEXT: Location description: 50 93 04
; Make sure we don't produce any relocations in any .dwo section (though in particular, debug_info.dwo)