Skip to content

Commit 7ca32bd

Browse files
int3int3
committed
Reland "[lld-macho] Overhaul map file code"
This reverts commit 38d6202. Differential Revision: https://reviews.llvm.org/D137368
1 parent 3ed6311 commit 7ca32bd

File tree

2 files changed

+126
-95
lines changed

2 files changed

+126
-95
lines changed

‎lld/MachO/MapFile.cpp

Lines changed: 88 additions & 65 deletions
Original file line numberDiff line numberDiff line change
@@ -6,9 +6,10 @@
66
//
77
//===----------------------------------------------------------------------===//
88
//
9-
// This file implements the -map option. It shows lists in order and
10-
// hierarchically the outputFile, arch, input files, output sections and
11-
// symbols:
9+
// This file implements the -map option, which maps address ranges to their
10+
// respective contents, plus the input file these contents were originally from.
11+
// The contents (typically symbols) are listed in address order. Dead-stripped
12+
// contents are included as well.
1213
//
1314
// # Path: test
1415
// # Arch: x86_84
@@ -28,15 +29,16 @@
2829
//===----------------------------------------------------------------------===//
2930

3031
#include "MapFile.h"
32+
#include "ConcatOutputSection.h"
3133
#include "Config.h"
3234
#include "InputFiles.h"
3335
#include "InputSection.h"
34-
#include "OutputSection.h"
3536
#include "OutputSegment.h"
3637
#include "Symbols.h"
3738
#include "SyntheticSections.h"
3839
#include "Target.h"
3940
#include "lld/Common/ErrorHandler.h"
41+
#include "llvm/ADT/DenseMap.h"
4042
#include "llvm/Support/Parallel.h"
4143
#include "llvm/Support/TimeProfiler.h"
4244

@@ -45,69 +47,75 @@ using namespace llvm::sys;
4547
using namespace lld;
4648
using namespace lld::macho;
4749

50+
struct CStringInfo {
51+
uint32_t fileIndex;
52+
StringRef str;
53+
};
54+
4855
struct MapInfo {
4956
SmallVector<InputFile *> files;
50-
SmallVector<Defined *> liveSymbols;
5157
SmallVector<Defined *> deadSymbols;
58+
DenseMap<const OutputSection *,
59+
SmallVector<std::pair<uint64_t /*addr*/, CStringInfo>>>
60+
liveCStringsForSection;
61+
SmallVector<CStringInfo> deadCStrings;
5262
};
5363

5464
static MapInfo gatherMapInfo() {
5565
MapInfo info;
5666
for (InputFile *file : inputFiles)
5767
if (isa<ObjFile>(file) || isa<BitcodeFile>(file)) {
58-
bool hasEmittedSymbol = false;
68+
uint32_t fileIndex = info.files.size() + 1;
69+
bool isReferencedFile = false;
70+
71+
// Gather the dead symbols. We don't have to bother with the live ones
72+
// because we will pick them up as we iterate over the OutputSections
73+
// later.
5974
for (Symbol *sym : file->symbols) {
6075
if (auto *d = dyn_cast_or_null<Defined>(sym))
61-
if (d->isec && d->getFile() == file) {
62-
if (d->isLive()) {
63-
assert(!shouldOmitFromOutput(d->isec));
64-
info.liveSymbols.push_back(d);
65-
} else {
76+
// Only emit the prevailing definition of a symbol. Also, don't emit
77+
// the symbol if it is part of a cstring section (we use the literal
78+
// value instead, similar to ld64)
79+
if (d->isec && d->getFile() == file &&
80+
!isa<CStringInputSection>(d->isec)) {
81+
isReferencedFile = true;
82+
if (!d->isLive())
6683
info.deadSymbols.push_back(d);
84+
}
85+
}
86+
87+
// Gather all the cstrings (both live and dead). A CString(Output)Section
88+
// doesn't provide us a way of figuring out which InputSections its
89+
// cstring contents came from, so we need to build up that mapping here.
90+
for (const Section *sec : file->sections) {
91+
for (const Subsection &subsec : sec->subsections) {
92+
if (auto isec = dyn_cast<CStringInputSection>(subsec.isec)) {
93+
auto &liveCStrings = info.liveCStringsForSection[isec->parent];
94+
for (const auto &[i, piece] : llvm::enumerate(isec->pieces)) {
95+
if (piece.live)
96+
liveCStrings.push_back({isec->parent->addr + piece.outSecOff,
97+
{fileIndex, isec->getStringRef(i)}});
98+
else
99+
info.deadCStrings.push_back({fileIndex, isec->getStringRef(i)});
100+
isReferencedFile = true;
67101
}
68-
hasEmittedSymbol = true;
102+
} else {
103+
break;
69104
}
105+
}
70106
}
71-
if (hasEmittedSymbol)
72-
info.files.push_back(file);
73-
}
74-
parallelSort(info.liveSymbols.begin(), info.liveSymbols.end(),
75-
[](Defined *a, Defined *b) { return a->getVA() < b->getVA(); });
76-
return info;
77-
}
78107

79-
// Construct a map from symbols to their stringified representations.
80-
// Demangling symbols (which is what toString() does) is slow, so
81-
// we do that in batch using parallel-for.
82-
static DenseMap<Symbol *, std::string>
83-
getSymbolStrings(ArrayRef<Defined *> syms) {
84-
std::vector<std::string> str(syms.size());
85-
parallelFor(0, syms.size(), [&](size_t i) {
86-
raw_string_ostream os(str[i]);
87-
Defined *sym = syms[i];
88-
89-
switch (sym->isec->kind()) {
90-
case InputSection::CStringLiteralKind: {
91-
// Output "literal string: <string literal>"
92-
const auto *isec = cast<CStringInputSection>(sym->isec);
93-
const StringPiece &piece = isec->getStringPiece(sym->value);
94-
assert(
95-
sym->value == piece.inSecOff &&
96-
"We expect symbols to always point to the start of a StringPiece.");
97-
StringRef str = isec->getStringRef(&piece - &(*isec->pieces.begin()));
98-
(os << "literal string: ").write_escaped(str);
99-
break;
100-
}
101-
case InputSection::ConcatKind:
102-
case InputSection::WordLiteralKind:
103-
os << toString(*sym);
108+
if (isReferencedFile)
109+
info.files.push_back(file);
104110
}
105-
});
106111

107-
DenseMap<Symbol *, std::string> ret;
108-
for (size_t i = 0, e = syms.size(); i < e; ++i)
109-
ret[syms[i]] = std::move(str[i]);
110-
return ret;
112+
// cstrings are not stored in sorted order in their OutputSections, so we sort
113+
// them here.
114+
for (auto &liveCStrings : info.liveCStringsForSection)
115+
parallelSort(liveCStrings.second, [](const auto &p1, const auto &p2) {
116+
return p1.first < p2.first;
117+
});
118+
return info;
111119
}
112120

113121
void macho::writeMapFile() {
@@ -124,16 +132,12 @@ void macho::writeMapFile() {
124132
return;
125133
}
126134

127-
// Dump output path.
128135
os << format("# Path: %s\n", config->outputFile.str().c_str());
129-
130-
// Dump output architecture.
131136
os << format("# Arch: %s\n",
132137
getArchitectureName(config->arch()).str().c_str());
133138

134139
MapInfo info = gatherMapInfo();
135140

136-
// Dump table of object files.
137141
os << "# Object files:\n";
138142
os << format("[%3u] %s\n", 0, (const char *)"linker synthesized");
139143
uint32_t fileIndex = 1;
@@ -143,7 +147,6 @@ void macho::writeMapFile() {
143147
readerToFileOrdinal[file] = fileIndex++;
144148
}
145149

146-
// Dump table of sections
147150
os << "# Sections:\n";
148151
os << "# Address\tSize \tSegment\tSection\n";
149152
for (OutputSegment *seg : outputSegments)
@@ -155,28 +158,48 @@ void macho::writeMapFile() {
155158
seg->name.str().c_str(), osec->name.str().c_str());
156159
}
157160

158-
// Dump table of symbols
159-
DenseMap<Symbol *, std::string> liveSymbolStrings =
160-
getSymbolStrings(info.liveSymbols);
161161
os << "# Symbols:\n";
162162
os << "# Address\tSize \tFile Name\n";
163-
for (Defined *sym : info.liveSymbols) {
164-
assert(sym->isLive());
165-
os << format("0x%08llX\t0x%08llX\t[%3u] %s\n", sym->getVA(), sym->size,
166-
readerToFileOrdinal[sym->getFile()],
167-
liveSymbolStrings[sym].c_str());
163+
for (const OutputSegment *seg : outputSegments) {
164+
for (const OutputSection *osec : seg->getSections()) {
165+
if (auto *concatOsec = dyn_cast<ConcatOutputSection>(osec)) {
166+
for (const InputSection *isec : concatOsec->inputs) {
167+
for (Defined *sym : isec->symbols)
168+
os << format("0x%08llX\t0x%08llX\t[%3u] %s\n", sym->getVA(),
169+
sym->size, readerToFileOrdinal[sym->getFile()],
170+
sym->getName().str().data());
171+
}
172+
} else if (osec == in.cStringSection || osec == in.objcMethnameSection) {
173+
const auto &liveCStrings = info.liveCStringsForSection.lookup(osec);
174+
uint64_t lastAddr = 0; // strings will never start at address 0, so this
175+
// is a sentinel value
176+
for (const auto &[addr, info] : liveCStrings) {
177+
uint64_t size = 0;
178+
if (addr != lastAddr)
179+
size = info.str.size() + 1; // include null terminator
180+
lastAddr = addr;
181+
os << format("0x%08llX\t0x%08llX\t[%3u] literal string: ", addr, size,
182+
info.fileIndex);
183+
os.write_escaped(info.str) << "\n";
184+
}
185+
}
186+
// TODO print other synthetic sections
187+
}
168188
}
169189

170190
if (config->deadStrip) {
171-
DenseMap<Symbol *, std::string> deadSymbolStrings =
172-
getSymbolStrings(info.deadSymbols);
173191
os << "# Dead Stripped Symbols:\n";
174192
os << "# \tSize \tFile Name\n";
175193
for (Defined *sym : info.deadSymbols) {
176194
assert(!sym->isLive());
177195
os << format("<<dead>>\t0x%08llX\t[%3u] %s\n", sym->size,
178196
readerToFileOrdinal[sym->getFile()],
179-
deadSymbolStrings[sym].c_str());
197+
sym->getName().str().data());
198+
}
199+
for (CStringInfo &cstrInfo : info.deadCStrings) {
200+
os << format("<<dead>>\t0x%08llX\t[%3u] literal string: ",
201+
cstrInfo.str.size() + 1, cstrInfo.fileIndex);
202+
os.write_escaped(cstrInfo.str) << "\n";
180203
}
181204
}
182205
}

‎lld/test/MachO/map-file.s

Lines changed: 38 additions & 30 deletions
Original file line numberDiff line numberDiff line change
@@ -4,23 +4,24 @@
44
# RUN: llvm-mc -filetype=obj -triple=x86_64-apple-darwin %t/test.s -o %t/test.o
55
# RUN: llvm-mc -filetype=obj -triple=x86_64-apple-darwin %t/c-string-literal.s -o %t/c-string-literal.o
66

7-
# RUN: %lld -map %t/map %t/test.o %t/foo.o %t/c-string-literal.o --time-trace -o %t/test
7+
# RUN: %lld -demangle -map %t/map %t/test.o %t/foo.o %t/c-string-literal.o \
8+
# RUN: --time-trace -o %t/test
89
# RUN: llvm-objdump --syms --section-headers %t/test > %t/objdump
9-
# RUN: cat %t/objdump %t/map > %t/out
10-
# RUN: FileCheck %s < %t/out
10+
## Check that symbols in cstring sections aren't emitted
11+
# RUN: cat %t/objdump %t/map | FileCheck %s --implicit-check-not _hello_world
1112
# RUN: FileCheck %s --check-prefix=MAPFILE < %t/test.time-trace
1213

1314
# CHECK: Sections:
14-
# CHECK-NEXT: Idx Name Size VMA Type
15-
# CHECK-NEXT: 0 __text {{[0-9a-f]+}} [[#%x,TEXT:]] TEXT
16-
# CHECK-NEXT: 1 obj {{[0-9a-f]+}} [[#%x,DATA:]] TEXT
17-
# CHECK-NEXT: 2 __cstring {{[0-9a-f]+}} [[#%x,CSTR:]] DATA
18-
# CHECK-NEXT: 3 __common {{[0-9a-f]+}} [[#%x,BSS:]] BSS
15+
# CHECK-NEXT: Idx Name Size VMA Type
16+
# CHECK-NEXT: 0 __text {{[0-9a-f]+}} [[#%x,TEXT:]] TEXT
17+
# CHECK-NEXT: 1 __cstring {{[0-9a-f]+}} [[#%x,CSTR:]] DATA
18+
# CHECK-NEXT: 2 __common {{[0-9a-f]+}} [[#%x,BSS:]] BSS
1919

2020
# CHECK: SYMBOL TABLE:
2121
# CHECK-DAG: [[#%x,MAIN:]] g F __TEXT,__text _main
2222
# CHECK-DAG: [[#%x,NUMBER:]] g O __DATA,__common _number
23-
# CHECK-DAG: [[#%x,FOO:]] g F __TEXT,obj _foo
23+
# CHECK-DAG: [[#%x,BAR:]] g F __TEXT,__text _bar
24+
# CHECK-DAG: [[#%x,FOO:]] g F __TEXT,__text __ZTIN3foo3bar4MethE
2425
# CHECK-DAG: [[#%x,HIWORLD:]] g O __TEXT,__cstring _hello_world
2526
# CHECK-DAG: [[#%x,HIITSME:]] g O __TEXT,__cstring _hello_its_me
2627

@@ -35,56 +36,61 @@
3536
# CHECK-NEXT: # Sections:
3637
# CHECK-NEXT: # Address Size Segment Section
3738
# CHECK-NEXT: 0x[[#%X,TEXT]] 0x{{[0-9A-F]+}} __TEXT __text
38-
# CHECK-NEXT: 0x[[#%X,DATA]] 0x{{[0-9A-F]+}} __TEXT obj
3939
# CHECK-NEXT: 0x[[#%X,CSTR]] 0x{{[0-9A-F]+}} __TEXT __cstring
4040
# CHECK-NEXT: 0x[[#%X,BSS]] 0x{{[0-9A-F]+}} __DATA __common
4141

4242
# CHECK-NEXT: # Symbols:
43-
# CHECK-NEXT: # Address Size File Name
44-
# CHECK-DAG: 0x[[#%X,MAIN]] 0x00000001 [ 1] _main
45-
# CHECK-DAG: 0x[[#%X,FOO]] 0x00000001 [ 2] _foo
46-
# CHECK-DAG: 0x[[#%X,HIWORLD]] 0x0000000E [ 3] literal string: Hello world!\n
47-
# CHECK-DAG: 0x[[#%X,HIITSME]] 0x0000000F [ 3] literal string: Hello, it's me
48-
# CHECK-DAG: 0x[[#%X,NUMBER]] 0x00000001 [ 1] _number
43+
# CHECK-NEXT: # Address Size File Name
44+
# CHECK-DAG: 0x[[#%X,MAIN]] 0x00000001 [ 1] _main
45+
# CHECK-DAG: 0x[[#%X,BAR]] 0x00000001 [ 1] _bar
46+
# CHECK-DAG: 0x[[#%X,FOO]] 0x00000001 [ 2] __ZTIN3foo3bar4MethE
47+
# CHECK-DAG: 0x[[#%X,HIWORLD]] 0x0000000E [ 3] literal string: Hello world!\n
48+
# CHECK-DAG: 0x[[#%X,HIITSME]] 0x0000000F [ 3] literal string: Hello, it's me
49+
# CHECK-DAG: 0x[[#%X,HIITSME + 0xf]] 0x0000000E [ 3] literal string: Hello world!\n
50+
# CHECK-DAG: 0x[[#%X,NUMBER]] 0x00000001 [ 1] _number
4951

5052
# MAPFILE: "name":"Total Write map file"
5153

52-
# RUN: %lld -dead_strip -map %t/stripped-map %t/test.o %t/foo.o %t/c-string-literal.o -o %t/stripped
54+
# RUN: %lld -demangle -dead_strip -map %t/stripped-map %t/test.o %t/foo.o %t/c-string-literal.o -o %t/stripped
5355
# RUN: FileCheck --check-prefix=STRIPPED %s < %t/stripped-map
5456

5557
## C-string literals should be printed as "literal string: <C string literal>"
5658
# STRIPPED-LABEL: Dead Stripped Symbols:
57-
# STRIPPED-DAG: <<dead>> 0x00000001 [ 2] _foo
58-
# STRIPPED-DAG: <<dead>> 0x0000000E [ 3] literal string: Hello world!\n
59-
# STRIPPED-DAG: <<dead>> 0x0000000F [ 3] literal string: Hello, it's me
60-
# STRIPPED-DAG: <<dead>> 0x00000001 [ 1] _number
59+
# STRIPPED-DAG: <<dead>> 0x00000001 [ 1] _bar
60+
# STRIPPED-DAG: <<dead>> 0x00000001 [ 1] _number
61+
# STRIPPED-DAG: <<dead>> 0x00000001 [ 2] __ZTIN3foo3bar4MethE
62+
# STRIPPED-DAG: <<dead>> 0x0000000E [ 3] literal string: Hello world!\n
63+
# STRIPPED-DAG: <<dead>> 0x0000000F [ 3] literal string: Hello, it's me
64+
# STRIPPED-DAG: <<dead>> 0x0000000E [ 3] literal string: Hello world!\n
6165

6266
# RUN: %lld --icf=all -map %t/icf-map %t/test.o %t/foo.o %t/c-string-literal.o -o %t/icf
6367
# RUN: FileCheck --check-prefix=ICF %s < %t/icf-map
6468

69+
## Verify that folded symbols and cstrings have size zero. Note that ld64 prints
70+
## folded symbols but not folded cstrings; we print both.
71+
6572
# ICF: Symbols:
66-
# ICF-DAG: 0x[[#%X,FOO:]] 0x00000000 [ 2] _foo
67-
# ICF-DAG: 0x[[#FOO]] 0x00000001 [ 1] _bar
73+
# ICF-DAG: 0x[[#%X,FOO:]] 0x00000000 [ 2] __ZTIN3foo3bar4MethE
74+
# ICF-DAG: 0x[[#FOO]] 0x00000001 [ 1] _bar
75+
# ICF-DAG: 0x[[#%X,HIWORLD:]] 0x0000000E [ 3] literal string: Hello world!\n
76+
# ICF-DAG: 0x[[#%X,HIWORLD]] 0x00000000 [ 3] literal string: Hello world!\n
6877

6978
#--- foo.s
70-
## ICF will only fold sections marked as pure_instructions
71-
.section __TEXT,obj,regular,pure_instructions
72-
.globl _foo
73-
.alt_entry _alt_foo
74-
_foo:
79+
.globl __ZTIN3foo3bar4MethE
80+
## This C++ symbol makes it clear that we do not print the demangled name in
81+
## the map file, even if `-demangle` is passed.
82+
__ZTIN3foo3bar4MethE:
7583
nop
7684

7785
.subsections_via_symbols
7886

7987
#--- test.s
8088
.comm _number, 1
8189
.globl _main, _bar
82-
.alt_entry _alt_bar
8390

8491
_main:
8592
ret
8693

87-
.section __TEXT,obj,regular,pure_instructions
8894
_bar:
8995
nop
9096

@@ -101,4 +107,6 @@ _hello_world:
101107
_hello_its_me:
102108
.asciz "Hello, it's me"
103109

110+
.asciz "Hello world!\n"
111+
104112
.subsections_via_symbols

0 commit comments

Comments
 (0)