1
1
mirror of https://github.com/rui314/mold.git synced 2024-09-22 02:20:51 +03:00
mold/main.cc

1235 lines
35 KiB
C++
Raw Normal View History

2020-10-20 08:54:35 +03:00
#include "mold.h"
2020-10-02 07:28:26 +03:00
2020-12-21 13:35:33 +03:00
#include <functional>
2021-01-16 05:57:36 +03:00
#include <map>
2020-12-24 08:39:02 +03:00
#include <signal.h>
2021-01-16 05:57:36 +03:00
#include <tbb/global_control.h>
#include <tbb/parallel_do.h>
#include <tbb/parallel_for_each.h>
#include <tbb/task_group.h>
2020-11-20 07:54:29 +03:00
#include <unordered_set>
2020-09-29 09:05:29 +03:00
2020-12-22 14:10:04 +03:00
static tbb::task_group parser_tg;
static bool preloading;
2020-12-22 11:37:49 +03:00
static bool is_text_file(MemoryMappedFile *mb) {
return mb->size() >= 4 &&
isprint(mb->data()[0]) &&
isprint(mb->data()[1]) &&
isprint(mb->data()[2]) &&
isprint(mb->data()[3]);
2020-12-12 07:08:40 +03:00
}
2020-12-10 16:51:38 +03:00
2021-01-09 13:21:28 +03:00
enum class FileType { UNKNOWN, OBJ, DSO, AR, THIN_AR, TEXT };
2020-12-20 14:58:22 +03:00
2020-12-22 11:37:49 +03:00
static FileType get_file_type(MemoryMappedFile *mb) {
if (mb->size() >= 20 && memcmp(mb->data(), "\177ELF", 4) == 0) {
ElfEhdr &ehdr = *(ElfEhdr *)mb->data();
2020-12-20 14:58:22 +03:00
if (ehdr.e_type == ET_REL)
2021-01-09 13:21:28 +03:00
return FileType::OBJ;
2020-12-20 14:58:22 +03:00
if (ehdr.e_type == ET_DYN)
2021-01-09 13:21:28 +03:00
return FileType::DSO;
return FileType::UNKNOWN;
2020-12-20 14:58:22 +03:00
}
2020-12-22 11:37:49 +03:00
if (mb->size() >= 8 && memcmp(mb->data(), "!<arch>\n", 8) == 0)
2021-01-09 13:21:28 +03:00
return FileType::AR;
2020-12-22 11:37:49 +03:00
if (mb->size() >= 8 && memcmp(mb->data(), "!<thin>\n", 8) == 0)
2021-01-09 13:21:28 +03:00
return FileType::THIN_AR;
2020-12-20 14:58:22 +03:00
if (is_text_file(mb))
2021-01-09 13:21:28 +03:00
return FileType::TEXT;
return FileType::UNKNOWN;
2020-12-20 14:58:22 +03:00
}
2021-02-09 16:07:44 +03:00
static ObjectFile *new_object_file(MemoryMappedFile *mb,
std::string archive_name,
ReadContext &ctx) {
2021-02-10 08:21:41 +03:00
bool in_lib = (!archive_name.empty() && !ctx.whole_archive);
ObjectFile *file = new ObjectFile(mb, archive_name, in_lib);
2020-12-21 13:35:33 +03:00
parser_tg.run([=]() { file->parse(); });
return file;
}
2020-12-22 11:37:49 +03:00
static SharedFile *new_shared_file(MemoryMappedFile *mb, bool as_needed) {
2020-12-21 13:35:33 +03:00
SharedFile *file = new SharedFile(mb, as_needed);
parser_tg.run([=]() { file->parse(); });
return file;
}
2021-01-15 09:50:10 +03:00
template <typename T>
class FileCache {
public:
void store(MemoryMappedFile *mb, T *obj) {
Key k(mb->name, mb->size(), mb->mtime);
cache[k].push_back(obj);
}
2020-12-22 14:10:04 +03:00
2021-01-15 09:50:10 +03:00
std::vector<T *> get(MemoryMappedFile *mb) {
Key k(mb->name, mb->size(), mb->mtime);
std::vector<T *> objs = cache[k];
2020-12-22 14:20:31 +03:00
cache[k].clear();
return objs;
2021-01-15 09:50:10 +03:00
}
T *get_one(MemoryMappedFile *mb) {
std::vector<T *> objs = get(mb);
return objs.empty() ? nullptr : objs[0];
}
private:
2021-01-24 06:01:43 +03:00
typedef std::tuple<std::string, i64, i64> Key;
2021-01-15 09:50:10 +03:00
std::map<Key, std::vector<T *>> cache;
};
2021-02-09 16:07:44 +03:00
void read_file(MemoryMappedFile *mb, ReadContext &ctx) {
2021-01-15 09:50:10 +03:00
static FileCache<ObjectFile> obj_cache;
static FileCache<SharedFile> dso_cache;
2020-12-22 14:20:31 +03:00
2021-01-15 09:41:09 +03:00
if (preloading) {
switch (get_file_type(mb)) {
case FileType::OBJ:
2021-02-09 16:07:44 +03:00
obj_cache.store(mb, new_object_file(mb, "", ctx));
2020-12-22 14:32:49 +03:00
return;
2021-01-15 09:41:09 +03:00
case FileType::DSO:
2021-02-09 16:07:44 +03:00
dso_cache.store(mb, new_shared_file(mb, ctx.as_needed));
2021-01-15 09:41:09 +03:00
return;
case FileType::AR:
for (MemoryMappedFile *child : read_fat_archive_members(mb))
2021-02-09 16:07:44 +03:00
obj_cache.store(mb, new_object_file(child, mb->name, ctx));
2021-01-15 09:41:09 +03:00
return;
case FileType::THIN_AR:
for (MemoryMappedFile *child : read_thin_archive_members(mb))
2021-02-09 16:07:44 +03:00
obj_cache.store(child, new_object_file(child, mb->name, ctx));
2021-01-15 09:41:09 +03:00
return;
case FileType::TEXT:
2021-02-09 16:07:44 +03:00
parse_linker_script(mb, ctx);
2021-01-15 09:41:09 +03:00
return;
2020-12-22 14:32:49 +03:00
}
2021-01-15 09:41:09 +03:00
Fatal() << mb->name << ": unknown file type";
}
2020-12-22 14:32:49 +03:00
2021-01-15 09:41:09 +03:00
switch (get_file_type(mb)) {
case FileType::OBJ:
2021-01-15 09:50:10 +03:00
if (ObjectFile *obj = obj_cache.get_one(mb))
out::objs.push_back(obj);
2020-12-24 11:37:02 +03:00
else
2021-02-09 16:07:44 +03:00
out::objs.push_back(new_object_file(mb, "", ctx));
2020-12-20 14:58:22 +03:00
return;
2021-01-09 13:21:28 +03:00
case FileType::DSO:
2021-01-15 09:50:10 +03:00
if (SharedFile *obj = dso_cache.get_one(mb))
out::dsos.push_back(obj);
else
2021-02-09 16:07:44 +03:00
out::dsos.push_back(new_shared_file(mb, ctx.as_needed));
2020-12-20 14:58:22 +03:00
return;
2021-01-09 13:21:28 +03:00
case FileType::AR:
2021-01-15 09:50:10 +03:00
if (std::vector<ObjectFile *> objs = obj_cache.get(mb); !objs.empty()) {
2020-12-22 14:10:04 +03:00
append(out::objs, objs);
2020-12-24 11:37:02 +03:00
} else {
for (MemoryMappedFile *child : read_archive_members(mb))
2021-02-09 16:07:44 +03:00
out::objs.push_back(new_object_file(child, mb->name, ctx));
2020-12-22 14:10:04 +03:00
}
return;
2021-01-09 13:21:28 +03:00
case FileType::THIN_AR:
2020-12-22 14:32:49 +03:00
for (MemoryMappedFile *child : read_thin_archive_members(mb)) {
2021-01-15 09:50:10 +03:00
if (ObjectFile *obj = obj_cache.get_one(child))
out::objs.push_back(obj);
2020-12-24 11:37:02 +03:00
else
2021-02-09 16:07:44 +03:00
out::objs.push_back(new_object_file(child, mb->name, ctx));
2020-12-22 14:10:04 +03:00
}
return;
2021-01-09 13:21:28 +03:00
case FileType::TEXT:
2021-02-09 16:07:44 +03:00
parse_linker_script(mb, ctx);
return;
}
2021-01-15 09:41:09 +03:00
Fatal() << mb->name << ": unknown file type";
2020-10-10 06:47:12 +03:00
}
2020-10-28 07:42:05 +03:00
template <typename T>
2021-01-24 06:01:43 +03:00
static std::vector<std::span<T>> split(std::vector<T> &input, i64 unit) {
2020-12-21 10:32:43 +03:00
assert(input.size() > 0);
2020-12-10 14:20:46 +03:00
std::span<T> span(input);
std::vector<std::span<T>> vec;
2020-10-28 07:42:05 +03:00
2020-12-10 14:20:46 +03:00
while (span.size() >= unit) {
vec.push_back(span.subspan(0, unit));
span = span.subspan(unit);
2020-10-28 07:42:05 +03:00
}
2020-12-10 14:20:46 +03:00
if (!span.empty())
vec.push_back(span);
2020-10-28 07:42:05 +03:00
return vec;
}
2020-11-17 08:05:40 +03:00
static void resolve_symbols() {
2021-01-09 11:45:50 +03:00
Timer t("resolve_symbols");
2020-11-11 04:42:26 +03:00
2020-11-11 04:51:30 +03:00
// Register defined symbols
2021-01-24 07:34:12 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
file->resolve_symbols();
});
tbb::parallel_for_each(out::dsos, [](SharedFile *file) {
file->resolve_symbols();
});
2020-11-11 04:42:26 +03:00
2020-11-30 10:43:47 +03:00
// Mark reachable objects and DSOs to decide which files to include
// into an output.
2021-01-25 08:02:08 +03:00
std::vector<ObjectFile *> roots;
2020-11-24 07:56:14 +03:00
for (ObjectFile *file : out::objs)
2020-11-24 08:31:05 +03:00
if (file->is_alive)
2021-01-25 08:02:08 +03:00
roots.push_back(file);
2021-02-21 11:17:41 +03:00
for (std::string_view name : config.undefined)
if (InputFile *file = Symbol::intern(name)->file)
if (!file->is_alive.exchange(true) && !file->is_dso)
roots.push_back((ObjectFile *)file);
2021-01-25 08:02:08 +03:00
tbb::parallel_do(roots,
[&](ObjectFile *file,
tbb::parallel_do_feeder<ObjectFile *> &feeder) {
file->mark_live_objects(
[&](ObjectFile *obj) { feeder.add(obj); });
});
2020-11-11 04:42:26 +03:00
2020-11-30 10:43:47 +03:00
// Eliminate unused archive members and as-needed DSOs.
2020-12-20 10:42:17 +03:00
erase(out::objs, [](InputFile *file) { return !file->is_alive; });
erase(out::dsos, [](InputFile *file) { return !file->is_alive; });
2020-11-11 04:42:26 +03:00
}
2020-11-17 08:04:53 +03:00
static void eliminate_comdats() {
2021-01-09 11:45:50 +03:00
Timer t("comdat");
2020-11-11 04:42:26 +03:00
2020-11-24 07:56:14 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
2020-11-08 12:17:24 +03:00
file->resolve_comdat_groups();
});
2020-11-24 07:56:14 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
2020-11-08 12:17:24 +03:00
file->eliminate_duplicate_comdat_groups();
});
}
2020-11-17 08:04:53 +03:00
static void handle_mergeable_strings() {
2021-01-09 11:45:50 +03:00
Timer t("resolve_strings");
2020-11-11 04:42:26 +03:00
2021-01-22 09:10:32 +03:00
// Resolve mergeable string fragments
2020-11-24 07:56:14 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
for (MergeableSection *isec : file->mergeable_sections) {
for (SectionFragment *frag : isec->fragments) {
2021-01-25 07:47:48 +03:00
if (!frag->is_alive)
continue;
2021-01-22 09:10:32 +03:00
MergeableSection *cur = frag->isec;
while (!cur || cur->file->priority > isec->file->priority)
if (frag->isec.compare_exchange_weak(cur, isec))
2020-11-07 15:53:21 +03:00
break;
}
}
});
// Calculate the total bytes of mergeable strings for each input section.
2020-11-24 07:56:14 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
for (MergeableSection *isec : file->mergeable_sections) {
2021-01-24 06:01:43 +03:00
i64 offset = 0;
for (SectionFragment *frag : isec->fragments) {
2021-01-22 14:35:33 +03:00
if (frag->isec == isec && frag->offset == -1) {
2021-01-22 11:13:52 +03:00
offset = align_to(offset, frag->alignment);
2021-01-22 14:35:33 +03:00
frag->offset = offset;
2021-01-22 10:59:37 +03:00
offset += frag->data.size();
2020-11-07 15:53:21 +03:00
}
}
isec->size = offset;
2020-11-07 15:53:21 +03:00
}
});
// Assign each mergeable input section a unique index.
2020-11-24 07:56:14 +03:00
for (ObjectFile *file : out::objs) {
for (MergeableSection *isec : file->mergeable_sections) {
2021-01-24 06:01:43 +03:00
i64 offset = isec->parent.shdr.sh_size;
2021-02-27 07:44:30 +03:00
i64 alignment = isec->shdr->sh_addralign;
isec->padding = align_to(offset, alignment) - offset;
isec->offset = offset + isec->padding;
isec->parent.shdr.sh_size = offset + isec->padding + isec->size;
2021-01-22 13:47:11 +03:00
isec->parent.shdr.sh_addralign =
2021-02-27 07:44:30 +03:00
std::max(isec->parent.shdr.sh_addralign, isec->shdr->sh_addralign);
2020-11-07 15:53:21 +03:00
}
}
}
2020-11-11 04:51:30 +03:00
// So far, each input section has a pointer to its corresponding
// output section, but there's no reverse edge to get a list of
// input sections from an output section. This function creates it.
//
// An output section may contain millions of input sections.
// So, we append input sections to output sections in parallel.
2020-11-17 08:04:53 +03:00
static void bin_sections() {
2021-01-09 11:45:50 +03:00
Timer t("bin_sections");
2020-11-11 04:42:26 +03:00
2021-01-24 06:01:43 +03:00
i64 unit = (out::objs.size() + 127) / 128;
2020-12-10 14:20:46 +03:00
std::vector<std::span<ObjectFile *>> slices = split(out::objs, unit);
2020-10-26 07:36:56 +03:00
2021-01-24 06:01:43 +03:00
i64 num_osec = OutputSection::instances.size();
2020-11-08 04:05:59 +03:00
2021-01-17 10:40:35 +03:00
std::vector<std::vector<std::vector<InputSection *>>> groups(slices.size());
2021-01-24 06:01:43 +03:00
for (i64 i = 0; i < groups.size(); i++)
2020-11-08 04:05:59 +03:00
groups[i].resize(num_osec);
2020-10-28 08:06:35 +03:00
2021-01-24 06:01:43 +03:00
tbb::parallel_for((i64)0, (i64)slices.size(), [&](i64 i) {
2020-12-21 13:35:33 +03:00
for (ObjectFile *file : slices[i])
2020-12-20 10:52:10 +03:00
for (InputSection *isec : file->sections)
if (isec)
groups[i][isec->output_section->idx].push_back(isec);
2020-10-28 08:06:35 +03:00
});
2021-01-24 06:01:43 +03:00
std::vector<i64> sizes(num_osec);
2020-10-26 07:36:56 +03:00
2021-01-17 10:40:35 +03:00
for (std::span<std::vector<InputSection *>> group : groups)
2021-01-24 06:01:43 +03:00
for (i64 i = 0; i < group.size(); i++)
2020-10-28 08:22:25 +03:00
sizes[i] += group[i].size();
2020-11-08 03:44:27 +03:00
2021-01-24 06:01:43 +03:00
tbb::parallel_for((i64)0, num_osec, [&](i64 j) {
2020-11-08 06:42:40 +03:00
OutputSection::instances[j]->members.reserve(sizes[j]);
2021-01-24 06:01:43 +03:00
for (i64 i = 0; i < groups.size(); i++)
2020-12-20 10:52:10 +03:00
append(OutputSection::instances[j]->members, groups[i][j]);
2020-11-08 04:05:59 +03:00
});
2020-10-26 07:36:56 +03:00
}
2020-10-22 17:19:48 +03:00
2020-11-26 12:09:32 +03:00
static void check_duplicate_symbols() {
2021-01-10 10:37:24 +03:00
Timer t("check_dup_syms");
2020-12-20 11:08:27 +03:00
2020-11-30 12:19:33 +03:00
tbb::parallel_for_each(out::objs, [&](ObjectFile *file) {
2021-01-24 06:01:43 +03:00
for (i64 i = file->first_global; i < file->elf_syms.size(); i++) {
2021-01-10 07:52:59 +03:00
const ElfSym &esym = file->elf_syms[i];
Symbol &sym = *file->symbols[i];
bool is_weak = (esym.st_bind == STB_WEAK);
bool is_eliminated =
!esym.is_abs() && !esym.is_common() && !file->get_section(esym);
2021-01-10 07:52:59 +03:00
2021-01-10 08:23:41 +03:00
if (esym.is_defined() && !is_weak && !is_eliminated && sym.file != file)
2021-01-13 17:52:11 +03:00
Error() << "duplicate symbol: " << *file << ": " << *sym.file
2021-02-27 06:08:55 +03:00
<< ": " << sym;
2020-11-20 11:12:45 +03:00
}
});
2021-01-13 17:52:11 +03:00
Error::checkpoint();
2020-11-20 11:12:45 +03:00
}
2021-03-05 07:09:35 +03:00
static void compute_visibility() {
2021-03-05 09:36:08 +03:00
if (!config.shared)
return;
2021-03-05 07:09:35 +03:00
2021-03-05 09:36:08 +03:00
Timer t("compute_visibility");
2021-03-05 07:09:35 +03:00
2021-03-05 09:36:08 +03:00
tbb::parallel_for_each(out::objs, [&](ObjectFile *file) {
for (Symbol *sym : std::span(file->symbols).subspan(file->first_global)) {
if (sym->file != file)
continue;
u8 visibility = sym->visibility;
bool bsymbolic = config.Bsymbolic ||
(config.Bsymbolic_functions && sym->get_type() == STT_FUNC);
if (visibility == STV_DEFAULT && bsymbolic)
visibility = STV_PROTECTED;
switch (visibility) {
case STV_DEFAULT:
sym->is_imported = true;
sym->is_exported = true;
break;
case STV_PROTECTED:
sym->is_imported = false;
sym->is_exported = true;
break;
case STV_HIDDEN:
sym->is_imported = false;
sym->is_exported = false;
break;
default:
unreachable();
2021-03-05 07:09:35 +03:00
}
2021-03-05 09:36:08 +03:00
}
});
2021-03-05 07:09:35 +03:00
}
2020-10-26 08:18:00 +03:00
static void set_isec_offsets() {
2021-01-09 11:45:50 +03:00
Timer t("isec_offsets");
2020-11-11 04:42:26 +03:00
2020-11-07 18:47:34 +03:00
tbb::parallel_for_each(OutputSection::instances, [&](OutputSection *osec) {
2020-11-08 06:42:40 +03:00
if (osec->members.empty())
2020-10-27 07:52:10 +03:00
return;
2021-01-17 10:40:35 +03:00
std::vector<std::span<InputSection *>> slices = split(osec->members, 10000);
2021-01-24 06:01:43 +03:00
std::vector<i64> size(slices.size());
std::vector<i64> alignments(slices.size());
2020-10-26 08:18:00 +03:00
2021-01-24 06:01:43 +03:00
tbb::parallel_for((i64)0, (i64)slices.size(), [&](i64 i) {
i64 off = 0;
i64 align = 1;
2020-10-26 10:12:35 +03:00
2020-11-08 06:36:08 +03:00
for (InputChunk *isec : slices[i]) {
2021-02-27 07:44:30 +03:00
off = align_to(off, isec->shdr->sh_addralign);
2020-10-26 10:12:35 +03:00
isec->offset = off;
2021-02-27 07:44:30 +03:00
off += isec->shdr->sh_size;
align = std::max<i64>(align, isec->shdr->sh_addralign);
2020-10-26 10:12:35 +03:00
}
size[i] = off;
alignments[i] = align;
});
2021-01-24 06:01:43 +03:00
i64 align = *std::max_element(alignments.begin(), alignments.end());
2020-10-26 10:12:35 +03:00
2021-01-24 06:01:43 +03:00
std::vector<i64> start(slices.size());
for (i64 i = 1; i < slices.size(); i++)
2020-11-10 06:23:14 +03:00
start[i] = align_to(start[i - 1] + size[i - 1], align);
2020-10-26 10:58:49 +03:00
2021-01-24 06:01:43 +03:00
tbb::parallel_for((i64)1, (i64)slices.size(), [&](i64 i) {
2020-11-08 06:36:08 +03:00
for (InputChunk *isec : slices[i])
2020-10-26 10:12:35 +03:00
isec->offset += start[i];
});
osec->shdr.sh_size = start.back() + size.back();
2020-10-26 08:18:00 +03:00
osec->shdr.sh_addralign = align;
});
}
2021-03-01 10:24:37 +03:00
static void export_dynamic() {
if (config.export_dynamic || config.shared) {
Timer t("export_dynamic");
tbb::parallel_for((i64)0, (i64)out::objs.size(), [&](i64 i) {
ObjectFile *file = out::objs[i];
for (Symbol *sym : std::span(file->symbols).subspan(file->first_global))
if (sym->file == file && sym->esym->st_visibility == STV_DEFAULT)
sym->flags |= NEEDS_DYNSYM;
});
}
}
2020-11-29 05:10:33 +03:00
static void scan_rels() {
2021-01-09 11:45:50 +03:00
Timer t("scan_rels");
2020-11-15 10:19:21 +03:00
2020-11-29 05:10:33 +03:00
// Scan relocations to find dynamic symbols.
2020-11-24 07:56:14 +03:00
tbb::parallel_for_each(out::objs, [&](ObjectFile *file) {
2021-01-25 04:03:34 +03:00
file->scan_relocations();
2020-11-17 14:54:13 +03:00
});
2020-11-17 13:56:02 +03:00
2021-01-13 17:52:11 +03:00
// Exit if there was a relocation that refers an undefined symbol.
Error::checkpoint();
2020-11-26 12:09:32 +03:00
2021-02-20 18:23:53 +03:00
// Export symbols referenced by DSOs.
tbb::parallel_for_each(out::dsos, [&](SharedFile *file) {
for (Symbol *sym : file->undefs)
if (sym->file && !sym->file->is_dso)
2021-03-04 14:58:52 +03:00
sym->is_exported = true;
2021-02-20 18:23:53 +03:00
});
2021-03-04 10:16:37 +03:00
tbb::parallel_for_each(out::objs, [&](ObjectFile *file) {
for (Symbol *sym : std::span(file->symbols).subspan(file->first_global))
if (sym->file == file && sym->is_exported)
sym->flags |= NEEDS_DYNSYM;
});
2020-11-29 05:10:33 +03:00
// Aggregate dynamic symbols to a single vector.
2020-11-24 10:22:32 +03:00
std::vector<InputFile *> files;
2020-12-20 09:51:37 +03:00
append(files, out::objs);
append(files, out::dsos);
2020-11-24 10:22:32 +03:00
2020-11-24 08:31:05 +03:00
std::vector<std::vector<Symbol *>> vec(files.size());
2020-11-17 14:54:13 +03:00
2021-01-24 06:01:43 +03:00
tbb::parallel_for((i64)0, (i64)files.size(), [&](i64 i) {
2020-11-24 08:31:05 +03:00
for (Symbol *sym : files[i]->symbols)
2021-01-10 09:38:10 +03:00
if (sym->flags && sym->file == files[i])
vec[i].push_back(sym);
2020-11-15 07:01:38 +03:00
});
2020-11-06 06:01:52 +03:00
2020-11-29 05:10:33 +03:00
// Assign offsets in additional tables for each dynamic symbol.
for (Symbol *sym : flatten(vec)) {
2021-03-04 10:16:37 +03:00
if (sym->flags & NEEDS_DYNSYM)
2020-12-13 15:57:45 +03:00
out::dynsym->add_symbol(sym);
2020-12-16 12:38:13 +03:00
if (sym->flags & NEEDS_GOT)
2020-11-21 06:49:28 +03:00
out::got->add_got_symbol(sym);
2020-11-06 07:54:37 +03:00
2021-02-26 14:56:01 +03:00
if (sym->flags & NEEDS_PLT) {
if (sym->flags & NEEDS_GOT)
out::pltgot->add_symbol(sym);
else
out::plt->add_symbol(sym);
}
2020-11-17 14:22:52 +03:00
2020-12-16 12:38:13 +03:00
if (sym->flags & NEEDS_GOTTPOFF)
2020-11-21 04:48:51 +03:00
out::got->add_gottpoff_symbol(sym);
2020-11-17 14:22:52 +03:00
2020-12-16 12:38:13 +03:00
if (sym->flags & NEEDS_TLSGD)
2020-11-21 04:48:23 +03:00
out::got->add_tlsgd_symbol(sym);
2020-12-16 12:38:13 +03:00
if (sym->flags & NEEDS_TLSLD)
2021-01-14 13:58:21 +03:00
out::got->add_tlsld();
2020-11-25 11:20:48 +03:00
2020-12-16 12:38:13 +03:00
if (sym->flags & NEEDS_COPYREL) {
2020-11-25 14:35:04 +03:00
assert(sym->file->is_dso);
2021-02-28 17:19:29 +03:00
SharedFile *file = (SharedFile *)sym->file;
sym->is_readonly = file->is_readonly(sym);
2020-11-25 14:35:04 +03:00
2021-02-28 17:19:29 +03:00
if (sym->is_readonly)
out::copyrel_relro->add_symbol(sym);
else
out::copyrel->add_symbol(sym);
for (Symbol *alias : file->find_aliases(sym)) {
2021-01-14 14:31:39 +03:00
alias->has_copyrel = true;
alias->value = sym->value;
2021-02-28 17:19:29 +03:00
alias->is_readonly = sym->is_readonly;
2020-11-25 14:35:04 +03:00
out::dynsym->add_symbol(alias);
}
}
2020-11-18 15:45:49 +03:00
}
2020-11-16 16:25:58 +03:00
}
2020-11-11 11:25:00 +03:00
2021-03-07 14:36:10 +03:00
static void set_import_export() {
Timer t("set_import_export");
2021-03-07 07:43:58 +03:00
2021-03-07 14:36:10 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
for (Symbol *sym : std::span(file->symbols).subspan(file->first_global))
if (sym->file == file)
sym->is_exported = true;
});
for (std::pair<std::string_view, i16> pair : config.version_patterns) {
std::string_view pattern = pair.first;
i16 veridx = pair.second;
if (pattern.find('*') == pattern.npos)
Symbol::intern(pattern)->ver_idx = veridx;
else
Fatal() << "not supported: " << pattern;
}
2021-03-07 07:43:58 +03:00
}
static void fill_verdef() {
Timer t("fill_verdef");
if (config.version_definitions.empty())
return;
// Resize of .gnu.version
out::versym->contents.resize(out::dynsym->symbols.size(), 1);
out::versym->contents[0] = 0;
// Allocate a buffer for .gnu.version_d.
out::verdef->contents.resize((sizeof(ElfVerdef) + sizeof(ElfVerdaux)) *
(config.version_definitions.size() + 1));
u8 *buf = (u8 *)&out::verdef->contents[0];
u8 *ptr = buf;
ElfVerdef *verdef = nullptr;
auto write = [&](std::string_view verstr, i64 idx, i64 flags) {
out::verdef->shdr.sh_info++;
if (verdef)
verdef->vd_next = ptr - (u8 *)verdef;
verdef = (ElfVerdef *)ptr;
ptr += sizeof(ElfVerdef);
verdef->vd_version = 1;
verdef->vd_flags = flags;
verdef->vd_ndx = idx;
verdef->vd_cnt = 1;
verdef->vd_hash = elf_hash(verstr);
verdef->vd_aux = sizeof(ElfVerdef);
ElfVerdaux *aux = (ElfVerdaux *)ptr;
ptr += sizeof(ElfVerdaux);
aux->vda_name = out::dynstr->add_string(verstr);
};
std::string_view basename = config.soname.empty() ?
config.output : config.soname;
write(basename, 1, VER_FLG_BASE);
i64 idx = 2;
for (std::string_view verstr : config.version_definitions)
write(verstr, idx++, 0);
for (Symbol *sym : std::span(out::dynsym->symbols).subspan(1))
out::versym->contents[sym->dynsym_idx] = sym->ver_idx;
}
2021-03-06 14:49:40 +03:00
static void fill_verneed() {
Timer t("fill_verneed");
2020-11-29 05:40:57 +03:00
2020-11-29 05:57:58 +03:00
// Create a list of versioned symbols and sort by file and version.
2021-01-23 09:01:49 +03:00
std::vector<Symbol *> syms(out::dynsym->symbols.begin() + 1,
out::dynsym->symbols.end());
2021-03-06 05:51:03 +03:00
erase(syms, [](Symbol *sym) {
2021-03-06 14:49:40 +03:00
return !sym->file->is_dso || sym->ver_idx <= VER_NDX_LAST_RESERVED;
2021-03-06 05:51:03 +03:00
});
2020-11-29 05:40:57 +03:00
2020-11-29 05:57:58 +03:00
if (syms.empty())
return;
2020-12-21 12:42:14 +03:00
sort(syms, [](Symbol *a, Symbol *b) {
2021-01-20 07:09:54 +03:00
return std::tuple(((SharedFile *)a->file)->soname, a->ver_idx) <
std::tuple(((SharedFile *)b->file)->soname, b->ver_idx);
2020-11-29 05:40:57 +03:00
});
2021-03-06 14:49:40 +03:00
// Resize of .gnu.version
2021-01-23 09:01:49 +03:00
out::versym->contents.resize(out::dynsym->symbols.size(), 1);
2020-11-29 05:57:58 +03:00
out::versym->contents[0] = 0;
2021-03-07 07:43:58 +03:00
// Allocate a large enough buffer for .gnu.version_r.
2021-03-06 14:49:40 +03:00
out::verneed->contents.resize((sizeof(ElfVerneed) + sizeof(ElfVernaux)) *
syms.size());
2020-11-29 05:57:58 +03:00
// Fill .gnu.versoin_r.
u8 *buf = (u8 *)&out::verneed->contents[0];
2021-03-06 14:49:40 +03:00
u8 *ptr = buf;
2020-12-10 15:18:20 +03:00
ElfVerneed *verneed = nullptr;
ElfVernaux *aux = nullptr;
2020-11-29 06:59:08 +03:00
2021-03-07 07:43:58 +03:00
u16 veridx = VER_NDX_LAST_RESERVED + config.version_definitions.size();
2021-03-06 14:49:40 +03:00
auto start_group = [&](InputFile *file) {
out::verneed->shdr.sh_info++;
if (verneed)
verneed->vn_next = ptr - (u8 *)verneed;
2020-11-29 06:59:08 +03:00
2021-03-06 14:49:40 +03:00
verneed = (ElfVerneed *)ptr;
ptr += sizeof(*verneed);
verneed->vn_version = 1;
verneed->vn_file = out::dynstr->find_string(((SharedFile *)file)->soname);
verneed->vn_aux = sizeof(ElfVerneed);
aux = nullptr;
};
auto add_entry = [&](Symbol *sym) {
2021-01-09 16:27:53 +03:00
verneed->vn_cnt++;
2021-03-06 14:49:40 +03:00
2020-11-29 08:05:44 +03:00
if (aux)
2020-12-10 15:18:20 +03:00
aux->vna_next = sizeof(ElfVernaux);
2021-03-06 14:49:40 +03:00
aux = (ElfVernaux *)ptr;
ptr += sizeof(*aux);
2020-11-29 08:05:44 +03:00
2021-03-06 14:49:40 +03:00
std::string_view verstr = sym->get_version();
2020-11-29 06:59:08 +03:00
aux->vna_hash = elf_hash(verstr);
2021-03-06 14:49:40 +03:00
aux->vna_other = ++veridx;
2020-11-29 06:59:08 +03:00
aux->vna_name = out::dynstr->add_string(verstr);
};
2021-03-06 14:49:40 +03:00
for (i64 i = 0; i < syms.size(); i++) {
if (i == 0 || syms[i - 1]->file != syms[i]->file) {
start_group(syms[i]->file);
add_entry(syms[i]);
} else if (syms[i - 1]->ver_idx != syms[i]->ver_idx) {
add_entry(syms[i]);
}
2020-11-29 07:14:56 +03:00
2021-03-06 14:49:40 +03:00
out::versym->contents[syms[i]->dynsym_idx] = veridx;
2020-11-29 06:59:08 +03:00
}
2021-03-06 14:49:40 +03:00
// Resize .gnu.version_r to fit to its contents.
out::verneed->contents.resize(ptr - buf);
2020-11-29 05:40:57 +03:00
}
2021-01-24 06:01:43 +03:00
static void clear_padding(i64 filesize) {
2021-01-09 11:45:50 +03:00
Timer t("clear_padding");
2020-11-11 04:42:26 +03:00
2021-01-24 06:01:43 +03:00
auto zero = [](OutputChunk *chunk, i64 next_start) {
i64 pos = chunk->shdr.sh_offset;
2020-11-10 11:32:41 +03:00
if (chunk->shdr.sh_type != SHT_NOBITS)
pos += chunk->shdr.sh_size;
2020-11-17 07:59:24 +03:00
memset(out::buf + pos, 0, next_start - pos);
2020-11-09 15:50:47 +03:00
};
2021-01-24 06:01:43 +03:00
for (i64 i = 1; i < out::chunks.size(); i++)
2020-11-17 07:59:24 +03:00
zero(out::chunks[i - 1], out::chunks[i]->shdr.sh_offset);
zero(out::chunks.back(), filesize);
2020-11-09 15:50:47 +03:00
}
2020-10-22 12:54:51 +03:00
// We want to sort output sections in the following order.
//
2021-01-16 05:57:36 +03:00
// note
2020-10-22 17:19:48 +03:00
// alloc readonly data
// alloc readonly code
// alloc writable tdata
// alloc writable tbss
// alloc writable data
// alloc writable bss
// nonalloc
2021-01-24 06:01:43 +03:00
static i64 get_section_rank(const ElfShdr &shdr) {
2021-01-16 05:57:36 +03:00
bool note = shdr.sh_type == SHT_NOTE;
2020-10-29 12:31:06 +03:00
bool alloc = shdr.sh_flags & SHF_ALLOC;
bool writable = shdr.sh_flags & SHF_WRITE;
bool exec = shdr.sh_flags & SHF_EXECINSTR;
bool tls = shdr.sh_flags & SHF_TLS;
2020-10-30 06:47:35 +03:00
bool nobits = shdr.sh_type == SHT_NOBITS;
2021-01-16 05:57:36 +03:00
return (!note << 6) | (!alloc << 5) | (writable << 4) |
(exec << 3) | (!tls << 2) | nobits;
2020-10-22 12:54:51 +03:00
}
2021-01-24 06:01:43 +03:00
static i64 set_osec_offsets(std::span<OutputChunk *> chunks) {
2021-01-09 11:45:50 +03:00
Timer t("osec_offset");
2020-11-11 04:42:26 +03:00
2021-01-24 06:01:43 +03:00
i64 fileoff = 0;
i64 vaddr = config.image_base;
2020-10-26 08:16:13 +03:00
2020-11-11 08:13:39 +03:00
for (OutputChunk *chunk : chunks) {
2020-10-30 05:40:38 +03:00
if (chunk->starts_new_ptload)
2020-10-26 08:16:13 +03:00
vaddr = align_to(vaddr, PAGE_SIZE);
2020-11-24 16:54:21 +03:00
if (vaddr % PAGE_SIZE > fileoff % PAGE_SIZE)
fileoff += vaddr % PAGE_SIZE - fileoff % PAGE_SIZE;
else if (vaddr % PAGE_SIZE < fileoff % PAGE_SIZE)
fileoff = align_to(fileoff, PAGE_SIZE) + vaddr % PAGE_SIZE;
2020-10-30 05:40:38 +03:00
fileoff = align_to(fileoff, chunk->shdr.sh_addralign);
2020-10-26 08:16:13 +03:00
vaddr = align_to(vaddr, chunk->shdr.sh_addralign);
chunk->shdr.sh_offset = fileoff;
if (chunk->shdr.sh_flags & SHF_ALLOC)
chunk->shdr.sh_addr = vaddr;
2020-11-24 16:54:21 +03:00
bool is_bss = chunk->shdr.sh_type == SHT_NOBITS;
2020-11-03 14:37:27 +03:00
if (!is_bss)
2020-11-03 14:13:03 +03:00
fileoff += chunk->shdr.sh_size;
2020-10-30 05:40:38 +03:00
2020-11-03 14:37:27 +03:00
bool is_tbss = is_bss && (chunk->shdr.sh_flags & SHF_TLS);
2020-10-30 05:40:38 +03:00
if (!is_tbss)
2020-11-03 14:13:03 +03:00
vaddr += chunk->shdr.sh_size;
2020-10-26 08:16:13 +03:00
}
return fileoff;
}
2020-12-10 14:20:46 +03:00
static void fix_synthetic_symbols(std::span<OutputChunk *> chunks) {
2021-01-13 08:43:34 +03:00
auto start = [](Symbol *sym, OutputChunk *chunk) {
2021-02-11 16:50:49 +03:00
if (sym && chunk) {
2020-11-11 04:45:52 +03:00
sym->shndx = chunk->shndx;
sym->value = chunk->shdr.sh_addr;
}
};
2020-11-04 08:23:39 +03:00
2021-01-13 08:43:34 +03:00
auto stop = [](Symbol *sym, OutputChunk *chunk) {
2021-02-11 16:50:49 +03:00
if (sym && chunk) {
2020-11-11 04:45:52 +03:00
sym->shndx = chunk->shndx;
sym->value = chunk->shdr.sh_addr + chunk->shdr.sh_size;
}
};
2020-11-04 08:23:39 +03:00
// __bss_start
2020-11-11 08:13:39 +03:00
for (OutputChunk *chunk : chunks) {
2020-11-08 10:09:01 +03:00
if (chunk->kind == OutputChunk::REGULAR && chunk->name == ".bss") {
2021-01-13 08:43:34 +03:00
start(out::__bss_start, chunk);
2020-11-04 08:23:39 +03:00
break;
}
}
// __ehdr_start
2020-11-11 08:13:39 +03:00
for (OutputChunk *chunk : chunks) {
2020-11-04 08:23:39 +03:00
if (chunk->shndx == 1) {
2020-11-04 08:41:40 +03:00
out::__ehdr_start->shndx = 1;
2020-11-12 07:15:29 +03:00
out::__ehdr_start->value = out::ehdr->shdr.sh_addr;
2020-11-04 08:23:39 +03:00
break;
}
}
// __rela_iplt_start and __rela_iplt_end
2021-01-13 08:43:34 +03:00
start(out::__rela_iplt_start, out::relplt);
stop(out::__rela_iplt_end, out::relplt);
2020-11-04 08:23:39 +03:00
// __{init,fini}_array_{start,end}
2020-11-11 08:13:39 +03:00
for (OutputChunk *chunk : chunks) {
2020-11-04 08:23:39 +03:00
switch (chunk->shdr.sh_type) {
case SHT_INIT_ARRAY:
2021-01-13 08:43:34 +03:00
start(out::__init_array_start, chunk);
stop(out::__init_array_end, chunk);
2020-11-04 08:23:39 +03:00
break;
case SHT_FINI_ARRAY:
2021-01-13 08:43:34 +03:00
start(out::__fini_array_start, chunk);
stop(out::__fini_array_end, chunk);
2020-11-04 08:23:39 +03:00
break;
}
}
// _end, end, _etext, etext, _edata and edata
2020-11-11 08:13:39 +03:00
for (OutputChunk *chunk : chunks) {
2020-11-08 10:09:01 +03:00
if (chunk->kind == OutputChunk::HEADER)
2020-11-04 08:23:39 +03:00
continue;
2020-11-12 07:19:19 +03:00
if (chunk->shdr.sh_flags & SHF_ALLOC)
2021-01-13 08:43:34 +03:00
stop(out::_end, chunk);
2020-11-04 08:23:39 +03:00
2020-11-12 07:19:19 +03:00
if (chunk->shdr.sh_flags & SHF_EXECINSTR)
2021-01-13 08:43:34 +03:00
stop(out::_etext, chunk);
2020-11-04 08:23:39 +03:00
2020-11-12 07:19:19 +03:00
if (chunk->shdr.sh_type != SHT_NOBITS && chunk->shdr.sh_flags & SHF_ALLOC)
2021-01-13 08:43:34 +03:00
stop(out::_edata, chunk);
2020-11-04 08:23:39 +03:00
}
2020-11-13 04:19:47 +03:00
// _DYNAMIC
2021-02-11 16:50:49 +03:00
start(out::_DYNAMIC, out::dynamic);
2020-11-13 04:19:47 +03:00
// _GLOBAL_OFFSET_TABLE_
2021-02-11 16:50:49 +03:00
start(out::_GLOBAL_OFFSET_TABLE_, out::gotplt);
2021-02-11 16:17:22 +03:00
// __GNU_EH_FRAME_HDR
2021-02-11 16:50:49 +03:00
start(out::__GNU_EH_FRAME_HDR, out::eh_frame_hdr);
2021-02-11 16:17:22 +03:00
2020-11-04 08:23:39 +03:00
// __start_ and __stop_ symbols
2020-11-11 08:13:39 +03:00
for (OutputChunk *chunk : chunks) {
2020-11-11 08:45:17 +03:00
if (is_c_identifier(chunk->name)) {
2021-02-26 19:08:38 +03:00
start(Symbol::intern_alloc("__start_" + std::string(chunk->name)), chunk);
stop(Symbol::intern_alloc("__stop_" + std::string(chunk->name)), chunk);
2020-11-11 08:45:17 +03:00
}
2020-11-04 08:23:39 +03:00
}
}
2020-12-24 14:15:17 +03:00
void cleanup() {
2021-01-09 11:07:19 +03:00
if (OutputFile::tmpfile)
unlink(OutputFile::tmpfile);
2020-12-25 07:34:12 +03:00
if (socket_tmpfile)
unlink(socket_tmpfile);
2020-12-24 14:15:17 +03:00
}
2020-12-25 07:34:12 +03:00
static void signal_handler(int) {
2020-12-24 14:15:17 +03:00
cleanup();
2020-12-24 08:39:02 +03:00
_exit(1);
}
2021-01-09 16:30:14 +03:00
MemoryMappedFile *find_library(std::string name,
std::span<std::string_view> lib_paths) {
2020-12-21 00:47:10 +03:00
for (std::string_view dir : lib_paths) {
2020-12-10 07:44:58 +03:00
std::string root = dir.starts_with("/") ? config.sysroot : "";
std::string stem = root + std::string(dir) + "/lib" + name;
2020-11-19 12:03:26 +03:00
if (!config.is_static)
2020-12-22 12:33:16 +03:00
if (MemoryMappedFile *mb = MemoryMappedFile::open(stem + ".so"))
2020-12-22 11:37:49 +03:00
return mb;
2020-12-22 12:33:16 +03:00
if (MemoryMappedFile *mb = MemoryMappedFile::open(stem + ".a"))
2020-12-22 11:37:49 +03:00
return mb;
2020-11-19 12:03:26 +03:00
}
2021-01-13 17:52:11 +03:00
Fatal() << "library not found: " << name;
2020-11-19 12:03:26 +03:00
}
2020-12-22 14:12:41 +03:00
static void read_input_files(std::span<std::string_view> args) {
2021-02-09 16:07:44 +03:00
ReadContext ctx;
2020-12-22 14:12:41 +03:00
while (!args.empty()) {
std::string_view arg;
if (read_flag(args, "as-needed")) {
2021-02-09 16:07:44 +03:00
ctx.as_needed = true;
2020-12-22 14:12:41 +03:00
} else if (read_flag(args, "no-as-needed")) {
2021-02-09 16:07:44 +03:00
ctx.as_needed = false;
} else if (read_flag(args, "whole-archive")) {
ctx.whole_archive = true;
} else if (read_flag(args, "no-whole-archive")) {
ctx.whole_archive = false;
2020-12-22 14:12:41 +03:00
} else if (read_arg(args, arg, "l")) {
2021-02-09 16:07:44 +03:00
read_file(find_library(std::string(arg), config.library_paths), ctx);
2020-12-22 14:12:41 +03:00
} else {
2021-02-09 16:07:44 +03:00
read_file(MemoryMappedFile::must_open(std::string(args[0])), ctx);
2020-12-22 14:12:41 +03:00
args = args.subspan(1);
}
}
parser_tg.wait();
}
2020-12-21 11:51:20 +03:00
static void show_stats() {
for (ObjectFile *obj : out::objs) {
static Counter defined("defined_syms");
2021-01-29 15:44:46 +03:00
defined += obj->first_global - 1;
2020-12-21 11:51:20 +03:00
static Counter undefined("undefined_syms");
2021-01-29 15:44:46 +03:00
undefined += obj->symbols.size() - obj->first_global;
2020-12-21 11:51:20 +03:00
}
Counter num_input_sections("input_sections");
for (ObjectFile *file : out::objs)
2021-01-29 15:44:46 +03:00
num_input_sections += file->sections.size();
2020-12-21 11:51:20 +03:00
Counter num_output_chunks("output_out::chunks", out::chunks.size());
Counter num_objs("num_objs", out::objs.size());
Counter num_dsos("num_dsos", out::dsos.size());
Counter::print();
}
2020-12-21 10:32:43 +03:00
int main(int argc, char **argv) {
Timer t_all("all");
// Parse non-positional command line options
std::vector<std::string_view> arg_vector = expand_response_files(argv + 1);
std::vector<std::string_view> file_args;
2021-03-05 13:38:37 +03:00
parse_nonpositional_args(arg_vector, file_args);
2020-12-21 10:32:43 +03:00
if (config.output == "")
Fatal() << "-o option is missing";
2021-01-09 15:14:52 +03:00
if (!config.preload)
2021-01-24 06:01:43 +03:00
if (i64 code; resume_daemon(argv, &code))
2021-01-09 15:33:46 +03:00
exit(code);
2020-12-24 17:14:37 +03:00
2021-01-22 05:08:05 +03:00
tbb::global_control tbb_cont(tbb::global_control::max_allowed_parallelism,
config.thread_count);
2021-01-09 15:14:52 +03:00
signal(SIGINT, signal_handler);
signal(SIGTERM, signal_handler);
2020-12-24 17:14:37 +03:00
2020-12-24 17:25:07 +03:00
// Preload input files
std::function<void()> on_complete;
if (config.preload) {
2021-01-09 15:14:52 +03:00
std::function<void()> wait_for_client;
daemonize(argv, &wait_for_client, &on_complete);
2020-12-24 17:25:07 +03:00
preloading = true;
read_input_files(file_args);
2021-01-09 15:14:52 +03:00
wait_for_client();
2020-12-24 17:25:07 +03:00
} else if (config.fork) {
2020-12-20 03:20:24 +03:00
on_complete = fork_child();
2021-01-09 15:14:52 +03:00
}
2021-03-07 14:41:51 +03:00
if (config.print_stats)
Counter::enabled = true;
2020-12-21 10:32:43 +03:00
for (std::string_view arg : config.trace_symbol)
Symbol::intern(arg)->traced = true;
2020-12-20 03:20:24 +03:00
// Parse input files
{
2021-01-09 11:45:50 +03:00
Timer t("parse");
2020-12-22 14:20:31 +03:00
preloading = false;
2020-12-22 14:12:41 +03:00
read_input_files(file_args);
2020-12-21 10:32:43 +03:00
}
2020-12-17 15:36:38 +03:00
2020-11-20 07:54:29 +03:00
// Uniquify shared object files with soname
{
2020-11-29 16:05:34 +03:00
std::vector<SharedFile *> vec;
2020-12-11 06:46:03 +03:00
std::unordered_set<std::string_view> seen;
2020-11-29 16:05:34 +03:00
for (SharedFile *file : out::dsos)
if (seen.insert(file->soname).second)
vec.push_back(file);
out::dsos = vec;
2020-11-20 07:54:29 +03:00
}
2020-12-11 10:51:20 +03:00
Timer t_total("total");
Timer t_before_copy("before_copy");
2020-11-06 10:58:13 +03:00
2020-11-16 18:17:01 +03:00
out::ehdr = new OutputEhdr;
2020-11-16 18:23:51 +03:00
out::shdr = new OutputShdr;
2020-11-16 18:33:41 +03:00
out::phdr = new OutputPhdr;
2020-11-18 11:11:58 +03:00
out::got = new GotSection;
2020-11-13 06:43:59 +03:00
out::gotplt = new GotPltSection;
2020-11-17 07:34:02 +03:00
out::relplt = new RelPltSection;
2020-11-19 10:13:19 +03:00
out::strtab = new StrtabSection;
2020-11-17 06:20:56 +03:00
out::shstrtab = new ShstrtabSection;
2020-11-11 15:32:41 +03:00
out::plt = new PltSection;
2021-02-26 14:56:01 +03:00
out::pltgot = new PltGotSection;
2021-01-26 14:11:47 +03:00
if (!config.strip_all)
out::symtab = new SymtabSection;
2020-11-16 17:40:01 +03:00
out::dynsym = new DynsymSection;
2020-11-17 07:19:54 +03:00
out::dynstr = new DynstrSection;
2021-01-21 01:47:48 +03:00
out::eh_frame = new EhFrameSection;
2021-02-28 17:19:29 +03:00
out::copyrel = new CopyrelSection(".bss");
out::copyrel_relro = new CopyrelSection(".bss.rel.ro");
2021-01-21 01:47:48 +03:00
2021-01-26 13:46:34 +03:00
if (config.build_id != BuildIdKind::NONE)
2021-01-15 14:14:01 +03:00
out::buildid = new BuildIdSection;
2021-01-21 01:47:48 +03:00
if (config.eh_frame_hdr)
out::eh_frame_hdr = new EhFrameHdrSection;
2021-01-23 09:01:49 +03:00
if (config.hash_style_sysv)
out::hash = new HashSection;
if (config.hash_style_gnu)
out::gnu_hash = new GnuHashSection;
2020-11-12 08:40:39 +03:00
2020-11-10 13:33:27 +03:00
if (!config.is_static) {
2021-03-05 07:57:00 +03:00
if (!config.shared)
out::interp = new InterpSection;
2020-11-16 18:43:32 +03:00
out::dynamic = new DynamicSection;
2020-11-17 07:30:33 +03:00
out::reldyn = new RelDynSection;
2020-11-29 05:57:58 +03:00
out::versym = new VersymSection;
out::verneed = new VerneedSection;
2020-11-10 13:33:27 +03:00
}
2021-03-07 07:43:58 +03:00
if (!config.version_definitions.empty())
out::verdef = new VerdefSection;
2020-11-20 06:44:02 +03:00
out::chunks.push_back(out::got);
out::chunks.push_back(out::plt);
out::chunks.push_back(out::gotplt);
2021-02-26 14:56:01 +03:00
out::chunks.push_back(out::pltgot);
2020-11-20 06:44:02 +03:00
out::chunks.push_back(out::relplt);
out::chunks.push_back(out::reldyn);
out::chunks.push_back(out::dynamic);
out::chunks.push_back(out::dynsym);
out::chunks.push_back(out::dynstr);
out::chunks.push_back(out::shstrtab);
out::chunks.push_back(out::symtab);
out::chunks.push_back(out::strtab);
out::chunks.push_back(out::hash);
2021-01-23 09:01:49 +03:00
out::chunks.push_back(out::gnu_hash);
2021-01-21 02:53:07 +03:00
out::chunks.push_back(out::eh_frame_hdr);
2021-01-21 07:33:29 +03:00
out::chunks.push_back(out::eh_frame);
2020-11-25 11:20:48 +03:00
out::chunks.push_back(out::copyrel);
2021-02-28 17:19:29 +03:00
out::chunks.push_back(out::copyrel_relro);
2020-11-29 06:59:08 +03:00
out::chunks.push_back(out::versym);
out::chunks.push_back(out::verneed);
2021-03-07 07:43:58 +03:00
out::chunks.push_back(out::verdef);
2021-01-15 14:14:01 +03:00
out::chunks.push_back(out::buildid);
2020-11-20 06:44:02 +03:00
2020-11-22 05:10:38 +03:00
// Set priorities to files. File priority 1 is reserved for the internal file.
2021-01-24 06:01:43 +03:00
i64 priority = 2;
2020-11-24 07:56:14 +03:00
for (ObjectFile *file : out::objs)
2021-02-09 16:07:44 +03:00
if (!file->is_in_lib)
2020-10-28 04:15:05 +03:00
file->priority = priority++;
2020-11-24 07:56:14 +03:00
for (ObjectFile *file : out::objs)
2021-02-09 16:07:44 +03:00
if (file->is_in_lib)
2020-10-28 04:15:05 +03:00
file->priority = priority++;
2020-11-24 10:22:32 +03:00
for (SharedFile *file : out::dsos)
2020-11-24 08:31:05 +03:00
file->priority = priority++;
2020-10-18 13:05:28 +03:00
2020-11-11 04:42:26 +03:00
// Resolve symbols and fix the set of object files that are
// included to the final output.
2020-11-17 08:05:40 +03:00
resolve_symbols();
2020-10-19 15:50:33 +03:00
2020-12-11 06:43:14 +03:00
if (config.trace) {
2020-11-24 07:56:14 +03:00
for (ObjectFile *file : out::objs)
2021-01-09 06:49:39 +03:00
SyncOut() << *file;
2020-11-24 10:22:32 +03:00
for (SharedFile *file : out::dsos)
2021-01-09 06:49:39 +03:00
SyncOut() << *file;
2020-11-24 08:31:05 +03:00
}
2020-11-05 03:24:47 +03:00
2020-11-11 04:42:26 +03:00
// Remove redundant comdat sections (e.g. duplicate inline functions).
2020-11-17 08:04:53 +03:00
eliminate_comdats();
2020-10-10 06:47:12 +03:00
2020-10-27 06:50:25 +03:00
// Create .bss sections for common symbols.
{
2021-01-09 11:45:50 +03:00
Timer t("common");
2020-12-13 16:23:30 +03:00
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
file->convert_common_symbols();
});
2020-10-27 06:50:25 +03:00
}
2021-01-24 10:12:34 +03:00
// Garbage-collect unreachable sections.
if (config.gc_sections)
gc_sections();
2021-01-27 12:18:11 +03:00
// Merge identical read-only sections.
2021-01-27 16:11:47 +03:00
if (config.icf)
icf_sections();
2021-01-27 12:18:11 +03:00
2021-01-25 07:33:16 +03:00
// Merge string constants in SHF_MERGE sections.
handle_mergeable_strings();
2020-10-26 05:34:26 +03:00
// Bin input sections into output sections
2020-11-17 08:04:53 +03:00
bin_sections();
2020-10-23 04:27:11 +03:00
2020-10-29 12:19:10 +03:00
// Assign offsets within an output section to input sections.
2020-11-11 04:42:26 +03:00
set_isec_offsets();
2020-10-22 10:35:17 +03:00
2020-11-12 10:09:17 +03:00
// Sections are added to the section lists in an arbitrary order because
2020-12-13 16:23:30 +03:00
// they are created in parallel. Sort them to to make the output deterministic.
2020-11-12 09:19:30 +03:00
auto section_compare = [](OutputChunk *x, OutputChunk *y) {
2021-01-24 06:01:43 +03:00
return std::tuple(x->name, x->shdr.sh_type, x->shdr.sh_flags) <
std::tuple(y->name, y->shdr.sh_type, y->shdr.sh_flags);
2020-11-12 09:19:30 +03:00
};
2020-12-21 12:42:14 +03:00
sort(OutputSection::instances, section_compare);
sort(MergedSection::instances, section_compare);
2020-11-12 09:19:30 +03:00
2020-11-12 10:09:17 +03:00
// Add sections to the section lists
2020-11-04 04:39:17 +03:00
for (OutputSection *osec : OutputSection::instances)
2020-11-11 16:14:12 +03:00
if (osec->shdr.sh_size)
2020-11-17 07:48:11 +03:00
out::chunks.push_back(osec);
2020-11-07 15:53:21 +03:00
for (MergedSection *osec : MergedSection::instances)
2020-11-11 16:14:12 +03:00
if (osec->shdr.sh_size)
2020-11-17 07:48:11 +03:00
out::chunks.push_back(osec);
2020-11-07 14:31:09 +03:00
2020-12-13 16:31:50 +03:00
erase(out::chunks, [](OutputChunk *c) { return !c; });
2020-11-20 06:44:02 +03:00
// Sort the sections by section flags so that we'll have to create
// as few segments as possible.
2020-12-21 12:42:14 +03:00
sort(out::chunks, [](OutputChunk *a, OutputChunk *b) {
return get_section_rank(a->shdr) < get_section_rank(b->shdr);
});
2020-11-20 06:44:02 +03:00
2020-11-04 04:39:17 +03:00
// Create a dummy file containing linker-synthesized symbols
// (e.g. `__bss_start`).
2021-02-22 07:08:22 +03:00
out::internal_obj = new ObjectFile;
out::internal_obj->resolve_symbols();
out::objs.push_back(out::internal_obj);
2020-11-04 04:39:17 +03:00
2020-11-20 05:08:06 +03:00
// Convert weak symbols to absolute symbols with value 0.
2021-01-10 11:14:08 +03:00
{
Timer t("undef_weak");
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
file->handle_undefined_weak_symbols();
});
}
2020-11-20 05:08:06 +03:00
2021-02-25 14:26:10 +03:00
// If we are linking a .so file, remaining undefined symbols does
// not cause a linker error. Instead, they are treated as if they
// were imported symbols.
if (config.shared) {
Timer t("claim_unresolved_symbols");
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
file->claim_unresolved_symbols();
});
}
2020-11-09 05:31:00 +03:00
// Beyond this point, no new symbols will be added to the result.
2021-01-23 04:46:27 +03:00
// Make sure that all symbols have been resolved.
2021-01-26 13:12:16 +03:00
if (!config.allow_multiple_definition)
check_duplicate_symbols();
2021-01-23 04:46:27 +03:00
2021-03-05 07:09:35 +03:00
compute_visibility();
2020-12-13 16:23:30 +03:00
// Copy shared object name strings to .dynstr.
2020-11-24 10:22:32 +03:00
for (SharedFile *file : out::dsos)
2020-11-30 11:52:08 +03:00
out::dynstr->add_string(file->soname);
2020-11-12 18:14:15 +03:00
2021-03-01 07:17:52 +03:00
// Copy DT_RUNPATH string to .dynstr.
2021-01-15 05:17:52 +03:00
out::dynstr->add_string(config.rpaths);
2020-11-30 11:57:58 +03:00
2021-03-01 07:17:52 +03:00
// Copy DT_SONAME string to .dynstr.
if (!config.soname.empty())
out::dynstr->add_string(config.soname);
2020-11-09 03:47:58 +03:00
// Add headers and sections that have to be at the beginning
// or the ending of a file.
2020-11-17 07:48:11 +03:00
out::chunks.insert(out::chunks.begin(), out::ehdr);
out::chunks.insert(out::chunks.begin() + 1, out::phdr);
2020-11-11 15:32:41 +03:00
if (out::interp)
2020-11-17 07:48:11 +03:00
out::chunks.insert(out::chunks.begin() + 2, out::interp);
out::chunks.push_back(out::shdr);
2020-10-27 11:36:55 +03:00
2021-03-01 10:24:37 +03:00
// Put symbols to .dynsym.
export_dynamic();
2020-11-29 05:06:11 +03:00
// Scan relocations to find symbols that need entries in .got, .plt,
// .got.plt, .dynsym, .dynstr, etc.
2020-11-29 05:10:33 +03:00
scan_rels();
2020-11-20 06:44:02 +03:00
2021-01-18 10:36:13 +03:00
// Sort .dynsym contents. Beyond this point, no symbol should be
// added to .dynsym.
2021-01-14 11:23:14 +03:00
out::dynsym->sort_symbols();
2021-03-07 07:43:58 +03:00
// Apply version scripts
2021-03-07 14:36:10 +03:00
set_import_export();
2021-03-07 07:43:58 +03:00
// Fill .gnu.version_d section contents.
fill_verdef();
2021-03-06 14:49:40 +03:00
// Fill .gnu.version_r section contents.
fill_verneed();
2020-11-29 05:40:57 +03:00
2020-11-29 12:58:36 +03:00
// Compute .symtab and .strtab sizes for each file.
2021-01-25 10:26:00 +03:00
{
Timer t("compute_symtab");
tbb::parallel_for_each(out::objs, [](ObjectFile *file) {
file->compute_symtab();
});
}
2020-11-29 12:58:36 +03:00
2021-01-20 16:41:32 +03:00
// .eh_frame is a special section from the linker's point of view,
// as it's contents are parsed, consumed and reconstructed by the
// linker, unlike other sections that consist of just opaque bytes.
// Here, we transplant .eh_frame sections from a regular output
// section to the special EHFrameSection.
{
Timer t("eh_frame");
erase(out::chunks, [](OutputChunk *chunk) {
return chunk->kind == OutputChunk::REGULAR && chunk->name == ".eh_frame";
});
2021-01-21 01:47:48 +03:00
out::eh_frame->construct();
2021-01-20 16:41:32 +03:00
}
2020-11-19 10:20:09 +03:00
// Now that we have computed sizes for all sections and assigned
// section indices to them, so we can fix section header contents
// for all output sections.
2020-12-07 10:12:19 +03:00
for (OutputChunk *chunk : out::chunks)
chunk->update_shdr();
2020-12-13 16:31:50 +03:00
erase(out::chunks, [](OutputChunk *c) { return c->shdr.sh_size == 0; });
2020-12-07 10:12:19 +03:00
// Set section indices.
2021-01-24 06:01:43 +03:00
for (i64 i = 0, shndx = 1; i < out::chunks.size(); i++)
2020-12-07 10:12:19 +03:00
if (out::chunks[i]->kind != OutputChunk::HEADER)
out::chunks[i]->shndx = shndx++;
for (OutputChunk *chunk : out::chunks)
2020-12-07 09:50:13 +03:00
chunk->update_shdr();
2020-11-16 18:45:02 +03:00
2020-10-30 10:55:59 +03:00
// Assign offsets to output sections
2021-01-24 06:01:43 +03:00
i64 filesize = set_osec_offsets(out::chunks);
2020-10-19 17:37:29 +03:00
2020-11-09 03:58:35 +03:00
// At this point, file layout is fixed. Beyond this, you can assume
// that symbol addresses including their GOT/PLT/etc addresses have
// a correct final value.
2021-01-23 04:46:27 +03:00
// Fix linker-synthesized symbol addresses.
fix_synthetic_symbols(out::chunks);
2021-01-14 09:26:40 +03:00
// Some types of relocations for TLS symbols need the TLS segment
// address. Find it out now.
2021-01-14 07:19:21 +03:00
for (ElfPhdr phdr : create_phdr()) {
if (phdr.p_type == PT_TLS) {
out::tls_begin = phdr.p_vaddr;
2020-11-29 14:19:59 +03:00
out::tls_end = align_to(phdr.p_vaddr + phdr.p_memsz, phdr.p_align);
2021-01-23 04:46:27 +03:00
break;
2021-01-14 07:19:21 +03:00
}
}
2020-10-26 08:38:43 +03:00
2020-12-11 10:51:20 +03:00
t_before_copy.stop();
2020-11-09 10:41:26 +03:00
// Create an output file
OutputFile *file = OutputFile::open(config.output, filesize);
out::buf = file->buf;
2020-11-09 10:41:26 +03:00
2020-12-11 10:51:20 +03:00
Timer t_copy("copy");
2020-11-17 08:49:07 +03:00
// Copy input sections to the output file
2020-11-12 16:06:47 +03:00
{
2021-01-09 11:45:50 +03:00
Timer t("copy_buf");
2020-11-17 07:48:11 +03:00
tbb::parallel_for_each(out::chunks, [&](OutputChunk *chunk) {
2021-01-19 12:40:20 +03:00
chunk->copy_buf();
});
2021-01-13 17:52:11 +03:00
Error::checkpoint();
2020-10-30 05:40:38 +03:00
}
2020-10-20 03:20:52 +03:00
2020-11-09 15:50:47 +03:00
// Zero-clear paddings between sections
2020-11-17 07:59:24 +03:00
clear_padding(filesize);
2020-11-09 15:50:47 +03:00
2020-11-09 10:41:26 +03:00
// Commit
2021-01-24 05:41:36 +03:00
if (out::buildid) {
Timer t("build_id");
2021-01-26 13:46:34 +03:00
out::buildid->write_buildid(filesize);
2021-01-24 05:41:36 +03:00
}
file->close();
2020-10-14 12:41:09 +03:00
2020-12-11 10:51:20 +03:00
t_copy.stop();
2020-12-11 11:04:19 +03:00
t_total.stop();
2020-12-11 10:51:20 +03:00
t_all.stop();
2020-11-06 10:58:13 +03:00
2020-12-11 10:51:20 +03:00
if (config.print_map)
2020-11-24 08:31:05 +03:00
print_map();
2020-10-29 06:24:54 +03:00
2020-12-12 06:57:56 +03:00
// Show stats numbers
2021-03-01 09:39:14 +03:00
if (config.print_stats)
2020-12-21 11:51:20 +03:00
show_stats();
2020-12-13 19:28:43 +03:00
2021-03-01 09:39:14 +03:00
if (config.print_perf)
2020-12-13 16:41:50 +03:00
Timer::print();
std::cout << std::flush;
std::cerr << std::flush;
2020-12-21 12:52:45 +03:00
if (on_complete)
on_complete();
2021-01-22 15:29:28 +03:00
if (config.quick_exit)
std::quick_exit(0);
return 0;
2020-09-29 09:05:29 +03:00
}