Skip to content

Commit

Permalink
Merge pull request #7435 from roc-lang/revert-PR7424
Browse files Browse the repository at this point in the history 
Revert "Merge pull request #7424 from kubkon/macho-surgery"
  • Loading branch information
@Anton-4
Anton-4 authored Dec 30, 2024
2 parents eda4db0 + 2dec72a commit d599e8c
Show file tree
Hide file tree
Showing 4 changed files with 87 additions and 188 deletions.
11 changes: 6 additions & 5 deletions crates/compiler/build/src/link.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1041,13 +1041,14 @@ fn link_macos(
// "--gc-sections",
"-arch",
&arch,
// Suppress warnings, because otherwise it prints:
//
// ld: warning: -undefined dynamic_lookup may not work with chained fixups
//
// We can't disable that option without breaking either x64 mac or ARM mac
"-w",
"-macos_version_min",
&get_macos_version(),
// Suppress fixup chains to ease working out dynamic relocs by the
// surgical linker. In my experience, working with dyld opcodes is
// slightly easier than unpacking compressed info from the __got section
// and fixups load command.
"-no_fixup_chains",
])
.args(input_paths)
.args(extra_link_flags());
Expand Down
36 changes: 6 additions & 30 deletions crates/linker/src/generate_dylib/macho.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -62,7 +62,6 @@ pub fn create_dylib_macho(

let ld_flag_soname = "-install_name";
let ld_prefix_args = [big_sur_fix, "-lSystem", "-dylib"];
let macos_version = get_macos_version();

let output = Command::new("ld")
.args(ld_prefix_args)
Expand All @@ -72,15 +71,12 @@ pub fn create_dylib_macho(
dummy_obj_file.path().to_str().unwrap(),
"-o",
dummy_lib_file.to_str().unwrap(),
// Suppress fixup chains to ease working out dynamic relocs by the
// surgical linker. In my experience, working with dyld opcodes is
// slightly easier than unpacking compressed info from the __got section
// and fixups load command.
"-no_fixup_chains",
"-platform_version",
"macos",
&macos_version,
&macos_version,
// Suppress warnings, because otherwise it prints:
//
// ld: warning: -undefined dynamic_lookup may not work with chained fixups
//
// We can't disable that option without breaking either x64 mac or ARM mac
"-w",
])
.output()
.unwrap();
Expand All @@ -102,23 +98,3 @@ pub fn create_dylib_macho(

Ok(std::fs::read(dummy_lib_file).expect("Failed to load dummy library"))
}

fn get_macos_version() -> String {
let mut cmd = Command::new("sw_vers");
cmd.arg("-productVersion");

let cmd_stdout = cmd
.output()
.expect("Failed to execute command 'sw_vers -productVersion'")
.stdout;

let full_version_string = String::from_utf8(cmd_stdout)
.expect("Failed to convert output of command 'sw_vers -productVersion' into a utf8 string");

full_version_string
.trim_end()
.split('.')
.take(3)
.collect::<Vec<&str>>()
.join(".")
}
1 change: 0 additions & 1 deletion crates/linker/src/lib.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,6 @@ pub fn supported(link_type: LinkType, target: Target) -> bool {
Target::WinX64 => true,
// macho support is incomplete
Target::MacX64 => false,
Target::MacArm64 => true,
_ => false,
}
} else {
Expand Down
227 changes: 75 additions & 152 deletions crates/linker/src/macho.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -109,19 +109,20 @@ fn collect_roc_definitions<'a>(object: &object::File<'a, &'a [u8]>) -> MutMap<St
let mut vaddresses = MutMap::default();

for sym in object.symbols().filter(is_roc_definition) {
let name = sym.name().unwrap().trim_start_matches('_');
// remove potentially trailing "@version".
let name = sym
.name()
.unwrap()
.trim_start_matches('_')
.split('@')
.next()
.unwrap();

let address = sym.address();

// special exceptions for memcpy and memset.
let direct_mapping = match name {
"roc_memset" => Some("memset"),
"roc_memmove" => Some("memmove"),

_ => None,
};

if let Some(libc_symbol) = direct_mapping {
vaddresses.insert(libc_symbol.to_string(), address);
if name == "roc_memset" {
vaddresses.insert("memset".to_string(), address);
}

vaddresses.insert(name.to_string(), address);
Expand Down Expand Up @@ -1207,7 +1208,6 @@ fn surgery_macho_help(
let rodata_sections: Vec<Section> = app_obj
.sections()
.filter(|sec| sec.kind() == SectionKind::ReadOnlyData)
.filter(|sec| sec.name().unwrap_or("") != "__eh_frame") // ignore __eh_frame for now
.collect();

// bss section is like rodata section, but it has zero file size and non-zero virtual size.
Expand All @@ -1224,15 +1224,6 @@ fn surgery_macho_help(
internal_error!("No text sections found. This application has no code.");
}

if verbose {
println!();
println!("Roc symbol addresses: {:+x?}", md.roc_symbol_vaddresses);
println!("App functions: {:?}", md.app_functions);
println!("Dynamic symbol indices: {:+x?}", md.dynamic_symbol_indices);
println!("PLT addresses: {:+x?}", md.plt_addresses);
println!();
}

// Calculate addresses and load symbols.
// Note, it is important the bss sections come after the rodata sections.
for sec in rodata_sections
Expand All @@ -1249,7 +1240,7 @@ fn surgery_macho_help(
sec.name().unwrap(),
offset,
virt_offset
);
)
}
section_offset_map.insert(sec.index(), (offset, virt_offset));
for sym in symbols.iter() {
Expand All @@ -1268,7 +1259,7 @@ fn surgery_macho_help(
Some((_, size)) => size,
None => 0,
};
if sec.name().unwrap_or_default().starts_with("__bss") {
if sec.name().unwrap_or_default().starts_with("__BSS") {
// bss sections only modify the virtual size.
virt_offset += sec.size() as usize;
} else if section_size != sec.size() {
Expand All @@ -1295,137 +1286,98 @@ fn surgery_macho_help(
for sec in rodata_sections
.iter()
.chain(bss_sections.iter())
// TODO why do we even include uninitialized data if it cannot
// ever have any relocations in the first place?
.chain(text_sections.iter())
{
let data = sec.data().unwrap_or_else(|err| {
internal_error!(
"Failed to load data for section, {:+x?}: {err}",
sec.name().unwrap(),
);
});
let data = match sec.data() {
Ok(data) => data,
Err(err) => {
internal_error!(
"Failed to load data for section, {:+x?}: {}",
sec.name().unwrap(),
err
);
}
};
let (section_offset, section_virtual_offset) =
section_offset_map.get(&sec.index()).unwrap();
let (section_offset, section_virtual_offset) = (*section_offset, *section_virtual_offset);
exec_mmap[section_offset..section_offset + data.len()].copy_from_slice(data);
// Deal with definitions and relocations for this section.
if verbose {
let segname = sec
.segment_name()
.expect(
"valid segment
name",
)
.unwrap();
let sectname = sec.name().unwrap();
println!();
println!(
"Processing Relocations for Section '{segname},{sectname}': 0x{sec:+x?} @ {section_offset:+x} (virt: {section_virtual_offset:+x})"
"Processing Relocations for Section: 0x{sec:+x?} @ {section_offset:+x} (virt: {section_virtual_offset:+x})"
);
}

let mut subtractor: Option<SymbolIndex> = None;
for rel in sec.relocations() {
if verbose {
println!("\tFound Relocation: {rel:+x?}");
}
match rel.1.target() {
RelocationTarget::Symbol(index) => {
let target_offset = if let Some(target_offset) =
get_target_offset(index, &app_obj, md, &symbol_vaddr_map, verbose)
{
target_offset
} else if matches!(app_obj.symbol_by_index(index), Ok(sym) if ["__divti3", "__udivti3", "___divti3", "___udivti3"].contains(&sym.name().unwrap_or_default()))
{
// Explicitly ignore some symbols that are currently always linked.
continue;
} else if matches!(app_obj.symbol_by_index(index), Ok(sym) if ["_longjmp", "_setjmp"].contains(&sym.name().unwrap_or_default()))
{
// These symbols have to stay undefined as we dynamically link them from libSystem.dylib at runtime.
// TODO have a table of all known symbols; perhaps parse and use an Apple provided libSystem.tbd stub file?
let name = app_obj
let target_offset = if let Some(target_offset) = symbol_vaddr_map.get(&index) {
if verbose {
println!("\t\tRelocation targets symbol in app at: {target_offset:+x}");
}
Some(*target_offset as i64)
} else {
app_obj
.symbol_by_index(index)
.and_then(|sym| sym.name())
.ok()
.unwrap();
match rel.1.kind() {
RelocationKind::PltRelative => {
println!("\t\tTODO synthesise __stub entry for {name}")
.and_then(|name| {
md.roc_symbol_vaddresses.get(name).map(|address| {
let vaddr = (*address + md.added_byte_count) as i64;
if verbose {
println!(
"\t\tRelocation targets symbol in host: {name} @ {vaddr:+x}"
);
}
vaddr
})
})
};

if let Some(target_offset) = target_offset {
let virt_base = section_virtual_offset + rel.0 as usize;
let base = section_offset + rel.0 as usize;
let target: i64 = match rel.1.kind() {
RelocationKind::Relative | RelocationKind::PltRelative => {
target_offset - virt_base as i64 + rel.1.addend()
}
RelocationKind::Got => {
println!("\t\tTODO synthesise __got entry for {name}")
x => {
internal_error!("Relocation Kind not yet support: {:?}", x);
}
_ => internal_error!(
"Invalid relocation for libc symbol, {:+x?}: {name}",
rel
),
};
if verbose {
println!(
"\t\tRelocation base location: {base:+x} (virt: {virt_base:+x})"
);
println!("\t\tFinal relocation target offset: {target:+x}");
}
match rel.1.size() {
32 => {
let data = (target as i32).to_le_bytes();
exec_mmap[base..base + 4].copy_from_slice(&data);
}
64 => {
let data = target.to_le_bytes();
exec_mmap[base..base + 8].copy_from_slice(&data);
}
x => {
internal_error!("Relocation size not yet supported: {}", x);
}
}
} else if matches!(app_obj.symbol_by_index(index), Ok(sym) if ["__divti3", "__udivti3", "___divti3", "___udivti3"].contains(&sym.name().unwrap_or_default()))
{
// Explicitly ignore some symbols that are currently always linked.
continue;
} else {
internal_error!(
"Undefined Symbol in relocation, {:+x?}: {:+x?}",
rel,
app_obj.symbol_by_index(index)
)
};

let virt_base = section_virtual_offset + rel.0 as usize;
let base = section_offset + rel.0 as usize;
let target: i64 = match rel.1.kind() {
RelocationKind::Relative | RelocationKind::PltRelative => {
target_offset - virt_base as i64 + rel.1.addend()
}
RelocationKind::Absolute => {
target_offset + rel.1.addend()
- subtractor
.take()
.map(|index| {
get_target_offset(
index,
&app_obj,
md,
&symbol_vaddr_map,
verbose,
)
.unwrap_or(0)
})
.unwrap()
}
RelocationKind::MachO { value, relative: _ } => match value {
macho::ARM64_RELOC_SUBTRACTOR => {
if subtractor.is_some() {
internal_error!("Malformed object: SUBTRACTOR must not be followed by SUBTRACTOR");
} else {
subtractor = Some(index);
}
continue;
}
_ => {
println!("\t\tHandle other MachO relocs: {value}");
0
}
},
x => {
internal_error!("Relocation Kind not yet support: {:?}", x);
}
};
if verbose {
println!("\t\tRelocation base location: {base:+x} (virt: {virt_base:+x})");
println!("\t\tFinal relocation target offset: {target:+x}");
}
match rel.1.size() {
32 => {
let data = (target as i32).to_le_bytes();
exec_mmap[base..base + 4].copy_from_slice(&data);
}
64 => {
let data = target.to_le_bytes();
exec_mmap[base..base + 8].copy_from_slice(&data);
}
x => {
internal_error!("Relocation size not yet supported: {}", x);
}
);
}
}

Expand Down Expand Up @@ -1626,32 +1578,3 @@ fn surgery_macho_help(

*offset_ref = offset;
}

fn get_target_offset(
index: SymbolIndex,
app_obj: &object::File,
md: &Metadata,
symbol_vaddr_map: &MutMap<SymbolIndex, usize>,
verbose: bool,
) -> Option<i64> {
if let Some(target_offset) = symbol_vaddr_map.get(&index) {
if verbose {
println!("\t\tRelocation targets symbol in app at: {target_offset:+x}");
}
Some(*target_offset as i64)
} else {
app_obj
.symbol_by_index(index)
.and_then(|sym| sym.name().map(|name| name.trim_start_matches('_')))
.ok()
.and_then(|name| {
md.roc_symbol_vaddresses.get(name).map(|address| {
let vaddr = (*address + md.added_byte_count) as i64;
if verbose {
println!("\t\tRelocation targets symbol in host: {name} @ {vaddr:+x}");
}
vaddr
})
})
}
}

0 comments on commit d599e8c

Please sign in to comment.