* Reimplement how unwind information is stored This commit is a major refactoring of how unwind information is stored after compilation of a function has finished. Previously we would store the raw `UnwindInfo` as a result of compilation and this would get serialized/deserialized alongside the rest of the ELF object that compilation creates. Whenever functions were registered with `CodeMemory` this would also result in registering unwinding information dynamically at runtime, which in the case of Unix, for example, would dynamically created FDE/CIE entries on-the-fly. Eventually I'd like to support compiling Wasmtime without Cranelift, but this means that `UnwindInfo` wouldn't be easily available to decode into and create unwinding information from. To solve this I've changed the ELF object created to have the unwinding information encoded into it ahead-of-time so loading code into memory no longer needs to create unwinding tables. This change has two different implementations for Windows/Unix: * On Windows the implementation was much easier. The unwinding information on Windows is already stored after the function itself in the text section. This was actually slightly duplicated in object building and in code memory allocation. Now the object building continues to do the same, recording unwinding information after functions, and code memory no longer manually tracks this. Additionally Wasmtime will emit a special custom section in the object file with unwinding information which is the list of `RUNTIME_FUNCTION` structures that `RtlAddFunctionTable` expects. This means that the object file has all the information precompiled into it and registration at runtime is simply passing a few pointers around to the runtime. * Unix was a little bit more difficult than Windows. Today a `.eh_frame` section is created on-the-fly with offsets in FDEs specified as the absolute address that functions are loaded at. This absolute address hindered the ability to precompile the FDE into the object file itself. I've switched how addresses are encoded, though, to using `DW_EH_PE_pcrel` which means that FDE addresses are now specified relative to the FDE itself. This means that we can maintain a fixed offset between the `.eh_frame` loaded in memory and the beginning of code memory. When doing so this enables precompiling the `.eh_frame` section into the object file and at runtime when loading an object no further construction of unwinding information is needed. The overall result of this commit is that unwinding information is no longer stored in its cranelift-data-structure form on disk. This means that this unwinding information format is only present during compilation, which will make it that much easier to compile out cranelift in the future. This commit also significantly refactors `CodeMemory` since the way unwinding information is handled is not much different from before. Previously `CodeMemory` was suitable for incrementally adding more and more functions to it, but nowadays a `CodeMemory` either lives per module (in which case all functions are known up front) or it's created once-per-`Func::new` with two trampolines. In both cases we know all functions up front so the functionality of incrementally adding more and more segments is no longer needed. This commit removes the ability to add a function-at-a-time in `CodeMemory` and instead it can now only load objects in their entirety. A small helper function is added to build a small object file for trampolines in `Func::new` to handle allocation there. Finally, this commit also folds the `wasmtime-obj` crate directly into the `wasmtime-cranelift` crate and its builder structure to be more amenable to this strategy of managing unwinding tables. It is not intentional to have any real functional change as a result of this commit. This might accelerate loading a module from cache slightly since less work is needed to manage the unwinding information, but that's just a side benefit from the main goal of this commit which is to remove the dependence on cranelift unwinding information being available at runtime. * Remove isa reexport from wasmtime-environ * Trim down reexports of `cranelift-codegen` Remove everything non-essential so that only the bits which will need to be refactored out of cranelift remain. * Fix debug tests * Review comments
791 lines
26 KiB
Rust
791 lines
26 KiB
Rust
use cranelift_codegen::ir::SourceLoc;
|
|
use gimli::write;
|
|
use more_asserts::assert_le;
|
|
use std::collections::BTreeMap;
|
|
use std::iter::FromIterator;
|
|
use wasmtime_environ::entity::{EntityRef, PrimaryMap};
|
|
use wasmtime_environ::wasm::DefinedFuncIndex;
|
|
use wasmtime_environ::{CompiledFunctions, FunctionAddressMap, WasmFileInfo};
|
|
|
|
pub type GeneratedAddress = usize;
|
|
pub type WasmAddress = u64;
|
|
|
|
/// Contains mapping of the generated address to its original
|
|
/// source location.
|
|
#[derive(Debug)]
|
|
pub struct AddressMap {
|
|
pub generated: GeneratedAddress,
|
|
pub wasm: WasmAddress,
|
|
}
|
|
|
|
/// Information about generated function code: its body start,
|
|
/// length, and instructions addresses.
|
|
#[derive(Debug)]
|
|
pub struct FunctionMap {
|
|
pub offset: GeneratedAddress,
|
|
pub len: GeneratedAddress,
|
|
pub wasm_start: WasmAddress,
|
|
pub wasm_end: WasmAddress,
|
|
pub addresses: Box<[AddressMap]>,
|
|
}
|
|
|
|
/// Mapping of the source location to its generated code range.
|
|
#[derive(Debug)]
|
|
struct Position {
|
|
wasm_pos: WasmAddress,
|
|
gen_start: GeneratedAddress,
|
|
gen_end: GeneratedAddress,
|
|
}
|
|
|
|
/// Mapping of continuous range of source location to its generated
|
|
/// code. The positions are always in ascending order for search.
|
|
#[derive(Debug)]
|
|
struct Range {
|
|
wasm_start: WasmAddress,
|
|
wasm_end: WasmAddress,
|
|
gen_start: GeneratedAddress,
|
|
gen_end: GeneratedAddress,
|
|
positions: Box<[Position]>,
|
|
}
|
|
|
|
type RangeIndex = usize;
|
|
|
|
/// Helper function address lookup data. Contains ranges start positions
|
|
/// index and ranges data. The multiple ranges can include the same
|
|
/// original source position. The index (B-Tree) uses range start
|
|
/// position as a key. The index values reference the ranges array.
|
|
/// The item are ordered RangeIndex.
|
|
#[derive(Debug)]
|
|
struct FuncLookup {
|
|
index: Vec<(WasmAddress, Box<[RangeIndex]>)>,
|
|
ranges: Box<[Range]>,
|
|
}
|
|
|
|
/// Mapping of original functions to generated code locations/ranges.
|
|
#[derive(Debug)]
|
|
struct FuncTransform {
|
|
start: WasmAddress,
|
|
end: WasmAddress,
|
|
index: DefinedFuncIndex,
|
|
lookup: FuncLookup,
|
|
}
|
|
|
|
/// Module functions mapping to generated code.
|
|
#[derive(Debug)]
|
|
pub struct AddressTransform {
|
|
map: PrimaryMap<DefinedFuncIndex, FunctionMap>,
|
|
func: Vec<(WasmAddress, FuncTransform)>,
|
|
}
|
|
|
|
/// Returns a wasm bytecode offset in the code section from SourceLoc.
|
|
pub fn get_wasm_code_offset(loc: SourceLoc, code_section_offset: u64) -> WasmAddress {
|
|
// Code section size <= 4GB, allow wrapped SourceLoc to recover the overflow.
|
|
loc.bits().wrapping_sub(code_section_offset as u32) as WasmAddress
|
|
}
|
|
|
|
fn build_function_lookup(
|
|
ft: &FunctionAddressMap,
|
|
code_section_offset: u64,
|
|
) -> (WasmAddress, WasmAddress, FuncLookup) {
|
|
assert_le!(code_section_offset, ft.start_srcloc.bits() as u64);
|
|
let fn_start = get_wasm_code_offset(ft.start_srcloc, code_section_offset);
|
|
let fn_end = get_wasm_code_offset(ft.end_srcloc, code_section_offset);
|
|
assert_le!(fn_start, fn_end);
|
|
|
|
// Build ranges of continuous source locations. The new ranges starts when
|
|
// non-descending order is interrupted. Assuming the same origin location can
|
|
// be present in multiple ranges.
|
|
let mut range_wasm_start = fn_start;
|
|
let mut range_gen_start = ft.body_offset;
|
|
let mut last_wasm_pos = range_wasm_start;
|
|
let mut ranges = Vec::new();
|
|
let mut ranges_index = BTreeMap::new();
|
|
let mut current_range = Vec::new();
|
|
let mut last_gen_inst_empty = false;
|
|
for (i, t) in ft.instructions.iter().enumerate() {
|
|
if t.srcloc.is_default() {
|
|
continue;
|
|
}
|
|
|
|
let offset = get_wasm_code_offset(t.srcloc, code_section_offset);
|
|
assert_le!(fn_start, offset);
|
|
assert_le!(offset, fn_end);
|
|
|
|
let inst_gen_start = t.code_offset as usize;
|
|
let inst_gen_end = match ft.instructions.get(i + 1) {
|
|
Some(i) => i.code_offset as usize,
|
|
None => ft.body_len as usize,
|
|
};
|
|
|
|
if last_wasm_pos > offset {
|
|
// Start new range.
|
|
ranges_index.insert(range_wasm_start, ranges.len());
|
|
ranges.push(Range {
|
|
wasm_start: range_wasm_start,
|
|
wasm_end: last_wasm_pos,
|
|
gen_start: range_gen_start,
|
|
gen_end: inst_gen_start,
|
|
positions: current_range.into_boxed_slice(),
|
|
});
|
|
range_wasm_start = offset;
|
|
range_gen_start = inst_gen_start;
|
|
current_range = Vec::new();
|
|
last_gen_inst_empty = false;
|
|
}
|
|
if last_gen_inst_empty && current_range.last().unwrap().gen_start == inst_gen_start {
|
|
// It is possible that previous inst_gen_start == inst_gen_end, so
|
|
// make an attempt to merge all such positions with current one.
|
|
if inst_gen_start < inst_gen_end {
|
|
let last = current_range.last_mut().unwrap();
|
|
last.gen_end = inst_gen_end;
|
|
last_gen_inst_empty = false;
|
|
}
|
|
} else {
|
|
// Continue existing range: add new wasm->generated code position.
|
|
current_range.push(Position {
|
|
wasm_pos: offset,
|
|
gen_start: inst_gen_start,
|
|
gen_end: inst_gen_end,
|
|
});
|
|
// Track if last position was empty (see if-branch above).
|
|
last_gen_inst_empty = inst_gen_start == inst_gen_end;
|
|
}
|
|
last_wasm_pos = offset;
|
|
}
|
|
let last_gen_addr = ft.body_offset + ft.body_len as usize;
|
|
ranges_index.insert(range_wasm_start, ranges.len());
|
|
ranges.push(Range {
|
|
wasm_start: range_wasm_start,
|
|
wasm_end: fn_end,
|
|
gen_start: range_gen_start,
|
|
gen_end: last_gen_addr,
|
|
positions: current_range.into_boxed_slice(),
|
|
});
|
|
|
|
// Making ranges lookup faster by building index: B-tree with every range
|
|
// start position that maps into list of active ranges at this position.
|
|
let ranges = ranges.into_boxed_slice();
|
|
let mut active_ranges = Vec::new();
|
|
let mut index = BTreeMap::new();
|
|
let mut last_wasm_pos = None;
|
|
for (wasm_start, range_index) in ranges_index {
|
|
if Some(wasm_start) == last_wasm_pos {
|
|
active_ranges.push(range_index);
|
|
continue;
|
|
}
|
|
if let Some(position) = last_wasm_pos {
|
|
let mut sorted_ranges = active_ranges.clone();
|
|
sorted_ranges.sort();
|
|
index.insert(position, sorted_ranges.into_boxed_slice());
|
|
}
|
|
active_ranges.retain(|r| ranges[*r].wasm_end.cmp(&wasm_start) != std::cmp::Ordering::Less);
|
|
active_ranges.push(range_index);
|
|
last_wasm_pos = Some(wasm_start);
|
|
}
|
|
active_ranges.sort();
|
|
index.insert(last_wasm_pos.unwrap(), active_ranges.into_boxed_slice());
|
|
let index = Vec::from_iter(index.into_iter());
|
|
(fn_start, fn_end, FuncLookup { index, ranges })
|
|
}
|
|
|
|
fn build_function_addr_map(
|
|
funcs: &CompiledFunctions,
|
|
code_section_offset: u64,
|
|
) -> PrimaryMap<DefinedFuncIndex, FunctionMap> {
|
|
let mut map = PrimaryMap::new();
|
|
for (_, f) in funcs {
|
|
let ft = &f.address_map;
|
|
let mut fn_map = Vec::new();
|
|
for t in ft.instructions.iter() {
|
|
if t.srcloc.is_default() {
|
|
continue;
|
|
}
|
|
let offset = get_wasm_code_offset(t.srcloc, code_section_offset);
|
|
fn_map.push(AddressMap {
|
|
generated: t.code_offset as usize,
|
|
wasm: offset,
|
|
});
|
|
}
|
|
|
|
if cfg!(debug) {
|
|
// fn_map is sorted by the generated field -- see FunctionAddressMap::instructions.
|
|
for i in 1..fn_map.len() {
|
|
assert_le!(fn_map[i - 1].generated, fn_map[i].generated);
|
|
}
|
|
}
|
|
|
|
map.push(FunctionMap {
|
|
offset: ft.body_offset,
|
|
len: ft.body_len as usize,
|
|
wasm_start: get_wasm_code_offset(ft.start_srcloc, code_section_offset),
|
|
wasm_end: get_wasm_code_offset(ft.end_srcloc, code_section_offset),
|
|
addresses: fn_map.into_boxed_slice(),
|
|
});
|
|
}
|
|
map
|
|
}
|
|
|
|
// Utility iterator to find all ranges starts for specific Wasm address.
|
|
// The iterator returns generated addresses sorted by RangeIndex.
|
|
struct TransformRangeStartIter<'a> {
|
|
addr: WasmAddress,
|
|
indices: &'a [RangeIndex],
|
|
ranges: &'a [Range],
|
|
}
|
|
|
|
impl<'a> TransformRangeStartIter<'a> {
|
|
fn new(func: &'a FuncTransform, addr: WasmAddress) -> Self {
|
|
let found = match func
|
|
.lookup
|
|
.index
|
|
.binary_search_by(|entry| entry.0.cmp(&addr))
|
|
{
|
|
Ok(i) => Some(&func.lookup.index[i].1),
|
|
Err(i) => {
|
|
if i > 0 {
|
|
Some(&func.lookup.index[i - 1].1)
|
|
} else {
|
|
None
|
|
}
|
|
}
|
|
};
|
|
if let Some(range_indices) = found {
|
|
TransformRangeStartIter {
|
|
addr,
|
|
indices: range_indices,
|
|
ranges: &func.lookup.ranges,
|
|
}
|
|
} else {
|
|
unreachable!();
|
|
}
|
|
}
|
|
}
|
|
|
|
impl<'a> Iterator for TransformRangeStartIter<'a> {
|
|
type Item = (GeneratedAddress, RangeIndex);
|
|
fn next(&mut self) -> Option<Self::Item> {
|
|
if let Some((first, tail)) = self.indices.split_first() {
|
|
let range_index = *first;
|
|
let range = &self.ranges[range_index];
|
|
self.indices = tail;
|
|
let address = match range
|
|
.positions
|
|
.binary_search_by(|a| a.wasm_pos.cmp(&self.addr))
|
|
{
|
|
Ok(i) => range.positions[i].gen_start,
|
|
Err(i) => {
|
|
if i == 0 {
|
|
range.gen_start
|
|
} else {
|
|
range.positions[i - 1].gen_end
|
|
}
|
|
}
|
|
};
|
|
Some((address, range_index))
|
|
} else {
|
|
None
|
|
}
|
|
}
|
|
}
|
|
|
|
// Utility iterator to find all ranges ends for specific Wasm address.
|
|
// The iterator returns generated addresses sorted by RangeIndex.
|
|
struct TransformRangeEndIter<'a> {
|
|
addr: WasmAddress,
|
|
indices: &'a [RangeIndex],
|
|
ranges: &'a [Range],
|
|
}
|
|
|
|
impl<'a> TransformRangeEndIter<'a> {
|
|
fn new(func: &'a FuncTransform, addr: WasmAddress) -> Self {
|
|
let found = match func
|
|
.lookup
|
|
.index
|
|
.binary_search_by(|entry| entry.0.cmp(&addr))
|
|
{
|
|
Ok(i) => Some(&func.lookup.index[i].1),
|
|
Err(i) => {
|
|
if i > 0 {
|
|
Some(&func.lookup.index[i - 1].1)
|
|
} else {
|
|
None
|
|
}
|
|
}
|
|
};
|
|
if let Some(range_indices) = found {
|
|
TransformRangeEndIter {
|
|
addr,
|
|
indices: range_indices,
|
|
ranges: &func.lookup.ranges,
|
|
}
|
|
} else {
|
|
unreachable!();
|
|
}
|
|
}
|
|
}
|
|
|
|
impl<'a> Iterator for TransformRangeEndIter<'a> {
|
|
type Item = (GeneratedAddress, RangeIndex);
|
|
fn next(&mut self) -> Option<Self::Item> {
|
|
while let Some((first, tail)) = self.indices.split_first() {
|
|
let range_index = *first;
|
|
let range = &self.ranges[range_index];
|
|
self.indices = tail;
|
|
if range.wasm_start >= self.addr {
|
|
continue;
|
|
}
|
|
let address = match range
|
|
.positions
|
|
.binary_search_by(|a| a.wasm_pos.cmp(&self.addr))
|
|
{
|
|
Ok(i) => range.positions[i].gen_end,
|
|
Err(i) => {
|
|
if i == range.positions.len() {
|
|
range.gen_end
|
|
} else {
|
|
range.positions[i].gen_start
|
|
}
|
|
}
|
|
};
|
|
return Some((address, range_index));
|
|
}
|
|
None
|
|
}
|
|
}
|
|
|
|
// Utility iterator to iterate by translated function ranges.
|
|
pub struct TransformRangeIter<'a> {
|
|
func: &'a FuncTransform,
|
|
start_it: TransformRangeStartIter<'a>,
|
|
end_it: TransformRangeEndIter<'a>,
|
|
last_start: Option<(GeneratedAddress, RangeIndex)>,
|
|
last_end: Option<(GeneratedAddress, RangeIndex)>,
|
|
last_item: Option<(GeneratedAddress, GeneratedAddress)>,
|
|
}
|
|
|
|
impl<'a> TransformRangeIter<'a> {
|
|
fn new(func: &'a FuncTransform, start: WasmAddress, end: WasmAddress) -> Self {
|
|
let mut start_it = TransformRangeStartIter::new(func, start);
|
|
let last_start = start_it.next();
|
|
let mut end_it = TransformRangeEndIter::new(func, end);
|
|
let last_end = end_it.next();
|
|
TransformRangeIter {
|
|
func,
|
|
start_it,
|
|
end_it,
|
|
last_start,
|
|
last_end,
|
|
last_item: None,
|
|
}
|
|
}
|
|
}
|
|
|
|
impl<'a> Iterator for TransformRangeIter<'a> {
|
|
type Item = (GeneratedAddress, GeneratedAddress);
|
|
fn next(&mut self) -> Option<Self::Item> {
|
|
loop {
|
|
// Merge TransformRangeStartIter and TransformRangeEndIter data using
|
|
// FuncLookup index's field propery to be sorted by RangeIndex.
|
|
let (start, end, range_index): (
|
|
Option<GeneratedAddress>,
|
|
Option<GeneratedAddress>,
|
|
RangeIndex,
|
|
) = {
|
|
match (self.last_start.as_ref(), self.last_end.as_ref()) {
|
|
(Some((s, sri)), Some((e, eri))) => {
|
|
if sri == eri {
|
|
// Start and end RangeIndex matched.
|
|
(Some(*s), Some(*e), *sri)
|
|
} else if sri < eri {
|
|
(Some(*s), None, *sri)
|
|
} else {
|
|
(None, Some(*e), *eri)
|
|
}
|
|
}
|
|
(Some((s, sri)), None) => (Some(*s), None, *sri),
|
|
(None, Some((e, eri))) => (None, Some(*e), *eri),
|
|
(None, None) => {
|
|
// Reached ends for start and end iterators.
|
|
return None;
|
|
}
|
|
}
|
|
};
|
|
let range_start = match start {
|
|
Some(range_start) => {
|
|
// Consume start iterator.
|
|
self.last_start = self.start_it.next();
|
|
range_start
|
|
}
|
|
None => {
|
|
let range = &self.func.lookup.ranges[range_index];
|
|
range.gen_start
|
|
}
|
|
};
|
|
let range_end = match end {
|
|
Some(range_end) => {
|
|
// Consume end iterator.
|
|
self.last_end = self.end_it.next();
|
|
range_end
|
|
}
|
|
None => {
|
|
let range = &self.func.lookup.ranges[range_index];
|
|
range.gen_end
|
|
}
|
|
};
|
|
|
|
if cfg!(debug_assertions) {
|
|
match self.last_item.replace((range_start, range_end)) {
|
|
Some((_, last_end)) => debug_assert!(last_end <= range_start),
|
|
None => (),
|
|
}
|
|
}
|
|
|
|
if range_start < range_end {
|
|
return Some((range_start, range_end));
|
|
}
|
|
// Throw away empty ranges.
|
|
debug_assert!(range_start == range_end);
|
|
}
|
|
}
|
|
}
|
|
|
|
impl AddressTransform {
|
|
pub fn new(funcs: &CompiledFunctions, wasm_file: &WasmFileInfo) -> Self {
|
|
let code_section_offset = wasm_file.code_section_offset;
|
|
|
|
let mut func = BTreeMap::new();
|
|
for (i, f) in funcs {
|
|
let ft = &f.address_map;
|
|
let (fn_start, fn_end, lookup) = build_function_lookup(ft, code_section_offset);
|
|
|
|
func.insert(
|
|
fn_start,
|
|
FuncTransform {
|
|
start: fn_start,
|
|
end: fn_end,
|
|
index: i,
|
|
lookup,
|
|
},
|
|
);
|
|
}
|
|
|
|
let map = build_function_addr_map(funcs, code_section_offset);
|
|
let func = Vec::from_iter(func.into_iter());
|
|
AddressTransform { map, func }
|
|
}
|
|
|
|
fn find_func(&self, addr: u64) -> Option<&FuncTransform> {
|
|
// TODO check if we need to include end address
|
|
let func = match self.func.binary_search_by(|entry| entry.0.cmp(&addr)) {
|
|
Ok(i) => &self.func[i].1,
|
|
Err(i) => {
|
|
if i > 0 {
|
|
&self.func[i - 1].1
|
|
} else {
|
|
return None;
|
|
}
|
|
}
|
|
};
|
|
if addr >= func.start {
|
|
return Some(func);
|
|
}
|
|
None
|
|
}
|
|
|
|
pub fn find_func_index(&self, addr: u64) -> Option<DefinedFuncIndex> {
|
|
self.find_func(addr).map(|f| f.index)
|
|
}
|
|
|
|
pub fn translate_raw(&self, addr: u64) -> Option<(DefinedFuncIndex, GeneratedAddress)> {
|
|
if addr == 0 {
|
|
// It's normally 0 for debug info without the linked code.
|
|
return None;
|
|
}
|
|
if let Some(func) = self.find_func(addr) {
|
|
if addr == func.end {
|
|
// Clamp last address to the end to extend translation to the end
|
|
// of the function.
|
|
let map = &self.map[func.index];
|
|
return Some((func.index, map.len));
|
|
}
|
|
let first_result = TransformRangeStartIter::new(func, addr).next();
|
|
first_result.map(|(address, _)| (func.index, address))
|
|
} else {
|
|
// Address was not found: function was not compiled?
|
|
None
|
|
}
|
|
}
|
|
|
|
pub fn can_translate_address(&self, addr: u64) -> bool {
|
|
self.translate(addr).is_some()
|
|
}
|
|
|
|
pub fn translate(&self, addr: u64) -> Option<write::Address> {
|
|
self.translate_raw(addr)
|
|
.map(|(func_index, address)| write::Address::Symbol {
|
|
symbol: func_index.index(),
|
|
addend: address as i64,
|
|
})
|
|
}
|
|
|
|
pub fn translate_ranges_raw<'a>(
|
|
&'a self,
|
|
start: u64,
|
|
end: u64,
|
|
) -> Option<(DefinedFuncIndex, impl Iterator<Item = (usize, usize)> + 'a)> {
|
|
if start == 0 {
|
|
// It's normally 0 for debug info without the linked code.
|
|
return None;
|
|
}
|
|
if let Some(func) = self.find_func(start) {
|
|
let result = TransformRangeIter::new(func, start, end);
|
|
return Some((func.index, result));
|
|
}
|
|
// Address was not found: function was not compiled?
|
|
None
|
|
}
|
|
|
|
pub fn translate_ranges<'a>(
|
|
&'a self,
|
|
start: u64,
|
|
end: u64,
|
|
) -> impl Iterator<Item = (write::Address, u64)> + 'a {
|
|
enum TranslateRangesResult<'a> {
|
|
Empty,
|
|
Raw {
|
|
symbol: usize,
|
|
it: Box<dyn Iterator<Item = (usize, usize)> + 'a>,
|
|
},
|
|
}
|
|
impl<'a> Iterator for TranslateRangesResult<'a> {
|
|
type Item = (write::Address, u64);
|
|
fn next(&mut self) -> Option<Self::Item> {
|
|
match self {
|
|
TranslateRangesResult::Empty => None,
|
|
TranslateRangesResult::Raw { symbol, it } => match it.next() {
|
|
Some((start, end)) => {
|
|
debug_assert!(start < end);
|
|
Some((
|
|
write::Address::Symbol {
|
|
symbol: *symbol,
|
|
addend: start as i64,
|
|
},
|
|
(end - start) as u64,
|
|
))
|
|
}
|
|
None => None,
|
|
},
|
|
}
|
|
}
|
|
}
|
|
|
|
match self.translate_ranges_raw(start, end) {
|
|
Some((func_index, ranges)) => TranslateRangesResult::Raw {
|
|
symbol: func_index.index(),
|
|
it: Box::new(ranges),
|
|
},
|
|
None => TranslateRangesResult::Empty,
|
|
}
|
|
}
|
|
|
|
pub fn map(&self) -> &PrimaryMap<DefinedFuncIndex, FunctionMap> {
|
|
&self.map
|
|
}
|
|
|
|
pub fn func_range(&self, index: DefinedFuncIndex) -> (GeneratedAddress, GeneratedAddress) {
|
|
let map = &self.map[index];
|
|
(map.offset, map.offset + map.len)
|
|
}
|
|
|
|
pub fn func_source_range(&self, index: DefinedFuncIndex) -> (WasmAddress, WasmAddress) {
|
|
let map = &self.map[index];
|
|
(map.wasm_start, map.wasm_end)
|
|
}
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use super::{build_function_lookup, get_wasm_code_offset, AddressTransform};
|
|
use gimli::write::Address;
|
|
use std::iter::FromIterator;
|
|
use std::mem;
|
|
use wasmtime_environ::entity::PrimaryMap;
|
|
use wasmtime_environ::ir::SourceLoc;
|
|
use wasmtime_environ::{CompiledFunction, WasmFileInfo};
|
|
use wasmtime_environ::{CompiledFunctions, FunctionAddressMap, InstructionAddressMap};
|
|
|
|
#[test]
|
|
fn test_get_wasm_code_offset() {
|
|
let offset = get_wasm_code_offset(SourceLoc::new(3), 1);
|
|
assert_eq!(2, offset);
|
|
let offset = get_wasm_code_offset(SourceLoc::new(16), 0xF000_0000);
|
|
assert_eq!(0x1000_0010, offset);
|
|
let offset = get_wasm_code_offset(SourceLoc::new(1), 0x20_8000_0000);
|
|
assert_eq!(0x8000_0001, offset);
|
|
}
|
|
|
|
fn create_simple_func(wasm_offset: u32) -> FunctionAddressMap {
|
|
FunctionAddressMap {
|
|
instructions: vec![
|
|
InstructionAddressMap {
|
|
srcloc: SourceLoc::new(wasm_offset + 2),
|
|
code_offset: 5,
|
|
},
|
|
InstructionAddressMap {
|
|
srcloc: SourceLoc::default(),
|
|
code_offset: 8,
|
|
},
|
|
InstructionAddressMap {
|
|
srcloc: SourceLoc::new(wasm_offset + 7),
|
|
code_offset: 15,
|
|
},
|
|
InstructionAddressMap {
|
|
srcloc: SourceLoc::default(),
|
|
code_offset: 23,
|
|
},
|
|
]
|
|
.into(),
|
|
start_srcloc: SourceLoc::new(wasm_offset),
|
|
end_srcloc: SourceLoc::new(wasm_offset + 10),
|
|
body_offset: 0,
|
|
body_len: 30,
|
|
}
|
|
}
|
|
|
|
fn create_simple_module(address_map: FunctionAddressMap) -> CompiledFunctions {
|
|
PrimaryMap::from_iter(vec![CompiledFunction {
|
|
address_map,
|
|
..Default::default()
|
|
}])
|
|
}
|
|
|
|
#[test]
|
|
fn test_build_function_lookup_simple() {
|
|
let input = create_simple_func(11);
|
|
let (start, end, lookup) = build_function_lookup(&input, 1);
|
|
assert_eq!(10, start);
|
|
assert_eq!(20, end);
|
|
|
|
assert_eq!(1, lookup.index.len());
|
|
let index_entry = lookup.index.into_iter().next().unwrap();
|
|
assert_eq!((10u64, vec![0].into_boxed_slice()), index_entry);
|
|
assert_eq!(1, lookup.ranges.len());
|
|
let range = &lookup.ranges[0];
|
|
assert_eq!(10, range.wasm_start);
|
|
assert_eq!(20, range.wasm_end);
|
|
assert_eq!(0, range.gen_start);
|
|
assert_eq!(30, range.gen_end);
|
|
let positions = &range.positions;
|
|
assert_eq!(2, positions.len());
|
|
assert_eq!(12, positions[0].wasm_pos);
|
|
assert_eq!(5, positions[0].gen_start);
|
|
assert_eq!(8, positions[0].gen_end);
|
|
assert_eq!(17, positions[1].wasm_pos);
|
|
assert_eq!(15, positions[1].gen_start);
|
|
assert_eq!(23, positions[1].gen_end);
|
|
}
|
|
|
|
#[test]
|
|
fn test_build_function_lookup_two_ranges() {
|
|
let mut input = create_simple_func(11);
|
|
// append instruction with same srcloc as input.instructions[0]
|
|
let mut list = Vec::from(mem::take(&mut input.instructions));
|
|
list.push(InstructionAddressMap {
|
|
srcloc: SourceLoc::new(11 + 2),
|
|
code_offset: 23,
|
|
});
|
|
list.push(InstructionAddressMap {
|
|
srcloc: SourceLoc::default(),
|
|
code_offset: 26,
|
|
});
|
|
input.instructions = list.into();
|
|
let (start, end, lookup) = build_function_lookup(&input, 1);
|
|
assert_eq!(10, start);
|
|
assert_eq!(20, end);
|
|
|
|
assert_eq!(2, lookup.index.len());
|
|
let index_entries = Vec::from_iter(lookup.index.into_iter());
|
|
assert_eq!((10u64, vec![0].into_boxed_slice()), index_entries[0]);
|
|
assert_eq!((12u64, vec![0, 1].into_boxed_slice()), index_entries[1]);
|
|
assert_eq!(2, lookup.ranges.len());
|
|
|
|
let range = &lookup.ranges[0];
|
|
assert_eq!(10, range.wasm_start);
|
|
assert_eq!(17, range.wasm_end);
|
|
assert_eq!(0, range.gen_start);
|
|
assert_eq!(23, range.gen_end);
|
|
let positions = &range.positions;
|
|
assert_eq!(2, positions.len());
|
|
assert_eq!(12, positions[0].wasm_pos);
|
|
assert_eq!(5, positions[0].gen_start);
|
|
assert_eq!(8, positions[0].gen_end);
|
|
assert_eq!(17, positions[1].wasm_pos);
|
|
assert_eq!(15, positions[1].gen_start);
|
|
assert_eq!(23, positions[1].gen_end);
|
|
|
|
let range = &lookup.ranges[1];
|
|
assert_eq!(12, range.wasm_start);
|
|
assert_eq!(20, range.wasm_end);
|
|
assert_eq!(23, range.gen_start);
|
|
assert_eq!(30, range.gen_end);
|
|
let positions = &range.positions;
|
|
assert_eq!(1, positions.len());
|
|
assert_eq!(12, positions[0].wasm_pos);
|
|
assert_eq!(23, positions[0].gen_start);
|
|
assert_eq!(26, positions[0].gen_end);
|
|
}
|
|
|
|
#[test]
|
|
fn test_addr_translate() {
|
|
let input = create_simple_module(create_simple_func(11));
|
|
let at = AddressTransform::new(
|
|
&input,
|
|
&WasmFileInfo {
|
|
path: None,
|
|
code_section_offset: 1,
|
|
imported_func_count: 0,
|
|
funcs: Vec::new(),
|
|
},
|
|
);
|
|
|
|
let addr = at.translate(10);
|
|
assert_eq!(
|
|
Some(Address::Symbol {
|
|
symbol: 0,
|
|
addend: 0,
|
|
}),
|
|
addr
|
|
);
|
|
|
|
let addr = at.translate(20);
|
|
assert_eq!(
|
|
Some(Address::Symbol {
|
|
symbol: 0,
|
|
addend: 30,
|
|
}),
|
|
addr
|
|
);
|
|
|
|
let addr = at.translate(0);
|
|
assert_eq!(None, addr);
|
|
|
|
let addr = at.translate(12);
|
|
assert_eq!(
|
|
Some(Address::Symbol {
|
|
symbol: 0,
|
|
addend: 5,
|
|
}),
|
|
addr
|
|
);
|
|
|
|
let addr = at.translate(18);
|
|
assert_eq!(
|
|
Some(Address::Symbol {
|
|
symbol: 0,
|
|
addend: 23,
|
|
}),
|
|
addr
|
|
);
|
|
}
|
|
}
|