wasmtime_cranelift/debug/transform/
simulate.rs

1use super::expression::{CompiledExpression, FunctionFrameInfo};
2use super::utils::{add_internal_types, append_vmctx_info, get_function_frame_info};
3use super::AddressTransform;
4use crate::debug::ModuleMemoryOffset;
5use crate::CompiledFunctions;
6use anyhow::{Context, Error};
7use cranelift_codegen::isa::TargetIsa;
8use cranelift_wasm::get_vmctx_value_label;
9use gimli::write;
10use gimli::{self, LineEncoding};
11use std::collections::{HashMap, HashSet};
12use std::path::PathBuf;
13use std::sync::atomic::{AtomicUsize, Ordering::SeqCst};
14use wasmparser::ValType as WasmType;
15use wasmtime_environ::{
16    DebugInfoData, DefinedFuncIndex, EntityRef, FuncIndex, FunctionMetadata, WasmFileInfo,
17};
18
19const PRODUCER_NAME: &str = "wasmtime";
20
21macro_rules! assert_dwarf_str {
22    ($s:expr) => {{
23        let s = $s;
24        if cfg!(debug_assertions) {
25            // Perform check the same way as gimli does it.
26            let bytes: Vec<u8> = s.clone().into();
27            debug_assert!(!bytes.contains(&0), "DWARF string shall not have NULL byte");
28        }
29        s
30    }};
31}
32
33fn generate_line_info(
34    addr_tr: &AddressTransform,
35    translated: &HashSet<DefinedFuncIndex>,
36    out_encoding: gimli::Encoding,
37    w: &WasmFileInfo,
38    comp_dir_id: write::StringId,
39    name_id: write::StringId,
40    name: &str,
41) -> Result<write::LineProgram, Error> {
42    let out_comp_dir = write::LineString::StringRef(comp_dir_id);
43    let out_comp_name = write::LineString::StringRef(name_id);
44
45    let line_encoding = LineEncoding::default();
46
47    let mut out_program = write::LineProgram::new(
48        out_encoding,
49        line_encoding,
50        out_comp_dir,
51        out_comp_name,
52        None,
53    );
54
55    let file_index = out_program.add_file(
56        write::LineString::String(name.as_bytes().to_vec()),
57        out_program.default_directory(),
58        None,
59    );
60
61    for (i, map) in addr_tr.map() {
62        let symbol = i.index();
63        if translated.contains(&i) {
64            continue;
65        }
66
67        let base_addr = map.offset;
68        out_program.begin_sequence(Some(write::Address::Symbol { symbol, addend: 0 }));
69        for addr_map in map.addresses.iter() {
70            let address_offset = (addr_map.generated - base_addr) as u64;
71            out_program.row().address_offset = address_offset;
72            out_program.row().op_index = 0;
73            out_program.row().file = file_index;
74            let wasm_offset = w.code_section_offset + addr_map.wasm as u64;
75            out_program.row().line = wasm_offset;
76            out_program.row().column = 0;
77            out_program.row().discriminator = 1;
78            out_program.row().is_statement = true;
79            out_program.row().basic_block = false;
80            out_program.row().prologue_end = false;
81            out_program.row().epilogue_begin = false;
82            out_program.row().isa = 0;
83            out_program.generate_row();
84        }
85        let end_addr = (map.offset + map.len - 1) as u64;
86        out_program.end_sequence(end_addr);
87    }
88
89    Ok(out_program)
90}
91
92fn check_invalid_chars_in_name(s: &str) -> Option<&str> {
93    if s.contains('\x00') {
94        None
95    } else {
96        Some(s)
97    }
98}
99
100fn autogenerate_dwarf_wasm_path(di: &DebugInfoData) -> PathBuf {
101    static NEXT_ID: AtomicUsize = AtomicUsize::new(0);
102    let module_name = di
103        .name_section
104        .module_name
105        .and_then(check_invalid_chars_in_name)
106        .map(|s| s.to_string())
107        .unwrap_or_else(|| format!("<gen-{}>", NEXT_ID.fetch_add(1, SeqCst)));
108    let path = format!("/<wasm-module>/{}.wasm", module_name);
109    PathBuf::from(path)
110}
111
112struct WasmTypesDieRefs {
113    vmctx: write::UnitEntryId,
114    i32: write::UnitEntryId,
115    i64: write::UnitEntryId,
116    f32: write::UnitEntryId,
117    f64: write::UnitEntryId,
118}
119
120fn add_wasm_types(
121    unit: &mut write::Unit,
122    root_id: write::UnitEntryId,
123    out_strings: &mut write::StringTable,
124    memory_offset: &ModuleMemoryOffset,
125) -> WasmTypesDieRefs {
126    let (_wp_die_id, vmctx_die_id) = add_internal_types(unit, root_id, out_strings, memory_offset);
127
128    macro_rules! def_type {
129        ($id:literal, $size:literal, $enc:path) => {{
130            let die_id = unit.add(root_id, gimli::DW_TAG_base_type);
131            let die = unit.get_mut(die_id);
132            die.set(
133                gimli::DW_AT_name,
134                write::AttributeValue::StringRef(out_strings.add($id)),
135            );
136            die.set(gimli::DW_AT_byte_size, write::AttributeValue::Data1($size));
137            die.set(gimli::DW_AT_encoding, write::AttributeValue::Encoding($enc));
138            die_id
139        }};
140    }
141
142    let i32_die_id = def_type!("i32", 4, gimli::DW_ATE_signed);
143    let i64_die_id = def_type!("i64", 8, gimli::DW_ATE_signed);
144    let f32_die_id = def_type!("f32", 4, gimli::DW_ATE_float);
145    let f64_die_id = def_type!("f64", 8, gimli::DW_ATE_float);
146
147    WasmTypesDieRefs {
148        vmctx: vmctx_die_id,
149        i32: i32_die_id,
150        i64: i64_die_id,
151        f32: f32_die_id,
152        f64: f64_die_id,
153    }
154}
155
156fn resolve_var_type(
157    index: usize,
158    wasm_types: &WasmTypesDieRefs,
159    func_meta: &FunctionMetadata,
160) -> Option<(write::UnitEntryId, bool)> {
161    let (ty, is_param) = if index < func_meta.params.len() {
162        (func_meta.params[index], true)
163    } else {
164        let mut i = (index - func_meta.params.len()) as u32;
165        let mut j = 0;
166        while j < func_meta.locals.len() && i >= func_meta.locals[j].0 {
167            i -= func_meta.locals[j].0;
168            j += 1;
169        }
170        if j >= func_meta.locals.len() {
171            // Ignore the var index out of bound.
172            return None;
173        }
174        (func_meta.locals[j].1, false)
175    };
176    let type_die_id = match ty {
177        WasmType::I32 => wasm_types.i32,
178        WasmType::I64 => wasm_types.i64,
179        WasmType::F32 => wasm_types.f32,
180        WasmType::F64 => wasm_types.f64,
181        _ => {
182            // Ignore unsupported types.
183            return None;
184        }
185    };
186    Some((type_die_id, is_param))
187}
188
189fn generate_vars(
190    unit: &mut write::Unit,
191    die_id: write::UnitEntryId,
192    addr_tr: &AddressTransform,
193    frame_info: &FunctionFrameInfo,
194    scope_ranges: &[(u64, u64)],
195    wasm_types: &WasmTypesDieRefs,
196    func_meta: &FunctionMetadata,
197    locals_names: Option<&HashMap<u32, &str>>,
198    out_strings: &mut write::StringTable,
199    isa: &dyn TargetIsa,
200) -> Result<(), Error> {
201    let vmctx_label = get_vmctx_value_label();
202
203    // Normalize order of ValueLabelsRanges keys to have reproducable results.
204    let mut vars = frame_info.value_ranges.keys().collect::<Vec<_>>();
205    vars.sort_by(|a, b| a.index().cmp(&b.index()));
206
207    for label in vars {
208        if label.index() == vmctx_label.index() {
209            append_vmctx_info(
210                unit,
211                die_id,
212                wasm_types.vmctx,
213                addr_tr,
214                Some(frame_info),
215                scope_ranges,
216                out_strings,
217                isa,
218            )?;
219        } else {
220            let var_index = label.index();
221            let (type_die_id, is_param) =
222                if let Some(result) = resolve_var_type(var_index, wasm_types, func_meta) {
223                    result
224                } else {
225                    // Skipping if type of local cannot be detected.
226                    continue;
227                };
228
229            let loc_list_id = {
230                let locs = CompiledExpression::from_label(*label)
231                    .build_with_locals(scope_ranges, addr_tr, Some(frame_info), isa)
232                    .map(|i| {
233                        i.map(|(begin, length, data)| write::Location::StartLength {
234                            begin,
235                            length,
236                            data,
237                        })
238                    })
239                    .collect::<Result<Vec<_>, _>>()?;
240                unit.locations.add(write::LocationList(locs))
241            };
242
243            let var_id = unit.add(
244                die_id,
245                if is_param {
246                    gimli::DW_TAG_formal_parameter
247                } else {
248                    gimli::DW_TAG_variable
249                },
250            );
251            let var = unit.get_mut(var_id);
252
253            let name_id = match locals_names
254                .and_then(|m| m.get(&(var_index as u32)))
255                .and_then(|s| check_invalid_chars_in_name(s))
256            {
257                Some(n) => out_strings.add(assert_dwarf_str!(n)),
258                None => out_strings.add(format!("var{}", var_index)),
259            };
260
261            var.set(gimli::DW_AT_name, write::AttributeValue::StringRef(name_id));
262            var.set(
263                gimli::DW_AT_type,
264                write::AttributeValue::UnitRef(type_die_id),
265            );
266            var.set(
267                gimli::DW_AT_location,
268                write::AttributeValue::LocationListRef(loc_list_id),
269            );
270        }
271    }
272    Ok(())
273}
274
275fn check_invalid_chars_in_path(path: PathBuf) -> Option<PathBuf> {
276    path.clone()
277        .to_str()
278        .and_then(move |s| if s.contains('\x00') { None } else { Some(path) })
279}
280
281pub fn generate_simulated_dwarf(
282    addr_tr: &AddressTransform,
283    di: &DebugInfoData,
284    memory_offset: &ModuleMemoryOffset,
285    funcs: &CompiledFunctions,
286    translated: &HashSet<DefinedFuncIndex>,
287    out_encoding: gimli::Encoding,
288    out_units: &mut write::UnitTable,
289    out_strings: &mut write::StringTable,
290    isa: &dyn TargetIsa,
291) -> Result<(), Error> {
292    let path = di
293        .wasm_file
294        .path
295        .to_owned()
296        .and_then(check_invalid_chars_in_path)
297        .unwrap_or_else(|| autogenerate_dwarf_wasm_path(di));
298
299    let func_names = &di.name_section.func_names;
300    let locals_names = &di.name_section.locals_names;
301    let imported_func_count = di.wasm_file.imported_func_count;
302
303    let (unit, root_id, name_id) = {
304        let comp_dir_id = out_strings.add(assert_dwarf_str!(path
305            .parent()
306            .context("path dir")?
307            .to_str()
308            .context("path dir encoding")?));
309        let name = path
310            .file_name()
311            .context("path name")?
312            .to_str()
313            .context("path name encoding")?;
314        let name_id = out_strings.add(assert_dwarf_str!(name));
315
316        let out_program = generate_line_info(
317            addr_tr,
318            translated,
319            out_encoding,
320            &di.wasm_file,
321            comp_dir_id,
322            name_id,
323            name,
324        )?;
325
326        let unit_id = out_units.add(write::Unit::new(out_encoding, out_program));
327        let unit = out_units.get_mut(unit_id);
328
329        let root_id = unit.root();
330        let root = unit.get_mut(root_id);
331
332        let id = out_strings.add(PRODUCER_NAME);
333        root.set(gimli::DW_AT_producer, write::AttributeValue::StringRef(id));
334        root.set(gimli::DW_AT_name, write::AttributeValue::StringRef(name_id));
335        root.set(
336            gimli::DW_AT_stmt_list,
337            write::AttributeValue::LineProgramRef,
338        );
339        root.set(
340            gimli::DW_AT_comp_dir,
341            write::AttributeValue::StringRef(comp_dir_id),
342        );
343        (unit, root_id, name_id)
344    };
345
346    let wasm_types = add_wasm_types(unit, root_id, out_strings, memory_offset);
347
348    for (i, map) in addr_tr.map().iter() {
349        let index = i.index();
350        if translated.contains(&i) {
351            continue;
352        }
353
354        let start = map.offset as u64;
355        let end = start + map.len as u64;
356        let die_id = unit.add(root_id, gimli::DW_TAG_subprogram);
357        let die = unit.get_mut(die_id);
358        die.set(
359            gimli::DW_AT_low_pc,
360            write::AttributeValue::Address(write::Address::Symbol {
361                symbol: index,
362                addend: start as i64,
363            }),
364        );
365        die.set(
366            gimli::DW_AT_high_pc,
367            write::AttributeValue::Udata((end - start) as u64),
368        );
369
370        let func_index = imported_func_count + (index as u32);
371        let id = match func_names
372            .get(&FuncIndex::from_u32(func_index))
373            .and_then(|s| check_invalid_chars_in_name(s))
374        {
375            Some(n) => out_strings.add(assert_dwarf_str!(n)),
376            None => out_strings.add(format!("wasm-function[{}]", func_index)),
377        };
378
379        die.set(gimli::DW_AT_name, write::AttributeValue::StringRef(id));
380
381        die.set(
382            gimli::DW_AT_decl_file,
383            write::AttributeValue::StringRef(name_id),
384        );
385
386        let f_start = map.addresses[0].wasm;
387        let wasm_offset = di.wasm_file.code_section_offset + f_start as u64;
388        die.set(
389            gimli::DW_AT_decl_file,
390            write::AttributeValue::Udata(wasm_offset),
391        );
392
393        if let Some(frame_info) = get_function_frame_info(memory_offset, funcs, i) {
394            let source_range = addr_tr.func_source_range(i);
395            generate_vars(
396                unit,
397                die_id,
398                addr_tr,
399                &frame_info,
400                &[(source_range.0, source_range.1)],
401                &wasm_types,
402                &di.wasm_file.funcs[index],
403                locals_names.get(&FuncIndex::from_u32(index as u32)),
404                out_strings,
405                isa,
406            )?;
407        }
408    }
409
410    Ok(())
411}