mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-06 12:20:52 +01:00
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/54284 In order to bring mobile deployment, via lite interpreter, on feature parity with JIT, with respect model level debug information we must make model level debug information available to mobile runtime. At the moment, model level debug information is stored in SourceRange which associates node's of graph to where the come from in original python source code. This information is serialized as part of debug_pkl and deserialized when JIT loads the model and reads the model code. On lite interpreter, we do not have access to all the functionality of JIT and hence we cannot load model in the same way as JIT, by reading code, constructing module hierarchy and graph corresponding module methods etc. Instead in, lite interpreter, only bytecode corresonding to the compiled graph, Code, is saved. Thus in order to annotate OPs in the bytecode with equivalent SourceRange information we do the following: 1. During model serialization, we create a unique tag for each source range of the model. 2. Create a map of <SourceRange, tag> 3. During debug_pkl serialization we save tag along with SourceRange, on top of byte offset. 4. During bytecode generation, the methods of the top module are lowered. During this process methods are inlined. In the inlined graph, when the node of a graph is lowered to bytecode, we query node's source range and look it up against the map. 5. Resulting source range tag is serialized in module_debug_info. 6. During model deserialization, we read all the debug_pkl records in the archieve and create a map of <tag, SourceRange> 7. This map can be used to find source code information. During mobile runtime: 1. We read all the debug_pkl records and create <tag=debug_handle, SourceRange> map. 1.1 This map, MobileDebugInfo, is a member of mobile Module. 2. Interpreter catches appropriate exceptions and sets the thread local debug handle and rethrows the exception. 3. In Function's run method we catch exception and query current debug handle where the exception happened. 4. Query MobileDebugInfo with debug handle to retrieve source range and augment error with source range info. This information is still incomplete as it does not contain entire callstack. In the following diffs we will serialize InlinedCallStack directly. Note that compilation is gated by SYMBOLICATE_MOBILE_DEBUG_HANDLE macro, so that mobile builds can avoid building MobileDebugInfo, source range and source range pickler/unpickler. Later we will add path where, if building without debug support stack trace will contain only debug handles. They can be symbolicated later. Test Plan: Ported bunch of source range tests from test_jit.py. Added on more test in test_lite_interpreter.py Imported from OSS Reviewed By: raziel Differential Revision: D27174722 fbshipit-source-id: a7b7c6088ce16dec37e823c7fefa4f0b61047e12
207 lines
6.6 KiB
C++
207 lines
6.6 KiB
C++
#include <torch/csrc/jit/frontend/source_range.h>
|
|
#include <torch/csrc/jit/serialization/source_range_serialization.h>
|
|
|
|
namespace torch {
|
|
namespace jit {
|
|
size_t SourceRangeHasher::operator()(const torch::jit::SourceRange& key) const {
|
|
return (
|
|
std::hash<uintptr_t>()(reinterpret_cast<uintptr_t>(key.source().get())) ^
|
|
std::hash<size_t>()(key.start()) ^ std::hash<size_t>()(key.end()));
|
|
}
|
|
|
|
c10::optional<SourceRange> Source::findSourceRangeThatGenerated(
|
|
const SourceRange& range) {
|
|
if (!gen_ranges_) {
|
|
return c10::nullopt;
|
|
}
|
|
return gen_ranges_->findSourceRangeThatGenerated(range);
|
|
}
|
|
|
|
C10_EXPORT void SourceRange::highlight(std::ostream& out) const {
|
|
// Retrieve original SourceRange, if present.
|
|
if (auto orig_source_range = findSourceRangeThatGenerated()) {
|
|
orig_source_range->highlight(out);
|
|
out << "Serialized ";
|
|
}
|
|
print_with_context(out, CONTEXT, true, "");
|
|
}
|
|
|
|
C10_EXPORT void format_stack_trace(
|
|
std::ostream& out,
|
|
const std::vector<StackEntry>& entries) {
|
|
bool has_orig_ranges = false;
|
|
std::vector<SourceRange> orig_ranges;
|
|
// gather original ranges. if we have a situation where we do not have orig
|
|
// ranges for some frames, we still want to report them for the frames we do
|
|
// have,
|
|
// so substitute the current range for that frame
|
|
for (const StackEntry& entry : entries) {
|
|
if (auto orig_source_range = entry.range.findSourceRangeThatGenerated()) {
|
|
orig_ranges.emplace_back(std::move(orig_source_range.value()));
|
|
has_orig_ranges = true;
|
|
} else {
|
|
orig_ranges.emplace_back(entry.range);
|
|
}
|
|
}
|
|
out << "Traceback of TorchScript";
|
|
if (has_orig_ranges) {
|
|
out << ", serialized code";
|
|
}
|
|
out << " (most recent call last):\n";
|
|
for (const StackEntry& entry : entries) {
|
|
entry.range.print_with_context(
|
|
out, SourceRange::CONTEXT, true, entry.filename);
|
|
}
|
|
if (has_orig_ranges) {
|
|
out << "\nTraceback of TorchScript, original code (most recent call last):\n";
|
|
auto it = entries.begin();
|
|
for (const SourceRange& range : orig_ranges) {
|
|
range.print_with_context(
|
|
out, SourceRange::CONTEXT, true, (*it++).filename);
|
|
}
|
|
}
|
|
}
|
|
|
|
C10_EXPORT void SourceRange::print_with_context(
|
|
std::ostream& out,
|
|
size_t context,
|
|
bool highlight,
|
|
const std::string& funcname) const {
|
|
// This is an empty SourceRange, used as a sentinel value.
|
|
if (!source_) {
|
|
return;
|
|
}
|
|
|
|
const std::string& str = source_->text();
|
|
if (size() == str.size()) {
|
|
// this is just the entire file, not a subset, so print it out.
|
|
// primarily used to print out python stack traces
|
|
out << str;
|
|
return;
|
|
}
|
|
|
|
size_t range_end =
|
|
(str.size() < end()
|
|
? str.size()
|
|
: end()); // use instead of 'end()' because some ranges extend past
|
|
// the length of the source
|
|
|
|
// determine CONTEXT line range
|
|
size_t begin_line = start(); // beginning of lines to highlight
|
|
size_t end_line = range_end;
|
|
while (begin_line > 0 && str[begin_line - 1] != '\n')
|
|
--begin_line;
|
|
while (end_line < str.size() && str[end_line] != '\n')
|
|
++end_line;
|
|
AT_ASSERT(begin_line == 0 || str[begin_line - 1] == '\n');
|
|
AT_ASSERT(end_line == str.size() || str[end_line] == '\n');
|
|
|
|
size_t begin_context = begin_line; // beginning of context, CONTEXT lines
|
|
// before the highlight lines
|
|
for (size_t i = 0; begin_context > 0; --begin_context) {
|
|
if (str[begin_context - 1] == '\n') {
|
|
++i;
|
|
}
|
|
if (i >= context) {
|
|
break;
|
|
}
|
|
}
|
|
AT_ASSERT(begin_context == 0 || str[begin_context - 1] == '\n');
|
|
|
|
size_t end_context =
|
|
end_line; // end of context, CONTEXT lines after the highlight lines
|
|
for (size_t i = 0; end_context < str.size(); ++end_context) {
|
|
if (str[end_context] == '\n') {
|
|
++i;
|
|
}
|
|
if (i >= context) {
|
|
break;
|
|
}
|
|
}
|
|
AT_ASSERT(end_context == str.size() || str[end_context] == '\n');
|
|
|
|
// print out location information
|
|
if (auto flc = file_line_col()) {
|
|
std::string filename;
|
|
// NOLINTNEXTLINE(cppcoreguidelines-init-variables)
|
|
size_t line, col;
|
|
std::tie(filename, line, col) = *flc;
|
|
out << " File \"" << filename << "\", line " << line;
|
|
if (funcname != "") {
|
|
out << ", in " << funcname;
|
|
}
|
|
out << "\n";
|
|
}
|
|
// print out inital context
|
|
out << str.substr(begin_context, start() - begin_context);
|
|
size_t line_start = start();
|
|
size_t line_end = range_end;
|
|
if (highlight) {
|
|
line_end = start();
|
|
while (line_start < range_end) {
|
|
// move line_end to end of line
|
|
while (str[line_end] != '\n' && line_end < str.size()) {
|
|
++line_end;
|
|
}
|
|
// print line of code
|
|
auto actual_line = str.substr(line_start, (line_end - line_start) + 1);
|
|
out << actual_line;
|
|
if (actual_line.back() != '\n') {
|
|
out << "\n";
|
|
}
|
|
|
|
size_t empty_space = 0;
|
|
size_t highlight_space = 0;
|
|
size_t hightlight_begin = line_start;
|
|
size_t highlight_end = line_start;
|
|
// determine length of line which is being highlighted
|
|
while (hightlight_begin > 0 && str[hightlight_begin - 1] != '\n') {
|
|
--hightlight_begin;
|
|
}
|
|
while (highlight_end < range_end && str[highlight_end] != '\n') {
|
|
++highlight_end;
|
|
}
|
|
AT_ASSERT(hightlight_begin == 0 || str[hightlight_begin - 1] == '\n');
|
|
AT_ASSERT(highlight_end == range_end || str[highlight_end] == '\n');
|
|
// determine amount of empty space vs highlighted space
|
|
for (size_t i = hightlight_begin; i < highlight_end; i++) {
|
|
if (str[i] == ' ' || i < start()) {
|
|
empty_space++;
|
|
} else {
|
|
break;
|
|
}
|
|
}
|
|
highlight_space = highlight_end - hightlight_begin - empty_space;
|
|
if (highlight_space > 0) {
|
|
// some ranges are off and include empty white space on new lines which
|
|
// don't need to be printed
|
|
bool more_lines = false;
|
|
for (size_t i = line_end; i <= range_end; i++) {
|
|
if (str[i] != '\n' && str[i] != ' ') {
|
|
more_lines = true;
|
|
}
|
|
}
|
|
out << std::string(empty_space, ' ');
|
|
out << std::string(highlight_space, '~');
|
|
out << (more_lines && line_end != range_end ? "\n" : " <--- HERE\n");
|
|
}
|
|
++line_end;
|
|
line_start = line_end;
|
|
}
|
|
} else {
|
|
// print out code with no highlight
|
|
out << str.substr(start(), range_end - start());
|
|
}
|
|
// print out ending context
|
|
if (line_end <= str.size()) {
|
|
auto line_substr = str.substr(line_end, end_context - line_end);
|
|
out << line_substr;
|
|
if (!line_substr.empty() && line_substr.back() != '\n') {
|
|
out << "\n";
|
|
}
|
|
}
|
|
}
|
|
|
|
} // namespace jit
|
|
} // namespace torch
|