1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
|
//===-------------- llvm-remark-size-diff/RemarkSizeDiff.cpp --------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
///
/// \file
/// Diffs instruction count and stack size remarks between two remark files.
///
/// This is intended for use by compiler developers who want to see how their
/// changes impact program code size.
///
//===----------------------------------------------------------------------===//
#include "llvm-c/Remarks.h"
#include "llvm/ADT/Optional.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/Remarks/Remark.h"
#include "llvm/Remarks/RemarkParser.h"
#include "llvm/Remarks/RemarkSerializer.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/FileSystem.h"
#include "llvm/Support/FormatVariadic.h"
#include "llvm/Support/InitLLVM.h"
#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/ToolOutputFile.h"
#include "llvm/Support/WithColor.h"
#include "llvm/Support/raw_ostream.h"
using namespace llvm;
enum ParserFormatOptions { yaml, bitstream };
enum ReportStyleOptions { human_output, json_output };
static cl::OptionCategory SizeDiffCategory("llvm-remark-size-diff options");
static cl::opt<std::string> InputFileNameA(cl::Positional, cl::Required,
cl::cat(SizeDiffCategory),
cl::desc("remarks_a"));
static cl::opt<std::string> InputFileNameB(cl::Positional, cl::Required,
cl::cat(SizeDiffCategory),
cl::desc("remarks_b"));
static cl::opt<std::string> OutputFilename("o", cl::init("-"),
cl::cat(SizeDiffCategory),
cl::desc("Output"),
cl::value_desc("file"));
static cl::opt<ParserFormatOptions>
ParserFormat("parser", cl::cat(SizeDiffCategory), cl::init(bitstream),
cl::desc("Set the remark parser format:"),
cl::values(clEnumVal(yaml, "YAML format"),
clEnumVal(bitstream, "Bitstream format")));
static cl::opt<ReportStyleOptions> ReportStyle(
"report_style", cl::cat(SizeDiffCategory),
cl::init(ReportStyleOptions::human_output),
cl::desc("Choose the report output format:"),
cl::values(clEnumValN(human_output, "human", "Human-readable format"),
clEnumValN(json_output, "json", "JSON format")));
static cl::opt<bool> PrettyPrint("pretty", cl::cat(SizeDiffCategory),
cl::init(false),
cl::desc("Pretty-print JSON"));
/// Contains information from size remarks.
// This is a little nicer to read than a std::pair.
struct InstCountAndStackSize {
int64_t InstCount = 0;
int64_t StackSize = 0;
};
/// Represents which files a function appeared in.
enum FilesPresent { A, B, BOTH };
/// Contains the data from the remarks in file A and file B for some function.
/// E.g. instruction count, stack size...
struct FunctionDiff {
/// Function name from the remark.
std::string FuncName;
// Idx 0 = A, Idx 1 = B.
int64_t InstCount[2] = {0, 0};
int64_t StackSize[2] = {0, 0};
// Calculate diffs between the first and second files.
int64_t getInstDiff() const { return InstCount[1] - InstCount[0]; }
int64_t getStackDiff() const { return StackSize[1] - StackSize[0]; }
// Accessors for the remarks from the first file.
int64_t getInstCountA() const { return InstCount[0]; }
int64_t getStackSizeA() const { return StackSize[0]; }
// Accessors for the remarks from the second file.
int64_t getInstCountB() const { return InstCount[1]; }
int64_t getStackSizeB() const { return StackSize[1]; }
/// \returns which files this function was present in.
FilesPresent getFilesPresent() const {
if (getInstCountA() == 0)
return B;
if (getInstCountB() == 0)
return A;
return BOTH;
}
FunctionDiff(StringRef FuncName, const InstCountAndStackSize &A,
const InstCountAndStackSize &B)
: FuncName(FuncName) {
InstCount[0] = A.InstCount;
InstCount[1] = B.InstCount;
StackSize[0] = A.StackSize;
StackSize[1] = B.StackSize;
}
};
/// Organizes the diffs into 3 categories:
/// - Functions which only appeared in the first file
/// - Functions which only appeared in the second file
/// - Functions which appeared in both files
struct DiffsCategorizedByFilesPresent {
/// Diffs for functions which only appeared in the first file.
SmallVector<FunctionDiff> OnlyInA;
/// Diffs for functions which only appeared in the second file.
SmallVector<FunctionDiff> OnlyInB;
/// Diffs for functions which appeared in both files.
SmallVector<FunctionDiff> InBoth;
/// Add a diff to the appropriate list.
void addDiff(FunctionDiff &FD) {
switch (FD.getFilesPresent()) {
case A:
OnlyInA.push_back(FD);
break;
case B:
OnlyInB.push_back(FD);
break;
case BOTH:
InBoth.push_back(FD);
break;
}
}
};
static void printFunctionDiff(const FunctionDiff &FD, llvm::raw_ostream &OS) {
// Describe which files the function had remarks in.
FilesPresent FP = FD.getFilesPresent();
const std::string &FuncName = FD.FuncName;
const int64_t InstDiff = FD.getInstDiff();
assert(InstDiff && "Shouldn't get functions with no size change?");
const int64_t StackDiff = FD.getStackDiff();
// Output an indicator denoting which files the function was present in.
switch (FP) {
case FilesPresent::A:
OS << "-- ";
break;
case FilesPresent::B:
OS << "++ ";
break;
case FilesPresent::BOTH:
OS << "== ";
break;
}
// Output an indicator denoting if a function changed in size.
if (InstDiff > 0)
OS << "> ";
else
OS << "< ";
OS << FuncName << ", ";
OS << InstDiff << " instrs, ";
OS << StackDiff << " stack B";
OS << "\n";
}
/// Print an item in the summary section.
///
/// \p TotalA - Total count of the metric in file A.
/// \p TotalB - Total count of the metric in file B.
/// \p Metric - Name of the metric we want to print (e.g. instruction
/// count).
/// \p OS - The output stream.
static void printSummaryItem(int64_t TotalA, int64_t TotalB, StringRef Metric,
llvm::raw_ostream &OS) {
OS << " " << Metric << ": ";
int64_t TotalDiff = TotalB - TotalA;
if (TotalDiff == 0) {
OS << "None\n";
return;
}
OS << TotalDiff << " (" << formatv("{0:p}", TotalDiff / (double)TotalA)
<< ")\n";
}
/// Print all contents of \p Diff and a high-level summary of the differences.
static void printDiffsCategorizedByFilesPresent(
DiffsCategorizedByFilesPresent &DiffsByFilesPresent,
llvm::raw_ostream &OS) {
int64_t InstrsA = 0;
int64_t InstrsB = 0;
int64_t StackA = 0;
int64_t StackB = 0;
// Helper lambda to sort + print a list of diffs.
auto PrintDiffList = [&](SmallVector<FunctionDiff> &FunctionDiffList) {
if (FunctionDiffList.empty())
return;
stable_sort(FunctionDiffList,
[](const FunctionDiff &LHS, const FunctionDiff &RHS) {
return LHS.getInstDiff() < RHS.getInstDiff();
});
for (const auto &FuncDiff : FunctionDiffList) {
// If there is a difference in instruction count, then print out info for
// the function.
if (FuncDiff.getInstDiff())
printFunctionDiff(FuncDiff, OS);
InstrsA += FuncDiff.getInstCountA();
InstrsB += FuncDiff.getInstCountB();
StackA += FuncDiff.getStackSizeA();
StackB += FuncDiff.getStackSizeB();
}
};
PrintDiffList(DiffsByFilesPresent.OnlyInA);
PrintDiffList(DiffsByFilesPresent.OnlyInB);
PrintDiffList(DiffsByFilesPresent.InBoth);
OS << "\n### Summary ###\n";
OS << "Total change: \n";
printSummaryItem(InstrsA, InstrsB, "instruction count", OS);
printSummaryItem(StackA, StackB, "stack byte usage", OS);
}
/// Collects an expected integer value from a given argument index in a remark.
///
/// \p Remark - The remark.
/// \p ArgIdx - The index where the integer value should be found.
/// \p ExpectedKeyName - The expected key name for the index
/// (e.g. "InstructionCount")
///
/// \returns the integer value at the index if it exists, and the key-value pair
/// is what is expected. Otherwise, returns an Error.
static Expected<int64_t> getIntValFromKey(const remarks::Remark &Remark,
unsigned ArgIdx,
StringRef ExpectedKeyName) {
auto KeyName = Remark.Args[ArgIdx].Key;
if (KeyName != ExpectedKeyName)
return createStringError(
inconvertibleErrorCode(),
Twine("Unexpected key at argument index " + std::to_string(ArgIdx) +
": Expected '" + ExpectedKeyName + "', got '" + KeyName + "'"));
long long Val;
auto ValStr = Remark.Args[ArgIdx].Val;
if (getAsSignedInteger(ValStr, 0, Val))
return createStringError(
inconvertibleErrorCode(),
Twine("Could not convert string to signed integer: " + ValStr));
return static_cast<int64_t>(Val);
}
/// Collects relevant size information from \p Remark if it is an size-related
/// remark of some kind (e.g. instruction count). Otherwise records nothing.
///
/// \p Remark - The remark.
/// \p FuncNameToSizeInfo - Maps function names to relevant size info.
/// \p NumInstCountRemarksParsed - Keeps track of the number of instruction
/// count remarks parsed. We need at least 1 in both files to produce a diff.
static Error processRemark(const remarks::Remark &Remark,
StringMap<InstCountAndStackSize> &FuncNameToSizeInfo,
unsigned &NumInstCountRemarksParsed) {
const auto &RemarkName = Remark.RemarkName;
const auto &PassName = Remark.PassName;
// Collect remarks which contain the number of instructions in a function.
if (PassName == "asm-printer" && RemarkName == "InstructionCount") {
// Expecting the 0-th argument to have the key "NumInstructions" and an
// integer value.
auto MaybeInstCount =
getIntValFromKey(Remark, /*ArgIdx = */ 0, "NumInstructions");
if (!MaybeInstCount)
return MaybeInstCount.takeError();
FuncNameToSizeInfo[Remark.FunctionName].InstCount = *MaybeInstCount;
++NumInstCountRemarksParsed;
}
// Collect remarks which contain the stack size of a function.
else if (PassName == "prologepilog" && RemarkName == "StackSize") {
// Expecting the 0-th argument to have the key "NumStackBytes" and an
// integer value.
auto MaybeStackSize =
getIntValFromKey(Remark, /*ArgIdx = */ 0, "NumStackBytes");
if (!MaybeStackSize)
return MaybeStackSize.takeError();
FuncNameToSizeInfo[Remark.FunctionName].StackSize = *MaybeStackSize;
}
// Either we collected a remark, or it's something we don't care about. In
// both cases, this is a success.
return Error::success();
}
/// Process all of the size-related remarks in a file.
///
/// \param[in] InputFileName - Name of file to read from.
/// \param[in, out] FuncNameToSizeInfo - Maps function names to relevant
/// size info.
static Error readFileAndProcessRemarks(
StringRef InputFileName,
StringMap<InstCountAndStackSize> &FuncNameToSizeInfo) {
auto Buf = MemoryBuffer::getFile(InputFileName);
if (auto EC = Buf.getError())
return createStringError(
EC, Twine("Cannot open file '" + InputFileName + "': " + EC.message()));
auto MaybeParser = remarks::createRemarkParserFromMeta(
ParserFormat == bitstream ? remarks::Format::Bitstream
: remarks::Format::YAML,
(*Buf)->getBuffer());
if (!MaybeParser)
return MaybeParser.takeError();
auto &Parser = **MaybeParser;
auto MaybeRemark = Parser.next();
unsigned NumInstCountRemarksParsed = 0;
for (; MaybeRemark; MaybeRemark = Parser.next()) {
if (auto E = processRemark(**MaybeRemark, FuncNameToSizeInfo,
NumInstCountRemarksParsed))
return E;
}
auto E = MaybeRemark.takeError();
if (!E.isA<remarks::EndOfFileError>())
return E;
consumeError(std::move(E));
// We need at least one instruction count remark in each file to produce a
// meaningful diff.
if (NumInstCountRemarksParsed == 0)
return createStringError(
inconvertibleErrorCode(),
"File '" + InputFileName +
"' did not contain any instruction-count remarks!");
return Error::success();
}
/// Wrapper function for readFileAndProcessRemarks which handles errors.
///
/// \param[in] InputFileName - Name of file to read from.
/// \param[out] FuncNameToSizeInfo - Populated with information from size
/// remarks in the input file.
///
/// \returns true if readFileAndProcessRemarks returned no errors. False
/// otherwise.
static bool tryReadFileAndProcessRemarks(
StringRef InputFileName,
StringMap<InstCountAndStackSize> &FuncNameToSizeInfo) {
if (Error E = readFileAndProcessRemarks(InputFileName, FuncNameToSizeInfo)) {
handleAllErrors(std::move(E), [&](const ErrorInfoBase &PE) {
PE.log(WithColor::error());
errs() << '\n';
});
return false;
}
return true;
}
/// Populates \p FuncDiffs with the difference between \p
/// FuncNameToSizeInfoA and \p FuncNameToSizeInfoB.
///
/// \param[in] FuncNameToSizeInfoA - Size info collected from the first
/// remarks file.
/// \param[in] FuncNameToSizeInfoB - Size info collected from
/// the second remarks file.
/// \param[out] DiffsByFilesPresent - Filled with the diff between \p
/// FuncNameToSizeInfoA and \p FuncNameToSizeInfoB.
static void
computeDiff(const StringMap<InstCountAndStackSize> &FuncNameToSizeInfoA,
const StringMap<InstCountAndStackSize> &FuncNameToSizeInfoB,
DiffsCategorizedByFilesPresent &DiffsByFilesPresent) {
SmallSet<std::string, 10> FuncNames;
for (const auto &FuncName : FuncNameToSizeInfoA.keys())
FuncNames.insert(FuncName.str());
for (const auto &FuncName : FuncNameToSizeInfoB.keys())
FuncNames.insert(FuncName.str());
for (const std::string &FuncName : FuncNames) {
const auto &SizeInfoA = FuncNameToSizeInfoA.lookup(FuncName);
const auto &SizeInfoB = FuncNameToSizeInfoB.lookup(FuncName);
FunctionDiff FuncDiff(FuncName, SizeInfoA, SizeInfoB);
DiffsByFilesPresent.addDiff(FuncDiff);
}
}
/// Attempt to get the output stream for writing the diff.
static ErrorOr<std::unique_ptr<ToolOutputFile>> getOutputStream() {
if (OutputFilename == "")
OutputFilename = "-";
std::error_code EC;
auto Out = std::make_unique<ToolOutputFile>(OutputFilename, EC,
sys::fs::OF_TextWithCRLF);
if (!EC)
return std::move(Out);
return EC;
}
/// \return a json::Array representing all FunctionDiffs in \p FunctionDiffs.
/// \p WhichFiles represents which files the functions in \p FunctionDiffs
/// appeared in (A, B, or both).
json::Array
getFunctionDiffListAsJSON(const SmallVector<FunctionDiff> &FunctionDiffs,
const FilesPresent &WhichFiles) {
json::Array FunctionDiffsAsJSON;
int64_t InstCountA, InstCountB, StackSizeA, StackSizeB;
for (auto &Diff : FunctionDiffs) {
InstCountA = InstCountB = StackSizeA = StackSizeB = 0;
switch (WhichFiles) {
case BOTH:
LLVM_FALLTHROUGH;
case A:
InstCountA = Diff.getInstCountA();
StackSizeA = Diff.getStackSizeA();
if (WhichFiles != BOTH)
break;
LLVM_FALLTHROUGH;
case B:
InstCountB = Diff.getInstCountB();
StackSizeB = Diff.getStackSizeB();
break;
}
// Each metric we care about is represented like:
// "Val": [A, B]
// This allows any consumer of the JSON to calculate the diff using B - A.
// This is somewhat wasteful for OnlyInA and OnlyInB (we only need A or B).
// However, this should make writing consuming tools easier, since the tool
// writer doesn't need to think about slightly different formats in each
// section.
json::Object FunctionObject({{"FunctionName", Diff.FuncName},
{"InstCount", {InstCountA, InstCountB}},
{"StackSize", {StackSizeA, StackSizeB}}});
FunctionDiffsAsJSON.push_back(std::move(FunctionObject));
}
return FunctionDiffsAsJSON;
}
/// Output all diffs in \p DiffsByFilesPresent as a JSON report. This is
/// intended for consumption by external tools.
///
/// \p InputFileNameA - File A used to produce the report.
/// \p InputFileNameB - File B used ot produce the report.
/// \p OS - Output stream.
///
/// JSON output includes:
/// - \p InputFileNameA and \p InputFileNameB under "Files".
/// - Functions present in both files under "InBoth".
/// - Functions present only in A in "OnlyInA".
/// - Functions present only in B in "OnlyInB".
/// - Instruction count and stack size differences for each function.
///
/// Differences are represented using [count_a, count_b]. The actual difference
/// can be computed via count_b - count_a.
static void
outputJSONForAllDiffs(StringRef InputFileNameA, StringRef InputFileNameB,
const DiffsCategorizedByFilesPresent &DiffsByFilesPresent,
llvm::raw_ostream &OS) {
json::Object Output;
// Include file names in the report.
json::Object Files(
{{"A", InputFileNameA.str()}, {"B", InputFileNameB.str()}});
Output["Files"] = std::move(Files);
Output["OnlyInA"] = getFunctionDiffListAsJSON(DiffsByFilesPresent.OnlyInA, A);
Output["OnlyInB"] = getFunctionDiffListAsJSON(DiffsByFilesPresent.OnlyInB, B);
Output["InBoth"] =
getFunctionDiffListAsJSON(DiffsByFilesPresent.InBoth, BOTH);
json::OStream JOS(OS, PrettyPrint ? 2 : 0);
JOS.value(std::move(Output));
OS << '\n';
}
/// Output all diffs in \p DiffsByFilesPresent using the desired output style.
/// \returns Error::success() on success, and an Error otherwise.
/// \p InputFileNameA - Name of input file A; may be used in the report.
/// \p InputFileNameB - Name of input file B; may be used in the report.
static Error
outputAllDiffs(StringRef InputFileNameA, StringRef InputFileNameB,
DiffsCategorizedByFilesPresent &DiffsByFilesPresent) {
auto MaybeOF = getOutputStream();
if (std::error_code EC = MaybeOF.getError())
return errorCodeToError(EC);
std::unique_ptr<ToolOutputFile> OF = std::move(*MaybeOF);
switch (ReportStyle) {
case human_output:
printDiffsCategorizedByFilesPresent(DiffsByFilesPresent, OF->os());
break;
case json_output:
outputJSONForAllDiffs(InputFileNameA, InputFileNameB, DiffsByFilesPresent,
OF->os());
break;
}
OF->keep();
return Error::success();
}
/// Boolean wrapper for outputDiff which handles errors.
static bool
tryOutputAllDiffs(StringRef InputFileNameA, StringRef InputFileNameB,
DiffsCategorizedByFilesPresent &DiffsByFilesPresent) {
if (Error E =
outputAllDiffs(InputFileNameA, InputFileNameB, DiffsByFilesPresent)) {
handleAllErrors(std::move(E), [&](const ErrorInfoBase &PE) {
PE.log(WithColor::error());
errs() << '\n';
});
return false;
}
return true;
}
int main(int argc, const char **argv) {
InitLLVM X(argc, argv);
cl::HideUnrelatedOptions(SizeDiffCategory);
cl::ParseCommandLineOptions(argc, argv,
"Diff instruction count and stack size remarks "
"between two remark files.\n");
StringMap<InstCountAndStackSize> FuncNameToSizeInfoA;
StringMap<InstCountAndStackSize> FuncNameToSizeInfoB;
if (!tryReadFileAndProcessRemarks(InputFileNameA, FuncNameToSizeInfoA) ||
!tryReadFileAndProcessRemarks(InputFileNameB, FuncNameToSizeInfoB))
return 1;
DiffsCategorizedByFilesPresent DiffsByFilesPresent;
computeDiff(FuncNameToSizeInfoA, FuncNameToSizeInfoB, DiffsByFilesPresent);
if (!tryOutputAllDiffs(InputFileNameA, InputFileNameB, DiffsByFilesPresent))
return 1;
}
|