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
|
use progvis:program;
/**
* Memory access.
*
* Used to report whether data was read from and/or written to. Also reports which threads read and
* wrote to the data for the model checker.
*
* The class makes a distinction between "new" reads/writes and "old" reads/writes. This is mainly
* for the benefit of the visualization.
*
* 'error' is used to indicate if an error was present.
*/
value MemAccess {
// Reads - 0=none, 1=old, 2>=new
Byte read;
// Writes - 0=none, 1=old, 2>=new
Byte write;
// Read error?
Bool readError;
// Write error?
Bool writeError;
// Exact reads+writes. This is mainly for the model checker. Contains the operations that are
// important for the current memory checker, this may not overlap entirely with the values in
// 'read' and 'write'. Stores thread ID:s of writing threads, followed by a zero, followed by
// thread ID:s of reading threads.
Nat[] threads;
// Create.
init() {
init {}
}
// Clear everything.
void clear() {
read = 0;
write = 0;
readError = false;
writeError = false;
threads.clear();
}
// Any error?
Bool error() {
readError | writeError;
}
}
/**
* Memory model to use.
*/
enum MemoryChecks {
// Don't use memory accesses at all in the visualization, not even for highlighting.
none,
// No checks at all, but visualizes accesses. Essentially sequential consistency.
sequential,
// Only check individual statements.
statements,
// Check between barriers. This is what the C/C++ standard requires (except for relaxed atomics)
barriers,
}
// Create a memory check.
MemCheck create(MemoryChecks checks) {
if (checks == MemoryChecks:none)
ZeroMemCheck();
else if (checks == MemoryChecks:sequential)
NoMemCheck();
else if (checks == MemoryChecks:statements)
StmtMemCheck();
else
FullMemCheck();
}
/**
* Object that keeps track of all reads and writes by threads during a traversal.
*
* It is possible to ask this object which threads read and write certain data, and to detect race
* conditions from those observations. Collects information on any instances where reads and writes
* cause race conditions.
*
* Abstract base class, there are implementations for the different models.
*/
class MemCheck {
// Memory trackers for each of the threads currently alive. Mapped thread-id->tracker.
protected Nat->MemTracker threads;
// Error messages we produced.
protected Str[] errors;
// Add a thread.
void add(Nat id, MemTracker tracker) {
threads.put(id, tracker);
}
// Get errors, if any.
ProgramError? error() {
if (errors.any)
return DataRaceError(errors);
else
return null;
}
// Query the memory access for some data. 'old' is used to inhibit old errors, for example.
MemAccess query(MemAccess old, unsafe:RawPtr object, Nat offset, Nat size) : abstract;
}
/**
* Mem check implementation that does not highlight anything. Mostly for programmatical use of
* Progvis, where the highlights would be confusing.
*/
class ZeroMemCheck extends MemCheck {
// Query the memory access for some data.
MemAccess query(MemAccess old, unsafe:RawPtr object, Nat offset, Nat size) : override {
return MemAccess();
}
}
/**
* Mem check implementation without checks, for sequential. Highlights reads and writes for clarity.
*/
class NoMemCheck extends MemCheck {
// Query the memory access for some data.
MemAccess query(MemAccess old, unsafe:RawPtr object, Nat offset, Nat size) : override {
MemAccess result;
// Check reads and writes for the visualization. We don't check correctness.
for (k, v in threads) {
if (v.newWrites.has(object, offset, size)) {
result.write = 2;
}
if (v.newReads.has(object, offset, size)) {
result.read = 2;
}
}
// Note: We don't touch "threads" as we don't check for errors.
result;
}
}
/**
* Mem check implementation that only considers new reads/writes.
*/
class StmtMemCheck extends MemCheck {
// Query the memory access for some data.
MemAccess query(MemAccess old, unsafe:RawPtr object, Nat offset, Nat size) : override {
MemAccess result;
// Check writes.
for (k, v in threads) {
if (v.newWrites.has(object, offset, size)) {
result.write = 2;
result.threads << k;
}
}
if (result.threads.count > 1) {
if (!old.writeError)
dualWriteError(result.threads);
result.writeError = true;
}
result.threads << 0;
Nat writtenBy = result.threads[0];
// Check reads.
for (k, v in threads) {
if (v.newReads.has(object, offset, size)) {
if ((writtenBy != 0) & (writtenBy != k))
result.readError = true;
result.read = 2;
result.threads << k;
}
}
if (result.readError) {
if (!old.readError) {
readWriteError(result.threads);
}
}
result;
}
// Add an error message concerning multiple writes.
private void dualWriteError(Nat[] writes) {
writes.sort();
StrBuf result;
result << "Threads " << join(writes, ", ") << " all wrote to the same value concurrently.";
errors << result.toS;
}
// Add an error message concerning read-write errors.
private void readWriteError(Nat[] threads) {
Nat[] writes;
Nat[] reads;
Bool zero = false;
for (v in threads) {
if (v == 0) {
zero = true;
} else if (zero) {
reads << v;
} else {
writes << v;
}
}
writes.sort();
reads.sort();
StrBuf result;
if (writes.count > 1)
result << "Threads ";
else
result << "Thread ";
result << join(writes, ", ") << " wrote to a value that ";
if (reads.count > 1)
result << "threads ";
else
result << "thread ";
result << join(reads, ", ") << " read from concurrently.";
errors << result.toS;
}
}
/**
* Object that keeps track of all reads and writes by threads during a traversal. Takes barrier
* information into account.
*/
class FullMemCheck extends MemCheck {
// Query the memory access for some data.
// This is the fast-path: it does not allocate memory in the case no errors happen
// (which is the normal case). It calls one of the error functions in case of an error.
// These will have to re-do parts of the work in order to provide better error messages,
// but as they are on the slow path, this is fine.
MemAccess query(MemAccess old, unsafe:RawPtr object, Nat offset, Nat size) : override {
MemAccess result;
// Check writes.
for (k, v in threads) {
if (v.writes.has(object, offset, size)) {
result.write |= 1;
result.threads << k;
if (v.newWrites.has(object, offset, size)) {
result.write = 2;
}
}
}
if (result.threads.count > 1) {
if (!old.writeError)
dualWriteError(result.threads);
result.writeError = true;
}
result.threads << 0;
Nat writtenBy = result.threads[0];
// Check reads.
for (k, v in threads) {
if (v.reads.has(object, offset, size)) {
if ((writtenBy != 0) & (writtenBy != k))
result.readError = true;
result.read |= 1;
result.threads << k;
if (v.newReads.has(object, offset, size)) {
result.read = 2;
}
}
}
if (result.readError) {
if (!old.readError) {
readWriteError(result.threads);
}
}
result;
}
// Add an error message concerning multiple writes.
private void dualWriteError(Nat[] writes) {
writes.sort();
StrBuf result;
result << "Threads " << join(writes, ", ") << " all wrote to the same value concurrently.";
errors << result.toS;
}
// Add an error message concerning read-write errors.
private void readWriteError(Nat[] threads) {
Nat[] writes;
Nat[] reads;
Bool zero = false;
for (v in threads) {
if (v == 0) {
zero = true;
} else if (zero) {
reads << v;
} else {
writes << v;
}
}
writes.sort();
reads.sort();
StrBuf result;
if (writes.count > 1)
result << "Threads ";
else
result << "Thread ";
result << join(writes, ", ") << " wrote to a value that ";
if (reads.count > 1)
result << "threads ";
else
result << "thread ";
result << join(reads, ", ") << " read from concurrently.";
errors << result.toS;
}
}
|