comparison Python/thread.c @ 48167:a8a8232cb541 legacy-trunk

Adds a sanity check to avoid a *very rare* infinite loop due to a corrupt tls key list data structure in the thread startup path. This change is a companion to r60148 which already successfully dealt with a similar issue on thread shutdown. In particular this loop has been observed happening from this call path: #0 in find_key () #1 in PyThread_set_key_value () #2 in _PyGILState_NoteThreadState () #3 in PyThreadState_New () #4 in t_bootstrap () #5 in pthread_start_thread () I don't know how this happens but it does, *very* rarely. On more than one hardware platform. I have not been able to reproduce it manually. (A flaky mutex implementation on the system in question is one hypothesis). As with r60148, the spinning we managed to observe in the wild was due to a single list element pointing back upon itself.
author Gregory P. Smith <greg@mad-scientist.com>
date Mon, 04 Aug 2008 07:33:37 +0000
parents db99f8752710
children e38335d4df91 8540268040ff
comparison
equal deleted inserted replaced
48166:ab4c9c384d31 48167:a8a8232cb541
262 * segfaults. Now we lock the whole routine. 262 * segfaults. Now we lock the whole routine.
263 */ 263 */
264 static struct key * 264 static struct key *
265 find_key(int key, void *value) 265 find_key(int key, void *value)
266 { 266 {
267 struct key *p; 267 struct key *p, *prev_p;
268 long id = PyThread_get_thread_ident(); 268 long id = PyThread_get_thread_ident();
269 269
270 if (!keymutex) 270 if (!keymutex)
271 return NULL; 271 return NULL;
272 PyThread_acquire_lock(keymutex, 1); 272 PyThread_acquire_lock(keymutex, 1);
273 prev_p = NULL;
273 for (p = keyhead; p != NULL; p = p->next) { 274 for (p = keyhead; p != NULL; p = p->next) {
274 if (p->id == id && p->key == key) 275 if (p->id == id && p->key == key)
275 goto Done; 276 goto Done;
277 /* Sanity check. These states should never happen but if
278 * they do we must abort. Otherwise we'll end up spinning in
279 * in a tight loop with the lock held. A similar check is done
280 * in pystate.c tstate_delete_common(). */
281 if (p == prev_p)
282 Py_FatalError("tls find_key: small circular list(!)");
283 prev_p = p;
284 if (p->next == keyhead)
285 Py_FatalError("tls find_key: circular list(!)");
276 } 286 }
277 if (value == NULL) { 287 if (value == NULL) {
278 assert(p == NULL); 288 assert(p == NULL);
279 goto Done; 289 goto Done;
280 } 290 }