bpo-30768: Recompute timeout on interrupted lock (GH-4103)

Fix the pthread+semaphore implementation of
PyThread_acquire_lock_timed() when called with timeout > 0 and
intr_flag=0: recompute the timeout if sem_timedwait() is interrupted
by a signal (EINTR).

See also the PEP 475.

The pthread implementation of PyThread_acquire_lock() now fails with
a fatal error if the timeout is larger than PY_TIMEOUT_MAX, as done
in the Windows implementation.

The check prevents any risk of overflow in PyThread_acquire_lock().

Add also PY_DWORD_MAX constant.
This commit is contained in:
Victor Stinner 2017-10-24 16:53:32 -07:00 committed by GitHub
parent 3557b05c5a
commit 850a18e03e
9 changed files with 86 additions and 31 deletions

View file

@ -283,12 +283,13 @@ PyThread_acquire_lock_timed(PyThread_type_lock aLock,
milliseconds = microseconds / 1000;
if (microseconds % 1000 > 0)
++milliseconds;
if ((DWORD) milliseconds != milliseconds)
Py_FatalError("Timeout too large for a DWORD, "
"please check PY_TIMEOUT_MAX");
if (milliseconds > PY_DWORD_MAX) {
Py_FatalError("Timeout larger than PY_TIMEOUT_MAX");
}
}
else
else {
milliseconds = INFINITE;
}
dprintf(("%lu: PyThread_acquire_lock_timed(%p, %lld) called\n",
PyThread_get_thread_ident(), aLock, microseconds));