Problem:
Upper bound on thread wakeup was set to 1.5 * (requested timeout).
On MacOS wakeup delay factors of 1.75 have been reported.
Solution:
Increase the bound to 5 * (requested timeout).
Refs: https://github.com/libuv/libuv/issues/1910
PR-URL: https://github.com/libuv/libuv/pull/1990
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Problem:
The condvar tests could pass without uv_cond_wait
(or uv_cond_timedwait) ever being invoked.
Solution:
Introduce semaphores to enforce ordering.
Now there will always be a thread waiting on the condition
when a signal() occurs.
Gotchas:
1. On Windows, waiting for a timeout may return earlier
than requested, depending on the granularity of timer ticks.
2. Timeout bounds are tuned based on our CI machines.
Bonuses:
1. I added additional test cases to complete the test matrix.
2. It seemed to me that several of the condvar tests were redundant,
because they used timing to probabilistically explore cases where there
would be "missed connections" (signal without a waiter).
Because it was not clear to me what the purpose of such tests were,
I have removed them.
Fixes: https://github.com/libuv/libuv/issues/1714
PR-URL: https://github.com/libuv/libuv/pull/1718
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Problem:
The "timeout" functionality of uv_cond_timedwait was not being tested.
The test (condvar_3) would use a worker that signaled the condition.
Solution:
Introduce a new condvar test case to ensure that the timeout also works.
PR-URL: https://github.com/libuv/libuv/pull/1713
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: John Barboza <jbarboza@ca.ibm.com>
Calling uv_cond_wait without uv_cond_signal/uv_cond_broadcast may
cause deadlock. This commit avoids this situation as well as tests
these functions.
PR-URL: https://github.com/libuv/libuv/pull/728
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
This commit changes the libuv API to return error codes directly rather
than storing them in a loop-global field.
A code snippet like this one:
if (uv_foo(loop) < 0) {
uv_err_t err = uv_last_error(loop);
fprintf(stderr, "%s\n", uv_strerror(err));
}
Should be rewritten like this:
int err = uv_foo(loop);
if (err < 0)
fprintf(stderr, "%s\n", uv_strerror(err));
The rationale for this change is that it should make creating bindings
for other languages a lot easier: dealing with struct return values is
painful with most FFIs and often downright buggy.
