Introduce pthread_condattr_setclock()

This is one of the few POSIX APIs that was missing. It lets you choose a
monotonic clock for your condition variables. This might improve perf on
some platforms. It might also grant more flexibility with NTP configs. I
know Qt is one project that believes it needs this. To introduce this, I
needed to change some the *NSYNC APIs, to support passing a clock param.
There's also new benchmarks, demonstrating Cosmopolitan's supremacy over
many libc implementations when it comes to mutex performance. Cygwin has
an alarmingly bad pthread_mutex_t implementation. It is so bad that they
would have been significantly better off if they'd used naive spinlocks.
This commit is contained in:
Justine Tunney 2024-09-02 23:37:50 -07:00
parent 79516bf08e
commit 3c61a541bd
No known key found for this signature in database
GPG key ID: BE714B4575D6E328
55 changed files with 449 additions and 155 deletions

View file

@ -25,6 +25,7 @@
#include "third_party/nsync/mu_semaphore.h"
#include "third_party/nsync/races.internal.h"
#include "third_party/nsync/wait_s.internal.h"
#include "libc/sysv/consts/clock.h"
#include "third_party/nsync/waiter.h"
__static_yoink("nsync_notice");
@ -100,7 +101,7 @@ uint32_t nsync_counter_wait (nsync_counter c, nsync_time abs_deadline) {
uint32_t result = 0;
waitable.v = c;
waitable.funcs = &nsync_counter_waitable_funcs;
if (nsync_wait_n (NULL, NULL, NULL, abs_deadline, 1, &pwaitable) != 0) {
if (nsync_wait_n (NULL, NULL, NULL, CLOCK_REALTIME, abs_deadline, 1, &pwaitable) != 0) {
IGNORE_RACES_START ();
result = ATM_LOAD_ACQ (&c->value);
IGNORE_RACES_END ();