mirrors
/
janet
mirror of https://github.com/janet-lang/janet
1
0
Fork 0
Code Issues 1 Releases Wiki Activity

Fixes for NetBSD (again) 

Minimum interval for a timer must be 1 or more (or we get EINVAL) and
Janet fails tests and halts events that the programmer may still be
interested in.
This commit is contained in:
llmII 2021-09-05 21:46:55 -05:00
parent bbc07c72d3
commit 121aa91139
No known key found for this signature in database
GPG Key ID: E3AD2E259F58A9A0
1 changed files with 17 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
src/core/ev.c
View File

@ -1577,31 +1577,32 @@ void janet_ev_deinit(void) {
#define EV_SETx(ev, a, b, c, d, e, f) EV_SET((ev), (a), (b), (c), (d), (e), ((__typeof__((ev)->udata))(f)))
#define JANET_KQUEUE_TF (EV_ADD | EV_ENABLE | EV_CLEAR | EV_ONESHOT)
/* NOTE:
* NetBSD doesn't like intervals less than 1 millisecond so lets make that the
* default anywhere JANET_KQUEUE_TS will be used. */
#ifdef __NetBSD__
#define JANET_KQUEUE_MIN_INTERVAL 1
#else
#define JANET_KQUEUE_MIN_INTERVAL 0
#endif
#ifdef __FreeBSD__
#define JANET_KQUEUE_TS(timestamp) (timestamp)
#else
/* NOTE:
* NetBSD and OpenBSD expect things are always intervals, so fake that we have
* abstime capability by changing how a timestamp is used in all kqueue calls
* and defining absent macros. */
#define JANET_KQUEUE_TS(timestamp) (timestamp - ts_now())
* and defining absent macros. Additionally NetBSD expects intervals be
* greater than 1 millisecond, so correct all intervals to be at least 1
* millisecond under NetBSD. */
JanetTimestamp fix_interval(const JanetTimestamp ts) {
return ts >= JANET_KQUEUE_MIN_INTERVAL ? ts : JANET_KQUEUE_MIN_INTERVAL;
}
#define JANET_KQUEUE_TS(timestamp) (fix_interval((timestamp - ts_now())))
#define NOTE_MSECONDS 0
#define NOTE_ABSTIME 0
#endif
/* NOTE:
* NetBSD fails to link when the EVSETx call is with EVFILT_TIMER and the data
* field is 0. I believe NetBSD supports 0 intervaled timeouts but when the
* value is a static value known at compile time, the macro does something
* that is odd enough to cause an issue unrelated to ev.c elsewhere in shell.c
* where linking just fails awfully. This is a kludge to fix it until someone
* smarter who understands why this is happening can create a better fix.
*/
#ifdef __NetBSD__
#define JANET_KQUEUE_INITIAL_WAIT 1
#else
#define JANET_KQUEUE_INITIAL_WAIT 0
#endif
/* TODO: make this available be we using kqueue or epoll, instead of
* redefinining it for kqueue and epoll separately? */
@ -1756,7 +1757,7 @@ void janet_ev_init(void) {
JANET_KQUEUE_TIMER_IDENT,
EVFILT_TIMER,
JANET_KQUEUE_TF,
NOTE_MSECONDS, JANET_KQUEUE_INITIAL_WAIT, &janet_vm.timer);
NOTE_MSECONDS, JANET_KQUEUE_MIN_INTERVAL, &janet_vm.timer);
EV_SETx(&events[1], janet_vm.selfpipe[0], EVFILT_READ, EV_ADD | EV_ENABLE, 0, 0, janet_vm.selfpipe);
add_kqueue_events(events, 2);
return;