skv/foot:timeout-nsec-overflow into master
When resize-delay-ms >= 1000, timerfd_settime fails with EINVAL (foot output: failed to arm TIOCSWINSZ timer). This PR adds a simple fix. I also skimmed over the code and it seems that any other usage of timerfd_settime is safe.
Thanks! See note below. In addition to that, I'd like to see a changelog entry for this.
@ -3722,0 +3722,4 @@
if (timeout.it_value.tv_nsec >= 1000000000) {
timeout.it_value.tv_sec += timeout.it_value.tv_nsec / 1000000000;
timeout.it_value.tv_nsec %= 1000000000;
}
We can do better than this: just calculate tv_sec and tv_nsec directly:
diff --git a/render.c b/render.c
index b285b120..f14911d4 100644
--- a/render.c
+++ b/render.c
@@ -3716,7 +3716,10 @@ send_dimensions_to_client(struct terminal *term)
if (fd >= 0) {
/* Reset timeout */
const struct itimerspec timeout = {
- .it_value = {.tv_sec = 0, .tv_nsec = delay_ms * 1000000},
+ .it_value = {
+ .tv_sec = delay_ms / 1000,
+ .tv_nsec = (delay_ms % 1000) * 1000000,
+ },
};
if (timerfd_settime(fd, 0, &timeout, NULL) < 0) {
bc9d8829f1
b1e2167655
Thanks! See note below. In addition to that, I'd like to see a changelog entry for this.
Done!
@dnkl Sorry, I must have accidentally closed this PR somehow, though I do not how.
Awesome, thanks! Merged in 37218be648
No due date set.
No dependencies set.
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?