Skip to content

fix: --count exists without waiting for extra input message #182

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
regalk13 wants to merge 1 commit into
base: next
Choose a base branch
from
Open

fix: --count exists without waiting for extra input message #182

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 12 additions & 22 deletions nmsg/io.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -741,29 +741,24 @@ io_write(struct nmsg_io_thr *iothr, struct nmsg_io_output *io_output,
nmsg_io_t io = iothr->io;
nmsg_res res;

/* It's possible a set "count" has been reached. */
check_close_event(iothr, io_output, 1);

if (io->stop) {
reset_close_event(iothr, io_output);
nmsg_message_destroy(&msg);
return (nmsg_res_stop);
}

res = nmsg_output_write(io_output->output, msg);
if (io_output->output->type != nmsg_output_type_callback)
nmsg_message_destroy(&msg);

/*
* Reset only after the write, in case another thread invokes
* check_close_event and makes changes to io_output in the meantime.
*/
reset_close_event(iothr, io_output);

if (res != nmsg_res_success)
return (res);

Copy link
Copy Markdown
Contributor

@cmikk cmikk May 29, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This commit adds stray whitespace on this and one other line, a common editor artifact.

git diff --check will show these before adding, and git diff --cached --check will show these before committing. I have the latter set up in a pre-commit hook script so I don't forget myself :)

atomic_fetch_add_explicit(&io->io_count_nmsg_payload_out, 1, memory_order_relaxed);

/* It's possible a set "count" has been reached. */
check_close_event(iothr, io_output, 1);
reset_close_event(iothr, io_output);
Copy link
Copy Markdown
Contributor

@cmikk cmikk May 29, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This change results in check_close_event and reset_close_event being called as a pair both places they are called, meaning that the reset_close_event code could be merged into check_close_event.

However, they were affirmatively separated at the last refactor of this code: before commit 77f40, check_close_event was a single function called after the top-level write. That commit split it into check_close_event and reset_close_event, with the underlying write call in between. I do not recall the precise reason I did that ~3 years ago, but it must have been sufficiently compelling to add the reset_close_event function.


if (io->stop) {
// nmsg_message_destroy(&msg);
return (nmsg_res_stop);
}

return (res);
}
Expand Down Expand Up @@ -797,6 +792,8 @@ check_close_event(struct nmsg_io_thr *iothr, struct nmsg_io_output *io_output, u
if (io->count > 0 && io_output->count_next_close == 0)
io_output->count_next_close = io->count;

io_output->count_nmsg_payload_out += count;

if (io->count > 0 &&
io_output->count_nmsg_payload_out == io_output->count_next_close)
{
Expand Down Expand Up @@ -856,13 +853,6 @@ check_close_event(struct nmsg_io_thr *iothr, struct nmsg_io_output *io_output, u
}

out:
/*
* This incr is implicitly locked IF it's used, and this counter is
* only used IF io->count > 0; that condition results in an acquired
* lock at the beginning of this function.
*/
io_output->count_nmsg_payload_out += count;

if (io->close_fp != NULL || io->count > 0)
pthread_mutex_unlock(&io_output->lock);
}
Expand Down