[systemd-devel] [systemd-commits] 7 commits - man/systemd.network.xml po/es.po po/LINGUAS src/journal-remote src/libsystemd src/libsystemd-network

[David Herrmann]

Hi

On Fri, Mar 13, 2015 at 5:27 AM, Zbigniew Jędrzejewski-Szmek
<zbyszek at kemper.freedesktop.org> wrote:
> commit f53f7c8fc43df4e38655f2a1f57777c5934fee06
> Author: Zbigniew Jędrzejewski-Szmek <zbyszek at in.waw.pl>
> Date:   Thu Mar 12 21:54:14 2015 -0400
>
>     journal-remote: check also for EWOULDBLOCK
>
>     This matches similar code elsewhere.
>
> diff --git a/src/journal-remote/journal-remote.c b/src/journal-remote/journal-remote.c
> index 01cfef5..e6aa072 100644
> --- a/src/journal-remote/journal-remote.c
> +++ b/src/journal-remote/journal-remote.c
> @@ -1022,7 +1022,7 @@ static int dispatch_raw_source_event(sd_event_source *event,
>          } else if (r == -E2BIG) {
>                  log_notice_errno(E2BIG, "Entry too big, skipped");
>                  return 1;
> -        } else if (r == -EAGAIN) {
> +        } else if (r == -EAGAIN || r == -EWOULDBLOCK) {
>                  return 0;
>          } else if (r < 0) {
>                  log_debug_errno(r, "Closing connection: %m");

What's the reason for this? EWOULDBLOCK is the same as EAGAIN on
linux. Why don't we get rid of all those EWOULDBLOCK checks in
systemd? The only other offender I see is udevd, but that can be
easily changed to EAGAIN.

Thanks
David