On Fri, 2016-10-07 at 17:52 +1100, Brian May wrote: > > Ben Hutchings <ben@decadent.org.uk> writes: > > > > It looks like this patch does three things > > > > > > * It removes "assert(n > 0)". > > > > > > * It removes the now unused n parameter from the > > > manager_invoke_notify_message() function. > > > > > > * It removes the return(0) if n==0. This looks like the only relevant part. > > > > > > For the first two changes, it looks like the > > > manager_invoke_notify_message() function and hence the assert was only > > > introduced in systemd in the following commit, before tag v209. This was > > > not in the wheezy version, so I don't think these parts are required. > > > > [...] > > > > Right. > > > This means patch 4 is just the inverse of patch 3, so both become > redundant. Oh, I didn't realise that. Let's drop those then. [...] > > > --- systemd-44/debian/patches/no_error_on_zero_len.patch 1970-01-01 10:00:00.000000000 +1000 > +++ systemd-44/debian/patches/no_error_on_zero_len.patch 2016-10-06 18:16:46.000000000 +1100 > @@ -0,0 +1,12 @@ > +--- a/src/manager.c > ++++ b/src/manager.c > +@@ -2030,9 +2030,6 @@ > + msghdr.msg_controllen = sizeof(control); > + > + if ((n = recvmsg(m->notify_watch.fd, &msghdr, MSG_DONTWAIT)) <= 0) { > +- if (n >= 0) > +- return -EIO; > +- > + if (errno == EAGAIN || errno == EINTR) > + break; > + [...] This patch also needs to change the '<= 0' to '< 0', doesn't it? Ben. -- Ben Hutchings Every program is either trivial or else contains at least one bug
Attachment:
signature.asc
Description: This is a digitally signed message part