Re: [PATCH 24/44] kdbus: Cleanup kdbus_conn_unicast()

[Sergei Zviagintsev]

On Thu, Oct 08, 2015 at 04:34:27PM +0200, David Herrmann wrote:
> Hi
> 
> On Thu, Oct 8, 2015 at 1:31 PM, Sergei Zviagintsev <sergei@xxxxxxxx> wrote:
> > Do not initialize `name' and `ret' as values are assigned to them at
> > the first use by kdbus_pin_dst(). Simplify handling of
> > kdbus_conn_entry_insert() return value and drop useless goto.
> >
> > Signed-off-by: Sergei Zviagintsev <sergei@xxxxxxxx>
> > ---
> >  ipc/kdbus/connection.c | 10 ++++------
> >  1 file changed, 4 insertions(+), 6 deletions(-)
> >
> > diff --git a/ipc/kdbus/connection.c b/ipc/kdbus/connection.c
> > index db49f282a1bf..b3c5f20a57d8 100644
> > --- a/ipc/kdbus/connection.c
> > +++ b/ipc/kdbus/connection.c
> > @@ -1229,12 +1229,12 @@ static int kdbus_conn_unicast(struct kdbus_conn *src,
> >                               struct kdbus_staging *staging)
> >  {
> >         const struct kdbus_msg *msg = staging->msg;
> > -       struct kdbus_name_entry *name = NULL;
> > +       struct kdbus_name_entry *name;
> >         struct kdbus_reply *wait = NULL;
> >         struct kdbus_conn *dst = NULL;
> >         struct kdbus_bus *bus = src->ep->bus;
> >         bool is_signal = (msg->flags & KDBUS_MSG_SIGNAL);
> > -       int ret = 0;
> > +       int ret;
> >
> >         if (WARN_ON(msg->dst_id == KDBUS_DST_ID_BROADCAST) ||
> >             WARN_ON(!(msg->flags & KDBUS_MSG_EXPECT_REPLY) &&
> > @@ -1245,7 +1245,6 @@ static int kdbus_conn_unicast(struct kdbus_conn *src,
> >         down_read(&bus->name_registry->rwlock);
> >
> >         /* find and pin destination */
> > -
> 
> If a comment is addressed to a whole following block, we usually put a
> newline after it. Only if the comment is only addressed at the next
> code-line, we don't.

Sorry, this change is here by mistake and shouldn't be in the patch at
all.

> 
> >         ret = kdbus_pin_dst(bus, staging, &name, &dst);
> >         if (ret < 0)
> >                 goto exit;
> > @@ -1276,11 +1275,10 @@ static int kdbus_conn_unicast(struct kdbus_conn *src,
> >                 kdbus_bus_eavesdrop(bus, src, staging);
> >
> >         ret = kdbus_conn_entry_insert(src, dst, staging, wait, name);
> > -       if (ret < 0 && !is_signal)
> > -               goto exit;
> >
> >         /* signals are treated like broadcasts, recv-errors are ignored */
> > -       ret = 0;
> > +       if (is_signal)
> > +               ret = 0;
> 
> Why? Just to reduce the line-count? You break the code-flow here, by
> making the success-path conditional, instead of the error-path.

IMO, it's easier to read as it's exacly what the comment says: ignore an
error in the case of signal. But I don't mind omitting this change from
the next submission.

> 
> Thanks
> David
> 
> >
> >  exit:
> >         up_read(&bus->name_registry->rwlock);
> > --
> > 1.8.3.1
> >
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/