On Thu, Nov 17, 2022 at 10:38:16AM +0100, Geert Uytterhoeven wrote:
> Hi Leon,
>
> On Thu, Nov 17, 2022 at 10:14 AM Leon Romanovsky <leon@xxxxxxxxxx> wrote:
> > On Thu, Nov 17, 2022 at 09:59:55AM +0100, Geert Uytterhoeven wrote:
> > > On Thu, Nov 17, 2022 at 9:58 AM Yoshihiro Shimoda
> > > <yoshihiro.shimoda.uh@xxxxxxxxxxx> wrote:
> > > > > From: Leon Romanovsky, Sent: Thursday, November 17, 2022 3:09 PM
> > > > > > --- a/drivers/net/ethernet/renesas/rswitch.c
> > > > > > +++ b/drivers/net/ethernet/renesas/rswitch.c
> > > > > > @@ -1714,7 +1714,7 @@ static int rswitch_init(struct rswitch_private *priv)
> > > > > > }
> > > > > >
> > > > > > for (i = 0; i < RSWITCH_NUM_PORTS; i++)
> > > > > > - netdev_info(priv->rdev[i]->ndev, "MAC address %pMn",
> > > > > > + netdev_info(priv->rdev[i]->ndev, "MAC address %pM\n",
> > > > >
> > > > > You can safely drop '\n' from here. It is not needed while printing one
> > > > > line.
> > > >
> > > > Oh, I didn't know that. I'll remove '\n' from here on v2 patch.
> > >
> > > Please don't remove it. The convention is to have the newlines.
> >
> > Can you please explain why?
>
> I'm quite sure this was discussed in the context of commits
> 5fd29d6ccbc98884 ("printk: clean up handling of log-levels and
> newlines") and 4bcc595ccd80decb ("printk: reinstate KERN_CONT for
> printing continuation lines"), but I couldn't find a pointer to an
> official statement.
Not a printk expert, but in first commit, Linus removed need of "\n"
together with KERN_CONT, and in second commit he returned KERN_CONT,
but didn't return need of "\n".
>
> I did find[1], which states:
>
> The printk subsystem will, for every printk, check
> if the last printk has a newline termination and if
> it doesn't and the current printk does not start with
> KERN_CONT will insert a newline.
>
> The negative to this approach is the last printk,
> if it does not have a newline, is buffered and not
> emitted until another printk occurs.
I have no idea if it is continue to be true in 2022.
>
> There is also the (now small) possibility that
> multiple concurrent kernel threads or processes
> could interleave printks without a terminating
> newline and a different process could emit a
> printk that starts with KERN_CONT and the emitted
> message could be garbled.
>
> [1] https://lore.kernel.org/all/b867ee8a02043ec6b18c9330bfe3a091d66c816c.camel@xxxxxxxxxxx
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds
