From mboxrd@z Thu Jan 1 00:00:00 1970 From: Uwe =?iso-8859-1?Q?Kleine-K=F6nig?= Subject: Re: Wrong GIT author ? also bugs: serial: Add uart driver for i.MX23/28 Date: Tue, 12 Apr 2011 09:28:43 +0200 Message-ID: <20110412072843.GE7771@pengutronix.de> References: <201103180302.p2I326tA020159@hera.kernel.org> <20110318112920.07148450@lxorguk.ukuu.org.uk> <20110329091456.GI30938@pengutronix.de> <20110329103335.7a5ff37d@lxorguk.ukuu.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Content-Disposition: inline In-Reply-To: <20110329103335.7a5ff37d@lxorguk.ukuu.org.uk> Sender: linux-kernel-owner@vger.kernel.org To: Alan Cox Cc: Linux Kernel Mailing List , Sascha Hauer , linux-serial@vger.kernel.org, torvalds@linux-foundation.org, kernel@pengutronix.de List-Id: linux-serial@vger.kernel.org On Tue, Mar 29, 2011 at 10:33:35AM +0100, Alan Cox wrote: > > I don't understand how to do this and didn't find something like th= at in > > other drivers. (I checked amba-pl011.c and imx.c.) Also > > Documentation/serial/driver doesn't describe this. (It is also sile= nt > > about CMSPAR and doesn't even advise to clear unsupported bits.) > > Can you be a bit more verbose here? >=20 > 8250.c is the best place to look. >=20 > Basically on return from your routine the _hardware_ bits of the term= ios > struct should be the ones set. >=20 > So eg if you only supported 8bit characters you'd wipe the CSx bits a= nd > set CS8. >=20 > For the speed there is a helper so you probably want >=20 > /* Don't rewrite B0 */ > if (tty_termios_baud_rate(termios)) > tty_termios_encode_baud_rate(termios, baud, baud); OK, I will review the function to assert that the other bits are update= d accordingly. > > @@ -286,6 +283,9 @@ static void mxs_auart_settermios(struct uart_po= rt *u, > > { > > u32 bm, ctrl, ctrl2, div; > > unsigned int cflag, baud; > > + unsigned long flags; > > + > > + spin_lock_irqsave(&u->lock, flags); >=20 > Termios you don't need the lock - just the IRQ handler and you still > really want the tty_port_tty_get()/tty_kref_put() if that wasn't alre= ady > dealt with. If I look at 8250.c spin_lock_irqsave seems OK :-) To be honest, it felt much better when a simpler driver than 8250 could be used to copy from. Documentation/serial/driver (which is IMHO very helpful) declares amba_pl011.c to be the reference implementation. It uses neither tty_termios_encode_baud_rate nor tty_port_tty_get()/tty_kref_put() :-( Anyhow, I'll do my best to implement both requests. Best regards Uwe --=20 Pengutronix e.K. | Uwe Kleine-K=F6nig = | Industrial Linux Solutions | http://www.pengutronix.de/= |