Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Tue, 12 Jun 2018 14:28:51 +0200
From:   Uwe =?iso-8859-1?Q?Kleine-K=F6nig?= 
        <u.kleine-koenig@pengutronix.de>
To:     Stefan Agner <stefan@agner.ch>
Cc:     gregkh@linuxfoundation.org, festevam@gmail.com, jslaby@suse.com,
        linux-serial@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] serial: imx: fix cached UCR2 read on software reset
Message-ID: <20180612122851.7xjneseskhkvn3ih@pengutronix.de>
References: <20180420124407.12892-1-stefan@agner.ch>
 <20180607075633.y3tm245jv7nkdrqx@pengutronix.de>
 <afda793c86ba7394233e17813589b44c@agner.ch>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <afda793c86ba7394233e17813589b44c@agner.ch>
User-Agent: NeoMutt/20170113 (1.7.2)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, Jun 12, 2018 at 02:11:44PM +0200, Stefan Agner wrote:
> On 07.06.2018 09:56, Uwe Kleine-K?nig wrote:
> > On Fri, Apr 20, 2018 at 02:44:07PM +0200, Stefan Agner wrote:
> >> To reset the UART the SRST needs be cleared (low active). According
> >> to the documentation the bit will remain active for 4 module clocks
> >> until it is cleared (set to 1).
> >>
> >> Hence the real register need to be read in case the cached register
> >> indicates that the SRST bit is zero.
> >>
> >> This bug lead to wrong baudrate because the baud rate register got
> >> restored before reset completed in imx_flush_buffer.
> >>
> >> Fixes: 3a0ab62f43de ("serial: imx: implement shadow registers for UCRx and UFCR")
> >> Signed-off-by: Stefan Agner <stefan@agner.ch>
> >> Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
> >> Reviewed-by: Uwe Kleine-K?nig <u.kleine-koenig@pengutronix.de>
> > 
> > For the record, there is a customer of mine who reports that this commit
> > breaks rs485 communication on i.MX25 because RTS stops to toggle as
> > intended.
> > 
> > (Some details: uart3, fsl,uart-has-rtscts, fsl,dte-mode,
> > linux,rs485-enabled-at-boot-time, native RTS.)
> > 
> > I didn't debug this yet.
> 
> I have seen your patch today "serial: imx: fix comment about UCR2_SRST
> and its handling for shadowing" so I assume you looked into this issue?
> 
> Was it related to that change?

I started looking into this issue, but didn't find the culprit yet.

My work in progress patch looks as follows:

diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
index 758a67f3c8b3..5270173721c6 100644
--- a/drivers/tty/serial/imx.c
+++ b/drivers/tty/serial/imx.c
@@ -1357,6 +1357,12 @@ static int imx_uart_startup(struct uart_port *port)
 	while (!(imx_uart_readl(sport, UCR2) & UCR2_SRST) && (--i > 0))
 		udelay(1);
 
+	if (imx_uart_readl(sport, UCR2) & UCR2_SRST) {
+		spin_unlock_irqrestore(&sport->port.lock, flags);
+		dev_warn(port->dev, "SRST didn't go away\n");
+		return -EIO;
+	}
+
 	/*
 	 * Finally, clear and enable interrupts
 	 */

which seems to trigger on both i.MX25 and i.MX6 at least occasionally.

Maybe 100 us is too short?

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |