[PATCH] serial: sh-sci: Fix HSCIF RX sampling point calculation

[PATCH] serial: sh-sci: Fix HSCIF RX sampling point calculation

[Geert Uytterhoeven]

There are several issues with the formula used for calculating the
deviation from the intended rate:
  1. While min_err and last_stop are signed, srr and baud are unsigned.
     Hence the signed values are promoted to unsigned, which will lead
     to a bogus value of deviation if min_err is negative,
  2. Srr is the register field value, which is one less than the actual
     sampling rate factor,
  3. The divisions do not use rounding.

Fix this by casting unsigned variables to int, adding one to srr, and
using a single DIV_ROUND_CLOSEST().

Fixes: 63ba1e00f178a448 ("serial: sh-sci: Support for HSCIF RX sampling point adjustment")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
Anyone with a good test setup for verifying this feature actually works
as advertised?
---
 drivers/tty/serial/sh-sci.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
index 2bdaeba5d527a6ce..8dea59edde34c224 100644
--- a/drivers/tty/serial/sh-sci.c
+++ b/drivers/tty/serial/sh-sci.c
@@ -2522,7 +2522,9 @@ static void sci_set_termios(struct uart_port *port, struct ktermios *termios,
 			 * center of the last stop bit in sampling clocks.
 			 */
 			int last_stop = bits * 2 - 1;
-			int deviation = min_err * srr * last_stop / 2 / baud;
+			int deviation = DIV_ROUND_CLOSEST(min_err * last_stop *
+							  (int)(srr + 1),
+							  2 * (int)baud);
 
 			if (abs(deviation) >= 2) {
 				/* At least two sampling clocks off at the
-- 
2.17.1

Re: [PATCH] serial: sh-sci: Fix HSCIF RX sampling point calculation

[Mukesh Ojha]

On 4/1/2019 4:55 PM, Geert Uytterhoeven wrote:
> There are several issues with the formula used for calculating the
> deviation from the intended rate:
>    1. While min_err and last_stop are signed, srr and baud are unsigned.
>       Hence the signed values are promoted to unsigned, which will lead
>       to a bogus value of deviation if min_err is negative,
>    2. Srr is the register field value, which is one less than the actual
>       sampling rate factor,
>    3. The divisions do not use rounding.
>
> Fix this by casting unsigned variables to int, adding one to srr, and
> using a single DIV_ROUND_CLOSEST().
>
> Fixes: 63ba1e00f178a448 ("serial: sh-sci: Support for HSCIF RX sampling point adjustment")
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>


Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>

Cheers,
-Mukesh

> ---
> Anyone with a good test setup for verifying this feature actually works
> as advertised?
> ---
>   drivers/tty/serial/sh-sci.c | 4 +++-
>   1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
> index 2bdaeba5d527a6ce..8dea59edde34c224 100644
> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -2522,7 +2522,9 @@ static void sci_set_termios(struct uart_port *port, struct ktermios *termios,
>   			 * center of the last stop bit in sampling clocks.
>   			 */
>   			int last_stop = bits * 2 - 1;
> -			int deviation = min_err * srr * last_stop / 2 / baud;
> +			int deviation = DIV_ROUND_CLOSEST(min_err * last_stop *
> +							  (int)(srr + 1),
> +							  2 * (int)baud);
>   
>   			if (abs(deviation) >= 2) {
>   				/* At least two sampling clocks off at the

Re: [PATCH] serial: sh-sci: Fix HSCIF RX sampling point calculation

[Ulrich Hecht]

> On April 1, 2019 at 1:25 PM Geert Uytterhoeven <geert+renesas@glider.be> wrote:
> 
> 
> There are several issues with the formula used for calculating the
> deviation from the intended rate:
>   1. While min_err and last_stop are signed, srr and baud are unsigned.
>      Hence the signed values are promoted to unsigned, which will lead
>      to a bogus value of deviation if min_err is negative,
>   2. Srr is the register field value, which is one less than the actual
>      sampling rate factor,
>   3. The divisions do not use rounding.
> 
> Fix this by casting unsigned variables to int, adding one to srr, and
> using a single DIV_ROUND_CLOSEST().
> 
> Fixes: 63ba1e00f178a448 ("serial: sh-sci: Support for HSCIF RX sampling point adjustment")
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>

Reviewed-by: Ulrich Hecht <uli+renesas@fpond.eu>

CU
Uli

Re: [PATCH] serial: sh-sci: Fix HSCIF RX sampling point calculation

[Eugeniu Rosca]

Hi Geert,

On Mon, Apr 01, 2019 at 01:25:10PM +0200, Geert Uytterhoeven wrote:
> There are several issues with the formula used for calculating the
> deviation from the intended rate:
>   1. While min_err and last_stop are signed, srr and baud are unsigned.
>      Hence the signed values are promoted to unsigned, which will lead
>      to a bogus value of deviation if min_err is negative,
>   2. Srr is the register field value, which is one less than the actual
>      sampling rate factor,
>   3. The divisions do not use rounding.
> 
> Fix this by casting unsigned variables to int, adding one to srr, and
> using a single DIV_ROUND_CLOSEST().
> 
> Fixes: 63ba1e00f178a448 ("serial: sh-sci: Support for HSCIF RX sampling point adjustment")
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> ---
> Anyone with a good test setup for verifying this feature actually works
> as advertised?
> ---
>  drivers/tty/serial/sh-sci.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
> index 2bdaeba5d527a6ce..8dea59edde34c224 100644
> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -2522,7 +2522,9 @@ static void sci_set_termios(struct uart_port *port, struct ktermios *termios,
>  			 * center of the last stop bit in sampling clocks.
>  			 */
>  			int last_stop = bits * 2 - 1;
> -			int deviation = min_err * srr * last_stop / 2 / baud;
> +			int deviation = DIV_ROUND_CLOSEST(min_err * last_stop *
> +							  (int)(srr + 1),
> +							  2 * (int)baud);
>  
>  			if (abs(deviation) >= 2) {
>  				/* At least two sampling clocks off at the

Just wanted to say thank you for your tremendous commitment and
continuity in supporting the R-Car platform (and beyond that), which
allowed us to gracefully overcome the problems encountered in the
project in the best possible time, specifically thanks to this patch.

Best regards,
Eugeniu.

Re: [PATCH] serial: sh-sci: Fix HSCIF RX sampling point calculation

[Eugeniu Rosca]

Hi Geert,

On Mon, Apr 01, 2019 at 01:25:10PM +0200, Geert Uytterhoeven wrote:
> There are several issues with the formula used for calculating the
> deviation from the intended rate:
>   1. While min_err and last_stop are signed, srr and baud are unsigned.
>      Hence the signed values are promoted to unsigned, which will lead
>      to a bogus value of deviation if min_err is negative,
>   2. Srr is the register field value, which is one less than the actual
>      sampling rate factor,
>   3. The divisions do not use rounding.
> 
> Fix this by casting unsigned variables to int, adding one to srr, and
> using a single DIV_ROUND_CLOSEST().
> 
> Fixes: 63ba1e00f178a448 ("serial: sh-sci: Support for HSCIF RX sampling point adjustment")
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> ---
> Anyone with a good test setup for verifying this feature actually works
> as advertised?
> ---
>  drivers/tty/serial/sh-sci.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
> index 2bdaeba5d527a6ce..8dea59edde34c224 100644
> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -2522,7 +2522,9 @@ static void sci_set_termios(struct uart_port *port, struct ktermios *termios,
>  			 * center of the last stop bit in sampling clocks.
>  			 */
>  			int last_stop = bits * 2 - 1;
> -			int deviation = min_err * srr * last_stop / 2 / baud;
> +			int deviation = DIV_ROUND_CLOSEST(min_err * last_stop *
> +							  (int)(srr + 1),
> +							  2 * (int)baud);
>  
>  			if (abs(deviation) >= 2) {
>  				/* At least two sampling clocks off at the

Just wanted to say thank you for your tremendous commitment and
continuity in supporting the R-Car platform (and beyond that), which
allowed us to gracefully overcome the problems encountered in the
project in the best possible time, specifically thanks to this patch.

Best regards,
Eugeniu.