diff mbox

seria: sirf: only use lookup table to set baudrate when ioclk=150MHz

Message ID 1356431162-8372-1-git-send-email-Barry.Song@csr.com (mailing list archive)
State New, archived
Headers show

Commit Message

Barry Song Dec. 25, 2012, 10:26 a.m. UTC
From: Barry Song <Baohua.Song@csr.com>

The fast lookup table to set baudrate is only right when ioclk
is 150MHz. for most platforms, ioclk is 150MHz, but some boards
might set ioclk to other frequency.

so re-calc the clk_div_reg when ioclk is not 150MHz.

Signed-off-by: Barry Song <Baohua.Song@csr.com>
---
 drivers/tty/serial/sirfsoc_uart.c |   18 +++++++++++++-----
 1 files changed, 13 insertions(+), 5 deletions(-)

Comments

Sergei Shtylyov Dec. 25, 2012, 6:53 p.m. UTC | #1
Hello.

On 12/25/2012 01:26 PM, Barry Song wrote:

> From: Barry Song <Baohua.Song@csr.com>

> The fast lookup table to set baudrate is only right when ioclk
> is 150MHz. for most platforms, ioclk is 150MHz, but some boards
> might set ioclk to other frequency.

> so re-calc the clk_div_reg when ioclk is not 150MHz.

> Signed-off-by: Barry Song <Baohua.Song@csr.com>
> ---
>  drivers/tty/serial/sirfsoc_uart.c |   18 +++++++++++++-----
>  1 files changed, 13 insertions(+), 5 deletions(-)

> diff --git a/drivers/tty/serial/sirfsoc_uart.c b/drivers/tty/serial/sirfsoc_uart.c
> index 142217c..fe28b1f 100644
> --- a/drivers/tty/serial/sirfsoc_uart.c
> +++ b/drivers/tty/serial/sirfsoc_uart.c
> @@ -375,7 +375,12 @@ static void sirfsoc_uart_set_termios(struct uart_port *port,
>  	int		threshold_div;
>  	int		temp;
>  
> -	ioclk_rate = 150000000;
> +	struct clk *clk = clk_get_sys("io", NULL);

   Please keep the coding style and insert emoty line here; you can also remove
preceding empty line (not to break the declaration block).

> +	BUG_ON(IS_ERR(clk));
> +
> +	ioclk_rate = clk_get_rate(clk);
> +	clk_put(clk);
> +
>  	switch (termios->c_cflag & CSIZE) {
>  	default:
>  	case CS8:

WBR, Sergei
Russell King - ARM Linux Dec. 26, 2012, 3:04 p.m. UTC | #2
On Tue, Dec 25, 2012 at 06:26:02PM +0800, Barry Song wrote:
> @@ -375,7 +375,12 @@ static void sirfsoc_uart_set_termios(struct uart_port *port,
>  	int		threshold_div;
>  	int		temp;
>  
> -	ioclk_rate = 150000000;
> +	struct clk *clk = clk_get_sys("io", NULL);
> +	BUG_ON(IS_ERR(clk));

No.  Really, no.  Stop using BUG_ON() as some kind of crappy assert().
BUG_ON() takes the entire kernel out when it fails.  There's absolutely
no need for this what so ever - especially here.

Get the clock at probe or port initialization time.  Save that pointer.
Only give it up when the port is torn down.  And treat it as any other
clock - prepare and enable it, and disable and unprepare it when you're
done with it.

And there's no need to use this clk_get_sys() crap in drivers.  Add the
necessary clkdev entries or deal with it in DT.  Absolutely do not use
clk_get_sys() in drivers; it's there for *PLATFORM* code to use when
there's no other possibility for them and NOT drivers.
Barry Song Dec. 31, 2012, 2:02 a.m. UTC | #3
2012/12/26, Russell King - ARM Linux <linux@arm.linux.org.uk>:
> On Tue, Dec 25, 2012 at 06:26:02PM +0800, Barry Song wrote:
>> @@ -375,7 +375,12 @@ static void sirfsoc_uart_set_termios(struct uart_port
>> *port,
>>  	int		threshold_div;
>>  	int		temp;
>>
>> -	ioclk_rate = 150000000;
>> +	struct clk *clk = clk_get_sys("io", NULL);
>> +	BUG_ON(IS_ERR(clk));
>
> No.  Really, no.  Stop using BUG_ON() as some kind of crappy assert().
> BUG_ON() takes the entire kernel out when it fails.  There's absolutely
> no need for this what so ever - especially here.
>
> Get the clock at probe or port initialization time.  Save that pointer.
> Only give it up when the port is torn down.  And treat it as any other
> clock - prepare and enable it, and disable and unprepare it when you're
> done with it.
>
> And there's no need to use this clk_get_sys() crap in drivers.  Add the
> necessary clkdev entries or deal with it in DT.  Absolutely do not use
> clk_get_sys() in drivers; it's there for *PLATFORM* code to use when
> there's no other possibility for them and NOT drivers.

Thanks, Russell.
what made this happen is that clkio is used in io bus and by all of
dmac, nand, audio, uart, i2c, spi, usp, pwm, pulse as an input, so we
took it as a global/system clock. this way should be not good for the
real users.
as i have enabled "clk: prima2: enable dt-binding clkdev mapping"
https://patchwork.kernel.org/patch/1898811/
i might simply let uart node in DT takes the uart clk output of clk
controller as your suggestion.

-barry
diff mbox

Patch

diff --git a/drivers/tty/serial/sirfsoc_uart.c b/drivers/tty/serial/sirfsoc_uart.c
index 142217c..fe28b1f 100644
--- a/drivers/tty/serial/sirfsoc_uart.c
+++ b/drivers/tty/serial/sirfsoc_uart.c
@@ -375,7 +375,12 @@  static void sirfsoc_uart_set_termios(struct uart_port *port,
 	int		threshold_div;
 	int		temp;
 
-	ioclk_rate = 150000000;
+	struct clk *clk = clk_get_sys("io", NULL);
+	BUG_ON(IS_ERR(clk));
+
+	ioclk_rate = clk_get_rate(clk);
+	clk_put(clk);
+
 	switch (termios->c_cflag & CSIZE) {
 	default:
 	case CS8:
@@ -431,10 +436,13 @@  static void sirfsoc_uart_set_termios(struct uart_port *port,
 			sirfsoc_uart_disable_ms(port);
 	}
 
-	/* common rate: fast calculation */
-	for (ic = 0; ic < SIRF_BAUD_RATE_SUPPORT_NR; ic++)
-		if (baud_rate == baudrate_to_regv[ic].baud_rate)
-			clk_div_reg = baudrate_to_regv[ic].reg_val;
+	if (ioclk_rate == 150000000) {
+		/* common rate: fast calculation */
+		for (ic = 0; ic < SIRF_BAUD_RATE_SUPPORT_NR; ic++)
+			if (baud_rate == baudrate_to_regv[ic].baud_rate)
+				clk_div_reg = baudrate_to_regv[ic].reg_val;
+	}
+
 	setted_baud = baud_rate;
 	/* arbitary rate setting */
 	if (unlikely(clk_div_reg == 0))