[PATCH v2] serial: 8250_dw: support high baudrates if possible

[PATCH v2] serial: 8250_dw: support high baudrates if possible

[Alex Elder]

Currently the Synopsys DesignWare 8250 driver assumes its UART clock
runs at a fixed rate.  If a "real" clock was set up using the common
clock framework, and that clock's rate is adjustable, it may be
possible to support a wider range of baud rates by changing the
UART clock rate.

This is done by setting up a uart_port->set_termios method that
requests a clock rate change if a different rate might make it
more likely to achieve a specified baud.  A new function
dw8250_adjustable_clk() determines whether such clock rate
adjustment is an option.

Signed-off-by: Alex Elder <elder@linaro.org>
---
v2: Limit clock changing to specific compatible devices.

 drivers/tty/serial/8250/8250_dw.c | 57 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 57 insertions(+)

diff --git a/drivers/tty/serial/8250/8250_dw.c b/drivers/tty/serial/8250/8250_dw.c
index 51b307a..77cae28 100644
--- a/drivers/tty/serial/8250/8250_dw.c
+++ b/drivers/tty/serial/8250/8250_dw.c
@@ -242,6 +242,43 @@ dw8250_do_pm(struct uart_port *port, unsigned int state, unsigned int old)
 		pm_runtime_put_sync_suspend(port->dev);
 }
 
+/*
+ * If our clock was configured through the common clock framework we
+ * might be able to support a wider range of baud rates by changing
+ * its frequency.  For the "traditional" baud rates (115200 or
+ * less), request the conventional 1.8432 MHz clock.  The clock
+ * frequency needs to be at least 16 times the baud rate, so
+ * use 16 times the requested rate for those higher than 115200.
+ */
+static void
+dw8250_set_termios(struct uart_port *p, struct ktermios *new,
+			struct ktermios *old)
+{
+	struct dw8250_data *data = p->private_data;
+
+	if (!IS_ERR(data->clk)) {
+		speed_t baud;
+		unsigned int rate;
+
+		baud = tty_termios_baud_rate(new);
+		BUG_ON(baud > (speed_t)UINT_MAX);
+		rate = 16 * max(115200U, (unsigned int)baud);
+
+		/*
+		 * Request a different clock rate if necessary, and
+		 * record it if successful.
+		 */
+		if (rate != p->uartclk) {
+			BUG_ON(!data->clk);
+			if (!clk_set_rate(data->clk, (unsigned long)rate))
+				p->uartclk = rate;
+		}
+	}
+
+	/* The standard code does most of the work. */
+	serial8250_do_set_termios(p, new, old);
+}
+
 static bool dw8250_dma_filter(struct dma_chan *chan, void *param)
 {
 	struct dw8250_data *data = param;
@@ -282,6 +319,21 @@ static void dw8250_setup_port(struct uart_8250_port *up)
 		up->capabilities |= UART_CAP_AFE;
 }
 
+/*
+ * Returns true if the UART clock's rate can be adjusted in order to
+ * achieve higher baud rates.
+ */
+static bool dw8250_adjustable_clk(struct uart_port *p)
+{
+	struct device_node *np = p->dev->of_node;
+
+	if (of_device_is_compatible(np, "brcm,bcm11351-dw-apb-uart"))
+		return true;
+	if (of_device_is_compatible(np, "brcm,bcm21664-dw-apb-uart"))
+		return true;
+	return false;
+}
+
 static int dw8250_probe_of(struct uart_port *p,
 			   struct dw8250_data *data)
 {
@@ -289,6 +341,9 @@ static int dw8250_probe_of(struct uart_port *p,
 	u32			val;
 	bool has_ucv = true;
 
+	if (dw8250_adjustable_clk(p))
+		p->set_termios = dw8250_set_termios;
+
 	if (of_device_is_compatible(np, "cavium,octeon-3860-uart")) {
 #ifdef __BIG_ENDIAN
 		/*
@@ -510,6 +565,8 @@ static const struct dev_pm_ops dw8250_pm_ops = {
 static const struct of_device_id dw8250_of_match[] = {
 	{ .compatible = "snps,dw-apb-uart" },
 	{ .compatible = "cavium,octeon-3860-uart" },
+	{ .compatible = "brcm,bcm11351-dw-apb-uart" },
+	{ .compatible = "brcm,bcm21664-dw-apb-uart" },
 	{ /* Sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, dw8250_of_match);
-- 
1.9.1

Re: [PATCH v2] serial: 8250_dw: support high baudrates if possible

[Greg KH]

On Tue, Jul 01, 2014 at 10:15:34AM -0500, Alex Elder wrote:
> Currently the Synopsys DesignWare 8250 driver assumes its UART clock
> runs at a fixed rate.  If a "real" clock was set up using the common
> clock framework, and that clock's rate is adjustable, it may be
> possible to support a wider range of baud rates by changing the
> UART clock rate.
> 
> This is done by setting up a uart_port->set_termios method that
> requests a clock rate change if a different rate might make it
> more likely to achieve a specified baud.  A new function
> dw8250_adjustable_clk() determines whether such clock rate
> adjustment is an option.
> 
> Signed-off-by: Alex Elder <elder@linaro.org>
> ---
> v2: Limit clock changing to specific compatible devices.
> 
>  drivers/tty/serial/8250/8250_dw.c | 57 +++++++++++++++++++++++++++++++++++++++
>  1 file changed, 57 insertions(+)

This no longer applies due to other changes in this driver.  Can you
please refresh it against my tty-next branch of the tty.git tree and
resend?

thanks,

greg k-h

Re: [PATCH v2] serial: 8250_dw: support high baudrates if possible

[Alex Elder]

On 07/10/2014 06:13 PM, Greg KH wrote:
> On Tue, Jul 01, 2014 at 10:15:34AM -0500, Alex Elder wrote:
>> Currently the Synopsys DesignWare 8250 driver assumes its UART clock
>> runs at a fixed rate.  If a "real" clock was set up using the common
>> clock framework, and that clock's rate is adjustable, it may be
>> possible to support a wider range of baud rates by changing the
>> UART clock rate.
>>
>> This is done by setting up a uart_port->set_termios method that
>> requests a clock rate change if a different rate might make it
>> more likely to achieve a specified baud.  A new function
>> dw8250_adjustable_clk() determines whether such clock rate
>> adjustment is an option.
>>
>> Signed-off-by: Alex Elder <elder@linaro.org>
>> ---
>> v2: Limit clock changing to specific compatible devices.
>>
>>  drivers/tty/serial/8250/8250_dw.c | 57 +++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 57 insertions(+)
> 
> This no longer applies due to other changes in this driver.  Can you
> please refresh it against my tty-next branch of the tty.git tree and
> resend?

Interesting.  Heikki seems to have generalized
byt_set_termios() into dw8250_set_termios(), such
that it's almost the same as what I did.

I think I can just reuse it and simplify my patch
considerably.  But I'll do that in the morning...

					-Alex


> thanks,
> 
> greg k-h
>