[PATCH 1/1] i3c: master: svc: adjust SDR timing according to i3c spec

[PATCH 1/1] i3c: master: svc: adjust SDR timing according to i3c spec

[Frank Li]

From: Carlos Song <carlos.song@nxp.com>

According to I3C Specification(Version 1.1) 5.1.2.4 "Use of Clock
Speed to Prevent Legacy I2C Devices From Seeing I3C traffic", when
slow i2c devices(FM/FM+ rate i2c frequency without 50ns filter)
works on i3c bus, i3c SDR should work at FM/FM+ rate.

Adjust timing for difference mode.

Signed-off-by: Clark Wang <xiaoning.wang@nxp.com>
Signed-off-by: Carlos Song <carlos.song@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
---
 drivers/i3c/master/svc-i3c-master.c | 66 +++++++++++++++++------------
 1 file changed, 40 insertions(+), 26 deletions(-)

diff --git a/drivers/i3c/master/svc-i3c-master.c b/drivers/i3c/master/svc-i3c-master.c
index bb299ce02cccb..26dc6d833c6e3 100644
--- a/drivers/i3c/master/svc-i3c-master.c
+++ b/drivers/i3c/master/svc-i3c-master.c
@@ -127,6 +127,8 @@
 
 /* This parameter depends on the implementation and may be tuned */
 #define SVC_I3C_FIFO_SIZE 16
+#define SVC_I3C_PPBAUD_MAX 15
+#define SVC_I3C_QUICK_I2C_CLK 4170000
 
 #define SVC_I3C_EVENT_IBI	BIT(0)
 #define SVC_I3C_EVENT_HOTJOIN	BIT(1)
@@ -535,6 +537,7 @@ static int svc_i3c_master_bus_init(struct i3c_master_controller *m)
 	struct i3c_bus *bus = i3c_master_get_bus(m);
 	struct i3c_device_info info = {};
 	unsigned long fclk_rate, fclk_period_ns;
+	unsigned long i2c_period_ns, i2c_scl_rate, i3c_scl_rate;
 	unsigned int high_period_ns, od_low_period_ns;
 	u32 ppbaud, pplow, odhpp, odbaud, odstop, i2cbaud, reg;
 	int ret;
@@ -555,44 +558,55 @@ static int svc_i3c_master_bus_init(struct i3c_master_controller *m)
 	}
 
 	fclk_period_ns = DIV_ROUND_UP(1000000000, fclk_rate);
-
-	/*
-	 * Using I3C Push-Pull mode, target is 12.5MHz/80ns period.
-	 * Simplest configuration is using a 50% duty-cycle of 40ns.
-	 */
-	ppbaud = DIV_ROUND_UP(40, fclk_period_ns) - 1;
-	pplow = 0;
-
-	/*
-	 * Using I3C Open-Drain mode, target is 4.17MHz/240ns with a
-	 * duty-cycle tuned so that high levels are filetered out by
-	 * the 50ns filter (target being 40ns).
-	 */
-	odhpp = 1;
-	high_period_ns = (ppbaud + 1) * fclk_period_ns;
-	odbaud = DIV_ROUND_UP(240 - high_period_ns, high_period_ns) - 1;
-	od_low_period_ns = (odbaud + 1) * high_period_ns;
+	i2c_period_ns = DIV_ROUND_UP(1000000000, bus->scl_rate.i2c);
+	i2c_scl_rate = bus->scl_rate.i2c;
+	i3c_scl_rate = bus->scl_rate.i3c;
 
 	switch (bus->mode) {
 	case I3C_BUS_MODE_PURE:
+		/* Using I3C Push-Pull mode and I2C OP 50% duty-cycle. */
+		pplow = 0;
+		ppbaud = DIV_ROUND_UP(fclk_rate / 2, i3c_scl_rate) - 1;
+		high_period_ns = (ppbaud + 1) * fclk_period_ns;
+		odbaud = DIV_ROUND_UP(fclk_rate, SVC_I3C_QUICK_I2C_CLK * (1 + ppbaud)) - 2;
+		od_low_period_ns = (odbaud + 1) * high_period_ns;
 		i2cbaud = 0;
 		odstop = 0;
 		break;
 	case I3C_BUS_MODE_MIXED_FAST:
-	case I3C_BUS_MODE_MIXED_LIMITED:
 		/*
-		 * Using I2C Fm+ mode, target is 1MHz/1000ns, the difference
-		 * between the high and low period does not really matter.
+		 * I3C in <= 12.5M PP + I3C OP + I2C OP in clk rate
+		 * keep I3C OD clk high period <= 40ns and use odbaud and pplow
+		 * to adjust the i2c/i3c duty cycle.
 		 */
-		i2cbaud = DIV_ROUND_UP(1000, od_low_period_ns) - 2;
+		ppbaud = DIV_ROUND_UP(fclk_rate / 2, I3C_BUS_TYP_I3C_SCL_RATE) - 1;
+		high_period_ns = (ppbaud + 1) * fclk_period_ns;
+		pplow =  DIV_ROUND_UP(fclk_rate, i3c_scl_rate) - (2 + 2 * ppbaud);
+
+		odhpp = 1;
+		odbaud = DIV_ROUND_UP(fclk_rate, SVC_I3C_QUICK_I2C_CLK * (1 + ppbaud)) - 2;
+
+		/* i2c FM/FM+ */
+		od_low_period_ns = (odbaud + 1) * high_period_ns;
+		i2cbaud = DIV_ROUND_UP(i2c_period_ns, od_low_period_ns) - 2;
 		odstop = 1;
 		break;
+	case I3C_BUS_MODE_MIXED_LIMITED:
 	case I3C_BUS_MODE_MIXED_SLOW:
-		/*
-		 * Using I2C Fm mode, target is 0.4MHz/2500ns, with the same
-		 * constraints as the FM+ mode.
-		 */
-		i2cbaud = DIV_ROUND_UP(2500, od_low_period_ns) - 2;
+		/* I3C PP + I3C OP + I2C OP both use i2c clk rate */
+		ppbaud = DIV_ROUND_UP(fclk_rate / 2, i3c_scl_rate) - 1;
+		pplow = 0;
+		if (ppbaud > SVC_I3C_PPBAUD_MAX) {
+			ppbaud = SVC_I3C_PPBAUD_MAX;
+			pplow =  DIV_ROUND_UP(fclk_rate, i3c_scl_rate) - (2 + 2 * ppbaud);
+		}
+
+		high_period_ns = (ppbaud + 1) * fclk_period_ns;
+		odhpp = 0;
+		odbaud = DIV_ROUND_UP(fclk_rate, i2c_scl_rate * (2 + 2 * ppbaud)) - 1;
+
+		od_low_period_ns = (odbaud + 1) * high_period_ns;
+		i2cbaud = DIV_ROUND_UP(i2c_period_ns, od_low_period_ns) - 2;
 		odstop = 1;
 		break;
 	default:
-- 
2.34.1


-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c

Re: [PATCH 1/1] i3c: master: svc: adjust SDR timing according to i3c spec

[Miquel Raynal]

Hi Frank,

Frank.Li@nxp.com wrote on Fri, 12 Jul 2024 12:46:02 -0400:

> From: Carlos Song <carlos.song@nxp.com>
> 
> According to I3C Specification(Version 1.1) 5.1.2.4 "Use of Clock
> Speed to Prevent Legacy I2C Devices From Seeing I3C traffic", when
> slow i2c devices(FM/FM+ rate i2c frequency without 50ns filter)
> works on i3c bus, i3c SDR should work at FM/FM+ rate.
> 
> Adjust timing for difference mode.

The calculations below look quite complex, but if you believe this is
right, I'm fine with it.

Acked-by: Miquel Raynal <miquel.raynal@bootlin.com>

> 
> Signed-off-by: Clark Wang <xiaoning.wang@nxp.com>
> Signed-off-by: Carlos Song <carlos.song@nxp.com>
> Signed-off-by: Frank Li <Frank.Li@nxp.com>
> ---

Thanks,
Miquèl

-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c