[PATCH V3 1/8] clk: vc5: Prevent division by zero on unconfigured outputs

[PATCH V3 1/8] clk: vc5: Prevent division by zero on unconfigured outputs

[Marek Vasut]

In case the initial values of the FOD registers are not configured in
the OTP or by the bootloader, it is possible that the FOD registers
will contain zeroes. The code in vc5_fod_recalc_rate() immediately
feeds the FOD divider value obtained from the FOD registers into the
div64_u64() and if the FOD divider value is zero, triggers division
by zero exception.

Check if the FOD divider value is zero and return the frequency of
the FOD output as 0 Hz if it is so. This prevents the division by
zero exception.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: None
V3: None
---
 drivers/clk/clk-versaclock5.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index ea7d552a2f2b..60bf4afb51bd 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -426,6 +426,10 @@ static unsigned long vc5_fod_recalc_rate(struct clk_hw *hw,
 	div_frc = (od_frc[0] << 22) | (od_frc[1] << 14) |
 		  (od_frc[2] << 6) | (od_frc[3] >> 2);
 
+	/* Avoid division by zero if the output is not configured. */
+	if ((div_int == 0) && (div_frc == 0))
+		return 0;
+
 	/* The PLL divider has 12 integer bits and 30 fractional bits */
 	return div64_u64((u64)f_in << 24ULL, ((u64)div_int << 24ULL) + div_frc);
 }
-- 
2.11.0

[PATCH V3 2/8] clk: vc5: Fix trivial typo

[Marek Vasut]

Fix trivial typo in vc5_clk_out_unprepare() , s/Enable/Disable/ .

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: None
V3: None
---
 drivers/clk/clk-versaclock5.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index 60bf4afb51bd..d835d48cf14d 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -520,7 +520,7 @@ static void vc5_clk_out_unprepare(struct clk_hw *hw)
 	struct vc5_hw_data *hwdata = container_of(hw, struct vc5_hw_data, hw);
 	struct vc5_driver_data *vc5 = hwdata->vc5;
 
-	/* Enable the clock buffer */
+	/* Disable the clock buffer */
 	regmap_update_bits(vc5->regmap, VC5_CLK_OUTPUT_CFG(hwdata->num, 1),
 			   VC5_CLK_OUTPUT_CFG1_EN_CLKBUF, 0);
 }
-- 
2.11.0

Re: [PATCH V3 2/8] clk: vc5: Fix trivial typo

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> Fix trivial typo in vc5_clk_out_unprepare() , s/Enable/Disable/ .
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH V3 3/8] clk: vc5: Do not warn about disabled output buffer input muxes

[Marek Vasut]

The output buffer input mux can be configured in either of three
states -- disabled, input from FOD, input from previous output.
If the output buffer input mux is set to disabled, the code in
vc5_clk_out_get_parent() would consider this an invalid setting
and warn about it, which is not necessarily the case.

In case the output buffer input mux is disabled, default to input
from FOD to have some parent and don't print the warning.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: Drop chunk adding useless newline
V3: None
---
 drivers/clk/clk-versaclock5.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index d835d48cf14d..248689d89632 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -541,6 +541,9 @@ static unsigned char vc5_clk_out_get_parent(struct clk_hw *hw)
 	regmap_read(vc5->regmap, VC5_OUT_DIV_CONTROL(hwdata->num), &src);
 	src &= mask;
 
+	if (src == 0)	/* Input mux set to DISABLED */
+		return 0;
+
 	if ((src & fodclkmask) == VC5_OUT_DIV_CONTROL_EN_FOD)
 		return 0;
 
-- 
2.11.0

Re: [PATCH V3 3/8] clk: vc5: Do not warn about disabled output buffer input muxes

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> The output buffer input mux can be configured in either of three
> states -- disabled, input from FOD, input from previous output.
> If the output buffer input mux is set to disabled, the code in
> vc5_clk_out_get_parent() would consider this an invalid setting
> and warn about it, which is not necessarily the case.
> 
> In case the output buffer input mux is disabled, default to input
> from FOD to have some parent and don't print the warning.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH V3 4/8] clk: vc5: Configure the output buffer input mux on prepare

[Marek Vasut]

The output buffer input mux can be configured in either of three
states -- disabled, input from FOD, input from previous output.
Once the .prepare() callback of the output buffer is called, the
output buffer input mux must be set to either input from FOD or
input from previous output, it cannot be set to Disabled anymore
or the output won't work.

Default to the input from FOD if the output buffer input mux was
Disabled and the .prepare() was called on it.

Note that we do not set the output buffer input mux back to Disabled
in the .unprepare() callback as there is no obvious benefit of doing
so. We disable the entire output buffer in the .unprepare() callback
already.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: None
V3: None
---
 drivers/clk/clk-versaclock5.c | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index 248689d89632..7bdfd34e8280 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -507,6 +507,25 @@ static int vc5_clk_out_prepare(struct clk_hw *hw)
 {
 	struct vc5_hw_data *hwdata = container_of(hw, struct vc5_hw_data, hw);
 	struct vc5_driver_data *vc5 = hwdata->vc5;
+	const u8 mask = VC5_OUT_DIV_CONTROL_SELB_NORM |
+			VC5_OUT_DIV_CONTROL_SEL_EXT |
+			VC5_OUT_DIV_CONTROL_EN_FOD;
+	unsigned int src;
+	int ret;
+
+	/*
+	 * If the input mux is disabled, enable it first and
+	 * select source from matching FOD.
+	 */
+	regmap_read(vc5->regmap, VC5_OUT_DIV_CONTROL(hwdata->num), &src);
+	if ((src & mask) == 0) {
+		src = VC5_OUT_DIV_CONTROL_RESET | VC5_OUT_DIV_CONTROL_EN_FOD;
+		ret = regmap_update_bits(vc5->regmap,
+					 VC5_OUT_DIV_CONTROL(hwdata->num),
+					 mask | VC5_OUT_DIV_CONTROL_RESET, src);
+		if (ret)
+			return ret;
+	}
 
 	/* Enable the clock buffer */
 	regmap_update_bits(vc5->regmap, VC5_CLK_OUTPUT_CFG(hwdata->num, 1),
-- 
2.11.0

Re: [PATCH V3 4/8] clk: vc5: Configure the output buffer input mux on prepare

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> The output buffer input mux can be configured in either of three
> states -- disabled, input from FOD, input from previous output.
> Once the .prepare() callback of the output buffer is called, the
> output buffer input mux must be set to either input from FOD or
> input from previous output, it cannot be set to Disabled anymore
> or the output won't work.
> 
> Default to the input from FOD if the output buffer input mux was
> Disabled and the .prepare() was called on it.
> 
> Note that we do not set the output buffer input mux back to Disabled
> in the .unprepare() callback as there is no obvious benefit of doing
> so. We disable the entire output buffer in the .unprepare() callback
> already.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH V3 5/8] clk: vc5: Split clock input mux and predivider

[Marek Vasut]

Split the VC5 clock input mux and the predivider to more accurately
model the hardware and fix the previously incorrect assumption that
both the OUT_SEL_I2CB and the PLL are fed from the predivider.

It is in fact the clock input mux output which is directly feeding
the clock into the OUT_SEL_I2CB output, while the clock input mux
output first passes through the predivider before it is fed into
the PLL.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: None
V3: None
---
 drivers/clk/clk-versaclock5.c | 46 ++++++++++++++++++++++++++++++++-----------
 1 file changed, 34 insertions(+), 12 deletions(-)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index 7bdfd34e8280..453c9f721b78 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -157,6 +157,7 @@ struct vc5_driver_data {
 	struct clk		*pin_clkin;
 	unsigned char		clk_mux_ins;
 	struct clk_hw		clk_mux;
+	struct clk_hw		clk_pfd;
 	struct vc5_hw_data	clk_pll;
 	struct vc5_hw_data	clk_fod[VC5_MAX_FOD_NUM];
 	struct vc5_hw_data	clk_out[VC5_MAX_CLK_OUT_NUM];
@@ -166,6 +167,10 @@ static const char * const vc5_mux_names[] = {
 	"mux"
 };
 
+static const char * const vc5_pfd_names[] = {
+	"pfd"
+};
+
 static const char * const vc5_pll_names[] = {
 	"pll"
 };
@@ -254,11 +259,16 @@ static int vc5_mux_set_parent(struct clk_hw *hw, u8 index)
 	return regmap_update_bits(vc5->regmap, VC5_PRIM_SRC_SHDN, mask, src);
 }
 
-static unsigned long vc5_mux_recalc_rate(struct clk_hw *hw,
+static const struct clk_ops vc5_mux_ops = {
+	.set_parent	= vc5_mux_set_parent,
+	.get_parent	= vc5_mux_get_parent,
+};
+
+static unsigned long vc5_pfd_recalc_rate(struct clk_hw *hw,
 					 unsigned long parent_rate)
 {
 	struct vc5_driver_data *vc5 =
-		container_of(hw, struct vc5_driver_data, clk_mux);
+		container_of(hw, struct vc5_driver_data, clk_pfd);
 	unsigned int prediv, div;
 
 	regmap_read(vc5->regmap, VC5_VCO_CTRL_AND_PREDIV, &prediv);
@@ -276,7 +286,7 @@ static unsigned long vc5_mux_recalc_rate(struct clk_hw *hw,
 		return parent_rate / VC5_REF_DIVIDER_REF_DIV(div);
 }
 
-static long vc5_mux_round_rate(struct clk_hw *hw, unsigned long rate,
+static long vc5_pfd_round_rate(struct clk_hw *hw, unsigned long rate,
 			       unsigned long *parent_rate)
 {
 	unsigned long idiv;
@@ -296,11 +306,11 @@ static long vc5_mux_round_rate(struct clk_hw *hw, unsigned long rate,
 	return *parent_rate / idiv;
 }
 
-static int vc5_mux_set_rate(struct clk_hw *hw, unsigned long rate,
+static int vc5_pfd_set_rate(struct clk_hw *hw, unsigned long rate,
 			    unsigned long parent_rate)
 {
 	struct vc5_driver_data *vc5 =
-		container_of(hw, struct vc5_driver_data, clk_mux);
+		container_of(hw, struct vc5_driver_data, clk_pfd);
 	unsigned long idiv;
 	u8 div;
 
@@ -328,12 +338,10 @@ static int vc5_mux_set_rate(struct clk_hw *hw, unsigned long rate,
 	return 0;
 }
 
-static const struct clk_ops vc5_mux_ops = {
-	.set_parent	= vc5_mux_set_parent,
-	.get_parent	= vc5_mux_get_parent,
-	.recalc_rate	= vc5_mux_recalc_rate,
-	.round_rate	= vc5_mux_round_rate,
-	.set_rate	= vc5_mux_set_rate,
+static const struct clk_ops vc5_pfd_ops = {
+	.recalc_rate	= vc5_pfd_recalc_rate,
+	.round_rate	= vc5_pfd_round_rate,
+	.set_rate	= vc5_pfd_set_rate,
 };
 
 /*
@@ -698,12 +706,26 @@ static int vc5_probe(struct i2c_client *client,
 		goto err_clk;
 	}
 
+	/* Register PFD */
+	memset(&init, 0, sizeof(init));
+	init.name = vc5_pfd_names[0];
+	init.ops = &vc5_pfd_ops;
+	init.flags = CLK_SET_RATE_PARENT;
+	init.parent_names = vc5_mux_names;
+	init.num_parents = 1;
+	vc5->clk_pfd.init = &init;
+	ret = devm_clk_hw_register(&client->dev, &vc5->clk_pfd);
+	if (ret) {
+		dev_err(&client->dev, "unable to register %s\n", init.name);
+		goto err_clk;
+	}
+
 	/* Register PLL */
 	memset(&init, 0, sizeof(init));
 	init.name = vc5_pll_names[0];
 	init.ops = &vc5_pll_ops;
 	init.flags = CLK_SET_RATE_PARENT;
-	init.parent_names = vc5_mux_names;
+	init.parent_names = vc5_pfd_names;
 	init.num_parents = 1;
 	vc5->clk_pll.num = 0;
 	vc5->clk_pll.vc5 = vc5;
-- 
2.11.0

Re: [PATCH V3 5/8] clk: vc5: Split clock input mux and predivider

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> Split the VC5 clock input mux and the predivider to more accurately
> model the hardware and fix the previously incorrect assumption that
> both the OUT_SEL_I2CB and the PLL are fed from the predivider.
> 
> It is in fact the clock input mux output which is directly feeding
> the clock into the OUT_SEL_I2CB output, while the clock input mux
> output first passes through the predivider before it is fed into
> the PLL.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH V3 6/8] clk: vc5: Add support for the input frequency doubler

[Marek Vasut]

The VersaClock 6 has an input frequency doubler between the input
clock mux and the predivider. Add new capability flag and support
for this frequency doubler block into the driver.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: None
V3: None
---
 drivers/clk/clk-versaclock5.c | 78 ++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 77 insertions(+), 1 deletion(-)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index 453c9f721b78..d0406b62133a 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -57,6 +57,7 @@
 #define VC5_PRIM_SRC_SHDN			0x10
 #define VC5_PRIM_SRC_SHDN_EN_XTAL		BIT(7)
 #define VC5_PRIM_SRC_SHDN_EN_CLKIN		BIT(6)
+#define VC5_PRIM_SRC_SHDN_EN_DOUBLE_XTAL_FREQ	BIT(3)
 #define VC5_PRIM_SRC_SHDN_SP			BIT(1)
 #define VC5_PRIM_SRC_SHDN_EN_GBL_SHDN		BIT(0)
 
@@ -122,6 +123,8 @@
 /* flags to describe chip features */
 /* chip has built-in oscilator */
 #define VC5_HAS_INTERNAL_XTAL	BIT(0)
+/* chip has PFD requency doubler */
+#define VC5_HAS_PFD_FREQ_DBL	BIT(1)
 
 /* Supported IDT VC5 models. */
 enum vc5_model {
@@ -157,6 +160,7 @@ struct vc5_driver_data {
 	struct clk		*pin_clkin;
 	unsigned char		clk_mux_ins;
 	struct clk_hw		clk_mux;
+	struct clk_hw		clk_mul;
 	struct clk_hw		clk_pfd;
 	struct vc5_hw_data	clk_pll;
 	struct vc5_hw_data	clk_fod[VC5_MAX_FOD_NUM];
@@ -167,6 +171,10 @@ static const char * const vc5_mux_names[] = {
 	"mux"
 };
 
+static const char * const vc5_dbl_names[] = {
+	"dbl"
+};
+
 static const char * const vc5_pfd_names[] = {
 	"pfd"
 };
@@ -264,6 +272,54 @@ static const struct clk_ops vc5_mux_ops = {
 	.get_parent	= vc5_mux_get_parent,
 };
 
+static unsigned long vc5_dbl_recalc_rate(struct clk_hw *hw,
+					 unsigned long parent_rate)
+{
+	struct vc5_driver_data *vc5 =
+		container_of(hw, struct vc5_driver_data, clk_mul);
+	unsigned int premul;
+
+	regmap_read(vc5->regmap, VC5_PRIM_SRC_SHDN, &premul);
+	if (premul & VC5_PRIM_SRC_SHDN_EN_DOUBLE_XTAL_FREQ)
+		parent_rate *= 2;
+
+	return parent_rate;
+}
+
+static long vc5_dbl_round_rate(struct clk_hw *hw, unsigned long rate,
+			       unsigned long *parent_rate)
+{
+	if ((*parent_rate == rate) || ((*parent_rate * 2) == rate))
+		return rate;
+	else
+		return -EINVAL;
+}
+
+static int vc5_dbl_set_rate(struct clk_hw *hw, unsigned long rate,
+			    unsigned long parent_rate)
+{
+	struct vc5_driver_data *vc5 =
+		container_of(hw, struct vc5_driver_data, clk_mul);
+	u32 mask;
+
+	if ((parent_rate * 2) == rate)
+		mask = VC5_PRIM_SRC_SHDN_EN_DOUBLE_XTAL_FREQ;
+	else
+		mask = 0;
+
+	regmap_update_bits(vc5->regmap, VC5_PRIM_SRC_SHDN,
+			   VC5_PRIM_SRC_SHDN_EN_DOUBLE_XTAL_FREQ,
+			   mask);
+
+	return 0;
+}
+
+static const struct clk_ops vc5_dbl_ops = {
+	.recalc_rate	= vc5_dbl_recalc_rate,
+	.round_rate	= vc5_dbl_round_rate,
+	.set_rate	= vc5_dbl_set_rate,
+};
+
 static unsigned long vc5_pfd_recalc_rate(struct clk_hw *hw,
 					 unsigned long parent_rate)
 {
@@ -706,12 +762,32 @@ static int vc5_probe(struct i2c_client *client,
 		goto err_clk;
 	}
 
+	if (vc5->chip_info->flags & VC5_HAS_PFD_FREQ_DBL) {
+		/* Register frequency doubler */
+		memset(&init, 0, sizeof(init));
+		init.name = vc5_dbl_names[0];
+		init.ops = &vc5_dbl_ops;
+		init.flags = CLK_SET_RATE_PARENT;
+		init.parent_names = vc5_mux_names;
+		init.num_parents = 1;
+		vc5->clk_mul.init = &init;
+		ret = devm_clk_hw_register(&client->dev, &vc5->clk_mul);
+		if (ret) {
+			dev_err(&client->dev, "unable to register %s\n",
+				init.name);
+			goto err_clk;
+		}
+	}
+
 	/* Register PFD */
 	memset(&init, 0, sizeof(init));
 	init.name = vc5_pfd_names[0];
 	init.ops = &vc5_pfd_ops;
 	init.flags = CLK_SET_RATE_PARENT;
-	init.parent_names = vc5_mux_names;
+	if (vc5->chip_info->flags & VC5_HAS_PFD_FREQ_DBL)
+		init.parent_names = vc5_dbl_names;
+	else
+		init.parent_names = vc5_mux_names;
 	init.num_parents = 1;
 	vc5->clk_pfd.init = &init;
 	ret = devm_clk_hw_register(&client->dev, &vc5->clk_pfd);
-- 
2.11.0

Re: [PATCH V3 6/8] clk: vc5: Add support for the input frequency doubler

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> The VersaClock 6 has an input frequency doubler between the input
> clock mux and the predivider. Add new capability flag and support
> for this frequency doubler block into the driver.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH V3 7/8] clk: vc5: Add bindings for IDT VersaClock 5P49V6901

[Marek Vasut]

IDT VersaClock 6 5P49V6901 has 4 clock outputs, 4 fractional dividers.
Input clock source can be taken from either external crystal or from
external reference clock.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Cc: devicetree@vger.kernel.org
Acked-by: Rob Herring <robh@kernel.org>
---
V2: Edit the top of the DT binding document to mention VersaClock 6
V3: Reformat the compatible entry to list one per line
---
 .../devicetree/bindings/clock/idt,versaclock5.txt  | 26 ++++++++++++++++------
 1 file changed, 19 insertions(+), 7 deletions(-)

diff --git a/Documentation/devicetree/bindings/clock/idt,versaclock5.txt b/Documentation/devicetree/bindings/clock/idt,versaclock5.txt
index 53d7e50ed875..66ef0a009b95 100644
--- a/Documentation/devicetree/bindings/clock/idt,versaclock5.txt
+++ b/Documentation/devicetree/bindings/clock/idt,versaclock5.txt
@@ -1,24 +1,29 @@
-Binding for IDT VersaClock5 programmable i2c clock generator.
+Binding for IDT VersaClock 5,6 programmable i2c clock generators.
 
-The IDT VersaClock5 are programmable i2c clock generators providing
-from 3 to 12 output clocks.
+The IDT VersaClock 5 and VersaClock 6 are programmable i2c clock
+generators providing from 3 to 12 output clocks.
 
 ==I2C device node==
 
 Required properties:
-- compatible:	shall be one of "idt,5p49v5923" , "idt,5p49v5933" ,
-		"idt,5p49v5935".
+- compatible:	shall be one of
+		"idt,5p49v5923"
+		"idt,5p49v5933"
+		"idt,5p49v5935"
+		"idt,5p49v6901"
 - reg:		i2c device address, shall be 0x68 or 0x6a.
 - #clock-cells:	from common clock binding; shall be set to 1.
 - clocks:	from common clock binding; list of parent clock handles,
-		- 5p49v5923: (required) either or both of XTAL or CLKIN
+		- 5p49v5923 and
+		  5p49v6901: (required) either or both of XTAL or CLKIN
 					reference clock.
 		- 5p49v5933 and
 		- 5p49v5935: (optional) property not present (internal
 					Xtal used) or CLKIN reference
 					clock.
 - clock-names:	from common clock binding; clock input names, can be
-		- 5p49v5923: (required) either or both of "xin", "clkin".
+		- 5p49v5923 and
+		  5p49v6901: (required) either or both of "xin", "clkin".
 		- 5p49v5933 and
 		- 5p49v5935: (optional) property not present or "clkin".
 
@@ -44,6 +49,13 @@ clock specifier, the following mapping applies:
 	3 -- OUT3
 	4 -- OUT4
 
+5P49V6901:
+	0 -- OUT0_SEL_I2CB
+	1 -- OUT1
+	2 -- OUT2
+	3 -- OUT3
+	4 -- OUT4
+
 ==Example==
 
 /* 25MHz reference crystal */
-- 
2.11.0

Re: [PATCH V3 7/8] clk: vc5: Add bindings for IDT VersaClock 5P49V6901

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> IDT VersaClock 6 5P49V6901 has 4 clock outputs, 4 fractional dividers.
> Input clock source can be taken from either external crystal or from
> external reference clock.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Rob Herring <robh@kernel.org>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Cc: devicetree@vger.kernel.org
> Acked-by: Rob Herring <robh@kernel.org>
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH V3 8/8] clk: vc5: Add support for IDT VersaClock 5P49V6901

[Marek Vasut]

Update IDT VersaClock 5 driver to support IDT VersaClock 6 5P49V6901.
This chip has two clock inputs (external XTAL or external CLKIN), four
fractional dividers (FODs) and five clock outputs (four universal clock
outputs and one reference clock output at OUT0_SELB_I2C).

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Alexey Firago <alexey_firago@mentor.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
on Salvator-XS with the display LVDS output.
---
V2: - Reword the commit message to explicitly mention the VC5 driver
      and the added support for VC6.
    - Mention VC6 in Kconfig
V3: None
---
 drivers/clk/Kconfig           |  6 +++---
 drivers/clk/clk-versaclock5.c | 11 +++++++++++
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/Kconfig b/drivers/clk/Kconfig
index d406b087553f..99adcb0d799c 100644
--- a/drivers/clk/Kconfig
+++ b/drivers/clk/Kconfig
@@ -210,14 +210,14 @@ config COMMON_CLK_OXNAS
 	  Support for the OXNAS SoC Family clocks.
 
 config COMMON_CLK_VC5
-	tristate "Clock driver for IDT VersaClock5 devices"
+	tristate "Clock driver for IDT VersaClock 5,6 devices"
 	depends on I2C
 	depends on OF
 	select REGMAP_I2C
 	help
 	---help---
-	  This driver supports the IDT VersaClock5 programmable clock
-	  generator.
+	  This driver supports the IDT VersaClock 5 and VersaClock 6
+	  programmable clock generators.
 
 source "drivers/clk/bcm/Kconfig"
 source "drivers/clk/hisilicon/Kconfig"
diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index d0406b62133a..c894db298cbd 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -131,6 +131,7 @@ enum vc5_model {
 	IDT_VC5_5P49V5923,
 	IDT_VC5_5P49V5933,
 	IDT_VC5_5P49V5935,
+	IDT_VC6_5P49V6901,
 };
 
 /* Structure to describe features of a particular VC5 model */
@@ -686,6 +687,7 @@ static int vc5_map_index_to_output(const enum vc5_model model,
 		return (n == 0) ? 0 : 3;
 	case IDT_VC5_5P49V5923:
 	case IDT_VC5_5P49V5935:
+	case IDT_VC6_5P49V6901:
 	default:
 		return n;
 	}
@@ -923,10 +925,18 @@ static const struct vc5_chip_info idt_5p49v5935_info = {
 	.flags = VC5_HAS_INTERNAL_XTAL,
 };
 
+static const struct vc5_chip_info idt_5p49v6901_info = {
+	.model = IDT_VC6_5P49V6901,
+	.clk_fod_cnt = 4,
+	.clk_out_cnt = 5,
+	.flags = VC5_HAS_PFD_FREQ_DBL,
+};
+
 static const struct i2c_device_id vc5_id[] = {
 	{ "5p49v5923", .driver_data = IDT_VC5_5P49V5923 },
 	{ "5p49v5933", .driver_data = IDT_VC5_5P49V5933 },
 	{ "5p49v5935", .driver_data = IDT_VC5_5P49V5935 },
+	{ "5p49v6901", .driver_data = IDT_VC6_5P49V6901 },
 	{ }
 };
 MODULE_DEVICE_TABLE(i2c, vc5_id);
@@ -935,6 +945,7 @@ static const struct of_device_id clk_vc5_of_match[] = {
 	{ .compatible = "idt,5p49v5923", .data = &idt_5p49v5923_info },
 	{ .compatible = "idt,5p49v5933", .data = &idt_5p49v5933_info },
 	{ .compatible = "idt,5p49v5935", .data = &idt_5p49v5935_info },
+	{ .compatible = "idt,5p49v6901", .data = &idt_5p49v6901_info },
 	{ },
 };
 MODULE_DEVICE_TABLE(of, clk_vc5_of_match);
-- 
2.11.0

Re: [PATCH V3 8/8] clk: vc5: Add support for IDT VersaClock 5P49V6901

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> Update IDT VersaClock 5 driver to support IDT VersaClock 6 5P49V6901.
> This chip has two clock inputs (external XTAL or external CLKIN), four
> fractional dividers (FODs) and five clock outputs (four universal clock
> outputs and one reference clock output at OUT0_SELB_I2C).
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH V3 1/8] clk: vc5: Prevent division by zero on unconfigured outputs

[Stephen Boyd]

On 07/09, Marek Vasut wrote:
> In case the initial values of the FOD registers are not configured in
> the OTP or by the bootloader, it is possible that the FOD registers
> will contain zeroes. The code in vc5_fod_recalc_rate() immediately
> feeds the FOD divider value obtained from the FOD registers into the
> div64_u64() and if the FOD divider value is zero, triggers division
> by zero exception.
> 
> Check if the FOD divider value is zero and return the frequency of
> the FOD output as 0 Hz if it is so. This prevents the division by
> zero exception.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Stephen Boyd <sboyd@codeaurora.org>
> Cc: Alexey Firago <alexey_firago@mentor.com>
> Cc: Michael Turquette <mturquette@baylibre.com>
> Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> on Salvator-XS with the display LVDS output.
> ---

Applied to clk-next

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project