[PATCH v4 0/3] clk: qcom: clk-rcg2: introduce support for multiple conf for same freq

[PATCH v4 0/3] clk: qcom: clk-rcg2: introduce support for multiple conf for same freq

[Christian Marangi]

This small series fix a current problem with ipq8074 where the 2 uniphy
port doesn't work in some corner case with some clk configuration. The
port to correctly work require a specific frequency, using the wrong one
results in the port not transmitting data.

With the current code with a requested freq of 125MHz, the frequency is
set to 105MHz. This is caused by the fact that there are 2 different
configuration to set 125MHz and it's always selected the first one that
results in 105MHz.

In the original QSDK code, the frequency configuration selection is
different and the CEIL FLOOR logic is not present. Instead it's used a
BEST approach where the frequency table is checked and then it's checked
if there are duplicate entry.

This proposed implementation is more specific and introduce an entire new
set of ops and a specific freq table to support this special configuration.

A union is introduced in rcg2 struct to not duplicate the struct.
A new set of ops clk_rcg2_fm_ops are introduced to support this new kind
of frequency table.

Changes v4:
- Drop suggested but wrong re-search patch
- Move everything to separate ops and struct to not affect current rcg2
  users.
Changes v3:
- Add qcom_find_freq_exact
- Drop re-search on rcg2_set_rate
- Rework multiple conf patch to follow new implementation
Changes v2:
- Out of RFC
- Fix compile warning from buildbot related to F redefinition

Christian Marangi (3):
  clk: qcom: clk-rcg: introduce support for multiple conf for same freq
  clk: qcom: clk-rcg2: add support for rcg2 freq multi ops
  clk: qcom: gcc-ipq8074: rework nss_port5/6 clock to multiple conf

 drivers/clk/qcom/clk-rcg.h     |  24 +++++-
 drivers/clk/qcom/clk-rcg2.c    | 152 +++++++++++++++++++++++++++++++++
 drivers/clk/qcom/common.c      |  18 ++++
 drivers/clk/qcom/common.h      |   2 +
 drivers/clk/qcom/gcc-ipq8074.c | 120 ++++++++++++++++----------
 5 files changed, 271 insertions(+), 45 deletions(-)

-- 
2.39.2

[PATCH v4 1/3] clk: qcom: clk-rcg: introduce support for multiple conf for same freq

[Christian Marangi]

Some RCG frequency can be reached by multiple configuration.

We currently declare multiple configuration for the same frequency but
that is not supported and always the first configuration will be taken.

These multiple configuration are needed as based on the current parent
configuration, it may be needed to use a different configuration to
reach the same frequency.

To handle this introduce 3 new macro, C, FM and FMS:

- C is used to declare a freq_conf where src, pre_div, m and n are
  provided.

- FM is used to declare a freq_multi_tbl with the frequency and an
  array of confs to insert all the config for the provided frequency.

- FMS is used to declare a freq_multi_tbl with the frequency and an
  array of a single conf with the provided src, pre_div, m and n.

Struct clk_rcg2 is changed to add a union type to reference a simple
freq_tbl or a complex freq_multi_tbl.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
---
 drivers/clk/qcom/clk-rcg.h | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)

diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
index 01581f4d2c39..dc85b46b0d79 100644
--- a/drivers/clk/qcom/clk-rcg.h
+++ b/drivers/clk/qcom/clk-rcg.h
@@ -17,6 +17,23 @@ struct freq_tbl {
 	u16 n;
 };
 
+#define C(s, h, m, n) { (s), (2 * (h) - 1), (m), (n) }
+#define FM(f, confs) { (f), ARRAY_SIZE(confs), (confs) }
+#define FMS(f, s, h, m, n) { (f), 1, (const struct freq_conf []){ C(s, h, m, n) } }
+
+struct freq_conf {
+	u8 src;
+	u8 pre_div;
+	u16 m;
+	u16 n;
+};
+
+struct freq_multi_tbl {
+	unsigned long freq;
+	int num_confs;
+	const struct freq_conf *confs;
+};
+
 /**
  * struct mn - M/N:D counter
  * @mnctr_en_bit: bit to enable mn counter
@@ -138,6 +155,7 @@ extern const struct clk_ops clk_dyn_rcg_ops;
  * @safe_src_index: safe src index value
  * @parent_map: map from software's parent index to hardware's src_sel field
  * @freq_tbl: frequency table
+ * @freq_multi_tbl: frequency table for clocks reachable with multiple RCGs conf
  * @clkr: regmap clock handle
  * @cfg_off: defines the cfg register offset from the CMD_RCGR + CFG_REG
  * @parked_cfg: cached value of the CFG register for parked RCGs
@@ -148,7 +166,10 @@ struct clk_rcg2 {
 	u8			hid_width;
 	u8			safe_src_index;
 	const struct parent_map	*parent_map;
-	const struct freq_tbl	*freq_tbl;
+	union {
+		const struct freq_tbl		*freq_tbl;
+		const struct freq_multi_tbl	*freq_multi_tbl;
+	};
 	struct clk_regmap	clkr;
 	u8			cfg_off;
 	u32			parked_cfg;
-- 
2.39.2

[PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Christian Marangi]

Some RCG frequency can be reached by multiple configuration.

Add clk_rcg2_fm_ops ops to support these special RCG configurations.

These alternative ops will select the frequency using a CEIL policy.

When the correct frequency is found, the correct config is selected by
calculating the final rate (by checking the defined parent and values
in the config that is being checked) and deciding based on the one that
is less different than the requested one.

These check are skipped if there is just on config for the requested
freq.

qcom_find_freq_multi is added to search the freq with the new struct
freq_multi_tbl.
__clk_rcg2_select_conf is used to select the correct conf by simulating
the final clock.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
---
 drivers/clk/qcom/clk-rcg.h  |   1 +
 drivers/clk/qcom/clk-rcg2.c | 152 ++++++++++++++++++++++++++++++++++++
 drivers/clk/qcom/common.c   |  18 +++++
 drivers/clk/qcom/common.h   |   2 +
 4 files changed, 173 insertions(+)

diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
index dc85b46b0d79..f8ec989ed3d9 100644
--- a/drivers/clk/qcom/clk-rcg.h
+++ b/drivers/clk/qcom/clk-rcg.h
@@ -188,6 +188,7 @@ struct clk_rcg2_gfx3d {
 
 extern const struct clk_ops clk_rcg2_ops;
 extern const struct clk_ops clk_rcg2_floor_ops;
+extern const struct clk_ops clk_rcg2_fm_ops;
 extern const struct clk_ops clk_rcg2_mux_closest_ops;
 extern const struct clk_ops clk_edp_pixel_ops;
 extern const struct clk_ops clk_byte_ops;
diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
index 76551534f10d..4f2fe012ef5f 100644
--- a/drivers/clk/qcom/clk-rcg2.c
+++ b/drivers/clk/qcom/clk-rcg2.c
@@ -266,6 +266,104 @@ static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f,
 	return 0;
 }
 
+static const struct freq_conf *
+__clk_rcg2_select_conf(struct clk_hw *hw, const struct freq_multi_tbl *f,
+		       unsigned long req_rate)
+{
+	unsigned long best_rate = 0, parent_rate, rate;
+	const struct freq_conf *conf, *best_conf;
+	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
+	struct clk_hw *p;
+	int index, i;
+
+	/* Exit early if only one config is defined */
+	if (f->num_confs == 1)
+		return f->confs;
+
+	/* Search in each provided config the one that is near the wanted rate */
+	for (i = 0, conf = f->confs; i < f->num_confs; i++, conf++) {
+		index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
+		if (index < 0)
+			continue;
+
+		p = clk_hw_get_parent_by_index(hw, index);
+		if (!p)
+			continue;
+
+		parent_rate =  clk_hw_get_rate(p);
+		rate = calc_rate(parent_rate, conf->n, conf->m, conf->n, conf->pre_div);
+
+		if (rate == req_rate) {
+			best_conf = conf;
+			break;
+		}
+
+		if (abs(req_rate - rate) < abs(best_rate - rate)) {
+			best_rate = rate;
+			best_conf = conf;
+		}
+	}
+
+	/*
+	 * Very unlikely.
+	 * Force the first conf if we can't find a correct config.
+	 */
+	if (unlikely(i == f->num_confs))
+		best_conf = f->confs;
+
+	return best_conf;
+}
+
+static int _freq_tbl_fm_determine_rate(struct clk_hw *hw, const struct freq_multi_tbl *f,
+				       struct clk_rate_request *req)
+{
+	unsigned long clk_flags, rate = req->rate;
+	const struct freq_conf *conf;
+	struct clk_hw *p;
+	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
+	int index;
+
+	f = qcom_find_freq_multi(f, rate);
+	if (!f || !f->confs)
+		return -EINVAL;
+
+	conf = __clk_rcg2_select_conf(hw, f, rate);
+	index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
+	if (index < 0)
+		return index;
+
+	clk_flags = clk_hw_get_flags(hw);
+	p = clk_hw_get_parent_by_index(hw, index);
+	if (!p)
+		return -EINVAL;
+
+	if (clk_flags & CLK_SET_RATE_PARENT) {
+		rate = f->freq;
+		if (conf->pre_div) {
+			if (!rate)
+				rate = req->rate;
+			rate /= 2;
+			rate *= conf->pre_div + 1;
+		}
+
+		if (conf->n) {
+			u64 tmp = rate;
+
+			tmp = tmp * conf->n;
+			do_div(tmp, conf->m);
+			rate = tmp;
+		}
+	} else {
+		rate =  clk_hw_get_rate(p);
+	}
+
+	req->best_parent_hw = p;
+	req->best_parent_rate = rate;
+	req->rate = f->freq;
+
+	return 0;
+}
+
 static int clk_rcg2_determine_rate(struct clk_hw *hw,
 				   struct clk_rate_request *req)
 {
@@ -282,6 +380,14 @@ static int clk_rcg2_determine_floor_rate(struct clk_hw *hw,
 	return _freq_tbl_determine_rate(hw, rcg->freq_tbl, req, FLOOR);
 }
 
+static int clk_rcg2_fm_determine_rate(struct clk_hw *hw,
+				      struct clk_rate_request *req)
+{
+	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
+
+	return _freq_tbl_fm_determine_rate(hw, rcg->freq_multi_tbl, req);
+}
+
 static int __clk_rcg2_configure(struct clk_rcg2 *rcg, const struct freq_tbl *f,
 				u32 *_cfg)
 {
@@ -375,6 +481,27 @@ static int __clk_rcg2_set_rate(struct clk_hw *hw, unsigned long rate,
 	return clk_rcg2_configure(rcg, f);
 }
 
+static int __clk_rcg2_fm_set_rate(struct clk_hw *hw, unsigned long rate)
+{
+	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
+	const struct freq_multi_tbl *f;
+	const struct freq_conf *conf;
+	struct freq_tbl f_tbl;
+
+	f = qcom_find_freq_multi(rcg->freq_multi_tbl, rate);
+	if (!f || !f->confs)
+		return -EINVAL;
+
+	conf = __clk_rcg2_select_conf(hw, f, rate);
+	f_tbl.freq = f->freq;
+	f_tbl.src = conf->src;
+	f_tbl.pre_div = conf->pre_div;
+	f_tbl.m = conf->m;
+	f_tbl.n = conf->n;
+
+	return clk_rcg2_configure(rcg, &f_tbl);
+}
+
 static int clk_rcg2_set_rate(struct clk_hw *hw, unsigned long rate,
 			    unsigned long parent_rate)
 {
@@ -387,6 +514,12 @@ static int clk_rcg2_set_floor_rate(struct clk_hw *hw, unsigned long rate,
 	return __clk_rcg2_set_rate(hw, rate, FLOOR);
 }
 
+static int clk_rcg2_fm_set_rate(struct clk_hw *hw, unsigned long rate,
+				unsigned long parent_rate)
+{
+	return __clk_rcg2_fm_set_rate(hw, rate);
+}
+
 static int clk_rcg2_set_rate_and_parent(struct clk_hw *hw,
 		unsigned long rate, unsigned long parent_rate, u8 index)
 {
@@ -399,6 +532,12 @@ static int clk_rcg2_set_floor_rate_and_parent(struct clk_hw *hw,
 	return __clk_rcg2_set_rate(hw, rate, FLOOR);
 }
 
+static int clk_rcg2_fm_set_rate_and_parent(struct clk_hw *hw,
+		unsigned long rate, unsigned long parent_rate, u8 index)
+{
+	return __clk_rcg2_fm_set_rate(hw, rate);
+}
+
 static int clk_rcg2_get_duty_cycle(struct clk_hw *hw, struct clk_duty *duty)
 {
 	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
@@ -509,6 +648,19 @@ const struct clk_ops clk_rcg2_floor_ops = {
 };
 EXPORT_SYMBOL_GPL(clk_rcg2_floor_ops);
 
+const struct clk_ops clk_rcg2_fm_ops = {
+	.is_enabled = clk_rcg2_is_enabled,
+	.get_parent = clk_rcg2_get_parent,
+	.set_parent = clk_rcg2_set_parent,
+	.recalc_rate = clk_rcg2_recalc_rate,
+	.determine_rate = clk_rcg2_fm_determine_rate,
+	.set_rate = clk_rcg2_fm_set_rate,
+	.set_rate_and_parent = clk_rcg2_fm_set_rate_and_parent,
+	.get_duty_cycle = clk_rcg2_get_duty_cycle,
+	.set_duty_cycle = clk_rcg2_set_duty_cycle,
+};
+EXPORT_SYMBOL_GPL(clk_rcg2_fm_ops);
+
 const struct clk_ops clk_rcg2_mux_closest_ops = {
 	.determine_rate = __clk_mux_determine_rate_closest,
 	.get_parent = clk_rcg2_get_parent,
diff --git a/drivers/clk/qcom/common.c b/drivers/clk/qcom/common.c
index 75f09e6e057e..48f81e3a5e80 100644
--- a/drivers/clk/qcom/common.c
+++ b/drivers/clk/qcom/common.c
@@ -41,6 +41,24 @@ struct freq_tbl *qcom_find_freq(const struct freq_tbl *f, unsigned long rate)
 }
 EXPORT_SYMBOL_GPL(qcom_find_freq);
 
+const struct freq_multi_tbl *qcom_find_freq_multi(const struct freq_multi_tbl *f,
+						  unsigned long rate)
+{
+	if (!f)
+		return NULL;
+
+	if (!f->freq)
+		return f;
+
+	for (; f->freq; f++)
+		if (rate <= f->freq)
+			return f;
+
+	/* Default to our fastest rate */
+	return f - 1;
+}
+EXPORT_SYMBOL_GPL(qcom_find_freq_multi);
+
 const struct freq_tbl *qcom_find_freq_floor(const struct freq_tbl *f,
 					    unsigned long rate)
 {
diff --git a/drivers/clk/qcom/common.h b/drivers/clk/qcom/common.h
index 9c8f7b798d9f..2d4a8a837e6c 100644
--- a/drivers/clk/qcom/common.h
+++ b/drivers/clk/qcom/common.h
@@ -45,6 +45,8 @@ extern const struct freq_tbl *qcom_find_freq(const struct freq_tbl *f,
 					     unsigned long rate);
 extern const struct freq_tbl *qcom_find_freq_floor(const struct freq_tbl *f,
 						   unsigned long rate);
+extern const struct freq_multi_tbl *qcom_find_freq_multi(const struct freq_multi_tbl *f,
+							 unsigned long rate);
 extern void
 qcom_pll_set_fsm_mode(struct regmap *m, u32 reg, u8 bias_count, u8 lock_count);
 extern int qcom_find_src_index(struct clk_hw *hw, const struct parent_map *map,
-- 
2.39.2

[PATCH v4 3/3] clk: qcom: gcc-ipq8074: rework nss_port5/6 clock to multiple conf

[Christian Marangi]

Rework nss_port5/6 to use the new multiple configuration implementation
and correctly fix the clocks for these port under some corner case.

This is particularly relevant for device that have 2.5G or 10G port
connected to port5 or port 6 on ipq8074. As the parent are shared
across multiple port it may be required to select the correct
configuration to accomplish the desired clock. Without this patch such
port doesn't work in some specific ethernet speed as the clock will be
set to the wrong frequency as we just select the first configuration for
the related frequency instead of selecting the best one.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
---
 drivers/clk/qcom/gcc-ipq8074.c | 120 +++++++++++++++++++++------------
 1 file changed, 76 insertions(+), 44 deletions(-)

diff --git a/drivers/clk/qcom/gcc-ipq8074.c b/drivers/clk/qcom/gcc-ipq8074.c
index 6541d98c0348..bce459cecb2d 100644
--- a/drivers/clk/qcom/gcc-ipq8074.c
+++ b/drivers/clk/qcom/gcc-ipq8074.c
@@ -1682,15 +1682,23 @@ static struct clk_regmap_div nss_port4_tx_div_clk_src = {
 	},
 };
 
-static const struct freq_tbl ftbl_nss_port5_rx_clk_src[] = {
-	F(19200000, P_XO, 1, 0, 0),
-	F(25000000, P_UNIPHY1_RX, 12.5, 0, 0),
-	F(25000000, P_UNIPHY0_RX, 5, 0, 0),
-	F(78125000, P_UNIPHY1_RX, 4, 0, 0),
-	F(125000000, P_UNIPHY1_RX, 2.5, 0, 0),
-	F(125000000, P_UNIPHY0_RX, 1, 0, 0),
-	F(156250000, P_UNIPHY1_RX, 2, 0, 0),
-	F(312500000, P_UNIPHY1_RX, 1, 0, 0),
+static const struct freq_conf ftbl_nss_port5_rx_clk_src_25[] = {
+	C(P_UNIPHY1_RX, 12.5, 0, 0),
+	C(P_UNIPHY0_RX, 5, 0, 0),
+};
+
+static const struct freq_conf ftbl_nss_port5_rx_clk_src_125[] = {
+	C(P_UNIPHY1_RX, 2.5, 0, 0),
+	C(P_UNIPHY0_RX, 1, 0, 0),
+};
+
+static const struct freq_multi_tbl ftbl_nss_port5_rx_clk_src[] = {
+	FMS(19200000, P_XO, 1, 0, 0),
+	FM(25000000, ftbl_nss_port5_rx_clk_src_25),
+	FMS(78125000, P_UNIPHY1_RX, 4, 0, 0),
+	FM(125000000, ftbl_nss_port5_rx_clk_src_125),
+	FMS(156250000, P_UNIPHY1_RX, 2, 0, 0),
+	FMS(312500000, P_UNIPHY1_RX, 1, 0, 0),
 	{ }
 };
 
@@ -1717,14 +1725,14 @@ gcc_xo_uniphy0_rx_tx_uniphy1_rx_tx_ubi32_bias_map[] = {
 
 static struct clk_rcg2 nss_port5_rx_clk_src = {
 	.cmd_rcgr = 0x68060,
-	.freq_tbl = ftbl_nss_port5_rx_clk_src,
+	.freq_multi_tbl = ftbl_nss_port5_rx_clk_src,
 	.hid_width = 5,
 	.parent_map = gcc_xo_uniphy0_rx_tx_uniphy1_rx_tx_ubi32_bias_map,
 	.clkr.hw.init = &(struct clk_init_data){
 		.name = "nss_port5_rx_clk_src",
 		.parent_data = gcc_xo_uniphy0_rx_tx_uniphy1_rx_tx_ubi32_bias,
 		.num_parents = ARRAY_SIZE(gcc_xo_uniphy0_rx_tx_uniphy1_rx_tx_ubi32_bias),
-		.ops = &clk_rcg2_ops,
+		.ops = &clk_rcg2_fm_ops,
 	},
 };
 
@@ -1744,15 +1752,23 @@ static struct clk_regmap_div nss_port5_rx_div_clk_src = {
 	},
 };
 
-static const struct freq_tbl ftbl_nss_port5_tx_clk_src[] = {
-	F(19200000, P_XO, 1, 0, 0),
-	F(25000000, P_UNIPHY1_TX, 12.5, 0, 0),
-	F(25000000, P_UNIPHY0_TX, 5, 0, 0),
-	F(78125000, P_UNIPHY1_TX, 4, 0, 0),
-	F(125000000, P_UNIPHY1_TX, 2.5, 0, 0),
-	F(125000000, P_UNIPHY0_TX, 1, 0, 0),
-	F(156250000, P_UNIPHY1_TX, 2, 0, 0),
-	F(312500000, P_UNIPHY1_TX, 1, 0, 0),
+static const struct freq_conf ftbl_nss_port5_tx_clk_src_25[] = {
+	C(P_UNIPHY1_TX, 12.5, 0, 0),
+	C(P_UNIPHY0_TX, 5, 0, 0),
+};
+
+static const struct freq_conf ftbl_nss_port5_tx_clk_src_125[] = {
+	C(P_UNIPHY1_TX, 2.5, 0, 0),
+	C(P_UNIPHY0_TX, 1, 0, 0),
+};
+
+static const struct freq_multi_tbl ftbl_nss_port5_tx_clk_src[] = {
+	FMS(19200000, P_XO, 1, 0, 0),
+	FM(25000000, ftbl_nss_port5_tx_clk_src_25),
+	FMS(78125000, P_UNIPHY1_TX, 4, 0, 0),
+	FM(125000000, ftbl_nss_port5_tx_clk_src_125),
+	FMS(156250000, P_UNIPHY1_TX, 2, 0, 0),
+	FMS(312500000, P_UNIPHY1_TX, 1, 0, 0),
 	{ }
 };
 
@@ -1779,14 +1795,14 @@ gcc_xo_uniphy0_tx_rx_uniphy1_tx_rx_ubi32_bias_map[] = {
 
 static struct clk_rcg2 nss_port5_tx_clk_src = {
 	.cmd_rcgr = 0x68068,
-	.freq_tbl = ftbl_nss_port5_tx_clk_src,
+	.freq_multi_tbl = ftbl_nss_port5_tx_clk_src,
 	.hid_width = 5,
 	.parent_map = gcc_xo_uniphy0_tx_rx_uniphy1_tx_rx_ubi32_bias_map,
 	.clkr.hw.init = &(struct clk_init_data){
 		.name = "nss_port5_tx_clk_src",
 		.parent_data = gcc_xo_uniphy0_tx_rx_uniphy1_tx_rx_ubi32_bias,
 		.num_parents = ARRAY_SIZE(gcc_xo_uniphy0_tx_rx_uniphy1_tx_rx_ubi32_bias),
-		.ops = &clk_rcg2_ops,
+		.ops = &clk_rcg2_fm_ops,
 	},
 };
 
@@ -1806,15 +1822,23 @@ static struct clk_regmap_div nss_port5_tx_div_clk_src = {
 	},
 };
 
-static const struct freq_tbl ftbl_nss_port6_rx_clk_src[] = {
-	F(19200000, P_XO, 1, 0, 0),
-	F(25000000, P_UNIPHY2_RX, 5, 0, 0),
-	F(25000000, P_UNIPHY2_RX, 12.5, 0, 0),
-	F(78125000, P_UNIPHY2_RX, 4, 0, 0),
-	F(125000000, P_UNIPHY2_RX, 1, 0, 0),
-	F(125000000, P_UNIPHY2_RX, 2.5, 0, 0),
-	F(156250000, P_UNIPHY2_RX, 2, 0, 0),
-	F(312500000, P_UNIPHY2_RX, 1, 0, 0),
+static const struct freq_conf ftbl_nss_port6_rx_clk_src_25[] = {
+	C(P_UNIPHY2_RX, 5, 0, 0),
+	C(P_UNIPHY2_RX, 12.5, 0, 0),
+};
+
+static const struct freq_conf ftbl_nss_port6_rx_clk_src_125[] = {
+	C(P_UNIPHY2_RX, 1, 0, 0),
+	C(P_UNIPHY2_RX, 2.5, 0, 0),
+};
+
+static const struct freq_multi_tbl ftbl_nss_port6_rx_clk_src[] = {
+	FMS(19200000, P_XO, 1, 0, 0),
+	FM(25000000, ftbl_nss_port6_rx_clk_src_25),
+	FMS(78125000, P_UNIPHY2_RX, 4, 0, 0),
+	FM(125000000, ftbl_nss_port6_rx_clk_src_125),
+	FMS(156250000, P_UNIPHY2_RX, 2, 0, 0),
+	FMS(312500000, P_UNIPHY2_RX, 1, 0, 0),
 	{ }
 };
 
@@ -1836,14 +1860,14 @@ static const struct parent_map gcc_xo_uniphy2_rx_tx_ubi32_bias_map[] = {
 
 static struct clk_rcg2 nss_port6_rx_clk_src = {
 	.cmd_rcgr = 0x68070,
-	.freq_tbl = ftbl_nss_port6_rx_clk_src,
+	.freq_multi_tbl = ftbl_nss_port6_rx_clk_src,
 	.hid_width = 5,
 	.parent_map = gcc_xo_uniphy2_rx_tx_ubi32_bias_map,
 	.clkr.hw.init = &(struct clk_init_data){
 		.name = "nss_port6_rx_clk_src",
 		.parent_data = gcc_xo_uniphy2_rx_tx_ubi32_bias,
 		.num_parents = ARRAY_SIZE(gcc_xo_uniphy2_rx_tx_ubi32_bias),
-		.ops = &clk_rcg2_ops,
+		.ops = &clk_rcg2_fm_ops,
 	},
 };
 
@@ -1863,15 +1887,23 @@ static struct clk_regmap_div nss_port6_rx_div_clk_src = {
 	},
 };
 
-static const struct freq_tbl ftbl_nss_port6_tx_clk_src[] = {
-	F(19200000, P_XO, 1, 0, 0),
-	F(25000000, P_UNIPHY2_TX, 5, 0, 0),
-	F(25000000, P_UNIPHY2_TX, 12.5, 0, 0),
-	F(78125000, P_UNIPHY2_TX, 4, 0, 0),
-	F(125000000, P_UNIPHY2_TX, 1, 0, 0),
-	F(125000000, P_UNIPHY2_TX, 2.5, 0, 0),
-	F(156250000, P_UNIPHY2_TX, 2, 0, 0),
-	F(312500000, P_UNIPHY2_TX, 1, 0, 0),
+static const struct freq_conf ftbl_nss_port6_tx_clk_src_25[] = {
+	C(P_UNIPHY2_TX, 5, 0, 0),
+	C(P_UNIPHY2_TX, 12.5, 0, 0),
+};
+
+static const struct freq_conf ftbl_nss_port6_tx_clk_src_125[] = {
+	C(P_UNIPHY2_TX, 1, 0, 0),
+	C(P_UNIPHY2_TX, 2.5, 0, 0),
+};
+
+static const struct freq_multi_tbl ftbl_nss_port6_tx_clk_src[] = {
+	FMS(19200000, P_XO, 1, 0, 0),
+	FM(25000000, ftbl_nss_port6_tx_clk_src_25),
+	FMS(78125000, P_UNIPHY1_RX, 4, 0, 0),
+	FM(125000000, ftbl_nss_port6_tx_clk_src_125),
+	FMS(156250000, P_UNIPHY1_RX, 2, 0, 0),
+	FMS(312500000, P_UNIPHY1_RX, 1, 0, 0),
 	{ }
 };
 
@@ -1893,14 +1925,14 @@ static const struct parent_map gcc_xo_uniphy2_tx_rx_ubi32_bias_map[] = {
 
 static struct clk_rcg2 nss_port6_tx_clk_src = {
 	.cmd_rcgr = 0x68078,
-	.freq_tbl = ftbl_nss_port6_tx_clk_src,
+	.freq_multi_tbl = ftbl_nss_port6_tx_clk_src,
 	.hid_width = 5,
 	.parent_map = gcc_xo_uniphy2_tx_rx_ubi32_bias_map,
 	.clkr.hw.init = &(struct clk_init_data){
 		.name = "nss_port6_tx_clk_src",
 		.parent_data = gcc_xo_uniphy2_tx_rx_ubi32_bias,
 		.num_parents = ARRAY_SIZE(gcc_xo_uniphy2_tx_rx_ubi32_bias),
-		.ops = &clk_rcg2_ops,
+		.ops = &clk_rcg2_fm_ops,
 	},
 };
 
-- 
2.39.2

Re: [PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Konrad Dybcio]

On 27.04.2023 17:07, Christian Marangi wrote:
> Some RCG frequency can be reached by multiple configuration.
> 
> Add clk_rcg2_fm_ops ops to support these special RCG configurations.
> 
> These alternative ops will select the frequency using a CEIL policy.
> 
> When the correct frequency is found, the correct config is selected by
> calculating the final rate (by checking the defined parent and values
> in the config that is being checked) and deciding based on the one that
> is less different than the requested one.
> 
> These check are skipped if there is just on config for the requested
> freq.
> 
> qcom_find_freq_multi is added to search the freq with the new struct
> freq_multi_tbl.
> __clk_rcg2_select_conf is used to select the correct conf by simulating
> the final clock.
> 
> Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
> ---
>  drivers/clk/qcom/clk-rcg.h  |   1 +
>  drivers/clk/qcom/clk-rcg2.c | 152 ++++++++++++++++++++++++++++++++++++
>  drivers/clk/qcom/common.c   |  18 +++++
>  drivers/clk/qcom/common.h   |   2 +
>  4 files changed, 173 insertions(+)
> 
> diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
> index dc85b46b0d79..f8ec989ed3d9 100644
> --- a/drivers/clk/qcom/clk-rcg.h
> +++ b/drivers/clk/qcom/clk-rcg.h
> @@ -188,6 +188,7 @@ struct clk_rcg2_gfx3d {
>  
>  extern const struct clk_ops clk_rcg2_ops;
>  extern const struct clk_ops clk_rcg2_floor_ops;
> +extern const struct clk_ops clk_rcg2_fm_ops;
>  extern const struct clk_ops clk_rcg2_mux_closest_ops;
>  extern const struct clk_ops clk_edp_pixel_ops;
>  extern const struct clk_ops clk_byte_ops;
> diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
> index 76551534f10d..4f2fe012ef5f 100644
> --- a/drivers/clk/qcom/clk-rcg2.c
> +++ b/drivers/clk/qcom/clk-rcg2.c
> @@ -266,6 +266,104 @@ static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f,
>  	return 0;
>  }
>  
> +static const struct freq_conf *
> +__clk_rcg2_select_conf(struct clk_hw *hw, const struct freq_multi_tbl *f,
> +		       unsigned long req_rate)
> +{
> +	unsigned long best_rate = 0, parent_rate, rate;
> +	const struct freq_conf *conf, *best_conf;
> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
> +	struct clk_hw *p;
> +	int index, i;
> +
> +	/* Exit early if only one config is defined */
> +	if (f->num_confs == 1)
> +		return f->confs;
> +
> +	/* Search in each provided config the one that is near the wanted rate */
> +	for (i = 0, conf = f->confs; i < f->num_confs; i++, conf++) {
> +		index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
> +		if (index < 0)
> +			continue;
> +
> +		p = clk_hw_get_parent_by_index(hw, index);
> +		if (!p)
> +			continue;
> +
> +		parent_rate =  clk_hw_get_rate(p);
> +		rate = calc_rate(parent_rate, conf->n, conf->m, conf->n, conf->pre_div);
> +
> +		if (rate == req_rate) {
> +			best_conf = conf;
> +			break;
> +		}
> +
> +		if (abs(req_rate - rate) < abs(best_rate - rate)) {
Shouldn't this be:

if (abs(req_rate - rate) < abs(best_rate - req_rate)

?

this way it'd say

"if this iteration's rate is closer to the requested one than the
best one we've found yet, it's better"

> +			best_rate = rate;
> +			best_conf = conf;
> +		}
> +	}
> +
> +	/*
> +	 * Very unlikely.
> +	 * Force the first conf if we can't find a correct config.
> +	 */
> +	if (unlikely(i == f->num_confs))
> +		best_conf = f->confs;
Is that a supported scenario or would it be a device driver / clock
driver error?

> +
> +	return best_conf;
> +}
> +
> +static int _freq_tbl_fm_determine_rate(struct clk_hw *hw, const struct freq_multi_tbl *f,
> +				       struct clk_rate_request *req)
> +{
> +	unsigned long clk_flags, rate = req->rate;
> +	const struct freq_conf *conf;
> +	struct clk_hw *p;
> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
swap lines 2, 3, 4 to 4, 2, 3 and you'll get a revers-Christmas-tree!

Konrad

Re: [PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Christian Marangi]

On Sat, May 27, 2023 at 06:11:16PM +0200, Konrad Dybcio wrote:
> 
> 
> On 27.04.2023 17:07, Christian Marangi wrote:
> > Some RCG frequency can be reached by multiple configuration.
> > 
> > Add clk_rcg2_fm_ops ops to support these special RCG configurations.
> > 
> > These alternative ops will select the frequency using a CEIL policy.
> > 
> > When the correct frequency is found, the correct config is selected by
> > calculating the final rate (by checking the defined parent and values
> > in the config that is being checked) and deciding based on the one that
> > is less different than the requested one.
> > 
> > These check are skipped if there is just on config for the requested
> > freq.
> > 
> > qcom_find_freq_multi is added to search the freq with the new struct
> > freq_multi_tbl.
> > __clk_rcg2_select_conf is used to select the correct conf by simulating
> > the final clock.
> > 
> > Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
> > ---
> >  drivers/clk/qcom/clk-rcg.h  |   1 +
> >  drivers/clk/qcom/clk-rcg2.c | 152 ++++++++++++++++++++++++++++++++++++
> >  drivers/clk/qcom/common.c   |  18 +++++
> >  drivers/clk/qcom/common.h   |   2 +
> >  4 files changed, 173 insertions(+)
> > 
> > diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
> > index dc85b46b0d79..f8ec989ed3d9 100644
> > --- a/drivers/clk/qcom/clk-rcg.h
> > +++ b/drivers/clk/qcom/clk-rcg.h
> > @@ -188,6 +188,7 @@ struct clk_rcg2_gfx3d {
> >  
> >  extern const struct clk_ops clk_rcg2_ops;
> >  extern const struct clk_ops clk_rcg2_floor_ops;
> > +extern const struct clk_ops clk_rcg2_fm_ops;
> >  extern const struct clk_ops clk_rcg2_mux_closest_ops;
> >  extern const struct clk_ops clk_edp_pixel_ops;
> >  extern const struct clk_ops clk_byte_ops;
> > diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
> > index 76551534f10d..4f2fe012ef5f 100644
> > --- a/drivers/clk/qcom/clk-rcg2.c
> > +++ b/drivers/clk/qcom/clk-rcg2.c
> > @@ -266,6 +266,104 @@ static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f,
> >  	return 0;
> >  }
> >  
> > +static const struct freq_conf *
> > +__clk_rcg2_select_conf(struct clk_hw *hw, const struct freq_multi_tbl *f,
> > +		       unsigned long req_rate)
> > +{
> > +	unsigned long best_rate = 0, parent_rate, rate;
> > +	const struct freq_conf *conf, *best_conf;
> > +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
> > +	struct clk_hw *p;
> > +	int index, i;
> > +
> > +	/* Exit early if only one config is defined */
> > +	if (f->num_confs == 1)
> > +		return f->confs;
> > +
> > +	/* Search in each provided config the one that is near the wanted rate */
> > +	for (i = 0, conf = f->confs; i < f->num_confs; i++, conf++) {
> > +		index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
> > +		if (index < 0)
> > +			continue;
> > +
> > +		p = clk_hw_get_parent_by_index(hw, index);
> > +		if (!p)
> > +			continue;
> > +
> > +		parent_rate =  clk_hw_get_rate(p);
> > +		rate = calc_rate(parent_rate, conf->n, conf->m, conf->n, conf->pre_div);
> > +
> > +		if (rate == req_rate) {
> > +			best_conf = conf;
> > +			break;
> > +		}
> > +
> > +		if (abs(req_rate - rate) < abs(best_rate - rate)) {
> Shouldn't this be:
> 
> if (abs(req_rate - rate) < abs(best_rate - req_rate)
> 
> ?
> 
> this way it'd say
> 
> "if this iteration's rate is closer to the requested one than the
> best one we've found yet, it's better"
>

Hi, thanks for the review!

I wonder if even better would be something where we save the best rate
diff and just compare that.

rate_diff = abs(req_rate - rate)
if (rate_diff < best_rate_diff) {
	best_rate_diff = rate_diff;
	best_conf = conf;
}

And best_rate_diff init to ULONG_MAX?

> > +			best_rate = rate;
> > +			best_conf = conf;
> > +		}
> > +	}
> > +
> > +	/*
> > +	 * Very unlikely.
> > +	 * Force the first conf if we can't find a correct config.
> > +	 */
> > +	if (unlikely(i == f->num_confs))
> > +		best_conf = f->confs;
> Is that a supported scenario or would it be a device driver / clock
> driver error?
> 

It's to handle case for the 2 continue in the loop and arriving in a
situation where best_conf was never set?

Should we return a warning and an ERR_PTR? Idea was to provide a best
effort selection.

> > +
> > +	return best_conf;
> > +}
> > +
> > +static int _freq_tbl_fm_determine_rate(struct clk_hw *hw, const struct freq_multi_tbl *f,
> > +				       struct clk_rate_request *req)
> > +{
> > +	unsigned long clk_flags, rate = req->rate;
> > +	const struct freq_conf *conf;
> > +	struct clk_hw *p;
> > +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
> swap lines 2, 3, 4 to 4, 2, 3 and you'll get a revers-Christmas-tree!
> 

Thanks, didn't notice this. Will do in v5.

-- 
	Ansuel

Re: [PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Konrad Dybcio]

On 28.05.2023 14:37, Christian Marangi wrote:
> On Sat, May 27, 2023 at 06:11:16PM +0200, Konrad Dybcio wrote:
>>
>>
>> On 27.04.2023 17:07, Christian Marangi wrote:
>>> Some RCG frequency can be reached by multiple configuration.
>>>
>>> Add clk_rcg2_fm_ops ops to support these special RCG configurations.
>>>
>>> These alternative ops will select the frequency using a CEIL policy.
>>>
>>> When the correct frequency is found, the correct config is selected by
>>> calculating the final rate (by checking the defined parent and values
>>> in the config that is being checked) and deciding based on the one that
>>> is less different than the requested one.
>>>
>>> These check are skipped if there is just on config for the requested
>>> freq.
>>>
>>> qcom_find_freq_multi is added to search the freq with the new struct
>>> freq_multi_tbl.
>>> __clk_rcg2_select_conf is used to select the correct conf by simulating
>>> the final clock.
>>>
>>> Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
>>> ---
>>>  drivers/clk/qcom/clk-rcg.h  |   1 +
>>>  drivers/clk/qcom/clk-rcg2.c | 152 ++++++++++++++++++++++++++++++++++++
>>>  drivers/clk/qcom/common.c   |  18 +++++
>>>  drivers/clk/qcom/common.h   |   2 +
>>>  4 files changed, 173 insertions(+)
>>>
>>> diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
>>> index dc85b46b0d79..f8ec989ed3d9 100644
>>> --- a/drivers/clk/qcom/clk-rcg.h
>>> +++ b/drivers/clk/qcom/clk-rcg.h
>>> @@ -188,6 +188,7 @@ struct clk_rcg2_gfx3d {
>>>  
>>>  extern const struct clk_ops clk_rcg2_ops;
>>>  extern const struct clk_ops clk_rcg2_floor_ops;
>>> +extern const struct clk_ops clk_rcg2_fm_ops;
>>>  extern const struct clk_ops clk_rcg2_mux_closest_ops;
>>>  extern const struct clk_ops clk_edp_pixel_ops;
>>>  extern const struct clk_ops clk_byte_ops;
>>> diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
>>> index 76551534f10d..4f2fe012ef5f 100644
>>> --- a/drivers/clk/qcom/clk-rcg2.c
>>> +++ b/drivers/clk/qcom/clk-rcg2.c
>>> @@ -266,6 +266,104 @@ static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f,
>>>  	return 0;
>>>  }
>>>  
>>> +static const struct freq_conf *
>>> +__clk_rcg2_select_conf(struct clk_hw *hw, const struct freq_multi_tbl *f,
>>> +		       unsigned long req_rate)
>>> +{
>>> +	unsigned long best_rate = 0, parent_rate, rate;
>>> +	const struct freq_conf *conf, *best_conf;
>>> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
>>> +	struct clk_hw *p;
>>> +	int index, i;
>>> +
>>> +	/* Exit early if only one config is defined */
>>> +	if (f->num_confs == 1)
>>> +		return f->confs;
>>> +
>>> +	/* Search in each provided config the one that is near the wanted rate */
>>> +	for (i = 0, conf = f->confs; i < f->num_confs; i++, conf++) {
>>> +		index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
>>> +		if (index < 0)
>>> +			continue;
>>> +
>>> +		p = clk_hw_get_parent_by_index(hw, index);
>>> +		if (!p)
>>> +			continue;
>>> +
>>> +		parent_rate =  clk_hw_get_rate(p);
>>> +		rate = calc_rate(parent_rate, conf->n, conf->m, conf->n, conf->pre_div);
>>> +
>>> +		if (rate == req_rate) {
>>> +			best_conf = conf;
>>> +			break;
>>> +		}
>>> +
>>> +		if (abs(req_rate - rate) < abs(best_rate - rate)) {
>> Shouldn't this be:
>>
>> if (abs(req_rate - rate) < abs(best_rate - req_rate)
>>
>> ?
>>
>> this way it'd say
>>
>> "if this iteration's rate is closer to the requested one than the
>> best one we've found yet, it's better"
>>
> 
> Hi, thanks for the review!
> 
> I wonder if even better would be something where we save the best rate
> diff and just compare that.
> 
> rate_diff = abs(req_rate - rate)
> if (rate_diff < best_rate_diff) {
> 	best_rate_diff = rate_diff;
> 	best_conf = conf;
> }
> 
> And best_rate_diff init to ULONG_MAX?
Yeah that would be more readable!

> 
>>> +			best_rate = rate;
>>> +			best_conf = conf;
>>> +		}
>>> +	}
>>> +
>>> +	/*
>>> +	 * Very unlikely.
>>> +	 * Force the first conf if we can't find a correct config.
>>> +	 */
>>> +	if (unlikely(i == f->num_confs))
>>> +		best_conf = f->confs;
>> Is that a supported scenario or would it be a device driver / clock
>> driver error?
>>
> 
> It's to handle case for the 2 continue in the loop and arriving in a
> situation where best_conf was never set?
> 
> Should we return a warning and an ERR_PTR? Idea was to provide a best
> effort selection.
Hm.. I'm not sure what's the expected behavior here.. Stephen?

Konrad
> 
>>> +
>>> +	return best_conf;
>>> +}
>>> +
>>> +static int _freq_tbl_fm_determine_rate(struct clk_hw *hw, const struct freq_multi_tbl *f,
>>> +				       struct clk_rate_request *req)
>>> +{
>>> +	unsigned long clk_flags, rate = req->rate;
>>> +	const struct freq_conf *conf;
>>> +	struct clk_hw *p;
>>> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
>> swap lines 2, 3, 4 to 4, 2, 3 and you'll get a revers-Christmas-tree!
>>
> 
> Thanks, didn't notice this. Will do in v5.
> 

Re: [PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Christian Marangi]

On Mon, May 29, 2023 at 02:12:23PM +0200, Konrad Dybcio wrote:
> 
> 
> On 28.05.2023 14:37, Christian Marangi wrote:
> > On Sat, May 27, 2023 at 06:11:16PM +0200, Konrad Dybcio wrote:
> >>
> >>
> >> On 27.04.2023 17:07, Christian Marangi wrote:
> >>> Some RCG frequency can be reached by multiple configuration.
> >>>
> >>> Add clk_rcg2_fm_ops ops to support these special RCG configurations.
> >>>
> >>> These alternative ops will select the frequency using a CEIL policy.
> >>>
> >>> When the correct frequency is found, the correct config is selected by
> >>> calculating the final rate (by checking the defined parent and values
> >>> in the config that is being checked) and deciding based on the one that
> >>> is less different than the requested one.
> >>>
> >>> These check are skipped if there is just on config for the requested
> >>> freq.
> >>>
> >>> qcom_find_freq_multi is added to search the freq with the new struct
> >>> freq_multi_tbl.
> >>> __clk_rcg2_select_conf is used to select the correct conf by simulating
> >>> the final clock.
> >>>
> >>> Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
> >>> ---
> >>>  drivers/clk/qcom/clk-rcg.h  |   1 +
> >>>  drivers/clk/qcom/clk-rcg2.c | 152 ++++++++++++++++++++++++++++++++++++
> >>>  drivers/clk/qcom/common.c   |  18 +++++
> >>>  drivers/clk/qcom/common.h   |   2 +
> >>>  4 files changed, 173 insertions(+)
> >>>
> >>> diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
> >>> index dc85b46b0d79..f8ec989ed3d9 100644
> >>> --- a/drivers/clk/qcom/clk-rcg.h
> >>> +++ b/drivers/clk/qcom/clk-rcg.h
> >>> @@ -188,6 +188,7 @@ struct clk_rcg2_gfx3d {
> >>>  
> >>>  extern const struct clk_ops clk_rcg2_ops;
> >>>  extern const struct clk_ops clk_rcg2_floor_ops;
> >>> +extern const struct clk_ops clk_rcg2_fm_ops;
> >>>  extern const struct clk_ops clk_rcg2_mux_closest_ops;
> >>>  extern const struct clk_ops clk_edp_pixel_ops;
> >>>  extern const struct clk_ops clk_byte_ops;
> >>> diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
> >>> index 76551534f10d..4f2fe012ef5f 100644
> >>> --- a/drivers/clk/qcom/clk-rcg2.c
> >>> +++ b/drivers/clk/qcom/clk-rcg2.c
> >>> @@ -266,6 +266,104 @@ static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f,
> >>>  	return 0;
> >>>  }
> >>>  
> >>> +static const struct freq_conf *
> >>> +__clk_rcg2_select_conf(struct clk_hw *hw, const struct freq_multi_tbl *f,
> >>> +		       unsigned long req_rate)
> >>> +{
> >>> +	unsigned long best_rate = 0, parent_rate, rate;
> >>> +	const struct freq_conf *conf, *best_conf;
> >>> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
> >>> +	struct clk_hw *p;
> >>> +	int index, i;
> >>> +
> >>> +	/* Exit early if only one config is defined */
> >>> +	if (f->num_confs == 1)
> >>> +		return f->confs;
> >>> +
> >>> +	/* Search in each provided config the one that is near the wanted rate */
> >>> +	for (i = 0, conf = f->confs; i < f->num_confs; i++, conf++) {
> >>> +		index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
> >>> +		if (index < 0)
> >>> +			continue;
> >>> +
> >>> +		p = clk_hw_get_parent_by_index(hw, index);
> >>> +		if (!p)
> >>> +			continue;
> >>> +
> >>> +		parent_rate =  clk_hw_get_rate(p);
> >>> +		rate = calc_rate(parent_rate, conf->n, conf->m, conf->n, conf->pre_div);
> >>> +
> >>> +		if (rate == req_rate) {
> >>> +			best_conf = conf;
> >>> +			break;
> >>> +		}
> >>> +
> >>> +		if (abs(req_rate - rate) < abs(best_rate - rate)) {
> >> Shouldn't this be:
> >>
> >> if (abs(req_rate - rate) < abs(best_rate - req_rate)
> >>
> >> ?
> >>
> >> this way it'd say
> >>
> >> "if this iteration's rate is closer to the requested one than the
> >> best one we've found yet, it's better"
> >>
> > 
> > Hi, thanks for the review!
> > 
> > I wonder if even better would be something where we save the best rate
> > diff and just compare that.
> > 
> > rate_diff = abs(req_rate - rate)
> > if (rate_diff < best_rate_diff) {
> > 	best_rate_diff = rate_diff;
> > 	best_conf = conf;
> > }
> > 
> > And best_rate_diff init to ULONG_MAX?
> Yeah that would be more readable!
> 
> > 
> >>> +			best_rate = rate;
> >>> +			best_conf = conf;
> >>> +		}
> >>> +	}
> >>> +
> >>> +	/*
> >>> +	 * Very unlikely.
> >>> +	 * Force the first conf if we can't find a correct config.
> >>> +	 */
> >>> +	if (unlikely(i == f->num_confs))
> >>> +		best_conf = f->confs;
> >> Is that a supported scenario or would it be a device driver / clock
> >> driver error?
> >>
> > 
> > It's to handle case for the 2 continue in the loop and arriving in a
> > situation where best_conf was never set?
> > 
> > Should we return a warning and an ERR_PTR? Idea was to provide a best
> > effort selection.
> Hm.. I'm not sure what's the expected behavior here.. Stephen?
> 

I have this implementation rady, if you want I can send this revision
and discuss that in v5 directly. It's WARN and returning -EINVAL.

> > 
> >>> +
> >>> +	return best_conf;
> >>> +}
> >>> +
> >>> +static int _freq_tbl_fm_determine_rate(struct clk_hw *hw, const struct freq_multi_tbl *f,
> >>> +				       struct clk_rate_request *req)
> >>> +{
> >>> +	unsigned long clk_flags, rate = req->rate;
> >>> +	const struct freq_conf *conf;
> >>> +	struct clk_hw *p;
> >>> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
> >> swap lines 2, 3, 4 to 4, 2, 3 and you'll get a revers-Christmas-tree!
> >>
> > 
> > Thanks, didn't notice this. Will do in v5.
> > 

-- 
	Ansuel

Re: [PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Stephen Boyd]

Quoting Christian Marangi (2023-05-29 05:34:57)
> On Mon, May 29, 2023 at 02:12:23PM +0200, Konrad Dybcio wrote:
> > On 28.05.2023 14:37, Christian Marangi wrote:
> > > On Sat, May 27, 2023 at 06:11:16PM +0200, Konrad Dybcio wrote:
> > >> On 27.04.2023 17:07, Christian Marangi wrote:
> > >>> +  * Force the first conf if we can't find a correct config.
> > >>> +  */
> > >>> + if (unlikely(i == f->num_confs))
> > >>> +         best_conf = f->confs;
> > >> Is that a supported scenario or would it be a device driver / clock
> > >> driver error?
> > >>
> > > 
> > > It's to handle case for the 2 continue in the loop and arriving in a
> > > situation where best_conf was never set?
> > > 
> > > Should we return a warning and an ERR_PTR? Idea was to provide a best
> > > effort selection.
> > Hm.. I'm not sure what's the expected behavior here.. Stephen?
> > 
> 
> I have this implementation rady, if you want I can send this revision
> and discuss that in v5 directly. It's WARN and returning -EINVAL.

I'd only have a WARN if you never expect to hit that case. Otherwise, it
should return -EINVAL and not warn. At a quick glance it sounds like
some sort of rounding policy, so just make sure the
round_rate/determine_rate implementation agrees with what set_rate()
will do and it should be good.

Re: [PATCH v4 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq multi ops

[Christian Marangi]

On Wed, Jun 14, 2023 at 05:28:08PM -0700, Stephen Boyd wrote:
> Quoting Christian Marangi (2023-05-29 05:34:57)
> > On Mon, May 29, 2023 at 02:12:23PM +0200, Konrad Dybcio wrote:
> > > On 28.05.2023 14:37, Christian Marangi wrote:
> > > > On Sat, May 27, 2023 at 06:11:16PM +0200, Konrad Dybcio wrote:
> > > >> On 27.04.2023 17:07, Christian Marangi wrote:
> > > >>> +  * Force the first conf if we can't find a correct config.
> > > >>> +  */
> > > >>> + if (unlikely(i == f->num_confs))
> > > >>> +         best_conf = f->confs;
> > > >> Is that a supported scenario or would it be a device driver / clock
> > > >> driver error?
> > > >>
> > > > 
> > > > It's to handle case for the 2 continue in the loop and arriving in a
> > > > situation where best_conf was never set?
> > > > 
> > > > Should we return a warning and an ERR_PTR? Idea was to provide a best
> > > > effort selection.
> > > Hm.. I'm not sure what's the expected behavior here.. Stephen?
> > > 
> > 
> > I have this implementation rady, if you want I can send this revision
> > and discuss that in v5 directly. It's WARN and returning -EINVAL.
> 
> I'd only have a WARN if you never expect to hit that case. Otherwise, it
> should return -EINVAL and not warn. At a quick glance it sounds like
> some sort of rounding policy, so just make sure the
> round_rate/determine_rate implementation agrees with what set_rate()
> will do and it should be good.

Hi, in theory the WAN path should never happen, as it means no parent
for the clk are present. So I guess with your logic printing a WARN is
correct.

About the rounding policy this is more or less problematic, it's a
CLOSEST policy, so not a CEIL or FLOOR one. The determine_rate apply the
very same selection logic of set_rate so also that should be good.

I sent v5 some time ago with the concern here addressed so if you have
time it would be good if you can check that.

-- 
	Ansuel