* [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller
@ 2022-08-05 8:57 Yu Tu
2022-08-05 8:57 ` [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings Yu Tu
` (5 more replies)
0 siblings, 6 replies; 41+ messages in thread
From: Yu Tu @ 2022-08-05 8:57 UTC (permalink / raw)
To: linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Cc: Yu Tu
1. Add PLL and Peripheral clock controller driver for S4 SOC.
Yu Tu (6):
dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings
arm64: dts: meson: add S4 Soc PLL clock controller in DT
clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
dt-bindings: clk: meson: add S4 SoC peripheral clock controller
bindings
arm64: dts: meson: add S4 Soc Peripheral clock controller in DT
clk: meson: s4: add s4 SoC peripheral clock controller driver
V2 -> V3: Use two clock controller.
V1 -> V2: Change format as discussed in the email.
Link:https://lore.kernel.org/all/20220728054202.6981-1-yu.tu@amlogic.com/
.../bindings/clock/amlogic,s4-clkc.yaml | 92 +
.../bindings/clock/amlogic,s4-pll-clkc.yaml | 51 +
MAINTAINERS | 1 +
arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 34 +
drivers/clk/meson/Kconfig | 25 +
drivers/clk/meson/Makefile | 2 +
drivers/clk/meson/s4-pll.c | 891 ++++
drivers/clk/meson/s4-pll.h | 88 +
drivers/clk/meson/s4.c | 3878 +++++++++++++++++
drivers/clk/meson/s4.h | 232 +
include/dt-bindings/clock/amlogic,s4-clkc.h | 131 +
.../dt-bindings/clock/amlogic,s4-pll-clkc.h | 30 +
12 files changed, 5455 insertions(+)
create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
create mode 100644 drivers/clk/meson/s4-pll.c
create mode 100644 drivers/clk/meson/s4-pll.h
create mode 100644 drivers/clk/meson/s4.c
create mode 100644 drivers/clk/meson/s4.h
create mode 100644 include/dt-bindings/clock/amlogic,s4-clkc.h
create mode 100644 include/dt-bindings/clock/amlogic,s4-pll-clkc.h
base-commit: 08fc500fe3d4b1f0603fb97ad353f246a3d52d2d
--
2.33.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings
2022-08-05 8:57 [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller Yu Tu
@ 2022-08-05 8:57 ` Yu Tu
2022-08-05 9:13 ` Krzysztof Kozlowski
2022-08-05 8:57 ` [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT Yu Tu
` (4 subsequent siblings)
5 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 8:57 UTC (permalink / raw)
To: linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Cc: Yu Tu
Add the documentation to support Amlogic S4 SoC PLL clock driver and
add S4 SoC PLL clock controller bindings.
Signed-off-by: Yu Tu <yu.tu@amlogic.com>
---
.../bindings/clock/amlogic,s4-pll-clkc.yaml | 51 +++++++++++++++++++
MAINTAINERS | 1 +
.../dt-bindings/clock/amlogic,s4-pll-clkc.h | 30 +++++++++++
3 files changed, 82 insertions(+)
create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
create mode 100644 include/dt-bindings/clock/amlogic,s4-pll-clkc.h
diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
new file mode 100644
index 000000000000..079ae905b69e
--- /dev/null
+++ b/Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
@@ -0,0 +1,51 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/clock/amlogic,s4-pll-clkc.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Amlogic Meson S serials PLL Clock Controller Device Tree Bindings
+
+maintainers:
+ - Neil Armstrong <narmstrong@baylibre.com>
+ - Jerome Brunet <jbrunet@baylibre.com>
+ - Yu Tu <yu.hu@amlogic.com>
+
+
+properties:
+ compatible:
+ const: amlogic,s4-pll-clkc
+
+ reg:
+ maxItems: 1
+
+ clocks:
+ maxItems: 1
+
+ clock-names:
+ items:
+ - const: xtal
+
+ "#clock-cells":
+ const: 1
+
+required:
+ - compatible
+ - reg
+ - clocks
+ - clock-names
+ - "#clock-cells"
+
+additionalProperties: false
+
+examples:
+ - |
+ clkc_pll: pll-clock-controller@fe008000 {
+ compatible = "amlogic,s4-pll-clkc";
+ reg = <0xfe008000 0x348>;
+ clocks = <&xtal>;
+ clock-names = "xtal";
+ #clock-cells = <1>;
+ };
+
+...
diff --git a/MAINTAINERS b/MAINTAINERS
index 64379c699903..b039cf953520 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1774,6 +1774,7 @@ L: linux-amlogic@lists.infradead.org
S: Maintained
F: Documentation/devicetree/bindings/clock/amlogic*
F: drivers/clk/meson/
+F: include/dt-bindings/clock/amlogic*
F: include/dt-bindings/clock/gxbb*
F: include/dt-bindings/clock/meson*
diff --git a/include/dt-bindings/clock/amlogic,s4-pll-clkc.h b/include/dt-bindings/clock/amlogic,s4-pll-clkc.h
new file mode 100644
index 000000000000..08b7c5c5ba01
--- /dev/null
+++ b/include/dt-bindings/clock/amlogic,s4-pll-clkc.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
+/*
+ * Copyright (c) 2021 Amlogic, Inc. All rights reserved.
+ * Author: Yu Tu <yu.tu@amlogic.com>
+ */
+
+#ifndef _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
+#define _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
+
+/*
+ * CLKID index values
+ */
+
+#define CLKID_FIXED_PLL 1
+#define CLKID_FCLK_DIV2 3
+#define CLKID_FCLK_DIV3 5
+#define CLKID_FCLK_DIV4 7
+#define CLKID_FCLK_DIV5 9
+#define CLKID_FCLK_DIV7 11
+#define CLKID_FCLK_DIV2P5 13
+#define CLKID_GP0_PLL 15
+#define CLKID_HIFI_PLL 17
+#define CLKID_HDMI_PLL 20
+#define CLKID_MPLL_50M 22
+#define CLKID_MPLL0 25
+#define CLKID_MPLL1 27
+#define CLKID_MPLL2 29
+#define CLKID_MPLL3 31
+
+#endif /* _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H */
--
2.33.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 41+ messages in thread
* [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-05 8:57 [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller Yu Tu
2022-08-05 8:57 ` [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings Yu Tu
@ 2022-08-05 8:57 ` Yu Tu
2022-08-05 9:16 ` Krzysztof Kozlowski
2022-08-05 8:57 ` [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver Yu Tu
` (3 subsequent siblings)
5 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 8:57 UTC (permalink / raw)
To: linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Cc: Yu Tu
Added information about the S4 SOC PLL Clock controller in DT.
Signed-off-by: Yu Tu <yu.tu@amlogic.com>
---
arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
index ff213618a598..a816b1f7694b 100644
--- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
+++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
@@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
#size-cells = <2>;
ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
+ clkc_pll: pll-clock-controller@8000 {
+ compatible = "amlogic,s4-pll-clkc";
+ reg = <0x0 0x8000 0x0 0x348>;
+ clocks = <&xtal>;
+ clock-names = "xtal";
+ #clock-cells = <1>;
+ };
+
periphs_pinctrl: pinctrl@4000 {
compatible = "amlogic,meson-s4-periphs-pinctrl";
#address-cells = <2>;
--
2.33.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 41+ messages in thread
* [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-05 8:57 [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller Yu Tu
2022-08-05 8:57 ` [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings Yu Tu
2022-08-05 8:57 ` [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT Yu Tu
@ 2022-08-05 8:57 ` Yu Tu
2022-08-10 13:47 ` Jerome Brunet
2022-08-05 8:57 ` [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings Yu Tu
` (2 subsequent siblings)
5 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 8:57 UTC (permalink / raw)
To: linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Cc: Yu Tu
Add the S4 PLL clock controller found in the s4 SoC family.
Signed-off-by: Yu Tu <yu.tu@amlogic.com>
---
drivers/clk/meson/Kconfig | 12 +
drivers/clk/meson/Makefile | 1 +
drivers/clk/meson/s4-pll.c | 891 +++++++++++++++++++++++++++++++++++++
drivers/clk/meson/s4-pll.h | 88 ++++
4 files changed, 992 insertions(+)
create mode 100644 drivers/clk/meson/s4-pll.c
create mode 100644 drivers/clk/meson/s4-pll.h
diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
index fc002c155bc3..f4244edc7b28 100644
--- a/drivers/clk/meson/Kconfig
+++ b/drivers/clk/meson/Kconfig
@@ -115,4 +115,16 @@ config COMMON_CLK_G12A
help
Support for the clock controller on Amlogic S905D2, S905X2 and S905Y2
devices, aka g12a. Say Y if you want peripherals to work.
+
+config COMMON_CLK_S4_PLL
+ tristate "S4 SoC PLL clock controllers support"
+ depends on ARM64
+ default y
+ select COMMON_CLK_MESON_MPLL
+ select COMMON_CLK_MESON_PLL
+ select COMMON_CLK_MESON_REGMAP
+ help
+ Support for the pll clock controller on Amlogic S805X2 and S905Y4 devices,
+ aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and AQ229.
+ Say Y if you want peripherals and CPU frequency scaling to work.
endmenu
diff --git a/drivers/clk/meson/Makefile b/drivers/clk/meson/Makefile
index 6eca2a406ee3..376f49cc13f1 100644
--- a/drivers/clk/meson/Makefile
+++ b/drivers/clk/meson/Makefile
@@ -19,3 +19,4 @@ obj-$(CONFIG_COMMON_CLK_AXG_AUDIO) += axg-audio.o
obj-$(CONFIG_COMMON_CLK_GXBB) += gxbb.o gxbb-aoclk.o
obj-$(CONFIG_COMMON_CLK_G12A) += g12a.o g12a-aoclk.o
obj-$(CONFIG_COMMON_CLK_MESON8B) += meson8b.o meson8-ddr.o
+obj-$(CONFIG_COMMON_CLK_S4_PLL) += s4-pll.o
diff --git a/drivers/clk/meson/s4-pll.c b/drivers/clk/meson/s4-pll.c
new file mode 100644
index 000000000000..478c78b5ed46
--- /dev/null
+++ b/drivers/clk/meson/s4-pll.c
@@ -0,0 +1,891 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Amlogic Meson-S4 Clock Controller Driver
+ *
+ * Copyright (c) 2021 Amlogic, inc.
+ * Author: Yu Tu <yu.tu@amlogic.com>
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+
+#include "clk-mpll.h"
+#include "clk-pll.h"
+#include "clk-regmap.h"
+#include "s4-pll.h"
+
+static DEFINE_SPINLOCK(meson_clk_lock);
+
+static struct clk_regmap s4_fixed_pll_dco = {
+ .data = &(struct meson_clk_pll_data){
+ .en = {
+ .reg_off = ANACTRL_FIXPLL_CTRL0,
+ .shift = 28,
+ .width = 1,
+ },
+ .m = {
+ .reg_off = ANACTRL_FIXPLL_CTRL0,
+ .shift = 0,
+ .width = 8,
+ },
+ .n = {
+ .reg_off = ANACTRL_FIXPLL_CTRL0,
+ .shift = 10,
+ .width = 5,
+ },
+ .frac = {
+ .reg_off = ANACTRL_FIXPLL_CTRL1,
+ .shift = 0,
+ .width = 17,
+ },
+ .l = {
+ .reg_off = ANACTRL_FIXPLL_CTRL0,
+ .shift = 31,
+ .width = 1,
+ },
+ .rst = {
+ .reg_off = ANACTRL_FIXPLL_CTRL0,
+ .shift = 29,
+ .width = 1,
+ },
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fixed_pll_dco",
+ .ops = &meson_clk_pll_ro_ops,
+ .parent_data = (const struct clk_parent_data []) {
+ { .fw_name = "xtal", }
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fixed_pll = {
+ .data = &(struct clk_regmap_div_data){
+ .offset = ANACTRL_FIXPLL_CTRL0,
+ .shift = 16,
+ .width = 2,
+ .flags = CLK_DIVIDER_POWER_OF_TWO,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fixed_pll",
+ .ops = &clk_regmap_divider_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fixed_pll_dco.hw
+ },
+ .num_parents = 1,
+ /*
+ * This clock won't ever change at runtime so
+ * CLK_SET_RATE_PARENT is not required
+ */
+ },
+};
+
+static struct clk_fixed_factor s4_fclk_div2_div = {
+ .mult = 1,
+ .div = 2,
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div2_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fclk_div2 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_FIXPLL_CTRL1,
+ .bit_idx = 24,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div2",
+ .ops = &clk_regmap_gate_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fclk_div2_div.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_fixed_factor s4_fclk_div3_div = {
+ .mult = 1,
+ .div = 3,
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div3_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fclk_div3 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_FIXPLL_CTRL1,
+ .bit_idx = 20,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div3",
+ .ops = &clk_regmap_gate_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fclk_div3_div.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_fixed_factor s4_fclk_div4_div = {
+ .mult = 1,
+ .div = 4,
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div4_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fclk_div4 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_FIXPLL_CTRL1,
+ .bit_idx = 21,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div4",
+ .ops = &clk_regmap_gate_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fclk_div4_div.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_fixed_factor s4_fclk_div5_div = {
+ .mult = 1,
+ .div = 5,
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div5_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fclk_div5 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_FIXPLL_CTRL1,
+ .bit_idx = 22,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div5",
+ .ops = &clk_regmap_gate_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fclk_div5_div.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_fixed_factor s4_fclk_div7_div = {
+ .mult = 1,
+ .div = 7,
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div7_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fclk_div7 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_FIXPLL_CTRL1,
+ .bit_idx = 23,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div7",
+ .ops = &clk_regmap_gate_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fclk_div7_div.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_fixed_factor s4_fclk_div2p5_div = {
+ .mult = 2,
+ .div = 5,
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div2p5_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fixed_pll.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_fclk_div2p5 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_FIXPLL_CTRL1,
+ .bit_idx = 25,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "fclk_div2p5",
+ .ops = &clk_regmap_gate_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fclk_div2p5_div.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static const struct pll_mult_range s4_gp0_pll_mult_range = {
+ .min = 125,
+ .max = 250,
+};
+
+/*
+ * Internal gp0 pll emulation configuration parameters
+ */
+static const struct reg_sequence s4_gp0_init_regs[] = {
+ { .reg = ANACTRL_GP0PLL_CTRL1, .def = 0x00000000 },
+ { .reg = ANACTRL_GP0PLL_CTRL2, .def = 0x00000000 },
+ { .reg = ANACTRL_GP0PLL_CTRL3, .def = 0x48681c00 },
+ { .reg = ANACTRL_GP0PLL_CTRL4, .def = 0x88770290 },
+ { .reg = ANACTRL_GP0PLL_CTRL5, .def = 0x39272000 },
+ { .reg = ANACTRL_GP0PLL_CTRL6, .def = 0x56540000 }
+};
+
+static struct clk_regmap s4_gp0_pll_dco = {
+ .data = &(struct meson_clk_pll_data){
+ .en = {
+ .reg_off = ANACTRL_GP0PLL_CTRL0,
+ .shift = 28,
+ .width = 1,
+ },
+ .m = {
+ .reg_off = ANACTRL_GP0PLL_CTRL0,
+ .shift = 0,
+ .width = 8,
+ },
+ .n = {
+ .reg_off = ANACTRL_GP0PLL_CTRL0,
+ .shift = 10,
+ .width = 5,
+ },
+ .frac = {
+ .reg_off = ANACTRL_GP0PLL_CTRL1,
+ .shift = 0,
+ .width = 17,
+ },
+ .l = {
+ .reg_off = ANACTRL_GP0PLL_CTRL0,
+ .shift = 31,
+ .width = 1,
+ },
+ .rst = {
+ .reg_off = ANACTRL_GP0PLL_CTRL0,
+ .shift = 29,
+ .width = 1,
+ },
+ .range = &s4_gp0_pll_mult_range,
+ .init_regs = s4_gp0_init_regs,
+ .init_count = ARRAY_SIZE(s4_gp0_init_regs),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "gp0_pll_dco",
+ .ops = &meson_clk_pll_ops,
+ .parent_data = (const struct clk_parent_data []) {
+ { .fw_name = "xtal", }
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_gp0_pll = {
+ .data = &(struct clk_regmap_div_data){
+ .offset = ANACTRL_GP0PLL_CTRL0,
+ .shift = 16,
+ .width = 3,
+ .flags = (CLK_DIVIDER_POWER_OF_TWO |
+ CLK_DIVIDER_ROUND_CLOSEST),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "gp0_pll",
+ .ops = &clk_regmap_divider_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_gp0_pll_dco.hw
+ },
+ .num_parents = 1,
+ .flags = CLK_SET_RATE_PARENT,
+ },
+};
+
+/*
+ * Internal hifi pll emulation configuration parameters
+ */
+static const struct reg_sequence s4_hifi_init_regs[] = {
+ { .reg = ANACTRL_HIFIPLL_CTRL1, .def = 0x00010e56 },
+ { .reg = ANACTRL_HIFIPLL_CTRL2, .def = 0x00000000 },
+ { .reg = ANACTRL_HIFIPLL_CTRL3, .def = 0x6a285c00 },
+ { .reg = ANACTRL_HIFIPLL_CTRL4, .def = 0x65771290 },
+ { .reg = ANACTRL_HIFIPLL_CTRL5, .def = 0x39272000 },
+ { .reg = ANACTRL_HIFIPLL_CTRL6, .def = 0x56540000 }
+};
+
+static struct clk_regmap s4_hifi_pll_dco = {
+ .data = &(struct meson_clk_pll_data){
+ .en = {
+ .reg_off = ANACTRL_HIFIPLL_CTRL0,
+ .shift = 28,
+ .width = 1,
+ },
+ .m = {
+ .reg_off = ANACTRL_HIFIPLL_CTRL0,
+ .shift = 0,
+ .width = 8,
+ },
+ .n = {
+ .reg_off = ANACTRL_HIFIPLL_CTRL0,
+ .shift = 10,
+ .width = 5,
+ },
+ .frac = {
+ .reg_off = ANACTRL_HIFIPLL_CTRL1,
+ .shift = 0,
+ .width = 17,
+ },
+ .l = {
+ .reg_off = ANACTRL_HIFIPLL_CTRL0,
+ .shift = 31,
+ .width = 1,
+ },
+ .rst = {
+ .reg_off = ANACTRL_HIFIPLL_CTRL0,
+ .shift = 29,
+ .width = 1,
+ },
+ .range = &s4_gp0_pll_mult_range,
+ .init_regs = s4_hifi_init_regs,
+ .init_count = ARRAY_SIZE(s4_hifi_init_regs),
+ .flags = CLK_MESON_PLL_ROUND_CLOSEST,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "hifi_pll_dco",
+ .ops = &meson_clk_pll_ops,
+ .parent_data = (const struct clk_parent_data []) {
+ { .fw_name = "xtal", }
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_hifi_pll = {
+ .data = &(struct clk_regmap_div_data){
+ .offset = ANACTRL_HIFIPLL_CTRL0,
+ .shift = 16,
+ .width = 2,
+ .flags = (CLK_DIVIDER_POWER_OF_TWO |
+ CLK_DIVIDER_ROUND_CLOSEST),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "hifi_pll",
+ .ops = &clk_regmap_divider_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_hifi_pll_dco.hw
+ },
+ .num_parents = 1,
+ .flags = CLK_SET_RATE_PARENT,
+ },
+};
+
+static struct clk_regmap s4_hdmi_pll_dco = {
+ .data = &(struct meson_clk_pll_data){
+ .en = {
+ .reg_off = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 28,
+ .width = 1,
+ },
+ .m = {
+ .reg_off = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 0,
+ .width = 8,
+ },
+ .n = {
+ .reg_off = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 10,
+ .width = 5,
+ },
+ .frac = {
+ .reg_off = ANACTRL_HDMIPLL_CTRL1,
+ .shift = 0,
+ .width = 17,
+ },
+ .l = {
+ .reg_off = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 31,
+ .width = 1,
+ },
+ .rst = {
+ .reg_off = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 29,
+ .width = 1,
+ },
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "hdmi_pll_dco",
+ .ops = &meson_clk_pll_ro_ops,
+ .parent_data = (const struct clk_parent_data []) {
+ { .fw_name = "xtal", }
+ },
+ .num_parents = 1,
+ /*
+ * Display directly handle hdmi pll registers ATM, we need
+ * NOCACHE to keep our view of the clock as accurate as
+ * possible
+ */
+ .flags = CLK_GET_RATE_NOCACHE,
+ },
+};
+
+static struct clk_regmap s4_hdmi_pll_od = {
+ .data = &(struct clk_regmap_div_data){
+ .offset = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 16,
+ .width = 4,
+ .flags = CLK_DIVIDER_POWER_OF_TWO,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "hdmi_pll_od",
+ .ops = &clk_regmap_divider_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_hdmi_pll_dco.hw
+ },
+ .num_parents = 1,
+ /*
+ * Display directly handle hdmi pll registers ATM, we need
+ * NOCACHE to keep our view of the clock as accurate as
+ * possible
+ */
+ .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
+ },
+};
+
+static struct clk_regmap s4_hdmi_pll = {
+ .data = &(struct clk_regmap_div_data){
+ .offset = ANACTRL_HDMIPLL_CTRL0,
+ .shift = 20,
+ .width = 2,
+ .flags = CLK_DIVIDER_POWER_OF_TWO,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "hdmi_pll",
+ .ops = &clk_regmap_divider_ro_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_hdmi_pll_od.hw
+ },
+ .num_parents = 1,
+ /*
+ * Display directly handle hdmi pll registers ATM, we need
+ * NOCACHE to keep our view of the clock as accurate as
+ * possible
+ */
+ .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
+ },
+};
+
+static struct clk_fixed_factor s4_mpll_50m_div = {
+ .mult = 1,
+ .div = 80,
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll_50m_div",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fixed_pll_dco.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_mpll_50m = {
+ .data = &(struct clk_regmap_mux_data){
+ .offset = ANACTRL_FIXPLL_CTRL3,
+ .mask = 0x1,
+ .shift = 5,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll_50m",
+ .ops = &clk_regmap_mux_ro_ops,
+ .parent_data = (const struct clk_parent_data []) {
+ { .fw_name = "xtal", },
+ { .hw = &s4_mpll_50m_div.hw },
+ },
+ .num_parents = 2,
+ },
+};
+
+static struct clk_fixed_factor s4_mpll_prediv = {
+ .mult = 1,
+ .div = 2,
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll_prediv",
+ .ops = &clk_fixed_factor_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_fixed_pll_dco.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static const struct reg_sequence s4_mpll0_init_regs[] = {
+ { .reg = ANACTRL_MPLL_CTRL2, .def = 0x40000033 }
+};
+
+static struct clk_regmap s4_mpll0_div = {
+ .data = &(struct meson_clk_mpll_data){
+ .sdm = {
+ .reg_off = ANACTRL_MPLL_CTRL1,
+ .shift = 0,
+ .width = 14,
+ },
+ .sdm_en = {
+ .reg_off = ANACTRL_MPLL_CTRL1,
+ .shift = 30,
+ .width = 1,
+ },
+ .n2 = {
+ .reg_off = ANACTRL_MPLL_CTRL1,
+ .shift = 20,
+ .width = 9,
+ },
+ .ssen = {
+ .reg_off = ANACTRL_MPLL_CTRL1,
+ .shift = 29,
+ .width = 1,
+ },
+ .lock = &meson_clk_lock,
+ .init_regs = s4_mpll0_init_regs,
+ .init_count = ARRAY_SIZE(s4_mpll0_init_regs),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll0_div",
+ .ops = &meson_clk_mpll_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_mpll_prediv.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_mpll0 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_MPLL_CTRL1,
+ .bit_idx = 31,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll0",
+ .ops = &clk_regmap_gate_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_mpll0_div.hw },
+ .num_parents = 1,
+ .flags = CLK_SET_RATE_PARENT,
+ },
+};
+
+static const struct reg_sequence s4_mpll1_init_regs[] = {
+ { .reg = ANACTRL_MPLL_CTRL4, .def = 0x40000033 }
+};
+
+static struct clk_regmap s4_mpll1_div = {
+ .data = &(struct meson_clk_mpll_data){
+ .sdm = {
+ .reg_off = ANACTRL_MPLL_CTRL3,
+ .shift = 0,
+ .width = 14,
+ },
+ .sdm_en = {
+ .reg_off = ANACTRL_MPLL_CTRL3,
+ .shift = 30,
+ .width = 1,
+ },
+ .n2 = {
+ .reg_off = ANACTRL_MPLL_CTRL3,
+ .shift = 20,
+ .width = 9,
+ },
+ .ssen = {
+ .reg_off = ANACTRL_MPLL_CTRL3,
+ .shift = 29,
+ .width = 1,
+ },
+ .lock = &meson_clk_lock,
+ .init_regs = s4_mpll1_init_regs,
+ .init_count = ARRAY_SIZE(s4_mpll1_init_regs),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll1_div",
+ .ops = &meson_clk_mpll_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_mpll_prediv.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_mpll1 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_MPLL_CTRL3,
+ .bit_idx = 31,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll1",
+ .ops = &clk_regmap_gate_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_mpll1_div.hw },
+ .num_parents = 1,
+ .flags = CLK_SET_RATE_PARENT,
+ },
+};
+
+static const struct reg_sequence s4_mpll2_init_regs[] = {
+ { .reg = ANACTRL_MPLL_CTRL6, .def = 0x40000033 }
+};
+
+static struct clk_regmap s4_mpll2_div = {
+ .data = &(struct meson_clk_mpll_data){
+ .sdm = {
+ .reg_off = ANACTRL_MPLL_CTRL5,
+ .shift = 0,
+ .width = 14,
+ },
+ .sdm_en = {
+ .reg_off = ANACTRL_MPLL_CTRL5,
+ .shift = 30,
+ .width = 1,
+ },
+ .n2 = {
+ .reg_off = ANACTRL_MPLL_CTRL5,
+ .shift = 20,
+ .width = 9,
+ },
+ .ssen = {
+ .reg_off = ANACTRL_MPLL_CTRL5,
+ .shift = 29,
+ .width = 1,
+ },
+ .lock = &meson_clk_lock,
+ .init_regs = s4_mpll2_init_regs,
+ .init_count = ARRAY_SIZE(s4_mpll2_init_regs),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll2_div",
+ .ops = &meson_clk_mpll_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_mpll_prediv.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_mpll2 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_MPLL_CTRL5,
+ .bit_idx = 31,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll2",
+ .ops = &clk_regmap_gate_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_mpll2_div.hw },
+ .num_parents = 1,
+ .flags = CLK_SET_RATE_PARENT,
+ },
+};
+
+static const struct reg_sequence s4_mpll3_init_regs[] = {
+ { .reg = ANACTRL_MPLL_CTRL8, .def = 0x40000033 }
+};
+
+static struct clk_regmap s4_mpll3_div = {
+ .data = &(struct meson_clk_mpll_data){
+ .sdm = {
+ .reg_off = ANACTRL_MPLL_CTRL7,
+ .shift = 0,
+ .width = 14,
+ },
+ .sdm_en = {
+ .reg_off = ANACTRL_MPLL_CTRL7,
+ .shift = 30,
+ .width = 1,
+ },
+ .n2 = {
+ .reg_off = ANACTRL_MPLL_CTRL7,
+ .shift = 20,
+ .width = 9,
+ },
+ .ssen = {
+ .reg_off = ANACTRL_MPLL_CTRL7,
+ .shift = 29,
+ .width = 1,
+ },
+ .lock = &meson_clk_lock,
+ .init_regs = s4_mpll3_init_regs,
+ .init_count = ARRAY_SIZE(s4_mpll3_init_regs),
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll3_div",
+ .ops = &meson_clk_mpll_ops,
+ .parent_hws = (const struct clk_hw *[]) {
+ &s4_mpll_prediv.hw
+ },
+ .num_parents = 1,
+ },
+};
+
+static struct clk_regmap s4_mpll3 = {
+ .data = &(struct clk_regmap_gate_data){
+ .offset = ANACTRL_MPLL_CTRL7,
+ .bit_idx = 31,
+ },
+ .hw.init = &(struct clk_init_data){
+ .name = "mpll3",
+ .ops = &clk_regmap_gate_ops,
+ .parent_hws = (const struct clk_hw *[]) { &s4_mpll3_div.hw },
+ .num_parents = 1,
+ .flags = CLK_SET_RATE_PARENT,
+ },
+};
+
+/* Array of all clocks provided by this provider */
+static struct clk_hw_onecell_data s4_pll_hw_onecell_data = {
+ .hws = {
+ [CLKID_FIXED_PLL_DCO] = &s4_fixed_pll_dco.hw,
+ [CLKID_FIXED_PLL] = &s4_fixed_pll.hw,
+ [CLKID_FCLK_DIV2_DIV] = &s4_fclk_div2_div.hw,
+ [CLKID_FCLK_DIV2] = &s4_fclk_div2.hw,
+ [CLKID_FCLK_DIV3_DIV] = &s4_fclk_div3_div.hw,
+ [CLKID_FCLK_DIV3] = &s4_fclk_div3.hw,
+ [CLKID_FCLK_DIV4_DIV] = &s4_fclk_div4_div.hw,
+ [CLKID_FCLK_DIV4] = &s4_fclk_div4.hw,
+ [CLKID_FCLK_DIV5_DIV] = &s4_fclk_div5_div.hw,
+ [CLKID_FCLK_DIV5] = &s4_fclk_div5.hw,
+ [CLKID_FCLK_DIV7_DIV] = &s4_fclk_div7_div.hw,
+ [CLKID_FCLK_DIV7] = &s4_fclk_div7.hw,
+ [CLKID_FCLK_DIV2P5_DIV] = &s4_fclk_div2p5_div.hw,
+ [CLKID_FCLK_DIV2P5] = &s4_fclk_div2p5.hw,
+ [CLKID_GP0_PLL_DCO] = &s4_gp0_pll_dco.hw,
+ [CLKID_GP0_PLL] = &s4_gp0_pll.hw,
+ [CLKID_HIFI_PLL_DCO] = &s4_hifi_pll_dco.hw,
+ [CLKID_HIFI_PLL] = &s4_hifi_pll.hw,
+ [CLKID_HDMI_PLL_DCO] = &s4_hdmi_pll_dco.hw,
+ [CLKID_HDMI_PLL_OD] = &s4_hdmi_pll_od.hw,
+ [CLKID_HDMI_PLL] = &s4_hdmi_pll.hw,
+ [CLKID_MPLL_50M_DIV] = &s4_mpll_50m_div.hw,
+ [CLKID_MPLL_50M] = &s4_mpll_50m.hw,
+ [CLKID_MPLL_PREDIV] = &s4_mpll_prediv.hw,
+ [CLKID_MPLL0_DIV] = &s4_mpll0_div.hw,
+ [CLKID_MPLL0] = &s4_mpll0.hw,
+ [CLKID_MPLL1_DIV] = &s4_mpll1_div.hw,
+ [CLKID_MPLL1] = &s4_mpll1.hw,
+ [CLKID_MPLL2_DIV] = &s4_mpll2_div.hw,
+ [CLKID_MPLL2] = &s4_mpll2.hw,
+ [CLKID_MPLL3_DIV] = &s4_mpll3_div.hw,
+ [CLKID_MPLL3] = &s4_mpll3.hw,
+
+ [NR_PLL_CLKS] = NULL
+ },
+ .num = NR_PLL_CLKS,
+};
+
+static struct clk_regmap *const s4_pll_clk_regmaps[] = {
+ &s4_fixed_pll_dco,
+ &s4_fixed_pll,
+ &s4_fclk_div2,
+ &s4_fclk_div3,
+ &s4_fclk_div4,
+ &s4_fclk_div5,
+ &s4_fclk_div7,
+ &s4_fclk_div2p5,
+ &s4_gp0_pll_dco,
+ &s4_gp0_pll,
+ &s4_hifi_pll_dco,
+ &s4_hifi_pll,
+ &s4_hdmi_pll_dco,
+ &s4_hdmi_pll_od,
+ &s4_hdmi_pll,
+ &s4_mpll_50m,
+ &s4_mpll0_div,
+ &s4_mpll0,
+ &s4_mpll1_div,
+ &s4_mpll1,
+ &s4_mpll2_div,
+ &s4_mpll2,
+ &s4_mpll3_div,
+ &s4_mpll3,
+};
+
+static const struct reg_sequence s4_init_regs[] = {
+ { .reg = ANACTRL_MPLL_CTRL0, .def = 0x00000543 },
+};
+
+static struct regmap_config clkc_regmap_config = {
+ .reg_bits = 32,
+ .val_bits = 32,
+ .reg_stride = 4,
+};
+
+static int meson_s4_pll_probe(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct regmap *regmap;
+ void __iomem *base;
+ int ret, i;
+
+ base = devm_platform_ioremap_resource(pdev, 0);
+ if (IS_ERR(base))
+ return PTR_ERR(base);
+
+ regmap = devm_regmap_init_mmio(dev, base, &clkc_regmap_config);
+ if (IS_ERR(regmap))
+ return PTR_ERR(regmap);
+
+ ret = regmap_multi_reg_write(regmap, s4_init_regs, ARRAY_SIZE(s4_init_regs));
+ if (ret) {
+ dev_err(dev, "Failed to init registers\n");
+ return ret;
+ }
+
+ /* Populate regmap for the regmap backed clocks */
+ for (i = 0; i < ARRAY_SIZE(s4_pll_clk_regmaps); i++)
+ s4_pll_clk_regmaps[i]->map = regmap;
+
+ for (i = 0; i < s4_pll_hw_onecell_data.num; i++) {
+ /* array might be sparse */
+ if (!s4_pll_hw_onecell_data.hws[i])
+ continue;
+
+ ret = devm_clk_hw_register(dev, s4_pll_hw_onecell_data.hws[i]);
+ if (ret) {
+ dev_err(dev, "Clock registration failed\n");
+ return ret;
+ }
+ }
+
+ return devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
+ &s4_pll_hw_onecell_data);
+}
+
+static const struct of_device_id clkc_match_table[] = {
+ {
+ .compatible = "amlogic,s4-pll-clkc",
+ },
+ {}
+};
+
+static struct platform_driver s4_driver = {
+ .probe = meson_s4_pll_probe,
+ .driver = {
+ .name = "s4-pll-clkc",
+ .of_match_table = clkc_match_table,
+ },
+};
+
+module_platform_driver(s4_driver);
+MODULE_LICENSE("GPL");
diff --git a/drivers/clk/meson/s4-pll.h b/drivers/clk/meson/s4-pll.h
new file mode 100644
index 000000000000..41dc6de978c1
--- /dev/null
+++ b/drivers/clk/meson/s4-pll.h
@@ -0,0 +1,88 @@
+/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
+/*
+ * Copyright (c) 2021 Amlogic, inc.
+ * Author: Yu Tu <yu.tu@amlogic.com>
+ */
+
+#ifndef __MESON_S4_PLL_H__
+#define __MESON_S4_PLL_H__
+
+/* ANA_CTRL - Registers
+ * REG_BASE: REGISTER_BASE_ADDR = 0xfe008000
+ */
+#define ANACTRL_FIXPLL_CTRL0 (0x0010 << 2)
+#define ANACTRL_FIXPLL_CTRL1 (0x0011 << 2)
+#define ANACTRL_FIXPLL_CTRL2 (0x0012 << 2)
+#define ANACTRL_FIXPLL_CTRL3 (0x0013 << 2)
+#define ANACTRL_FIXPLL_CTRL4 (0x0014 << 2)
+#define ANACTRL_FIXPLL_CTRL5 (0x0015 << 2)
+#define ANACTRL_FIXPLL_CTRL6 (0x0016 << 2)
+#define ANACTRL_FIXPLL_STS (0x0017 << 2)
+#define ANACTRL_GP0PLL_CTRL0 (0x0020 << 2)
+#define ANACTRL_GP0PLL_CTRL1 (0x0021 << 2)
+#define ANACTRL_GP0PLL_CTRL2 (0x0022 << 2)
+#define ANACTRL_GP0PLL_CTRL3 (0x0023 << 2)
+#define ANACTRL_GP0PLL_CTRL4 (0x0024 << 2)
+#define ANACTRL_GP0PLL_CTRL5 (0x0025 << 2)
+#define ANACTRL_GP0PLL_CTRL6 (0x0026 << 2)
+#define ANACTRL_GP0PLL_STS (0x0027 << 2)
+#define ANACTRL_HIFIPLL_CTRL0 (0x0040 << 2)
+#define ANACTRL_HIFIPLL_CTRL1 (0x0041 << 2)
+#define ANACTRL_HIFIPLL_CTRL2 (0x0042 << 2)
+#define ANACTRL_HIFIPLL_CTRL3 (0x0043 << 2)
+#define ANACTRL_HIFIPLL_CTRL4 (0x0044 << 2)
+#define ANACTRL_HIFIPLL_CTRL5 (0x0045 << 2)
+#define ANACTRL_HIFIPLL_CTRL6 (0x0046 << 2)
+#define ANACTRL_HIFIPLL_STS (0x0047 << 2)
+#define ANACTRL_MPLL_CTRL0 (0x0060 << 2)
+#define ANACTRL_MPLL_CTRL1 (0x0061 << 2)
+#define ANACTRL_MPLL_CTRL2 (0x0062 << 2)
+#define ANACTRL_MPLL_CTRL3 (0x0063 << 2)
+#define ANACTRL_MPLL_CTRL4 (0x0064 << 2)
+#define ANACTRL_MPLL_CTRL5 (0x0065 << 2)
+#define ANACTRL_MPLL_CTRL6 (0x0066 << 2)
+#define ANACTRL_MPLL_CTRL7 (0x0067 << 2)
+#define ANACTRL_MPLL_CTRL8 (0x0068 << 2)
+#define ANACTRL_MPLL_STS (0x0069 << 2)
+#define ANACTRL_HDMIPLL_CTRL0 (0x0070 << 2)
+#define ANACTRL_HDMIPLL_CTRL1 (0x0071 << 2)
+#define ANACTRL_HDMIPLL_CTRL2 (0x0072 << 2)
+#define ANACTRL_HDMIPLL_CTRL3 (0x0073 << 2)
+#define ANACTRL_HDMIPLL_CTRL4 (0x0074 << 2)
+#define ANACTRL_HDMIPLL_CTRL5 (0x0075 << 2)
+#define ANACTRL_HDMIPLL_CTRL6 (0x0076 << 2)
+#define ANACTRL_HDMIPLL_STS (0x0077 << 2)
+#define ANACTRL_HDMIPLL_VLOCK (0x0079 << 2)
+
+/*
+ * CLKID index values
+ *
+ * These indices are entirely contrived and do not map onto the hardware.
+ * It has now been decided to expose everything by default in the DT header:
+ * include/dt-bindings/clock/axg-clkc.h. Only the clocks ids we don't want
+ * to expose, such as the internal muxes and dividers of composite clocks,
+ * will remain defined here.
+ */
+#define CLKID_FIXED_PLL_DCO 0
+#define CLKID_FCLK_DIV2_DIV 2
+#define CLKID_FCLK_DIV3_DIV 4
+#define CLKID_FCLK_DIV4_DIV 6
+#define CLKID_FCLK_DIV5_DIV 8
+#define CLKID_FCLK_DIV7_DIV 10
+#define CLKID_FCLK_DIV2P5_DIV 12
+#define CLKID_GP0_PLL_DCO 14
+#define CLKID_HIFI_PLL_DCO 16
+#define CLKID_HDMI_PLL_DCO 18
+#define CLKID_HDMI_PLL_OD 19
+#define CLKID_MPLL_50M_DIV 21
+#define CLKID_MPLL_PREDIV 23
+#define CLKID_MPLL0_DIV 24
+#define CLKID_MPLL1_DIV 26
+#define CLKID_MPLL2_DIV 28
+#define CLKID_MPLL3_DIV 30
+
+#define NR_PLL_CLKS 32
+/* include the CLKIDs that have been made part of the DT binding */
+#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
+
+#endif /* __MESON_S4_PLL_H__ */
--
2.33.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 41+ messages in thread
* [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings
2022-08-05 8:57 [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller Yu Tu
` (2 preceding siblings ...)
2022-08-05 8:57 ` [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver Yu Tu
@ 2022-08-05 8:57 ` Yu Tu
2022-08-05 9:15 ` Krzysztof Kozlowski
2022-08-05 8:57 ` [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT Yu Tu
[not found] ` <20220805085716.5635-7-yu.tu@amlogic.com>
5 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 8:57 UTC (permalink / raw)
To: linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Cc: Yu Tu
Add peripheral clock controller compatible and dt-bindings header for
the of the S4 SoC.
Signed-off-by: Yu Tu <yu.tu@amlogic.com>
---
.../bindings/clock/amlogic,s4-clkc.yaml | 92 ++++++++++++
include/dt-bindings/clock/amlogic,s4-clkc.h | 131 ++++++++++++++++++
2 files changed, 223 insertions(+)
create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
create mode 100644 include/dt-bindings/clock/amlogic,s4-clkc.h
diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
new file mode 100644
index 000000000000..2471276afda9
--- /dev/null
+++ b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
@@ -0,0 +1,92 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/clock/amlogic,s4-clkc.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Amlogic Meson S serials Peripheral Clock Controller Device Tree Bindings
+
+maintainers:
+ - Neil Armstrong <narmstrong@baylibre.com>
+ - Jerome Brunet <jbrunet@baylibre.com>
+ - Yu Tu <yu.hu@amlogic.com>
+
+
+properties:
+ compatible:
+ const: amlogic,s4-periphs-clkc
+
+ reg:
+ maxItems: 1
+
+ clocks:
+ items:
+ - description: input fixed pll div2
+ - description: input fixed pll div2p5
+ - description: input fixed pll div3
+ - description: input fixed pll div4
+ - description: input fixed pll div5
+ - description: input fixed pll div7
+ - description: input hifi pll
+ - description: input gp0 pll
+ - description: input mpll0
+ - description: input mpll1
+ - description: input mpll2
+ - description: input mpll3
+ - description: input hdmi pll
+ - description: input oscillator (usually at 24MHz)
+
+ clock-names:
+ items:
+ - const: fclk_div2
+ - const: fclk_div2p5
+ - const: fclk_div3
+ - const: fclk_div4
+ - const: fclk_div5
+ - const: fclk_div7
+ - const: hifi_pll
+ - const: gp0_pll
+ - const: mpll0
+ - const: mpll1
+ - const: mpll2
+ - const: mpll3
+ - const: hdmi_pll
+ - const: xtal
+
+ "#clock-cells":
+ const: 1
+
+required:
+ - compatible
+ - reg
+ - clocks
+ - clock-names
+ - "#clock-cells"
+
+additionalProperties: false
+
+examples:
+ - |
+ clkc_periphs: periphs-clock-controller@fe000000 {
+ compatible = "amlogic,s4-periphs-clkc";
+ reg = <0xfe000000 0x49c>;
+ clocks = <&clkc_pll 3>,
+ <&clkc_pll 13>,
+ <&clkc_pll 5>,
+ <&clkc_pll 7>,
+ <&clkc_pll 9>,
+ <&clkc_pll 11>,
+ <&clkc_pll 17>,
+ <&clkc_pll 15>,
+ <&clkc_pll 25>,
+ <&clkc_pll 27>,
+ <&clkc_pll 29>,
+ <&clkc_pll 31>,
+ <&clkc_pll 20>,
+ <&xtal>;
+ clock-names = "fclk_div2", "fclk_div2p5", "fclk_div3", "fclk_div4",
+ "fclk_div5", "fclk_div7", "hifi_pll", "gp0_pll",
+ "mpll0", "mpll1", "mpll2", "mpll3", "hdmi_pll", "xtal";
+ #clock-cells = <1>;
+ };
+...
diff --git a/include/dt-bindings/clock/amlogic,s4-clkc.h b/include/dt-bindings/clock/amlogic,s4-clkc.h
new file mode 100644
index 000000000000..d203b9bbf29e
--- /dev/null
+++ b/include/dt-bindings/clock/amlogic,s4-clkc.h
@@ -0,0 +1,131 @@
+/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
+/*
+ * Copyright (c) 2021 Amlogic, Inc. All rights reserved.
+ * Author: Yu Tu <yu.tu@amlogic.com>
+ */
+
+#ifndef _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
+#define _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
+
+/*
+ * CLKID index values
+ */
+
+#define CLKID_RTC_CLK 1
+#define CLKID_SYS_CLK_B_GATE 2
+#define CLKID_SYS_CLK_A_GATE 3
+#define CLKID_SYS_CLK 4
+#define CLKID_CECA_32K_CLKOUT 5
+#define CLKID_CECB_32K_CLKOUT 6
+#define CLKID_SC_CLK_GATE 7
+#define CLKID_12_24M_CLK_SEL 8
+#define CLKID_VID_PLL 9
+#define CLKID_VCLK 10
+#define CLKID_VCLK2 11
+#define CLKID_VCLK_DIV1 12
+#define CLKID_VCLK2_DIV1 13
+#define CLKID_VCLK_DIV2 14
+#define CLKID_VCLK_DIV4 15
+#define CLKID_VCLK_DIV6 16
+#define CLKID_VCLK_DIV12 17
+#define CLKID_VCLK2_DIV2 18
+#define CLKID_VCLK2_DIV4 19
+#define CLKID_VCLK2_DIV6 20
+#define CLKID_VCLK2_DIV12 21
+#define CLKID_CTS_ENCI 22
+#define CLKID_CTS_ENCP 23
+#define CLKID_CTS_VDAC 24
+#define CLKID_HDMI 25
+#define CLKID_TS_CLK_GATE 26
+#define CLKID_MALI_0 27
+#define CLKID_MALI_1 28
+#define CLKID_MALI 29
+#define CLKID_VDEC_P0 30
+#define CLKID_VDEC_P1 31
+#define CLKID_VDEC_SEL 32
+#define CLKID_HEVCF_P0 33
+#define CLKID_HEVCF_P1 34
+#define CLKID_HEVCF_SEL 35
+#define CLKID_VPU_0 36
+#define CLKID_VPU_1 37
+#define CLKID_VPU 38
+#define CLKID_VPU_CLKB_TMP 39
+#define CLKID_VPU_CLKB 40
+#define CLKID_VPU_CLKC_P0 41
+#define CLKID_VPU_CLKC_P1 42
+#define CLKID_VPU_CLKC_SEL 43
+#define CLKID_VAPB_0 44
+#define CLKID_VAPB_1 45
+#define CLKID_VAPB 46
+#define CLKID_GE2D 47
+#define CLKID_VDIN_MEAS_GATE 48
+#define CLKID_SD_EMMC_C_CLK 49
+#define CLKID_SD_EMMC_A_CLK 50
+#define CLKID_SD_EMMC_B_CLK 51
+#define CLKID_SPICC0_GATE 52
+#define CLKID_PWM_A_GATE 53
+#define CLKID_PWM_B_GATE 54
+#define CLKID_PWM_C_GATE 55
+#define CLKID_PWM_D_GATE 56
+#define CLKID_PWM_E_GATE 57
+#define CLKID_PWM_F_GATE 58
+#define CLKID_PWM_G_GATE 59
+#define CLKID_PWM_H_GATE 60
+#define CLKID_PWM_I_GATE 61
+#define CLKID_PWM_J_GATE 62
+#define CLKID_SARADC_GATE 63
+#define CLKID_GEN_GATE 64
+#define CLKID_DDR 65
+#define CLKID_DOS 66
+#define CLKID_ETHPHY 67
+#define CLKID_MALI_GATE 68
+#define CLKID_AOCPU 69
+#define CLKID_AUCPU 70
+#define CLKID_CEC 71
+#define CLKID_SD_EMMC_A 72
+#define CLKID_SD_EMMC_B 73
+#define CLKID_NAND 74
+#define CLKID_SMARTCARD 75
+#define CLKID_ACODEC 76
+#define CLKID_SPIFC 77
+#define CLKID_MSR_CLK 78
+#define CLKID_IR_CTRL 79
+#define CLKID_AUDIO 80
+#define CLKID_ETH 81
+#define CLKID_UART_A 82
+#define CLKID_UART_B 83
+#define CLKID_UART_C 84
+#define CLKID_UART_D 85
+#define CLKID_UART_E 86
+#define CLKID_AIFIFO 87
+#define CLKID_TS_DDR 88
+#define CLKID_TS_PLL 89
+#define CLKID_G2D 90
+#define CLKID_SPICC0 91
+#define CLKID_SPICC1 92
+#define CLKID_USB 93
+#define CLKID_I2C_M_A 94
+#define CLKID_I2C_M_B 95
+#define CLKID_I2C_M_C 96
+#define CLKID_I2C_M_D 97
+#define CLKID_I2C_M_E 98
+#define CLKID_HDMITX_APB 99
+#define CLKID_I2C_S_A 100
+#define CLKID_USB1_TO_DDR 101
+#define CLKID_HDCP22 102
+#define CLKID_MMC_APB 103
+#define CLKID_RSA 104
+#define CLKID_CPU_DEBUG 105
+#define CLKID_VPU_INTR 106
+#define CLKID_DEMOD 107
+#define CLKID_SAR_ADC 108
+#define CLKID_GIC 109
+#define CLKID_PWM_AB 110
+#define CLKID_PWM_CD 111
+#define CLKID_PWM_EF 112
+#define CLKID_PWM_GH 113
+#define CLKID_PWM_IJ 114
+#define CLKID_HDCP22_ESMCLK_GATE 115
+#define CLKID_HDCP22_SKPCLK_GATE 116
+
+#endif /* _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H */
--
2.33.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 41+ messages in thread
* [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT
2022-08-05 8:57 [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller Yu Tu
` (3 preceding siblings ...)
2022-08-05 8:57 ` [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings Yu Tu
@ 2022-08-05 8:57 ` Yu Tu
2022-08-05 9:16 ` Krzysztof Kozlowski
[not found] ` <20220805085716.5635-7-yu.tu@amlogic.com>
5 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 8:57 UTC (permalink / raw)
To: linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Cc: Yu Tu
Added information about the S4 SOC Peripheral Clock controller in DT.
Signed-off-by: Yu Tu <yu.tu@amlogic.com>
---
arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 26 +++++++++++++++++++++++
1 file changed, 26 insertions(+)
diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
index a816b1f7694b..71be1dda15a2 100644
--- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
+++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
@@ -6,6 +6,8 @@
#include <dt-bindings/interrupt-controller/irq.h>
#include <dt-bindings/interrupt-controller/arm-gic.h>
#include <dt-bindings/gpio/gpio.h>
+#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
+#include <dt-bindings/clock/amlogic,s4-clkc.h>
/ {
cpus {
@@ -100,6 +102,30 @@ clkc_pll: pll-clock-controller@8000 {
#clock-cells = <1>;
};
+ clkc_periphs: periphs-clock-controller {
+ compatible = "amlogic,s4-periphs-clkc";
+ reg = <0x0 0x0 0x0 0x49c>;
+ clocks = <&clkc_pll CLKID_FCLK_DIV2>,
+ <&clkc_pll CLKID_FCLK_DIV2P5>,
+ <&clkc_pll CLKID_FCLK_DIV3>,
+ <&clkc_pll CLKID_FCLK_DIV4>,
+ <&clkc_pll CLKID_FCLK_DIV5>,
+ <&clkc_pll CLKID_FCLK_DIV7>,
+ <&clkc_pll CLKID_HIFI_PLL>,
+ <&clkc_pll CLKID_GP0_PLL>,
+ <&clkc_pll CLKID_MPLL0>,
+ <&clkc_pll CLKID_MPLL1>,
+ <&clkc_pll CLKID_MPLL2>,
+ <&clkc_pll CLKID_MPLL3>,
+ <&clkc_pll CLKID_HDMI_PLL>,
+ <&xtal>;
+ clock-names = "fclk_div2", "fclk_div2p5", "fclk_div3",
+ "fclk_div4", "fclk_div5", "fclk_div7",
+ "hifi_pll", "gp0_pll", "mpll0", "mpll1",
+ "mpll2", "mpll3", "hdmi_pll", "xtal";
+ #clock-cells = <1>;
+ };
+
periphs_pinctrl: pinctrl@4000 {
compatible = "amlogic,meson-s4-periphs-pinctrl";
#address-cells = <2>;
--
2.33.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 41+ messages in thread
* Re: [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings
2022-08-05 8:57 ` [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings Yu Tu
@ 2022-08-05 9:13 ` Krzysztof Kozlowski
0 siblings, 0 replies; 41+ messages in thread
From: Krzysztof Kozlowski @ 2022-08-05 9:13 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 05/08/2022 10:57, Yu Tu wrote:
> Add the documentation to support Amlogic S4 SoC PLL clock driver and
> add S4 SoC PLL clock controller bindings.
>
> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
> ---
> .../bindings/clock/amlogic,s4-pll-clkc.yaml | 51 +++++++++++++++++++
> MAINTAINERS | 1 +
> .../dt-bindings/clock/amlogic,s4-pll-clkc.h | 30 +++++++++++
> 3 files changed, 82 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
> create mode 100644 include/dt-bindings/clock/amlogic,s4-pll-clkc.h
>
> diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
> new file mode 100644
> index 000000000000..079ae905b69e
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/clock/amlogic,s4-pll-clkc.yaml
> @@ -0,0 +1,51 @@
> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/clock/amlogic,s4-pll-clkc.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Amlogic Meson S serials PLL Clock Controller Device Tree Bindings
s/Device Tree Bindings//
With above:
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Best regards,
Krzysztof
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings
2022-08-05 8:57 ` [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings Yu Tu
@ 2022-08-05 9:15 ` Krzysztof Kozlowski
2022-08-05 9:33 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Krzysztof Kozlowski @ 2022-08-05 9:15 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 05/08/2022 10:57, Yu Tu wrote:
> Add peripheral clock controller compatible and dt-bindings header for
> the of the S4 SoC.
>
> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
> ---
> .../bindings/clock/amlogic,s4-clkc.yaml | 92 ++++++++++++
> include/dt-bindings/clock/amlogic,s4-clkc.h | 131 ++++++++++++++++++
> 2 files changed, 223 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
> create mode 100644 include/dt-bindings/clock/amlogic,s4-clkc.h
>
> diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
> new file mode 100644
> index 000000000000..2471276afda9
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
Filename should be based on compatible, so amlogic,s4-periphs-clkc.yaml
> @@ -0,0 +1,92 @@
> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/clock/amlogic,s4-clkc.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Amlogic Meson S serials Peripheral Clock Controller Device Tree Bindings
s/Device Tree Bindings//
> +
> +maintainers:
> + - Neil Armstrong <narmstrong@baylibre.com>
> + - Jerome Brunet <jbrunet@baylibre.com>
> + - Yu Tu <yu.hu@amlogic.com>
> +
> +
> +properties:
> + compatible:
> + const: amlogic,s4-periphs-clkc
> +
> + reg:
> + maxItems: 1
> +
> + clocks:
> + items:
> + - description: input fixed pll div2
> + - description: input fixed pll div2p5
> + - description: input fixed pll div3
> + - description: input fixed pll div4
> + - description: input fixed pll div5
> + - description: input fixed pll div7
> + - description: input hifi pll
> + - description: input gp0 pll
> + - description: input mpll0
> + - description: input mpll1
> + - description: input mpll2
> + - description: input mpll3
> + - description: input hdmi pll
> + - description: input oscillator (usually at 24MHz)
> +
> + clock-names:
> + items:
> + - const: fclk_div2
> + - const: fclk_div2p5
> + - const: fclk_div3
> + - const: fclk_div4
> + - const: fclk_div5
> + - const: fclk_div7
> + - const: hifi_pll
> + - const: gp0_pll
> + - const: mpll0
> + - const: mpll1
> + - const: mpll2
> + - const: mpll3
> + - const: hdmi_pll
> + - const: xtal
> +
> + "#clock-cells":
> + const: 1
> +
> +required:
> + - compatible
> + - reg
> + - clocks
> + - clock-names
> + - "#clock-cells"
> +
> +additionalProperties: false
> +
> +examples:
> + - |
> + clkc_periphs: periphs-clock-controller@fe000000 {
> + compatible = "amlogic,s4-periphs-clkc";
> + reg = <0xfe000000 0x49c>;
> + clocks = <&clkc_pll 3>,
> + <&clkc_pll 13>,
> + <&clkc_pll 5>,
> + <&clkc_pll 7>,
> + <&clkc_pll 9>,
> + <&clkc_pll 11>,
> + <&clkc_pll 17>,
> + <&clkc_pll 15>,
> + <&clkc_pll 25>,
> + <&clkc_pll 27>,
> + <&clkc_pll 29>,
> + <&clkc_pll 31>,
> + <&clkc_pll 20>,
> + <&xtal>;
> + clock-names = "fclk_div2", "fclk_div2p5", "fclk_div3", "fclk_div4",
> + "fclk_div5", "fclk_div7", "hifi_pll", "gp0_pll",
> + "mpll0", "mpll1", "mpll2", "mpll3", "hdmi_pll", "xtal";
> + #clock-cells = <1>;
> + };
> +...
> diff --git a/include/dt-bindings/clock/amlogic,s4-clkc.h b/include/dt-bindings/clock/amlogic,s4-clkc.h
> new file mode 100644
> index 000000000000..d203b9bbf29e
> --- /dev/null
> +++ b/include/dt-bindings/clock/amlogic,s4-clkc.h
Probably this should be then amlogic,s4-periphs-clkc.h
> @@ -0,0 +1,131 @@
> +/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
> +/*
> + * Copyright (c) 2021 Amlogic, Inc. All rights reserved.
> + * Author: Yu Tu <yu.tu@amlogic.com>
> + */
> +
> +#ifndef _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
> +#define _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
Best regards,
Krzysztof
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT
2022-08-05 8:57 ` [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT Yu Tu
@ 2022-08-05 9:16 ` Krzysztof Kozlowski
2022-08-05 9:36 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Krzysztof Kozlowski @ 2022-08-05 9:16 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 05/08/2022 10:57, Yu Tu wrote:
> Added information about the S4 SOC Peripheral Clock controller in DT.
>
> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
> ---
> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 26 +++++++++++++++++++++++
> 1 file changed, 26 insertions(+)
>
> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
> index a816b1f7694b..71be1dda15a2 100644
> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
> @@ -6,6 +6,8 @@
> #include <dt-bindings/interrupt-controller/irq.h>
> #include <dt-bindings/interrupt-controller/arm-gic.h>
> #include <dt-bindings/gpio/gpio.h>
> +#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
> +#include <dt-bindings/clock/amlogic,s4-clkc.h>
>
> / {
> cpus {
> @@ -100,6 +102,30 @@ clkc_pll: pll-clock-controller@8000 {
> #clock-cells = <1>;
> };
>
> + clkc_periphs: periphs-clock-controller {
Node names should be generic, so "clock-controller"
https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
You miss here unit address. Test your DTS with dtbs check and with
regular compile with W=1.
Best regards,
Krzysztof
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-05 8:57 ` [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT Yu Tu
@ 2022-08-05 9:16 ` Krzysztof Kozlowski
2022-08-05 9:39 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Krzysztof Kozlowski @ 2022-08-05 9:16 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 05/08/2022 10:57, Yu Tu wrote:
> Added information about the S4 SOC PLL Clock controller in DT.
>
> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
> ---
> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
> index ff213618a598..a816b1f7694b 100644
> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
> #size-cells = <2>;
> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>
> + clkc_pll: pll-clock-controller@8000 {
Node names should be generic - clock-controller.
https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
Best regards,
Krzysztof
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings
2022-08-05 9:15 ` Krzysztof Kozlowski
@ 2022-08-05 9:33 ` Yu Tu
2022-08-08 6:16 ` Krzysztof Kozlowski
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 9:33 UTC (permalink / raw)
To: Krzysztof Kozlowski, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Jerome Brunet, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Krzysztof,
Thank you for your reply.
On 2022/8/5 17:15, Krzysztof Kozlowski wrote:
> [ EXTERNAL EMAIL ]
>
> On 05/08/2022 10:57, Yu Tu wrote:
>> Add peripheral clock controller compatible and dt-bindings header for
>> the of the S4 SoC.
>>
>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>> ---
>> .../bindings/clock/amlogic,s4-clkc.yaml | 92 ++++++++++++
>> include/dt-bindings/clock/amlogic,s4-clkc.h | 131 ++++++++++++++++++
>> 2 files changed, 223 insertions(+)
>> create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>> create mode 100644 include/dt-bindings/clock/amlogic,s4-clkc.h
>>
>> diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>> new file mode 100644
>> index 000000000000..2471276afda9
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>
> Filename should be based on compatible, so amlogic,s4-periphs-clkc.yaml
I will correct as you suggest in the next version.
>
>> @@ -0,0 +1,92 @@
>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>> +%YAML 1.2
>> +---
>> +$id: http://devicetree.org/schemas/clock/amlogic,s4-clkc.yaml#
>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> +
>> +title: Amlogic Meson S serials Peripheral Clock Controller Device Tree Bindings
>
> s/Device Tree Bindings//
what's mean?
>
>> +
>> +maintainers:
>> + - Neil Armstrong <narmstrong@baylibre.com>
>> + - Jerome Brunet <jbrunet@baylibre.com>
>> + - Yu Tu <yu.hu@amlogic.com>
>> +
>> +
>> +properties:
>> + compatible:
>> + const: amlogic,s4-periphs-clkc
>> +
>> + reg:
>> + maxItems: 1
>> +
>> + clocks:
>> + items:
>> + - description: input fixed pll div2
>> + - description: input fixed pll div2p5
>> + - description: input fixed pll div3
>> + - description: input fixed pll div4
>> + - description: input fixed pll div5
>> + - description: input fixed pll div7
>> + - description: input hifi pll
>> + - description: input gp0 pll
>> + - description: input mpll0
>> + - description: input mpll1
>> + - description: input mpll2
>> + - description: input mpll3
>> + - description: input hdmi pll
>> + - description: input oscillator (usually at 24MHz)
>> +
>> + clock-names:
>> + items:
>> + - const: fclk_div2
>> + - const: fclk_div2p5
>> + - const: fclk_div3
>> + - const: fclk_div4
>> + - const: fclk_div5
>> + - const: fclk_div7
>> + - const: hifi_pll
>> + - const: gp0_pll
>> + - const: mpll0
>> + - const: mpll1
>> + - const: mpll2
>> + - const: mpll3
>> + - const: hdmi_pll
>> + - const: xtal
>> +
>> + "#clock-cells":
>> + const: 1
>> +
>> +required:
>> + - compatible
>> + - reg
>> + - clocks
>> + - clock-names
>> + - "#clock-cells"
>> +
>> +additionalProperties: false
>> +
>> +examples:
>> + - |
>> + clkc_periphs: periphs-clock-controller@fe000000 {
>> + compatible = "amlogic,s4-periphs-clkc";
>> + reg = <0xfe000000 0x49c>;
>> + clocks = <&clkc_pll 3>,
>> + <&clkc_pll 13>,
>> + <&clkc_pll 5>,
>> + <&clkc_pll 7>,
>> + <&clkc_pll 9>,
>> + <&clkc_pll 11>,
>> + <&clkc_pll 17>,
>> + <&clkc_pll 15>,
>> + <&clkc_pll 25>,
>> + <&clkc_pll 27>,
>> + <&clkc_pll 29>,
>> + <&clkc_pll 31>,
>> + <&clkc_pll 20>,
>> + <&xtal>;
>> + clock-names = "fclk_div2", "fclk_div2p5", "fclk_div3", "fclk_div4",
>> + "fclk_div5", "fclk_div7", "hifi_pll", "gp0_pll",
>> + "mpll0", "mpll1", "mpll2", "mpll3", "hdmi_pll", "xtal";
>> + #clock-cells = <1>;
>> + };
>> +...
>> diff --git a/include/dt-bindings/clock/amlogic,s4-clkc.h b/include/dt-bindings/clock/amlogic,s4-clkc.h
>> new file mode 100644
>> index 000000000000..d203b9bbf29e
>> --- /dev/null
>> +++ b/include/dt-bindings/clock/amlogic,s4-clkc.h
>
> Probably this should be then amlogic,s4-periphs-clkc.h
I will correct as you suggest in the next version.
>
>> @@ -0,0 +1,131 @@
>> +/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
>> +/*
>> + * Copyright (c) 2021 Amlogic, Inc. All rights reserved.
>> + * Author: Yu Tu <yu.tu@amlogic.com>
>> + */
>> +
>> +#ifndef _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
>> +#define _DT_BINDINGS_CLOCK_AMLOGIC_S4_CLKC_H
>
> Best regards,
> Krzysztof
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT
2022-08-05 9:16 ` Krzysztof Kozlowski
@ 2022-08-05 9:36 ` Yu Tu
2022-08-08 6:17 ` Krzysztof Kozlowski
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 9:36 UTC (permalink / raw)
To: Krzysztof Kozlowski, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Jerome Brunet, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Krzysztof,
Thank you for your reply.
On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
> [ EXTERNAL EMAIL ]
>
> On 05/08/2022 10:57, Yu Tu wrote:
>> Added information about the S4 SOC Peripheral Clock controller in DT.
>>
>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>> ---
>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 26 +++++++++++++++++++++++
>> 1 file changed, 26 insertions(+)
>>
>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>> index a816b1f7694b..71be1dda15a2 100644
>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>> @@ -6,6 +6,8 @@
>> #include <dt-bindings/interrupt-controller/irq.h>
>> #include <dt-bindings/interrupt-controller/arm-gic.h>
>> #include <dt-bindings/gpio/gpio.h>
>> +#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
>> +#include <dt-bindings/clock/amlogic,s4-clkc.h>
>>
>> / {
>> cpus {
>> @@ -100,6 +102,30 @@ clkc_pll: pll-clock-controller@8000 {
>> #clock-cells = <1>;
>> };
>>
>> + clkc_periphs: periphs-clock-controller {
>
> Node names should be generic, so "clock-controller"
> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>
> You miss here unit address. Test your DTS with dtbs check and with
> regular compile with W=1.
I will change to clkc_periphs: clock-controller@0 {.
Is that okay?
>
>
> Best regards,
> Krzysztof
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-05 9:16 ` Krzysztof Kozlowski
@ 2022-08-05 9:39 ` Yu Tu
2022-08-10 13:32 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-05 9:39 UTC (permalink / raw)
To: Krzysztof Kozlowski, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Jerome Brunet, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Krzysztof,
Thank you for your reply.
On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
> [ EXTERNAL EMAIL ]
>
> On 05/08/2022 10:57, Yu Tu wrote:
>> Added information about the S4 SOC PLL Clock controller in DT.
>>
>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>> ---
>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>> 1 file changed, 8 insertions(+)
>>
>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>> index ff213618a598..a816b1f7694b 100644
>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>> #size-cells = <2>;
>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>
>> + clkc_pll: pll-clock-controller@8000 {
>
> Node names should be generic - clock-controller.
> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>
I will change to clkc_pll: clock-controller@8000, in next version.
>
> Best regards,
> Krzysztof
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings
2022-08-05 9:33 ` Yu Tu
@ 2022-08-08 6:16 ` Krzysztof Kozlowski
2022-08-08 10:00 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Krzysztof Kozlowski @ 2022-08-08 6:16 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 05/08/2022 11:33, Yu Tu wrote:
> Hi Krzysztof,
> Thank you for your reply.
>
> On 2022/8/5 17:15, Krzysztof Kozlowski wrote:
>> [ EXTERNAL EMAIL ]
>>
>> On 05/08/2022 10:57, Yu Tu wrote:
>>> Add peripheral clock controller compatible and dt-bindings header for
>>> the of the S4 SoC.
>>>
>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>> ---
>>> .../bindings/clock/amlogic,s4-clkc.yaml | 92 ++++++++++++
>>> include/dt-bindings/clock/amlogic,s4-clkc.h | 131 ++++++++++++++++++
>>> 2 files changed, 223 insertions(+)
>>> create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>>> create mode 100644 include/dt-bindings/clock/amlogic,s4-clkc.h
>>>
>>> diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>>> new file mode 100644
>>> index 000000000000..2471276afda9
>>> --- /dev/null
>>> +++ b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>>
>> Filename should be based on compatible, so amlogic,s4-periphs-clkc.yaml
> I will correct as you suggest in the next version.
>>
>>> @@ -0,0 +1,92 @@
>>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>>> +%YAML 1.2
>>> +---
>>> +$id: http://devicetree.org/schemas/clock/amlogic,s4-clkc.yaml#
>>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>>> +
>>> +title: Amlogic Meson S serials Peripheral Clock Controller Device Tree Bindings
>>
>> s/Device Tree Bindings//
> what's mean?
It's a expression (simplified not accurate) from many Linux tools, e.g.
sed, vim etc
s/SEARCH/REPLACE/
so just drop that piece I wrote above.
Best regards,
Krzysztof
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT
2022-08-05 9:36 ` Yu Tu
@ 2022-08-08 6:17 ` Krzysztof Kozlowski
2022-08-08 10:02 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Krzysztof Kozlowski @ 2022-08-08 6:17 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Jerome Brunet,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 05/08/2022 11:36, Yu Tu wrote:
>>> @@ -100,6 +102,30 @@ clkc_pll: pll-clock-controller@8000 {
>>> #clock-cells = <1>;
>>> };
>>>
>>> + clkc_periphs: periphs-clock-controller {
>>
>> Node names should be generic, so "clock-controller"
>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>
>> You miss here unit address. Test your DTS with dtbs check and with
>> regular compile with W=1.
> I will change to clkc_periphs: clock-controller@0 {.
> Is that okay?
I cut the context, so if "0" was the value in reg, then it is OK.
Best regards,
Krzysztof
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings
2022-08-08 6:16 ` Krzysztof Kozlowski
@ 2022-08-08 10:00 ` Yu Tu
0 siblings, 0 replies; 41+ messages in thread
From: Yu Tu @ 2022-08-08 10:00 UTC (permalink / raw)
To: Krzysztof Kozlowski, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Jerome Brunet, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Krzysztof,
On 2022/8/8 14:16, Krzysztof Kozlowski wrote:
> [ EXTERNAL EMAIL ]
>
> On 05/08/2022 11:33, Yu Tu wrote:
>> Hi Krzysztof,
>> Thank you for your reply.
>>
>> On 2022/8/5 17:15, Krzysztof Kozlowski wrote:
>>> [ EXTERNAL EMAIL ]
>>>
>>> On 05/08/2022 10:57, Yu Tu wrote:
>>>> Add peripheral clock controller compatible and dt-bindings header for
>>>> the of the S4 SoC.
>>>>
>>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>>> ---
>>>> .../bindings/clock/amlogic,s4-clkc.yaml | 92 ++++++++++++
>>>> include/dt-bindings/clock/amlogic,s4-clkc.h | 131 ++++++++++++++++++
>>>> 2 files changed, 223 insertions(+)
>>>> create mode 100644 Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>>>> create mode 100644 include/dt-bindings/clock/amlogic,s4-clkc.h
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>>>> new file mode 100644
>>>> index 000000000000..2471276afda9
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/clock/amlogic,s4-clkc.yaml
>>>
>>> Filename should be based on compatible, so amlogic,s4-periphs-clkc.yaml
>> I will correct as you suggest in the next version.
>>>
>>>> @@ -0,0 +1,92 @@
>>>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>>>> +%YAML 1.2
>>>> +---
>>>> +$id: http://devicetree.org/schemas/clock/amlogic,s4-clkc.yaml#
>>>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>>>> +
>>>> +title: Amlogic Meson S serials Peripheral Clock Controller Device Tree Bindings
>>>
>>> s/Device Tree Bindings//
>> what's mean?
>
> It's a expression (simplified not accurate) from many Linux tools, e.g.
> sed, vim etc
> s/SEARCH/REPLACE/
> so just drop that piece I wrote above.
Do you mean you need me to delete "Device Tree Bindings"?
>
> Best regards,
> Krzysztof
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT
2022-08-08 6:17 ` Krzysztof Kozlowski
@ 2022-08-08 10:02 ` Yu Tu
0 siblings, 0 replies; 41+ messages in thread
From: Yu Tu @ 2022-08-08 10:02 UTC (permalink / raw)
To: Krzysztof Kozlowski, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Jerome Brunet, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/8 14:17, Krzysztof Kozlowski wrote:
> [ EXTERNAL EMAIL ]
>
> On 05/08/2022 11:36, Yu Tu wrote:
>>>> @@ -100,6 +102,30 @@ clkc_pll: pll-clock-controller@8000 {
>>>> #clock-cells = <1>;
>>>> };
>>>>
>>>> + clkc_periphs: periphs-clock-controller {
>>>
>>> Node names should be generic, so "clock-controller"
>>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>>
>>> You miss here unit address. Test your DTS with dtbs check and with
>>> regular compile with W=1.
>> I will change to clkc_periphs: clock-controller@0 {.
>> Is that okay?
>
> I cut the context, so if "0" was the value in reg, then it is OK.
Ok, I got it.I will change in next version.
> Best regards,
> Krzysztof
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-05 9:39 ` Yu Tu
@ 2022-08-10 13:32 ` Jerome Brunet
2022-08-15 6:17 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-10 13:32 UTC (permalink / raw)
To: Yu Tu, Krzysztof Kozlowski, linux-clk, linux-arm-kernel,
linux-amlogic, linux-kernel, devicetree, Rob Herring,
Neil Armstrong, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On Fri 05 Aug 2022 at 17:39, Yu Tu <yu.tu@amlogic.com> wrote:
> Hi Krzysztof,
> Thank you for your reply.
>
> On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
>> [ EXTERNAL EMAIL ]
>> On 05/08/2022 10:57, Yu Tu wrote:
>>> Added information about the S4 SOC PLL Clock controller in DT.
>>>
>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>> ---
>>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>>> 1 file changed, 8 insertions(+)
>>>
>>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>> index ff213618a598..a816b1f7694b 100644
>>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>>> #size-cells = <2>;
>>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>> + clkc_pll: pll-clock-controller@8000 {
>> Node names should be generic - clock-controller.
>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>
> I will change to clkc_pll: clock-controller@8000, in next version.
Same comment applies to the binding doc.
Also it would be nice to split this in two series.
Bindings and drivers in one, arm64 dt in the other. These changes goes
in through different trees.
>> Best regards,
>> Krzysztof
>> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-05 8:57 ` [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver Yu Tu
@ 2022-08-10 13:47 ` Jerome Brunet
2022-08-15 6:34 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-10 13:47 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Fri 05 Aug 2022 at 16:57, Yu Tu <yu.tu@amlogic.com> wrote:
> Add the S4 PLL clock controller found in the s4 SoC family.
>
> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
> ---
> drivers/clk/meson/Kconfig | 12 +
> drivers/clk/meson/Makefile | 1 +
> drivers/clk/meson/s4-pll.c | 891 +++++++++++++++++++++++++++++++++++++
> drivers/clk/meson/s4-pll.h | 88 ++++
> 4 files changed, 992 insertions(+)
> create mode 100644 drivers/clk/meson/s4-pll.c
> create mode 100644 drivers/clk/meson/s4-pll.h
>
> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
> index fc002c155bc3..f4244edc7b28 100644
> --- a/drivers/clk/meson/Kconfig
> +++ b/drivers/clk/meson/Kconfig
> @@ -115,4 +115,16 @@ config COMMON_CLK_G12A
> help
> Support for the clock controller on Amlogic S905D2, S905X2 and S905Y2
> devices, aka g12a. Say Y if you want peripherals to work.
> +
> +config COMMON_CLK_S4_PLL
> + tristate "S4 SoC PLL clock controllers support"
> + depends on ARM64
> + default y
> + select COMMON_CLK_MESON_MPLL
> + select COMMON_CLK_MESON_PLL
> + select COMMON_CLK_MESON_REGMAP
> + help
> + Support for the pll clock controller on Amlogic S805X2 and S905Y4 devices,
> + aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and AQ229.
> + Say Y if you want peripherals and CPU frequency scaling to work.
Looks like this doc is copy/paste of the peripheral controller.
This probably needs to be updated
> endmenu
> diff --git a/drivers/clk/meson/Makefile b/drivers/clk/meson/Makefile
> index 6eca2a406ee3..376f49cc13f1 100644
> --- a/drivers/clk/meson/Makefile
> +++ b/drivers/clk/meson/Makefile
> @@ -19,3 +19,4 @@ obj-$(CONFIG_COMMON_CLK_AXG_AUDIO) += axg-audio.o
> obj-$(CONFIG_COMMON_CLK_GXBB) += gxbb.o gxbb-aoclk.o
> obj-$(CONFIG_COMMON_CLK_G12A) += g12a.o g12a-aoclk.o
> obj-$(CONFIG_COMMON_CLK_MESON8B) += meson8b.o meson8-ddr.o
> +obj-$(CONFIG_COMMON_CLK_S4_PLL) += s4-pll.o
> diff --git a/drivers/clk/meson/s4-pll.c b/drivers/clk/meson/s4-pll.c
> new file mode 100644
> index 000000000000..478c78b5ed46
> --- /dev/null
> +++ b/drivers/clk/meson/s4-pll.c
> @@ -0,0 +1,891 @@
> +// SPDX-License-Identifier: GPL-2.0+
> +/*
> + * Amlogic Meson-S4 Clock Controller Driver
> + *
> + * Copyright (c) 2021 Amlogic, inc.
> + * Author: Yu Tu <yu.tu@amlogic.com>
> + */
> +
> +#include <linux/clk-provider.h>
> +#include <linux/of_device.h>
> +#include <linux/platform_device.h>
> +
> +#include "clk-mpll.h"
> +#include "clk-pll.h"
> +#include "clk-regmap.h"
> +#include "s4-pll.h"
> +
> +static DEFINE_SPINLOCK(meson_clk_lock);
> +
> +static struct clk_regmap s4_fixed_pll_dco = {
> + .data = &(struct meson_clk_pll_data){
> + .en = {
> + .reg_off = ANACTRL_FIXPLL_CTRL0,
> + .shift = 28,
> + .width = 1,
> + },
> + .m = {
> + .reg_off = ANACTRL_FIXPLL_CTRL0,
> + .shift = 0,
> + .width = 8,
> + },
> + .n = {
> + .reg_off = ANACTRL_FIXPLL_CTRL0,
> + .shift = 10,
> + .width = 5,
> + },
> + .frac = {
> + .reg_off = ANACTRL_FIXPLL_CTRL1,
> + .shift = 0,
> + .width = 17,
> + },
> + .l = {
> + .reg_off = ANACTRL_FIXPLL_CTRL0,
> + .shift = 31,
> + .width = 1,
> + },
> + .rst = {
> + .reg_off = ANACTRL_FIXPLL_CTRL0,
> + .shift = 29,
> + .width = 1,
> + },
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fixed_pll_dco",
> + .ops = &meson_clk_pll_ro_ops,
> + .parent_data = (const struct clk_parent_data []) {
> + { .fw_name = "xtal", }
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fixed_pll = {
> + .data = &(struct clk_regmap_div_data){
> + .offset = ANACTRL_FIXPLL_CTRL0,
> + .shift = 16,
> + .width = 2,
> + .flags = CLK_DIVIDER_POWER_OF_TWO,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fixed_pll",
> + .ops = &clk_regmap_divider_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fixed_pll_dco.hw
> + },
> + .num_parents = 1,
> + /*
> + * This clock won't ever change at runtime so
> + * CLK_SET_RATE_PARENT is not required
> + */
> + },
> +};
> +
> +static struct clk_fixed_factor s4_fclk_div2_div = {
> + .mult = 1,
> + .div = 2,
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div2_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fclk_div2 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_FIXPLL_CTRL1,
> + .bit_idx = 24,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div2",
> + .ops = &clk_regmap_gate_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fclk_div2_div.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_fclk_div3_div = {
> + .mult = 1,
> + .div = 3,
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div3_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fclk_div3 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_FIXPLL_CTRL1,
> + .bit_idx = 20,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div3",
> + .ops = &clk_regmap_gate_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fclk_div3_div.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_fclk_div4_div = {
> + .mult = 1,
> + .div = 4,
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div4_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fclk_div4 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_FIXPLL_CTRL1,
> + .bit_idx = 21,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div4",
> + .ops = &clk_regmap_gate_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fclk_div4_div.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_fclk_div5_div = {
> + .mult = 1,
> + .div = 5,
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div5_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fclk_div5 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_FIXPLL_CTRL1,
> + .bit_idx = 22,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div5",
> + .ops = &clk_regmap_gate_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fclk_div5_div.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_fclk_div7_div = {
> + .mult = 1,
> + .div = 7,
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div7_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fclk_div7 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_FIXPLL_CTRL1,
> + .bit_idx = 23,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div7",
> + .ops = &clk_regmap_gate_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fclk_div7_div.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_fclk_div2p5_div = {
> + .mult = 2,
> + .div = 5,
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div2p5_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fixed_pll.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_fclk_div2p5 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_FIXPLL_CTRL1,
> + .bit_idx = 25,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "fclk_div2p5",
> + .ops = &clk_regmap_gate_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fclk_div2p5_div.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static const struct pll_mult_range s4_gp0_pll_mult_range = {
> + .min = 125,
> + .max = 250,
> +};
> +
> +/*
> + * Internal gp0 pll emulation configuration parameters
> + */
> +static const struct reg_sequence s4_gp0_init_regs[] = {
> + { .reg = ANACTRL_GP0PLL_CTRL1, .def = 0x00000000 },
> + { .reg = ANACTRL_GP0PLL_CTRL2, .def = 0x00000000 },
> + { .reg = ANACTRL_GP0PLL_CTRL3, .def = 0x48681c00 },
> + { .reg = ANACTRL_GP0PLL_CTRL4, .def = 0x88770290 },
> + { .reg = ANACTRL_GP0PLL_CTRL5, .def = 0x39272000 },
> + { .reg = ANACTRL_GP0PLL_CTRL6, .def = 0x56540000 }
> +};
> +
> +static struct clk_regmap s4_gp0_pll_dco = {
> + .data = &(struct meson_clk_pll_data){
> + .en = {
> + .reg_off = ANACTRL_GP0PLL_CTRL0,
> + .shift = 28,
> + .width = 1,
> + },
> + .m = {
> + .reg_off = ANACTRL_GP0PLL_CTRL0,
> + .shift = 0,
> + .width = 8,
> + },
> + .n = {
> + .reg_off = ANACTRL_GP0PLL_CTRL0,
> + .shift = 10,
> + .width = 5,
> + },
> + .frac = {
> + .reg_off = ANACTRL_GP0PLL_CTRL1,
> + .shift = 0,
> + .width = 17,
> + },
> + .l = {
> + .reg_off = ANACTRL_GP0PLL_CTRL0,
> + .shift = 31,
> + .width = 1,
> + },
> + .rst = {
> + .reg_off = ANACTRL_GP0PLL_CTRL0,
> + .shift = 29,
> + .width = 1,
> + },
> + .range = &s4_gp0_pll_mult_range,
> + .init_regs = s4_gp0_init_regs,
> + .init_count = ARRAY_SIZE(s4_gp0_init_regs),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "gp0_pll_dco",
> + .ops = &meson_clk_pll_ops,
> + .parent_data = (const struct clk_parent_data []) {
> + { .fw_name = "xtal", }
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_gp0_pll = {
> + .data = &(struct clk_regmap_div_data){
> + .offset = ANACTRL_GP0PLL_CTRL0,
> + .shift = 16,
> + .width = 3,
> + .flags = (CLK_DIVIDER_POWER_OF_TWO |
> + CLK_DIVIDER_ROUND_CLOSEST),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "gp0_pll",
> + .ops = &clk_regmap_divider_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_gp0_pll_dco.hw
> + },
> + .num_parents = 1,
> + .flags = CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +/*
> + * Internal hifi pll emulation configuration parameters
> + */
> +static const struct reg_sequence s4_hifi_init_regs[] = {
> + { .reg = ANACTRL_HIFIPLL_CTRL1, .def = 0x00010e56 },
> + { .reg = ANACTRL_HIFIPLL_CTRL2, .def = 0x00000000 },
> + { .reg = ANACTRL_HIFIPLL_CTRL3, .def = 0x6a285c00 },
> + { .reg = ANACTRL_HIFIPLL_CTRL4, .def = 0x65771290 },
> + { .reg = ANACTRL_HIFIPLL_CTRL5, .def = 0x39272000 },
> + { .reg = ANACTRL_HIFIPLL_CTRL6, .def = 0x56540000 }
> +};
> +
> +static struct clk_regmap s4_hifi_pll_dco = {
> + .data = &(struct meson_clk_pll_data){
> + .en = {
> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
> + .shift = 28,
> + .width = 1,
> + },
> + .m = {
> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
> + .shift = 0,
> + .width = 8,
> + },
> + .n = {
> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
> + .shift = 10,
> + .width = 5,
> + },
> + .frac = {
> + .reg_off = ANACTRL_HIFIPLL_CTRL1,
> + .shift = 0,
> + .width = 17,
> + },
> + .l = {
> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
> + .shift = 31,
> + .width = 1,
> + },
> + .rst = {
> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
> + .shift = 29,
> + .width = 1,
> + },
> + .range = &s4_gp0_pll_mult_range,
> + .init_regs = s4_hifi_init_regs,
> + .init_count = ARRAY_SIZE(s4_hifi_init_regs),
> + .flags = CLK_MESON_PLL_ROUND_CLOSEST,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "hifi_pll_dco",
> + .ops = &meson_clk_pll_ops,
> + .parent_data = (const struct clk_parent_data []) {
> + { .fw_name = "xtal", }
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_hifi_pll = {
> + .data = &(struct clk_regmap_div_data){
> + .offset = ANACTRL_HIFIPLL_CTRL0,
> + .shift = 16,
> + .width = 2,
> + .flags = (CLK_DIVIDER_POWER_OF_TWO |
> + CLK_DIVIDER_ROUND_CLOSEST),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "hifi_pll",
> + .ops = &clk_regmap_divider_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_hifi_pll_dco.hw
> + },
> + .num_parents = 1,
> + .flags = CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +static struct clk_regmap s4_hdmi_pll_dco = {
> + .data = &(struct meson_clk_pll_data){
> + .en = {
> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 28,
> + .width = 1,
> + },
> + .m = {
> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 0,
> + .width = 8,
> + },
> + .n = {
> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 10,
> + .width = 5,
> + },
> + .frac = {
> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
> + .shift = 0,
> + .width = 17,
> + },
> + .l = {
> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 31,
> + .width = 1,
> + },
> + .rst = {
> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 29,
> + .width = 1,
> + },
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "hdmi_pll_dco",
> + .ops = &meson_clk_pll_ro_ops,
> + .parent_data = (const struct clk_parent_data []) {
> + { .fw_name = "xtal", }
> + },
> + .num_parents = 1,
> + /*
> + * Display directly handle hdmi pll registers ATM, we need
> + * NOCACHE to keep our view of the clock as accurate as
> + * possible
> + */
Is it really ?
Given that HDMI support for the s4 is there yet, the
addresses have changes and the region is no longer a syscon, it is time
for the HDMI driver to get fixed.
> + .flags = CLK_GET_RATE_NOCACHE,
> + },
> +};
> +
> +static struct clk_regmap s4_hdmi_pll_od = {
> + .data = &(struct clk_regmap_div_data){
> + .offset = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 16,
> + .width = 4,
> + .flags = CLK_DIVIDER_POWER_OF_TWO,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "hdmi_pll_od",
> + .ops = &clk_regmap_divider_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_hdmi_pll_dco.hw
> + },
> + .num_parents = 1,
> + /*
> + * Display directly handle hdmi pll registers ATM, we need
> + * NOCACHE to keep our view of the clock as accurate as
> + * possible
> + */
> + .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +static struct clk_regmap s4_hdmi_pll = {
> + .data = &(struct clk_regmap_div_data){
> + .offset = ANACTRL_HDMIPLL_CTRL0,
> + .shift = 20,
> + .width = 2,
> + .flags = CLK_DIVIDER_POWER_OF_TWO,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "hdmi_pll",
> + .ops = &clk_regmap_divider_ro_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_hdmi_pll_od.hw
> + },
> + .num_parents = 1,
> + /*
> + * Display directly handle hdmi pll registers ATM, we need
> + * NOCACHE to keep our view of the clock as accurate as
> + * possible
> + */
> + .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_mpll_50m_div = {
> + .mult = 1,
> + .div = 80,
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll_50m_div",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fixed_pll_dco.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_mpll_50m = {
> + .data = &(struct clk_regmap_mux_data){
> + .offset = ANACTRL_FIXPLL_CTRL3,
> + .mask = 0x1,
> + .shift = 5,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll_50m",
> + .ops = &clk_regmap_mux_ro_ops,
> + .parent_data = (const struct clk_parent_data []) {
> + { .fw_name = "xtal", },
> + { .hw = &s4_mpll_50m_div.hw },
> + },
> + .num_parents = 2,
> + },
> +};
> +
> +static struct clk_fixed_factor s4_mpll_prediv = {
> + .mult = 1,
> + .div = 2,
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll_prediv",
> + .ops = &clk_fixed_factor_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_fixed_pll_dco.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static const struct reg_sequence s4_mpll0_init_regs[] = {
> + { .reg = ANACTRL_MPLL_CTRL2, .def = 0x40000033 }
> +};
> +
> +static struct clk_regmap s4_mpll0_div = {
> + .data = &(struct meson_clk_mpll_data){
> + .sdm = {
> + .reg_off = ANACTRL_MPLL_CTRL1,
> + .shift = 0,
> + .width = 14,
> + },
> + .sdm_en = {
> + .reg_off = ANACTRL_MPLL_CTRL1,
> + .shift = 30,
> + .width = 1,
> + },
> + .n2 = {
> + .reg_off = ANACTRL_MPLL_CTRL1,
> + .shift = 20,
> + .width = 9,
> + },
> + .ssen = {
> + .reg_off = ANACTRL_MPLL_CTRL1,
> + .shift = 29,
> + .width = 1,
> + },
> + .lock = &meson_clk_lock,
> + .init_regs = s4_mpll0_init_regs,
> + .init_count = ARRAY_SIZE(s4_mpll0_init_regs),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll0_div",
> + .ops = &meson_clk_mpll_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_mpll_prediv.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_mpll0 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_MPLL_CTRL1,
> + .bit_idx = 31,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll0",
> + .ops = &clk_regmap_gate_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll0_div.hw },
> + .num_parents = 1,
> + .flags = CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +static const struct reg_sequence s4_mpll1_init_regs[] = {
> + { .reg = ANACTRL_MPLL_CTRL4, .def = 0x40000033 }
> +};
> +
> +static struct clk_regmap s4_mpll1_div = {
> + .data = &(struct meson_clk_mpll_data){
> + .sdm = {
> + .reg_off = ANACTRL_MPLL_CTRL3,
> + .shift = 0,
> + .width = 14,
> + },
> + .sdm_en = {
> + .reg_off = ANACTRL_MPLL_CTRL3,
> + .shift = 30,
> + .width = 1,
> + },
> + .n2 = {
> + .reg_off = ANACTRL_MPLL_CTRL3,
> + .shift = 20,
> + .width = 9,
> + },
> + .ssen = {
> + .reg_off = ANACTRL_MPLL_CTRL3,
> + .shift = 29,
> + .width = 1,
> + },
> + .lock = &meson_clk_lock,
> + .init_regs = s4_mpll1_init_regs,
> + .init_count = ARRAY_SIZE(s4_mpll1_init_regs),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll1_div",
> + .ops = &meson_clk_mpll_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_mpll_prediv.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_mpll1 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_MPLL_CTRL3,
> + .bit_idx = 31,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll1",
> + .ops = &clk_regmap_gate_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll1_div.hw },
> + .num_parents = 1,
> + .flags = CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +static const struct reg_sequence s4_mpll2_init_regs[] = {
> + { .reg = ANACTRL_MPLL_CTRL6, .def = 0x40000033 }
> +};
> +
> +static struct clk_regmap s4_mpll2_div = {
> + .data = &(struct meson_clk_mpll_data){
> + .sdm = {
> + .reg_off = ANACTRL_MPLL_CTRL5,
> + .shift = 0,
> + .width = 14,
> + },
> + .sdm_en = {
> + .reg_off = ANACTRL_MPLL_CTRL5,
> + .shift = 30,
> + .width = 1,
> + },
> + .n2 = {
> + .reg_off = ANACTRL_MPLL_CTRL5,
> + .shift = 20,
> + .width = 9,
> + },
> + .ssen = {
> + .reg_off = ANACTRL_MPLL_CTRL5,
> + .shift = 29,
> + .width = 1,
> + },
> + .lock = &meson_clk_lock,
> + .init_regs = s4_mpll2_init_regs,
> + .init_count = ARRAY_SIZE(s4_mpll2_init_regs),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll2_div",
> + .ops = &meson_clk_mpll_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_mpll_prediv.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_mpll2 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_MPLL_CTRL5,
> + .bit_idx = 31,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll2",
> + .ops = &clk_regmap_gate_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll2_div.hw },
> + .num_parents = 1,
> + .flags = CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +static const struct reg_sequence s4_mpll3_init_regs[] = {
> + { .reg = ANACTRL_MPLL_CTRL8, .def = 0x40000033 }
> +};
> +
> +static struct clk_regmap s4_mpll3_div = {
> + .data = &(struct meson_clk_mpll_data){
> + .sdm = {
> + .reg_off = ANACTRL_MPLL_CTRL7,
> + .shift = 0,
> + .width = 14,
> + },
> + .sdm_en = {
> + .reg_off = ANACTRL_MPLL_CTRL7,
> + .shift = 30,
> + .width = 1,
> + },
> + .n2 = {
> + .reg_off = ANACTRL_MPLL_CTRL7,
> + .shift = 20,
> + .width = 9,
> + },
> + .ssen = {
> + .reg_off = ANACTRL_MPLL_CTRL7,
> + .shift = 29,
> + .width = 1,
> + },
> + .lock = &meson_clk_lock,
> + .init_regs = s4_mpll3_init_regs,
> + .init_count = ARRAY_SIZE(s4_mpll3_init_regs),
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll3_div",
> + .ops = &meson_clk_mpll_ops,
> + .parent_hws = (const struct clk_hw *[]) {
> + &s4_mpll_prediv.hw
> + },
> + .num_parents = 1,
> + },
> +};
> +
> +static struct clk_regmap s4_mpll3 = {
> + .data = &(struct clk_regmap_gate_data){
> + .offset = ANACTRL_MPLL_CTRL7,
> + .bit_idx = 31,
> + },
> + .hw.init = &(struct clk_init_data){
> + .name = "mpll3",
> + .ops = &clk_regmap_gate_ops,
> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll3_div.hw },
> + .num_parents = 1,
> + .flags = CLK_SET_RATE_PARENT,
> + },
> +};
> +
> +/* Array of all clocks provided by this provider */
> +static struct clk_hw_onecell_data s4_pll_hw_onecell_data = {
> + .hws = {
> + [CLKID_FIXED_PLL_DCO] = &s4_fixed_pll_dco.hw,
> + [CLKID_FIXED_PLL] = &s4_fixed_pll.hw,
> + [CLKID_FCLK_DIV2_DIV] = &s4_fclk_div2_div.hw,
> + [CLKID_FCLK_DIV2] = &s4_fclk_div2.hw,
> + [CLKID_FCLK_DIV3_DIV] = &s4_fclk_div3_div.hw,
> + [CLKID_FCLK_DIV3] = &s4_fclk_div3.hw,
> + [CLKID_FCLK_DIV4_DIV] = &s4_fclk_div4_div.hw,
> + [CLKID_FCLK_DIV4] = &s4_fclk_div4.hw,
> + [CLKID_FCLK_DIV5_DIV] = &s4_fclk_div5_div.hw,
> + [CLKID_FCLK_DIV5] = &s4_fclk_div5.hw,
> + [CLKID_FCLK_DIV7_DIV] = &s4_fclk_div7_div.hw,
> + [CLKID_FCLK_DIV7] = &s4_fclk_div7.hw,
> + [CLKID_FCLK_DIV2P5_DIV] = &s4_fclk_div2p5_div.hw,
> + [CLKID_FCLK_DIV2P5] = &s4_fclk_div2p5.hw,
> + [CLKID_GP0_PLL_DCO] = &s4_gp0_pll_dco.hw,
> + [CLKID_GP0_PLL] = &s4_gp0_pll.hw,
> + [CLKID_HIFI_PLL_DCO] = &s4_hifi_pll_dco.hw,
> + [CLKID_HIFI_PLL] = &s4_hifi_pll.hw,
> + [CLKID_HDMI_PLL_DCO] = &s4_hdmi_pll_dco.hw,
> + [CLKID_HDMI_PLL_OD] = &s4_hdmi_pll_od.hw,
> + [CLKID_HDMI_PLL] = &s4_hdmi_pll.hw,
> + [CLKID_MPLL_50M_DIV] = &s4_mpll_50m_div.hw,
> + [CLKID_MPLL_50M] = &s4_mpll_50m.hw,
> + [CLKID_MPLL_PREDIV] = &s4_mpll_prediv.hw,
> + [CLKID_MPLL0_DIV] = &s4_mpll0_div.hw,
> + [CLKID_MPLL0] = &s4_mpll0.hw,
> + [CLKID_MPLL1_DIV] = &s4_mpll1_div.hw,
> + [CLKID_MPLL1] = &s4_mpll1.hw,
> + [CLKID_MPLL2_DIV] = &s4_mpll2_div.hw,
> + [CLKID_MPLL2] = &s4_mpll2.hw,
> + [CLKID_MPLL3_DIV] = &s4_mpll3_div.hw,
> + [CLKID_MPLL3] = &s4_mpll3.hw,
> +
> + [NR_PLL_CLKS] = NULL
> + },
> + .num = NR_PLL_CLKS,
> +};
> +
> +static struct clk_regmap *const s4_pll_clk_regmaps[] = {
> + &s4_fixed_pll_dco,
> + &s4_fixed_pll,
> + &s4_fclk_div2,
> + &s4_fclk_div3,
> + &s4_fclk_div4,
> + &s4_fclk_div5,
> + &s4_fclk_div7,
> + &s4_fclk_div2p5,
> + &s4_gp0_pll_dco,
> + &s4_gp0_pll,
> + &s4_hifi_pll_dco,
> + &s4_hifi_pll,
> + &s4_hdmi_pll_dco,
> + &s4_hdmi_pll_od,
> + &s4_hdmi_pll,
> + &s4_mpll_50m,
> + &s4_mpll0_div,
> + &s4_mpll0,
> + &s4_mpll1_div,
> + &s4_mpll1,
> + &s4_mpll2_div,
> + &s4_mpll2,
> + &s4_mpll3_div,
> + &s4_mpll3,
> +};
> +
> +static const struct reg_sequence s4_init_regs[] = {
> + { .reg = ANACTRL_MPLL_CTRL0, .def = 0x00000543 },
> +};
> +
> +static struct regmap_config clkc_regmap_config = {
> + .reg_bits = 32,
> + .val_bits = 32,
> + .reg_stride = 4,
> +};
> +
> +static int meson_s4_pll_probe(struct platform_device *pdev)
> +{
> + struct device *dev = &pdev->dev;
> + struct regmap *regmap;
> + void __iomem *base;
> + int ret, i;
> +
> + base = devm_platform_ioremap_resource(pdev, 0);
> + if (IS_ERR(base))
> + return PTR_ERR(base);
> +
> + regmap = devm_regmap_init_mmio(dev, base, &clkc_regmap_config);
> + if (IS_ERR(regmap))
> + return PTR_ERR(regmap);
> +
> + ret = regmap_multi_reg_write(regmap, s4_init_regs, ARRAY_SIZE(s4_init_regs));
> + if (ret) {
> + dev_err(dev, "Failed to init registers\n");
> + return ret;
> + }
> +
> + /* Populate regmap for the regmap backed clocks */
> + for (i = 0; i < ARRAY_SIZE(s4_pll_clk_regmaps); i++)
> + s4_pll_clk_regmaps[i]->map = regmap;
> +
> + for (i = 0; i < s4_pll_hw_onecell_data.num; i++) {
> + /* array might be sparse */
> + if (!s4_pll_hw_onecell_data.hws[i])
> + continue;
> +
> + ret = devm_clk_hw_register(dev, s4_pll_hw_onecell_data.hws[i]);
> + if (ret) {
> + dev_err(dev, "Clock registration failed\n");
> + return ret;
> + }
> + }
> +
> + return devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
> + &s4_pll_hw_onecell_data);
> +}
> +
> +static const struct of_device_id clkc_match_table[] = {
> + {
> + .compatible = "amlogic,s4-pll-clkc",
> + },
> + {}
> +};
> +
> +static struct platform_driver s4_driver = {
> + .probe = meson_s4_pll_probe,
> + .driver = {
> + .name = "s4-pll-clkc",
> + .of_match_table = clkc_match_table,
> + },
> +};
> +
> +module_platform_driver(s4_driver);
> +MODULE_LICENSE("GPL");
> diff --git a/drivers/clk/meson/s4-pll.h b/drivers/clk/meson/s4-pll.h
> new file mode 100644
> index 000000000000..41dc6de978c1
> --- /dev/null
> +++ b/drivers/clk/meson/s4-pll.h
> @@ -0,0 +1,88 @@
> +/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
> +/*
> + * Copyright (c) 2021 Amlogic, inc.
> + * Author: Yu Tu <yu.tu@amlogic.com>
> + */
> +
> +#ifndef __MESON_S4_PLL_H__
> +#define __MESON_S4_PLL_H__
> +
> +/* ANA_CTRL - Registers
> + * REG_BASE: REGISTER_BASE_ADDR = 0xfe008000
> + */
> +#define ANACTRL_FIXPLL_CTRL0 (0x0010 << 2)
I already commented on the "<< 2" . Remove them please.
> +#define ANACTRL_FIXPLL_CTRL1 (0x0011 << 2)
> +#define ANACTRL_FIXPLL_CTRL2 (0x0012 << 2)
> +#define ANACTRL_FIXPLL_CTRL3 (0x0013 << 2)
> +#define ANACTRL_FIXPLL_CTRL4 (0x0014 << 2)
> +#define ANACTRL_FIXPLL_CTRL5 (0x0015 << 2)
> +#define ANACTRL_FIXPLL_CTRL6 (0x0016 << 2)
> +#define ANACTRL_FIXPLL_STS (0x0017 << 2)
> +#define ANACTRL_GP0PLL_CTRL0 (0x0020 << 2)
> +#define ANACTRL_GP0PLL_CTRL1 (0x0021 << 2)
> +#define ANACTRL_GP0PLL_CTRL2 (0x0022 << 2)
> +#define ANACTRL_GP0PLL_CTRL3 (0x0023 << 2)
> +#define ANACTRL_GP0PLL_CTRL4 (0x0024 << 2)
> +#define ANACTRL_GP0PLL_CTRL5 (0x0025 << 2)
> +#define ANACTRL_GP0PLL_CTRL6 (0x0026 << 2)
> +#define ANACTRL_GP0PLL_STS (0x0027 << 2)
> +#define ANACTRL_HIFIPLL_CTRL0 (0x0040 << 2)
> +#define ANACTRL_HIFIPLL_CTRL1 (0x0041 << 2)
> +#define ANACTRL_HIFIPLL_CTRL2 (0x0042 << 2)
> +#define ANACTRL_HIFIPLL_CTRL3 (0x0043 << 2)
> +#define ANACTRL_HIFIPLL_CTRL4 (0x0044 << 2)
> +#define ANACTRL_HIFIPLL_CTRL5 (0x0045 << 2)
> +#define ANACTRL_HIFIPLL_CTRL6 (0x0046 << 2)
> +#define ANACTRL_HIFIPLL_STS (0x0047 << 2)
> +#define ANACTRL_MPLL_CTRL0 (0x0060 << 2)
> +#define ANACTRL_MPLL_CTRL1 (0x0061 << 2)
> +#define ANACTRL_MPLL_CTRL2 (0x0062 << 2)
> +#define ANACTRL_MPLL_CTRL3 (0x0063 << 2)
> +#define ANACTRL_MPLL_CTRL4 (0x0064 << 2)
> +#define ANACTRL_MPLL_CTRL5 (0x0065 << 2)
> +#define ANACTRL_MPLL_CTRL6 (0x0066 << 2)
> +#define ANACTRL_MPLL_CTRL7 (0x0067 << 2)
> +#define ANACTRL_MPLL_CTRL8 (0x0068 << 2)
> +#define ANACTRL_MPLL_STS (0x0069 << 2)
> +#define ANACTRL_HDMIPLL_CTRL0 (0x0070 << 2)
> +#define ANACTRL_HDMIPLL_CTRL1 (0x0071 << 2)
> +#define ANACTRL_HDMIPLL_CTRL2 (0x0072 << 2)
> +#define ANACTRL_HDMIPLL_CTRL3 (0x0073 << 2)
> +#define ANACTRL_HDMIPLL_CTRL4 (0x0074 << 2)
> +#define ANACTRL_HDMIPLL_CTRL5 (0x0075 << 2)
> +#define ANACTRL_HDMIPLL_CTRL6 (0x0076 << 2)
> +#define ANACTRL_HDMIPLL_STS (0x0077 << 2)
> +#define ANACTRL_HDMIPLL_VLOCK (0x0079 << 2)
> +
> +/*
> + * CLKID index values
> + *
> + * These indices are entirely contrived and do not map onto the hardware.
> + * It has now been decided to expose everything by default in the DT header:
> + * include/dt-bindings/clock/axg-clkc.h. Only the clocks ids we don't want
> + * to expose, such as the internal muxes and dividers of composite clocks,
> + * will remain defined here.
> + */
> +#define CLKID_FIXED_PLL_DCO 0
> +#define CLKID_FCLK_DIV2_DIV 2
> +#define CLKID_FCLK_DIV3_DIV 4
> +#define CLKID_FCLK_DIV4_DIV 6
> +#define CLKID_FCLK_DIV5_DIV 8
> +#define CLKID_FCLK_DIV7_DIV 10
> +#define CLKID_FCLK_DIV2P5_DIV 12
> +#define CLKID_GP0_PLL_DCO 14
> +#define CLKID_HIFI_PLL_DCO 16
> +#define CLKID_HDMI_PLL_DCO 18
> +#define CLKID_HDMI_PLL_OD 19
> +#define CLKID_MPLL_50M_DIV 21
> +#define CLKID_MPLL_PREDIV 23
> +#define CLKID_MPLL0_DIV 24
> +#define CLKID_MPLL1_DIV 26
> +#define CLKID_MPLL2_DIV 28
> +#define CLKID_MPLL3_DIV 30
> +
> +#define NR_PLL_CLKS 32
> +/* include the CLKIDs that have been made part of the DT binding */
> +#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
> +
> +#endif /* __MESON_S4_PLL_H__ */
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-10 13:32 ` Jerome Brunet
@ 2022-08-15 6:17 ` Yu Tu
2022-08-29 9:43 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-15 6:17 UTC (permalink / raw)
To: Jerome Brunet, Krzysztof Kozlowski, linux-clk, linux-arm-kernel,
linux-amlogic, linux-kernel, devicetree, Rob Herring,
Neil Armstrong, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Jerome,
On 2022/8/10 21:32, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Fri 05 Aug 2022 at 17:39, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> Hi Krzysztof,
>> Thank you for your reply.
>>
>> On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
>>> [ EXTERNAL EMAIL ]
>>> On 05/08/2022 10:57, Yu Tu wrote:
>>>> Added information about the S4 SOC PLL Clock controller in DT.
>>>>
>>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>>> ---
>>>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>>>> 1 file changed, 8 insertions(+)
>>>>
>>>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>> index ff213618a598..a816b1f7694b 100644
>>>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>>>> #size-cells = <2>;
>>>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>>> + clkc_pll: pll-clock-controller@8000 {
>>> Node names should be generic - clock-controller.
>>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>>
>> I will change to clkc_pll: clock-controller@8000, in next version.
>
> Same comment applies to the binding doc.
OKay.
>
> Also it would be nice to split this in two series.
> Bindings and drivers in one, arm64 dt in the other. These changes goes
> in through different trees.
At present, Bindings, DTS and drivers are three series. Do you mean to
put Bindings and drivers together? If so, checkpatch.pl will report a
warning.
>
>>> Best regards,
>>> Krzysztof
>>> .
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-10 13:47 ` Jerome Brunet
@ 2022-08-15 6:34 ` Yu Tu
2022-08-15 13:20 ` Yu Tu
2022-08-29 9:46 ` Jerome Brunet
0 siblings, 2 replies; 41+ messages in thread
From: Yu Tu @ 2022-08-15 6:34 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Jerome,
On 2022/8/10 21:47, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Fri 05 Aug 2022 at 16:57, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> Add the S4 PLL clock controller found in the s4 SoC family.
>>
>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>> ---
>> drivers/clk/meson/Kconfig | 12 +
>> drivers/clk/meson/Makefile | 1 +
>> drivers/clk/meson/s4-pll.c | 891 +++++++++++++++++++++++++++++++++++++
>> drivers/clk/meson/s4-pll.h | 88 ++++
>> 4 files changed, 992 insertions(+)
>> create mode 100644 drivers/clk/meson/s4-pll.c
>> create mode 100644 drivers/clk/meson/s4-pll.h
>>
>> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
>> index fc002c155bc3..f4244edc7b28 100644
>> --- a/drivers/clk/meson/Kconfig
>> +++ b/drivers/clk/meson/Kconfig
>> @@ -115,4 +115,16 @@ config COMMON_CLK_G12A
>> help
>> Support for the clock controller on Amlogic S905D2, S905X2 and S905Y2
>> devices, aka g12a. Say Y if you want peripherals to work.
>> +
>> +config COMMON_CLK_S4_PLL
>> + tristate "S4 SoC PLL clock controllers support"
>> + depends on ARM64
>> + default y
>> + select COMMON_CLK_MESON_MPLL
>> + select COMMON_CLK_MESON_PLL
>> + select COMMON_CLK_MESON_REGMAP
>> + help
>> + Support for the pll clock controller on Amlogic S805X2 and S905Y4 devices,
>> + aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and AQ229.
>> + Say Y if you want peripherals and CPU frequency scaling to work.
>
> Looks like this doc is copy/paste of the peripheral controller.
> This probably needs to be updated
Okay.
>
>> endmenu
>> diff --git a/drivers/clk/meson/Makefile b/drivers/clk/meson/Makefile
>> index 6eca2a406ee3..376f49cc13f1 100644
>> --- a/drivers/clk/meson/Makefile
>> +++ b/drivers/clk/meson/Makefile
>> @@ -19,3 +19,4 @@ obj-$(CONFIG_COMMON_CLK_AXG_AUDIO) += axg-audio.o
>> obj-$(CONFIG_COMMON_CLK_GXBB) += gxbb.o gxbb-aoclk.o
>> obj-$(CONFIG_COMMON_CLK_G12A) += g12a.o g12a-aoclk.o
>> obj-$(CONFIG_COMMON_CLK_MESON8B) += meson8b.o meson8-ddr.o
>> +obj-$(CONFIG_COMMON_CLK_S4_PLL) += s4-pll.o
>> diff --git a/drivers/clk/meson/s4-pll.c b/drivers/clk/meson/s4-pll.c
>> new file mode 100644
>> index 000000000000..478c78b5ed46
>> --- /dev/null
>> +++ b/drivers/clk/meson/s4-pll.c
>> @@ -0,0 +1,891 @@
>> +// SPDX-License-Identifier: GPL-2.0+
>> +/*
>> + * Amlogic Meson-S4 Clock Controller Driver
>> + *
>> + * Copyright (c) 2021 Amlogic, inc.
>> + * Author: Yu Tu <yu.tu@amlogic.com>
>> + */
>> +
>> +#include <linux/clk-provider.h>
>> +#include <linux/of_device.h>
>> +#include <linux/platform_device.h>
>> +
>> +#include "clk-mpll.h"
>> +#include "clk-pll.h"
>> +#include "clk-regmap.h"
>> +#include "s4-pll.h"
>> +
>> +static DEFINE_SPINLOCK(meson_clk_lock);
>> +
>> +static struct clk_regmap s4_fixed_pll_dco = {
>> + .data = &(struct meson_clk_pll_data){
>> + .en = {
>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>> + .shift = 28,
>> + .width = 1,
>> + },
>> + .m = {
>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>> + .shift = 0,
>> + .width = 8,
>> + },
>> + .n = {
>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>> + .shift = 10,
>> + .width = 5,
>> + },
>> + .frac = {
>> + .reg_off = ANACTRL_FIXPLL_CTRL1,
>> + .shift = 0,
>> + .width = 17,
>> + },
>> + .l = {
>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>> + .shift = 31,
>> + .width = 1,
>> + },
>> + .rst = {
>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fixed_pll_dco",
>> + .ops = &meson_clk_pll_ro_ops,
>> + .parent_data = (const struct clk_parent_data []) {
>> + { .fw_name = "xtal", }
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fixed_pll = {
>> + .data = &(struct clk_regmap_div_data){
>> + .offset = ANACTRL_FIXPLL_CTRL0,
>> + .shift = 16,
>> + .width = 2,
>> + .flags = CLK_DIVIDER_POWER_OF_TWO,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fixed_pll",
>> + .ops = &clk_regmap_divider_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fixed_pll_dco.hw
>> + },
>> + .num_parents = 1,
>> + /*
>> + * This clock won't ever change at runtime so
>> + * CLK_SET_RATE_PARENT is not required
>> + */
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_fclk_div2_div = {
>> + .mult = 1,
>> + .div = 2,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div2_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fclk_div2 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_FIXPLL_CTRL1,
>> + .bit_idx = 24,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div2",
>> + .ops = &clk_regmap_gate_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fclk_div2_div.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_fclk_div3_div = {
>> + .mult = 1,
>> + .div = 3,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div3_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fclk_div3 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_FIXPLL_CTRL1,
>> + .bit_idx = 20,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div3",
>> + .ops = &clk_regmap_gate_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fclk_div3_div.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_fclk_div4_div = {
>> + .mult = 1,
>> + .div = 4,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div4_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fclk_div4 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_FIXPLL_CTRL1,
>> + .bit_idx = 21,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div4",
>> + .ops = &clk_regmap_gate_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fclk_div4_div.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_fclk_div5_div = {
>> + .mult = 1,
>> + .div = 5,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div5_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fclk_div5 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_FIXPLL_CTRL1,
>> + .bit_idx = 22,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div5",
>> + .ops = &clk_regmap_gate_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fclk_div5_div.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_fclk_div7_div = {
>> + .mult = 1,
>> + .div = 7,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div7_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fclk_div7 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_FIXPLL_CTRL1,
>> + .bit_idx = 23,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div7",
>> + .ops = &clk_regmap_gate_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fclk_div7_div.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_fclk_div2p5_div = {
>> + .mult = 2,
>> + .div = 5,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div2p5_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fixed_pll.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_fclk_div2p5 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_FIXPLL_CTRL1,
>> + .bit_idx = 25,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "fclk_div2p5",
>> + .ops = &clk_regmap_gate_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fclk_div2p5_div.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static const struct pll_mult_range s4_gp0_pll_mult_range = {
>> + .min = 125,
>> + .max = 250,
>> +};
>> +
>> +/*
>> + * Internal gp0 pll emulation configuration parameters
>> + */
>> +static const struct reg_sequence s4_gp0_init_regs[] = {
>> + { .reg = ANACTRL_GP0PLL_CTRL1, .def = 0x00000000 },
>> + { .reg = ANACTRL_GP0PLL_CTRL2, .def = 0x00000000 },
>> + { .reg = ANACTRL_GP0PLL_CTRL3, .def = 0x48681c00 },
>> + { .reg = ANACTRL_GP0PLL_CTRL4, .def = 0x88770290 },
>> + { .reg = ANACTRL_GP0PLL_CTRL5, .def = 0x39272000 },
>> + { .reg = ANACTRL_GP0PLL_CTRL6, .def = 0x56540000 }
>> +};
>> +
>> +static struct clk_regmap s4_gp0_pll_dco = {
>> + .data = &(struct meson_clk_pll_data){
>> + .en = {
>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>> + .shift = 28,
>> + .width = 1,
>> + },
>> + .m = {
>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>> + .shift = 0,
>> + .width = 8,
>> + },
>> + .n = {
>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>> + .shift = 10,
>> + .width = 5,
>> + },
>> + .frac = {
>> + .reg_off = ANACTRL_GP0PLL_CTRL1,
>> + .shift = 0,
>> + .width = 17,
>> + },
>> + .l = {
>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>> + .shift = 31,
>> + .width = 1,
>> + },
>> + .rst = {
>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + .range = &s4_gp0_pll_mult_range,
>> + .init_regs = s4_gp0_init_regs,
>> + .init_count = ARRAY_SIZE(s4_gp0_init_regs),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "gp0_pll_dco",
>> + .ops = &meson_clk_pll_ops,
>> + .parent_data = (const struct clk_parent_data []) {
>> + { .fw_name = "xtal", }
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_gp0_pll = {
>> + .data = &(struct clk_regmap_div_data){
>> + .offset = ANACTRL_GP0PLL_CTRL0,
>> + .shift = 16,
>> + .width = 3,
>> + .flags = (CLK_DIVIDER_POWER_OF_TWO |
>> + CLK_DIVIDER_ROUND_CLOSEST),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "gp0_pll",
>> + .ops = &clk_regmap_divider_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_gp0_pll_dco.hw
>> + },
>> + .num_parents = 1,
>> + .flags = CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +/*
>> + * Internal hifi pll emulation configuration parameters
>> + */
>> +static const struct reg_sequence s4_hifi_init_regs[] = {
>> + { .reg = ANACTRL_HIFIPLL_CTRL1, .def = 0x00010e56 },
>> + { .reg = ANACTRL_HIFIPLL_CTRL2, .def = 0x00000000 },
>> + { .reg = ANACTRL_HIFIPLL_CTRL3, .def = 0x6a285c00 },
>> + { .reg = ANACTRL_HIFIPLL_CTRL4, .def = 0x65771290 },
>> + { .reg = ANACTRL_HIFIPLL_CTRL5, .def = 0x39272000 },
>> + { .reg = ANACTRL_HIFIPLL_CTRL6, .def = 0x56540000 }
>> +};
>> +
>> +static struct clk_regmap s4_hifi_pll_dco = {
>> + .data = &(struct meson_clk_pll_data){
>> + .en = {
>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>> + .shift = 28,
>> + .width = 1,
>> + },
>> + .m = {
>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>> + .shift = 0,
>> + .width = 8,
>> + },
>> + .n = {
>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>> + .shift = 10,
>> + .width = 5,
>> + },
>> + .frac = {
>> + .reg_off = ANACTRL_HIFIPLL_CTRL1,
>> + .shift = 0,
>> + .width = 17,
>> + },
>> + .l = {
>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>> + .shift = 31,
>> + .width = 1,
>> + },
>> + .rst = {
>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + .range = &s4_gp0_pll_mult_range,
>> + .init_regs = s4_hifi_init_regs,
>> + .init_count = ARRAY_SIZE(s4_hifi_init_regs),
>> + .flags = CLK_MESON_PLL_ROUND_CLOSEST,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "hifi_pll_dco",
>> + .ops = &meson_clk_pll_ops,
>> + .parent_data = (const struct clk_parent_data []) {
>> + { .fw_name = "xtal", }
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_hifi_pll = {
>> + .data = &(struct clk_regmap_div_data){
>> + .offset = ANACTRL_HIFIPLL_CTRL0,
>> + .shift = 16,
>> + .width = 2,
>> + .flags = (CLK_DIVIDER_POWER_OF_TWO |
>> + CLK_DIVIDER_ROUND_CLOSEST),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "hifi_pll",
>> + .ops = &clk_regmap_divider_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_hifi_pll_dco.hw
>> + },
>> + .num_parents = 1,
>> + .flags = CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_hdmi_pll_dco = {
>> + .data = &(struct meson_clk_pll_data){
>> + .en = {
>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 28,
>> + .width = 1,
>> + },
>> + .m = {
>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 0,
>> + .width = 8,
>> + },
>> + .n = {
>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 10,
>> + .width = 5,
>> + },
>> + .frac = {
>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>> + .shift = 0,
>> + .width = 17,
>> + },
>> + .l = {
>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 31,
>> + .width = 1,
>> + },
>> + .rst = {
>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "hdmi_pll_dco",
>> + .ops = &meson_clk_pll_ro_ops,
>> + .parent_data = (const struct clk_parent_data []) {
>> + { .fw_name = "xtal", }
>> + },
>> + .num_parents = 1,
>> + /*
>> + * Display directly handle hdmi pll registers ATM, we need
>> + * NOCACHE to keep our view of the clock as accurate as
>> + * possible
>> + */
>
> Is it really ?
>
> Given that HDMI support for the s4 is there yet, the
> addresses have changes and the region is no longer a syscon, it is time
> for the HDMI driver to get fixed.
>
>> + .flags = CLK_GET_RATE_NOCACHE,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_hdmi_pll_od = {
>> + .data = &(struct clk_regmap_div_data){
>> + .offset = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 16,
>> + .width = 4,
>> + .flags = CLK_DIVIDER_POWER_OF_TWO,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "hdmi_pll_od",
>> + .ops = &clk_regmap_divider_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_hdmi_pll_dco.hw
>> + },
>> + .num_parents = 1,
>> + /*
>> + * Display directly handle hdmi pll registers ATM, we need
>> + * NOCACHE to keep our view of the clock as accurate as
>> + * possible
>> + */
>> + .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_hdmi_pll = {
>> + .data = &(struct clk_regmap_div_data){
>> + .offset = ANACTRL_HDMIPLL_CTRL0,
>> + .shift = 20,
>> + .width = 2,
>> + .flags = CLK_DIVIDER_POWER_OF_TWO,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "hdmi_pll",
>> + .ops = &clk_regmap_divider_ro_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_hdmi_pll_od.hw
>> + },
>> + .num_parents = 1,
>> + /*
>> + * Display directly handle hdmi pll registers ATM, we need
>> + * NOCACHE to keep our view of the clock as accurate as
>> + * possible
>> + */
>> + .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_mpll_50m_div = {
>> + .mult = 1,
>> + .div = 80,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll_50m_div",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fixed_pll_dco.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_mpll_50m = {
>> + .data = &(struct clk_regmap_mux_data){
>> + .offset = ANACTRL_FIXPLL_CTRL3,
>> + .mask = 0x1,
>> + .shift = 5,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll_50m",
>> + .ops = &clk_regmap_mux_ro_ops,
>> + .parent_data = (const struct clk_parent_data []) {
>> + { .fw_name = "xtal", },
>> + { .hw = &s4_mpll_50m_div.hw },
>> + },
>> + .num_parents = 2,
>> + },
>> +};
>> +
>> +static struct clk_fixed_factor s4_mpll_prediv = {
>> + .mult = 1,
>> + .div = 2,
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll_prediv",
>> + .ops = &clk_fixed_factor_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_fixed_pll_dco.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static const struct reg_sequence s4_mpll0_init_regs[] = {
>> + { .reg = ANACTRL_MPLL_CTRL2, .def = 0x40000033 }
>> +};
>> +
>> +static struct clk_regmap s4_mpll0_div = {
>> + .data = &(struct meson_clk_mpll_data){
>> + .sdm = {
>> + .reg_off = ANACTRL_MPLL_CTRL1,
>> + .shift = 0,
>> + .width = 14,
>> + },
>> + .sdm_en = {
>> + .reg_off = ANACTRL_MPLL_CTRL1,
>> + .shift = 30,
>> + .width = 1,
>> + },
>> + .n2 = {
>> + .reg_off = ANACTRL_MPLL_CTRL1,
>> + .shift = 20,
>> + .width = 9,
>> + },
>> + .ssen = {
>> + .reg_off = ANACTRL_MPLL_CTRL1,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + .lock = &meson_clk_lock,
>> + .init_regs = s4_mpll0_init_regs,
>> + .init_count = ARRAY_SIZE(s4_mpll0_init_regs),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll0_div",
>> + .ops = &meson_clk_mpll_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_mpll_prediv.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_mpll0 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_MPLL_CTRL1,
>> + .bit_idx = 31,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll0",
>> + .ops = &clk_regmap_gate_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll0_div.hw },
>> + .num_parents = 1,
>> + .flags = CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +static const struct reg_sequence s4_mpll1_init_regs[] = {
>> + { .reg = ANACTRL_MPLL_CTRL4, .def = 0x40000033 }
>> +};
>> +
>> +static struct clk_regmap s4_mpll1_div = {
>> + .data = &(struct meson_clk_mpll_data){
>> + .sdm = {
>> + .reg_off = ANACTRL_MPLL_CTRL3,
>> + .shift = 0,
>> + .width = 14,
>> + },
>> + .sdm_en = {
>> + .reg_off = ANACTRL_MPLL_CTRL3,
>> + .shift = 30,
>> + .width = 1,
>> + },
>> + .n2 = {
>> + .reg_off = ANACTRL_MPLL_CTRL3,
>> + .shift = 20,
>> + .width = 9,
>> + },
>> + .ssen = {
>> + .reg_off = ANACTRL_MPLL_CTRL3,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + .lock = &meson_clk_lock,
>> + .init_regs = s4_mpll1_init_regs,
>> + .init_count = ARRAY_SIZE(s4_mpll1_init_regs),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll1_div",
>> + .ops = &meson_clk_mpll_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_mpll_prediv.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_mpll1 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_MPLL_CTRL3,
>> + .bit_idx = 31,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll1",
>> + .ops = &clk_regmap_gate_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll1_div.hw },
>> + .num_parents = 1,
>> + .flags = CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +static const struct reg_sequence s4_mpll2_init_regs[] = {
>> + { .reg = ANACTRL_MPLL_CTRL6, .def = 0x40000033 }
>> +};
>> +
>> +static struct clk_regmap s4_mpll2_div = {
>> + .data = &(struct meson_clk_mpll_data){
>> + .sdm = {
>> + .reg_off = ANACTRL_MPLL_CTRL5,
>> + .shift = 0,
>> + .width = 14,
>> + },
>> + .sdm_en = {
>> + .reg_off = ANACTRL_MPLL_CTRL5,
>> + .shift = 30,
>> + .width = 1,
>> + },
>> + .n2 = {
>> + .reg_off = ANACTRL_MPLL_CTRL5,
>> + .shift = 20,
>> + .width = 9,
>> + },
>> + .ssen = {
>> + .reg_off = ANACTRL_MPLL_CTRL5,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + .lock = &meson_clk_lock,
>> + .init_regs = s4_mpll2_init_regs,
>> + .init_count = ARRAY_SIZE(s4_mpll2_init_regs),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll2_div",
>> + .ops = &meson_clk_mpll_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_mpll_prediv.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_mpll2 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_MPLL_CTRL5,
>> + .bit_idx = 31,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll2",
>> + .ops = &clk_regmap_gate_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll2_div.hw },
>> + .num_parents = 1,
>> + .flags = CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +static const struct reg_sequence s4_mpll3_init_regs[] = {
>> + { .reg = ANACTRL_MPLL_CTRL8, .def = 0x40000033 }
>> +};
>> +
>> +static struct clk_regmap s4_mpll3_div = {
>> + .data = &(struct meson_clk_mpll_data){
>> + .sdm = {
>> + .reg_off = ANACTRL_MPLL_CTRL7,
>> + .shift = 0,
>> + .width = 14,
>> + },
>> + .sdm_en = {
>> + .reg_off = ANACTRL_MPLL_CTRL7,
>> + .shift = 30,
>> + .width = 1,
>> + },
>> + .n2 = {
>> + .reg_off = ANACTRL_MPLL_CTRL7,
>> + .shift = 20,
>> + .width = 9,
>> + },
>> + .ssen = {
>> + .reg_off = ANACTRL_MPLL_CTRL7,
>> + .shift = 29,
>> + .width = 1,
>> + },
>> + .lock = &meson_clk_lock,
>> + .init_regs = s4_mpll3_init_regs,
>> + .init_count = ARRAY_SIZE(s4_mpll3_init_regs),
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll3_div",
>> + .ops = &meson_clk_mpll_ops,
>> + .parent_hws = (const struct clk_hw *[]) {
>> + &s4_mpll_prediv.hw
>> + },
>> + .num_parents = 1,
>> + },
>> +};
>> +
>> +static struct clk_regmap s4_mpll3 = {
>> + .data = &(struct clk_regmap_gate_data){
>> + .offset = ANACTRL_MPLL_CTRL7,
>> + .bit_idx = 31,
>> + },
>> + .hw.init = &(struct clk_init_data){
>> + .name = "mpll3",
>> + .ops = &clk_regmap_gate_ops,
>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll3_div.hw },
>> + .num_parents = 1,
>> + .flags = CLK_SET_RATE_PARENT,
>> + },
>> +};
>> +
>> +/* Array of all clocks provided by this provider */
>> +static struct clk_hw_onecell_data s4_pll_hw_onecell_data = {
>> + .hws = {
>> + [CLKID_FIXED_PLL_DCO] = &s4_fixed_pll_dco.hw,
>> + [CLKID_FIXED_PLL] = &s4_fixed_pll.hw,
>> + [CLKID_FCLK_DIV2_DIV] = &s4_fclk_div2_div.hw,
>> + [CLKID_FCLK_DIV2] = &s4_fclk_div2.hw,
>> + [CLKID_FCLK_DIV3_DIV] = &s4_fclk_div3_div.hw,
>> + [CLKID_FCLK_DIV3] = &s4_fclk_div3.hw,
>> + [CLKID_FCLK_DIV4_DIV] = &s4_fclk_div4_div.hw,
>> + [CLKID_FCLK_DIV4] = &s4_fclk_div4.hw,
>> + [CLKID_FCLK_DIV5_DIV] = &s4_fclk_div5_div.hw,
>> + [CLKID_FCLK_DIV5] = &s4_fclk_div5.hw,
>> + [CLKID_FCLK_DIV7_DIV] = &s4_fclk_div7_div.hw,
>> + [CLKID_FCLK_DIV7] = &s4_fclk_div7.hw,
>> + [CLKID_FCLK_DIV2P5_DIV] = &s4_fclk_div2p5_div.hw,
>> + [CLKID_FCLK_DIV2P5] = &s4_fclk_div2p5.hw,
>> + [CLKID_GP0_PLL_DCO] = &s4_gp0_pll_dco.hw,
>> + [CLKID_GP0_PLL] = &s4_gp0_pll.hw,
>> + [CLKID_HIFI_PLL_DCO] = &s4_hifi_pll_dco.hw,
>> + [CLKID_HIFI_PLL] = &s4_hifi_pll.hw,
>> + [CLKID_HDMI_PLL_DCO] = &s4_hdmi_pll_dco.hw,
>> + [CLKID_HDMI_PLL_OD] = &s4_hdmi_pll_od.hw,
>> + [CLKID_HDMI_PLL] = &s4_hdmi_pll.hw,
>> + [CLKID_MPLL_50M_DIV] = &s4_mpll_50m_div.hw,
>> + [CLKID_MPLL_50M] = &s4_mpll_50m.hw,
>> + [CLKID_MPLL_PREDIV] = &s4_mpll_prediv.hw,
>> + [CLKID_MPLL0_DIV] = &s4_mpll0_div.hw,
>> + [CLKID_MPLL0] = &s4_mpll0.hw,
>> + [CLKID_MPLL1_DIV] = &s4_mpll1_div.hw,
>> + [CLKID_MPLL1] = &s4_mpll1.hw,
>> + [CLKID_MPLL2_DIV] = &s4_mpll2_div.hw,
>> + [CLKID_MPLL2] = &s4_mpll2.hw,
>> + [CLKID_MPLL3_DIV] = &s4_mpll3_div.hw,
>> + [CLKID_MPLL3] = &s4_mpll3.hw,
>> +
>> + [NR_PLL_CLKS] = NULL
>> + },
>> + .num = NR_PLL_CLKS,
>> +};
>> +
>> +static struct clk_regmap *const s4_pll_clk_regmaps[] = {
>> + &s4_fixed_pll_dco,
>> + &s4_fixed_pll,
>> + &s4_fclk_div2,
>> + &s4_fclk_div3,
>> + &s4_fclk_div4,
>> + &s4_fclk_div5,
>> + &s4_fclk_div7,
>> + &s4_fclk_div2p5,
>> + &s4_gp0_pll_dco,
>> + &s4_gp0_pll,
>> + &s4_hifi_pll_dco,
>> + &s4_hifi_pll,
>> + &s4_hdmi_pll_dco,
>> + &s4_hdmi_pll_od,
>> + &s4_hdmi_pll,
>> + &s4_mpll_50m,
>> + &s4_mpll0_div,
>> + &s4_mpll0,
>> + &s4_mpll1_div,
>> + &s4_mpll1,
>> + &s4_mpll2_div,
>> + &s4_mpll2,
>> + &s4_mpll3_div,
>> + &s4_mpll3,
>> +};
>> +
>> +static const struct reg_sequence s4_init_regs[] = {
>> + { .reg = ANACTRL_MPLL_CTRL0, .def = 0x00000543 },
>> +};
>> +
>> +static struct regmap_config clkc_regmap_config = {
>> + .reg_bits = 32,
>> + .val_bits = 32,
>> + .reg_stride = 4,
>> +};
>> +
>> +static int meson_s4_pll_probe(struct platform_device *pdev)
>> +{
>> + struct device *dev = &pdev->dev;
>> + struct regmap *regmap;
>> + void __iomem *base;
>> + int ret, i;
>> +
>> + base = devm_platform_ioremap_resource(pdev, 0);
>> + if (IS_ERR(base))
>> + return PTR_ERR(base);
>> +
>> + regmap = devm_regmap_init_mmio(dev, base, &clkc_regmap_config);
>> + if (IS_ERR(regmap))
>> + return PTR_ERR(regmap);
>> +
>> + ret = regmap_multi_reg_write(regmap, s4_init_regs, ARRAY_SIZE(s4_init_regs));
>> + if (ret) {
>> + dev_err(dev, "Failed to init registers\n");
>> + return ret;
>> + }
>> +
>> + /* Populate regmap for the regmap backed clocks */
>> + for (i = 0; i < ARRAY_SIZE(s4_pll_clk_regmaps); i++)
>> + s4_pll_clk_regmaps[i]->map = regmap;
>> +
>> + for (i = 0; i < s4_pll_hw_onecell_data.num; i++) {
>> + /* array might be sparse */
>> + if (!s4_pll_hw_onecell_data.hws[i])
>> + continue;
>> +
>> + ret = devm_clk_hw_register(dev, s4_pll_hw_onecell_data.hws[i]);
>> + if (ret) {
>> + dev_err(dev, "Clock registration failed\n");
>> + return ret;
>> + }
>> + }
>> +
>> + return devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
>> + &s4_pll_hw_onecell_data);
>> +}
>> +
>> +static const struct of_device_id clkc_match_table[] = {
>> + {
>> + .compatible = "amlogic,s4-pll-clkc",
>> + },
>> + {}
>> +};
>> +
>> +static struct platform_driver s4_driver = {
>> + .probe = meson_s4_pll_probe,
>> + .driver = {
>> + .name = "s4-pll-clkc",
>> + .of_match_table = clkc_match_table,
>> + },
>> +};
>> +
>> +module_platform_driver(s4_driver);
>> +MODULE_LICENSE("GPL");
>> diff --git a/drivers/clk/meson/s4-pll.h b/drivers/clk/meson/s4-pll.h
>> new file mode 100644
>> index 000000000000..41dc6de978c1
>> --- /dev/null
>> +++ b/drivers/clk/meson/s4-pll.h
>> @@ -0,0 +1,88 @@
>> +/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
>> +/*
>> + * Copyright (c) 2021 Amlogic, inc.
>> + * Author: Yu Tu <yu.tu@amlogic.com>
>> + */
>> +
>> +#ifndef __MESON_S4_PLL_H__
>> +#define __MESON_S4_PLL_H__
>> +
>> +/* ANA_CTRL - Registers
>> + * REG_BASE: REGISTER_BASE_ADDR = 0xfe008000
>> + */
>> +#define ANACTRL_FIXPLL_CTRL0 (0x0010 << 2)
>
> I already commented on the "<< 2" . Remove them please.
Sorry, maybe I didn't pay attention to this comment earlier. A little
bit of a question why is this not okay? I understand isn't it better for
the compiler to help with this calculation itself?
>
>> +#define ANACTRL_FIXPLL_CTRL1 (0x0011 << 2)
>> +#define ANACTRL_FIXPLL_CTRL2 (0x0012 << 2)
>> +#define ANACTRL_FIXPLL_CTRL3 (0x0013 << 2)
>> +#define ANACTRL_FIXPLL_CTRL4 (0x0014 << 2)
>> +#define ANACTRL_FIXPLL_CTRL5 (0x0015 << 2)
>> +#define ANACTRL_FIXPLL_CTRL6 (0x0016 << 2)
>> +#define ANACTRL_FIXPLL_STS (0x0017 << 2)
>> +#define ANACTRL_GP0PLL_CTRL0 (0x0020 << 2)
>> +#define ANACTRL_GP0PLL_CTRL1 (0x0021 << 2)
>> +#define ANACTRL_GP0PLL_CTRL2 (0x0022 << 2)
>> +#define ANACTRL_GP0PLL_CTRL3 (0x0023 << 2)
>> +#define ANACTRL_GP0PLL_CTRL4 (0x0024 << 2)
>> +#define ANACTRL_GP0PLL_CTRL5 (0x0025 << 2)
>> +#define ANACTRL_GP0PLL_CTRL6 (0x0026 << 2)
>> +#define ANACTRL_GP0PLL_STS (0x0027 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL0 (0x0040 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL1 (0x0041 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL2 (0x0042 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL3 (0x0043 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL4 (0x0044 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL5 (0x0045 << 2)
>> +#define ANACTRL_HIFIPLL_CTRL6 (0x0046 << 2)
>> +#define ANACTRL_HIFIPLL_STS (0x0047 << 2)
>> +#define ANACTRL_MPLL_CTRL0 (0x0060 << 2)
>> +#define ANACTRL_MPLL_CTRL1 (0x0061 << 2)
>> +#define ANACTRL_MPLL_CTRL2 (0x0062 << 2)
>> +#define ANACTRL_MPLL_CTRL3 (0x0063 << 2)
>> +#define ANACTRL_MPLL_CTRL4 (0x0064 << 2)
>> +#define ANACTRL_MPLL_CTRL5 (0x0065 << 2)
>> +#define ANACTRL_MPLL_CTRL6 (0x0066 << 2)
>> +#define ANACTRL_MPLL_CTRL7 (0x0067 << 2)
>> +#define ANACTRL_MPLL_CTRL8 (0x0068 << 2)
>> +#define ANACTRL_MPLL_STS (0x0069 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL0 (0x0070 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL1 (0x0071 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL2 (0x0072 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL3 (0x0073 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL4 (0x0074 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL5 (0x0075 << 2)
>> +#define ANACTRL_HDMIPLL_CTRL6 (0x0076 << 2)
>> +#define ANACTRL_HDMIPLL_STS (0x0077 << 2)
>> +#define ANACTRL_HDMIPLL_VLOCK (0x0079 << 2)
>> +
>> +/*
>> + * CLKID index values
>> + *
>> + * These indices are entirely contrived and do not map onto the hardware.
>> + * It has now been decided to expose everything by default in the DT header:
>> + * include/dt-bindings/clock/axg-clkc.h. Only the clocks ids we don't want
>> + * to expose, such as the internal muxes and dividers of composite clocks,
>> + * will remain defined here.
>> + */
>> +#define CLKID_FIXED_PLL_DCO 0
>> +#define CLKID_FCLK_DIV2_DIV 2
>> +#define CLKID_FCLK_DIV3_DIV 4
>> +#define CLKID_FCLK_DIV4_DIV 6
>> +#define CLKID_FCLK_DIV5_DIV 8
>> +#define CLKID_FCLK_DIV7_DIV 10
>> +#define CLKID_FCLK_DIV2P5_DIV 12
>> +#define CLKID_GP0_PLL_DCO 14
>> +#define CLKID_HIFI_PLL_DCO 16
>> +#define CLKID_HDMI_PLL_DCO 18
>> +#define CLKID_HDMI_PLL_OD 19
>> +#define CLKID_MPLL_50M_DIV 21
>> +#define CLKID_MPLL_PREDIV 23
>> +#define CLKID_MPLL0_DIV 24
>> +#define CLKID_MPLL1_DIV 26
>> +#define CLKID_MPLL2_DIV 28
>> +#define CLKID_MPLL3_DIV 30
>> +
>> +#define NR_PLL_CLKS 32
>> +/* include the CLKIDs that have been made part of the DT binding */
>> +#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
>> +
>> +#endif /* __MESON_S4_PLL_H__ */
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-15 6:34 ` Yu Tu
@ 2022-08-15 13:20 ` Yu Tu
2022-08-29 9:48 ` Jerome Brunet
2022-08-29 9:46 ` Jerome Brunet
1 sibling, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-15 13:20 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/15 14:34, Yu Tu wrote:
> Hi Jerome,
>
> On 2022/8/10 21:47, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>>
>>
>> On Fri 05 Aug 2022 at 16:57, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>>> Add the S4 PLL clock controller found in the s4 SoC family.
>>>
>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>> ---
>>> drivers/clk/meson/Kconfig | 12 +
>>> drivers/clk/meson/Makefile | 1 +
>>> drivers/clk/meson/s4-pll.c | 891 +++++++++++++++++++++++++++++++++++++
>>> drivers/clk/meson/s4-pll.h | 88 ++++
>>> 4 files changed, 992 insertions(+)
>>> create mode 100644 drivers/clk/meson/s4-pll.c
>>> create mode 100644 drivers/clk/meson/s4-pll.h
>>>
>>> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
>>> index fc002c155bc3..f4244edc7b28 100644
>>> --- a/drivers/clk/meson/Kconfig
>>> +++ b/drivers/clk/meson/Kconfig
>>> @@ -115,4 +115,16 @@ config COMMON_CLK_G12A
>>> help
>>> Support for the clock controller on Amlogic S905D2, S905X2
>>> and S905Y2
>>> devices, aka g12a. Say Y if you want peripherals to work.
>>> +
>>> +config COMMON_CLK_S4_PLL
>>> + tristate "S4 SoC PLL clock controllers support"
>>> + depends on ARM64
>>> + default y
>>> + select COMMON_CLK_MESON_MPLL
>>> + select COMMON_CLK_MESON_PLL
>>> + select COMMON_CLK_MESON_REGMAP
>>> + help
>>> + Support for the pll clock controller on Amlogic S805X2 and
>>> S905Y4 devices,
>>> + aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and
>>> AQ229.
>>> + Say Y if you want peripherals and CPU frequency scaling to work.
>>
>> Looks like this doc is copy/paste of the peripheral controller.
>> This probably needs to be updated
> Okay.
>>
>>> endmenu
>>> diff --git a/drivers/clk/meson/Makefile b/drivers/clk/meson/Makefile
>>> index 6eca2a406ee3..376f49cc13f1 100644
>>> --- a/drivers/clk/meson/Makefile
>>> +++ b/drivers/clk/meson/Makefile
>>> @@ -19,3 +19,4 @@ obj-$(CONFIG_COMMON_CLK_AXG_AUDIO) += axg-audio.o
>>> obj-$(CONFIG_COMMON_CLK_GXBB) += gxbb.o gxbb-aoclk.o
>>> obj-$(CONFIG_COMMON_CLK_G12A) += g12a.o g12a-aoclk.o
>>> obj-$(CONFIG_COMMON_CLK_MESON8B) += meson8b.o meson8-ddr.o
>>> +obj-$(CONFIG_COMMON_CLK_S4_PLL) += s4-pll.o
>>> diff --git a/drivers/clk/meson/s4-pll.c b/drivers/clk/meson/s4-pll.c
>>> new file mode 100644
>>> index 000000000000..478c78b5ed46
>>> --- /dev/null
>>> +++ b/drivers/clk/meson/s4-pll.c
>>> @@ -0,0 +1,891 @@
>>> +// SPDX-License-Identifier: GPL-2.0+
>>> +/*
>>> + * Amlogic Meson-S4 Clock Controller Driver
>>> + *
>>> + * Copyright (c) 2021 Amlogic, inc.
>>> + * Author: Yu Tu <yu.tu@amlogic.com>
>>> + */
>>> +
>>> +#include <linux/clk-provider.h>
>>> +#include <linux/of_device.h>
>>> +#include <linux/platform_device.h>
>>> +
>>> +#include "clk-mpll.h"
>>> +#include "clk-pll.h"
>>> +#include "clk-regmap.h"
>>> +#include "s4-pll.h"
>>> +
>>> +static DEFINE_SPINLOCK(meson_clk_lock);
>>> +
>>> +static struct clk_regmap s4_fixed_pll_dco = {
>>> + .data = &(struct meson_clk_pll_data){
>>> + .en = {
>>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>>> + .shift = 28,
>>> + .width = 1,
>>> + },
>>> + .m = {
>>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>>> + .shift = 0,
>>> + .width = 8,
>>> + },
>>> + .n = {
>>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>>> + .shift = 10,
>>> + .width = 5,
>>> + },
>>> + .frac = {
>>> + .reg_off = ANACTRL_FIXPLL_CTRL1,
>>> + .shift = 0,
>>> + .width = 17,
>>> + },
>>> + .l = {
>>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>>> + .shift = 31,
>>> + .width = 1,
>>> + },
>>> + .rst = {
>>> + .reg_off = ANACTRL_FIXPLL_CTRL0,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fixed_pll_dco",
>>> + .ops = &meson_clk_pll_ro_ops,
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .fw_name = "xtal", }
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fixed_pll = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL0,
>>> + .shift = 16,
>>> + .width = 2,
>>> + .flags = CLK_DIVIDER_POWER_OF_TWO,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fixed_pll",
>>> + .ops = &clk_regmap_divider_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fixed_pll_dco.hw
>>> + },
>>> + .num_parents = 1,
>>> + /*
>>> + * This clock won't ever change at runtime so
>>> + * CLK_SET_RATE_PARENT is not required
>>> + */
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_fclk_div2_div = {
>>> + .mult = 1,
>>> + .div = 2,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div2_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fclk_div2 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL1,
>>> + .bit_idx = 24,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div2",
>>> + .ops = &clk_regmap_gate_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fclk_div2_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_fclk_div3_div = {
>>> + .mult = 1,
>>> + .div = 3,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div3_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fclk_div3 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL1,
>>> + .bit_idx = 20,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div3",
>>> + .ops = &clk_regmap_gate_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fclk_div3_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_fclk_div4_div = {
>>> + .mult = 1,
>>> + .div = 4,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div4_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fclk_div4 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL1,
>>> + .bit_idx = 21,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div4",
>>> + .ops = &clk_regmap_gate_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fclk_div4_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_fclk_div5_div = {
>>> + .mult = 1,
>>> + .div = 5,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div5_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fclk_div5 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL1,
>>> + .bit_idx = 22,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div5",
>>> + .ops = &clk_regmap_gate_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fclk_div5_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_fclk_div7_div = {
>>> + .mult = 1,
>>> + .div = 7,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div7_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_fixed_pll.hw },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fclk_div7 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL1,
>>> + .bit_idx = 23,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div7",
>>> + .ops = &clk_regmap_gate_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fclk_div7_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_fclk_div2p5_div = {
>>> + .mult = 2,
>>> + .div = 5,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div2p5_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fixed_pll.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_fclk_div2p5 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL1,
>>> + .bit_idx = 25,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "fclk_div2p5",
>>> + .ops = &clk_regmap_gate_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fclk_div2p5_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static const struct pll_mult_range s4_gp0_pll_mult_range = {
>>> + .min = 125,
>>> + .max = 250,
>>> +};
>>> +
>>> +/*
>>> + * Internal gp0 pll emulation configuration parameters
>>> + */
>>> +static const struct reg_sequence s4_gp0_init_regs[] = {
>>> + { .reg = ANACTRL_GP0PLL_CTRL1, .def = 0x00000000 },
>>> + { .reg = ANACTRL_GP0PLL_CTRL2, .def = 0x00000000 },
>>> + { .reg = ANACTRL_GP0PLL_CTRL3, .def = 0x48681c00 },
>>> + { .reg = ANACTRL_GP0PLL_CTRL4, .def = 0x88770290 },
>>> + { .reg = ANACTRL_GP0PLL_CTRL5, .def = 0x39272000 },
>>> + { .reg = ANACTRL_GP0PLL_CTRL6, .def = 0x56540000 }
>>> +};
>>> +
>>> +static struct clk_regmap s4_gp0_pll_dco = {
>>> + .data = &(struct meson_clk_pll_data){
>>> + .en = {
>>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>>> + .shift = 28,
>>> + .width = 1,
>>> + },
>>> + .m = {
>>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>>> + .shift = 0,
>>> + .width = 8,
>>> + },
>>> + .n = {
>>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>>> + .shift = 10,
>>> + .width = 5,
>>> + },
>>> + .frac = {
>>> + .reg_off = ANACTRL_GP0PLL_CTRL1,
>>> + .shift = 0,
>>> + .width = 17,
>>> + },
>>> + .l = {
>>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>>> + .shift = 31,
>>> + .width = 1,
>>> + },
>>> + .rst = {
>>> + .reg_off = ANACTRL_GP0PLL_CTRL0,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + .range = &s4_gp0_pll_mult_range,
>>> + .init_regs = s4_gp0_init_regs,
>>> + .init_count = ARRAY_SIZE(s4_gp0_init_regs),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "gp0_pll_dco",
>>> + .ops = &meson_clk_pll_ops,
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .fw_name = "xtal", }
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_gp0_pll = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = ANACTRL_GP0PLL_CTRL0,
>>> + .shift = 16,
>>> + .width = 3,
>>> + .flags = (CLK_DIVIDER_POWER_OF_TWO |
>>> + CLK_DIVIDER_ROUND_CLOSEST),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "gp0_pll",
>>> + .ops = &clk_regmap_divider_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_gp0_pll_dco.hw
>>> + },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +/*
>>> + * Internal hifi pll emulation configuration parameters
>>> + */
>>> +static const struct reg_sequence s4_hifi_init_regs[] = {
>>> + { .reg = ANACTRL_HIFIPLL_CTRL1, .def = 0x00010e56 },
>>> + { .reg = ANACTRL_HIFIPLL_CTRL2, .def = 0x00000000 },
>>> + { .reg = ANACTRL_HIFIPLL_CTRL3, .def = 0x6a285c00 },
>>> + { .reg = ANACTRL_HIFIPLL_CTRL4, .def = 0x65771290 },
>>> + { .reg = ANACTRL_HIFIPLL_CTRL5, .def = 0x39272000 },
>>> + { .reg = ANACTRL_HIFIPLL_CTRL6, .def = 0x56540000 }
>>> +};
>>> +
>>> +static struct clk_regmap s4_hifi_pll_dco = {
>>> + .data = &(struct meson_clk_pll_data){
>>> + .en = {
>>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>>> + .shift = 28,
>>> + .width = 1,
>>> + },
>>> + .m = {
>>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>>> + .shift = 0,
>>> + .width = 8,
>>> + },
>>> + .n = {
>>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>>> + .shift = 10,
>>> + .width = 5,
>>> + },
>>> + .frac = {
>>> + .reg_off = ANACTRL_HIFIPLL_CTRL1,
>>> + .shift = 0,
>>> + .width = 17,
>>> + },
>>> + .l = {
>>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>>> + .shift = 31,
>>> + .width = 1,
>>> + },
>>> + .rst = {
>>> + .reg_off = ANACTRL_HIFIPLL_CTRL0,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + .range = &s4_gp0_pll_mult_range,
>>> + .init_regs = s4_hifi_init_regs,
>>> + .init_count = ARRAY_SIZE(s4_hifi_init_regs),
>>> + .flags = CLK_MESON_PLL_ROUND_CLOSEST,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "hifi_pll_dco",
>>> + .ops = &meson_clk_pll_ops,
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .fw_name = "xtal", }
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_hifi_pll = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = ANACTRL_HIFIPLL_CTRL0,
>>> + .shift = 16,
>>> + .width = 2,
>>> + .flags = (CLK_DIVIDER_POWER_OF_TWO |
>>> + CLK_DIVIDER_ROUND_CLOSEST),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "hifi_pll",
>>> + .ops = &clk_regmap_divider_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_hifi_pll_dco.hw
>>> + },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>> + .data = &(struct meson_clk_pll_data){
>>> + .en = {
>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 28,
>>> + .width = 1,
>>> + },
>>> + .m = {
>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 0,
>>> + .width = 8,
>>> + },
>>> + .n = {
>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 10,
>>> + .width = 5,
>>> + },
>>> + .frac = {
>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>> + .shift = 0,
>>> + .width = 17,
>>> + },
>>> + .l = {
>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 31,
>>> + .width = 1,
>>> + },
>>> + .rst = {
>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "hdmi_pll_dco",
>>> + .ops = &meson_clk_pll_ro_ops,
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .fw_name = "xtal", }
>>> + },
>>> + .num_parents = 1,
>>> + /*
>>> + * Display directly handle hdmi pll registers ATM, we need
>>> + * NOCACHE to keep our view of the clock as accurate as
>>> + * possible
>>> + */
>>
>> Is it really ?
>>
>> Given that HDMI support for the s4 is there yet, the
>> addresses have changes and the region is no longer a syscon, it is time
>> for the HDMI driver to get fixed.
The HDMI PLL is configured in the Uboot phase and does not change the
frequency in the kernel phase. So we use the NOCACHE flag and "ro_ops".
>>
>>> + .flags = CLK_GET_RATE_NOCACHE,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_hdmi_pll_od = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 16,
>>> + .width = 4,
>>> + .flags = CLK_DIVIDER_POWER_OF_TWO,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "hdmi_pll_od",
>>> + .ops = &clk_regmap_divider_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_hdmi_pll_dco.hw
>>> + },
>>> + .num_parents = 1,
>>> + /*
>>> + * Display directly handle hdmi pll registers ATM, we need
>>> + * NOCACHE to keep our view of the clock as accurate as
>>> + * possible
>>> + */
>>> + .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_hdmi_pll = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = ANACTRL_HDMIPLL_CTRL0,
>>> + .shift = 20,
>>> + .width = 2,
>>> + .flags = CLK_DIVIDER_POWER_OF_TWO,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "hdmi_pll",
>>> + .ops = &clk_regmap_divider_ro_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_hdmi_pll_od.hw
>>> + },
>>> + .num_parents = 1,
>>> + /*
>>> + * Display directly handle hdmi pll registers ATM, we need
>>> + * NOCACHE to keep our view of the clock as accurate as
>>> + * possible
>>> + */
>>> + .flags = CLK_GET_RATE_NOCACHE | CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_mpll_50m_div = {
>>> + .mult = 1,
>>> + .div = 80,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll_50m_div",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fixed_pll_dco.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll_50m = {
>>> + .data = &(struct clk_regmap_mux_data){
>>> + .offset = ANACTRL_FIXPLL_CTRL3,
>>> + .mask = 0x1,
>>> + .shift = 5,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll_50m",
>>> + .ops = &clk_regmap_mux_ro_ops,
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .fw_name = "xtal", },
>>> + { .hw = &s4_mpll_50m_div.hw },
>>> + },
>>> + .num_parents = 2,
>>> + },
>>> +};
>>> +
>>> +static struct clk_fixed_factor s4_mpll_prediv = {
>>> + .mult = 1,
>>> + .div = 2,
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll_prediv",
>>> + .ops = &clk_fixed_factor_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_fixed_pll_dco.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static const struct reg_sequence s4_mpll0_init_regs[] = {
>>> + { .reg = ANACTRL_MPLL_CTRL2, .def = 0x40000033 }
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll0_div = {
>>> + .data = &(struct meson_clk_mpll_data){
>>> + .sdm = {
>>> + .reg_off = ANACTRL_MPLL_CTRL1,
>>> + .shift = 0,
>>> + .width = 14,
>>> + },
>>> + .sdm_en = {
>>> + .reg_off = ANACTRL_MPLL_CTRL1,
>>> + .shift = 30,
>>> + .width = 1,
>>> + },
>>> + .n2 = {
>>> + .reg_off = ANACTRL_MPLL_CTRL1,
>>> + .shift = 20,
>>> + .width = 9,
>>> + },
>>> + .ssen = {
>>> + .reg_off = ANACTRL_MPLL_CTRL1,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + .lock = &meson_clk_lock,
>>> + .init_regs = s4_mpll0_init_regs,
>>> + .init_count = ARRAY_SIZE(s4_mpll0_init_regs),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll0_div",
>>> + .ops = &meson_clk_mpll_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_mpll_prediv.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll0 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_MPLL_CTRL1,
>>> + .bit_idx = 31,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll0",
>>> + .ops = &clk_regmap_gate_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll0_div.hw },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static const struct reg_sequence s4_mpll1_init_regs[] = {
>>> + { .reg = ANACTRL_MPLL_CTRL4, .def = 0x40000033 }
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll1_div = {
>>> + .data = &(struct meson_clk_mpll_data){
>>> + .sdm = {
>>> + .reg_off = ANACTRL_MPLL_CTRL3,
>>> + .shift = 0,
>>> + .width = 14,
>>> + },
>>> + .sdm_en = {
>>> + .reg_off = ANACTRL_MPLL_CTRL3,
>>> + .shift = 30,
>>> + .width = 1,
>>> + },
>>> + .n2 = {
>>> + .reg_off = ANACTRL_MPLL_CTRL3,
>>> + .shift = 20,
>>> + .width = 9,
>>> + },
>>> + .ssen = {
>>> + .reg_off = ANACTRL_MPLL_CTRL3,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + .lock = &meson_clk_lock,
>>> + .init_regs = s4_mpll1_init_regs,
>>> + .init_count = ARRAY_SIZE(s4_mpll1_init_regs),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll1_div",
>>> + .ops = &meson_clk_mpll_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_mpll_prediv.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll1 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_MPLL_CTRL3,
>>> + .bit_idx = 31,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll1",
>>> + .ops = &clk_regmap_gate_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll1_div.hw },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static const struct reg_sequence s4_mpll2_init_regs[] = {
>>> + { .reg = ANACTRL_MPLL_CTRL6, .def = 0x40000033 }
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll2_div = {
>>> + .data = &(struct meson_clk_mpll_data){
>>> + .sdm = {
>>> + .reg_off = ANACTRL_MPLL_CTRL5,
>>> + .shift = 0,
>>> + .width = 14,
>>> + },
>>> + .sdm_en = {
>>> + .reg_off = ANACTRL_MPLL_CTRL5,
>>> + .shift = 30,
>>> + .width = 1,
>>> + },
>>> + .n2 = {
>>> + .reg_off = ANACTRL_MPLL_CTRL5,
>>> + .shift = 20,
>>> + .width = 9,
>>> + },
>>> + .ssen = {
>>> + .reg_off = ANACTRL_MPLL_CTRL5,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + .lock = &meson_clk_lock,
>>> + .init_regs = s4_mpll2_init_regs,
>>> + .init_count = ARRAY_SIZE(s4_mpll2_init_regs),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll2_div",
>>> + .ops = &meson_clk_mpll_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_mpll_prediv.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll2 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_MPLL_CTRL5,
>>> + .bit_idx = 31,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll2",
>>> + .ops = &clk_regmap_gate_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll2_div.hw },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static const struct reg_sequence s4_mpll3_init_regs[] = {
>>> + { .reg = ANACTRL_MPLL_CTRL8, .def = 0x40000033 }
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll3_div = {
>>> + .data = &(struct meson_clk_mpll_data){
>>> + .sdm = {
>>> + .reg_off = ANACTRL_MPLL_CTRL7,
>>> + .shift = 0,
>>> + .width = 14,
>>> + },
>>> + .sdm_en = {
>>> + .reg_off = ANACTRL_MPLL_CTRL7,
>>> + .shift = 30,
>>> + .width = 1,
>>> + },
>>> + .n2 = {
>>> + .reg_off = ANACTRL_MPLL_CTRL7,
>>> + .shift = 20,
>>> + .width = 9,
>>> + },
>>> + .ssen = {
>>> + .reg_off = ANACTRL_MPLL_CTRL7,
>>> + .shift = 29,
>>> + .width = 1,
>>> + },
>>> + .lock = &meson_clk_lock,
>>> + .init_regs = s4_mpll3_init_regs,
>>> + .init_count = ARRAY_SIZE(s4_mpll3_init_regs),
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll3_div",
>>> + .ops = &meson_clk_mpll_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_mpll_prediv.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_mpll3 = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = ANACTRL_MPLL_CTRL7,
>>> + .bit_idx = 31,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "mpll3",
>>> + .ops = &clk_regmap_gate_ops,
>>> + .parent_hws = (const struct clk_hw *[]) { &s4_mpll3_div.hw },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +/* Array of all clocks provided by this provider */
>>> +static struct clk_hw_onecell_data s4_pll_hw_onecell_data = {
>>> + .hws = {
>>> + [CLKID_FIXED_PLL_DCO] = &s4_fixed_pll_dco.hw,
>>> + [CLKID_FIXED_PLL] = &s4_fixed_pll.hw,
>>> + [CLKID_FCLK_DIV2_DIV] = &s4_fclk_div2_div.hw,
>>> + [CLKID_FCLK_DIV2] = &s4_fclk_div2.hw,
>>> + [CLKID_FCLK_DIV3_DIV] = &s4_fclk_div3_div.hw,
>>> + [CLKID_FCLK_DIV3] = &s4_fclk_div3.hw,
>>> + [CLKID_FCLK_DIV4_DIV] = &s4_fclk_div4_div.hw,
>>> + [CLKID_FCLK_DIV4] = &s4_fclk_div4.hw,
>>> + [CLKID_FCLK_DIV5_DIV] = &s4_fclk_div5_div.hw,
>>> + [CLKID_FCLK_DIV5] = &s4_fclk_div5.hw,
>>> + [CLKID_FCLK_DIV7_DIV] = &s4_fclk_div7_div.hw,
>>> + [CLKID_FCLK_DIV7] = &s4_fclk_div7.hw,
>>> + [CLKID_FCLK_DIV2P5_DIV] = &s4_fclk_div2p5_div.hw,
>>> + [CLKID_FCLK_DIV2P5] = &s4_fclk_div2p5.hw,
>>> + [CLKID_GP0_PLL_DCO] = &s4_gp0_pll_dco.hw,
>>> + [CLKID_GP0_PLL] = &s4_gp0_pll.hw,
>>> + [CLKID_HIFI_PLL_DCO] = &s4_hifi_pll_dco.hw,
>>> + [CLKID_HIFI_PLL] = &s4_hifi_pll.hw,
>>> + [CLKID_HDMI_PLL_DCO] = &s4_hdmi_pll_dco.hw,
>>> + [CLKID_HDMI_PLL_OD] = &s4_hdmi_pll_od.hw,
>>> + [CLKID_HDMI_PLL] = &s4_hdmi_pll.hw,
>>> + [CLKID_MPLL_50M_DIV] = &s4_mpll_50m_div.hw,
>>> + [CLKID_MPLL_50M] = &s4_mpll_50m.hw,
>>> + [CLKID_MPLL_PREDIV] = &s4_mpll_prediv.hw,
>>> + [CLKID_MPLL0_DIV] = &s4_mpll0_div.hw,
>>> + [CLKID_MPLL0] = &s4_mpll0.hw,
>>> + [CLKID_MPLL1_DIV] = &s4_mpll1_div.hw,
>>> + [CLKID_MPLL1] = &s4_mpll1.hw,
>>> + [CLKID_MPLL2_DIV] = &s4_mpll2_div.hw,
>>> + [CLKID_MPLL2] = &s4_mpll2.hw,
>>> + [CLKID_MPLL3_DIV] = &s4_mpll3_div.hw,
>>> + [CLKID_MPLL3] = &s4_mpll3.hw,
>>> +
>>> + [NR_PLL_CLKS] = NULL
>>> + },
>>> + .num = NR_PLL_CLKS,
>>> +};
>>> +
>>> +static struct clk_regmap *const s4_pll_clk_regmaps[] = {
>>> + &s4_fixed_pll_dco,
>>> + &s4_fixed_pll,
>>> + &s4_fclk_div2,
>>> + &s4_fclk_div3,
>>> + &s4_fclk_div4,
>>> + &s4_fclk_div5,
>>> + &s4_fclk_div7,
>>> + &s4_fclk_div2p5,
>>> + &s4_gp0_pll_dco,
>>> + &s4_gp0_pll,
>>> + &s4_hifi_pll_dco,
>>> + &s4_hifi_pll,
>>> + &s4_hdmi_pll_dco,
>>> + &s4_hdmi_pll_od,
>>> + &s4_hdmi_pll,
>>> + &s4_mpll_50m,
>>> + &s4_mpll0_div,
>>> + &s4_mpll0,
>>> + &s4_mpll1_div,
>>> + &s4_mpll1,
>>> + &s4_mpll2_div,
>>> + &s4_mpll2,
>>> + &s4_mpll3_div,
>>> + &s4_mpll3,
>>> +};
>>> +
>>> +static const struct reg_sequence s4_init_regs[] = {
>>> + { .reg = ANACTRL_MPLL_CTRL0, .def = 0x00000543 },
>>> +};
>>> +
>>> +static struct regmap_config clkc_regmap_config = {
>>> + .reg_bits = 32,
>>> + .val_bits = 32,
>>> + .reg_stride = 4,
>>> +};
>>> +
>>> +static int meson_s4_pll_probe(struct platform_device *pdev)
>>> +{
>>> + struct device *dev = &pdev->dev;
>>> + struct regmap *regmap;
>>> + void __iomem *base;
>>> + int ret, i;
>>> +
>>> + base = devm_platform_ioremap_resource(pdev, 0);
>>> + if (IS_ERR(base))
>>> + return PTR_ERR(base);
>>> +
>>> + regmap = devm_regmap_init_mmio(dev, base, &clkc_regmap_config);
>>> + if (IS_ERR(regmap))
>>> + return PTR_ERR(regmap);
>>> +
>>> + ret = regmap_multi_reg_write(regmap, s4_init_regs,
>>> ARRAY_SIZE(s4_init_regs));
>>> + if (ret) {
>>> + dev_err(dev, "Failed to init registers\n");
>>> + return ret;
>>> + }
>>> +
>>> + /* Populate regmap for the regmap backed clocks */
>>> + for (i = 0; i < ARRAY_SIZE(s4_pll_clk_regmaps); i++)
>>> + s4_pll_clk_regmaps[i]->map = regmap;
>>> +
>>> + for (i = 0; i < s4_pll_hw_onecell_data.num; i++) {
>>> + /* array might be sparse */
>>> + if (!s4_pll_hw_onecell_data.hws[i])
>>> + continue;
>>> +
>>> + ret = devm_clk_hw_register(dev, s4_pll_hw_onecell_data.hws[i]);
>>> + if (ret) {
>>> + dev_err(dev, "Clock registration failed\n");
>>> + return ret;
>>> + }
>>> + }
>>> +
>>> + return devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
>>> + &s4_pll_hw_onecell_data);
>>> +}
>>> +
>>> +static const struct of_device_id clkc_match_table[] = {
>>> + {
>>> + .compatible = "amlogic,s4-pll-clkc",
>>> + },
>>> + {}
>>> +};
>>> +
>>> +static struct platform_driver s4_driver = {
>>> + .probe = meson_s4_pll_probe,
>>> + .driver = {
>>> + .name = "s4-pll-clkc",
>>> + .of_match_table = clkc_match_table,
>>> + },
>>> +};
>>> +
>>> +module_platform_driver(s4_driver);
>>> +MODULE_LICENSE("GPL");
>>> diff --git a/drivers/clk/meson/s4-pll.h b/drivers/clk/meson/s4-pll.h
>>> new file mode 100644
>>> index 000000000000..41dc6de978c1
>>> --- /dev/null
>>> +++ b/drivers/clk/meson/s4-pll.h
>>> @@ -0,0 +1,88 @@
>>> +/* SPDX-License-Identifier: (GPL-2.0+ OR MIT) */
>>> +/*
>>> + * Copyright (c) 2021 Amlogic, inc.
>>> + * Author: Yu Tu <yu.tu@amlogic.com>
>>> + */
>>> +
>>> +#ifndef __MESON_S4_PLL_H__
>>> +#define __MESON_S4_PLL_H__
>>> +
>>> +/* ANA_CTRL - Registers
>>> + * REG_BASE: REGISTER_BASE_ADDR = 0xfe008000
>>> + */
>>> +#define ANACTRL_FIXPLL_CTRL0 (0x0010 << 2)
>>
>> I already commented on the "<< 2" . Remove them please.
> Sorry, maybe I didn't pay attention to this comment earlier. A little
> bit of a question why is this not okay? I understand isn't it better for
> the compiler to help with this calculation itself?
>>
>>> +#define ANACTRL_FIXPLL_CTRL1 (0x0011 << 2)
>>> +#define ANACTRL_FIXPLL_CTRL2 (0x0012 << 2)
>>> +#define ANACTRL_FIXPLL_CTRL3 (0x0013 << 2)
>>> +#define ANACTRL_FIXPLL_CTRL4 (0x0014 << 2)
>>> +#define ANACTRL_FIXPLL_CTRL5 (0x0015 << 2)
>>> +#define ANACTRL_FIXPLL_CTRL6 (0x0016 << 2)
>>> +#define ANACTRL_FIXPLL_STS (0x0017 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL0 (0x0020 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL1 (0x0021 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL2 (0x0022 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL3 (0x0023 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL4 (0x0024 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL5 (0x0025 << 2)
>>> +#define ANACTRL_GP0PLL_CTRL6 (0x0026 << 2)
>>> +#define ANACTRL_GP0PLL_STS (0x0027 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL0 (0x0040 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL1 (0x0041 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL2 (0x0042 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL3 (0x0043 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL4 (0x0044 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL5 (0x0045 << 2)
>>> +#define ANACTRL_HIFIPLL_CTRL6 (0x0046 << 2)
>>> +#define ANACTRL_HIFIPLL_STS (0x0047 << 2)
>>> +#define ANACTRL_MPLL_CTRL0 (0x0060 << 2)
>>> +#define ANACTRL_MPLL_CTRL1 (0x0061 << 2)
>>> +#define ANACTRL_MPLL_CTRL2 (0x0062 << 2)
>>> +#define ANACTRL_MPLL_CTRL3 (0x0063 << 2)
>>> +#define ANACTRL_MPLL_CTRL4 (0x0064 << 2)
>>> +#define ANACTRL_MPLL_CTRL5 (0x0065 << 2)
>>> +#define ANACTRL_MPLL_CTRL6 (0x0066 << 2)
>>> +#define ANACTRL_MPLL_CTRL7 (0x0067 << 2)
>>> +#define ANACTRL_MPLL_CTRL8 (0x0068 << 2)
>>> +#define ANACTRL_MPLL_STS (0x0069 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL0 (0x0070 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL1 (0x0071 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL2 (0x0072 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL3 (0x0073 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL4 (0x0074 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL5 (0x0075 << 2)
>>> +#define ANACTRL_HDMIPLL_CTRL6 (0x0076 << 2)
>>> +#define ANACTRL_HDMIPLL_STS (0x0077 << 2)
>>> +#define ANACTRL_HDMIPLL_VLOCK (0x0079 << 2)
>>> +
>>> +/*
>>> + * CLKID index values
>>> + *
>>> + * These indices are entirely contrived and do not map onto the
>>> hardware.
>>> + * It has now been decided to expose everything by default in the DT
>>> header:
>>> + * include/dt-bindings/clock/axg-clkc.h. Only the clocks ids we
>>> don't want
>>> + * to expose, such as the internal muxes and dividers of composite
>>> clocks,
>>> + * will remain defined here.
>>> + */
>>> +#define CLKID_FIXED_PLL_DCO 0
>>> +#define CLKID_FCLK_DIV2_DIV 2
>>> +#define CLKID_FCLK_DIV3_DIV 4
>>> +#define CLKID_FCLK_DIV4_DIV 6
>>> +#define CLKID_FCLK_DIV5_DIV 8
>>> +#define CLKID_FCLK_DIV7_DIV 10
>>> +#define CLKID_FCLK_DIV2P5_DIV 12
>>> +#define CLKID_GP0_PLL_DCO 14
>>> +#define CLKID_HIFI_PLL_DCO 16
>>> +#define CLKID_HDMI_PLL_DCO 18
>>> +#define CLKID_HDMI_PLL_OD 19
>>> +#define CLKID_MPLL_50M_DIV 21
>>> +#define CLKID_MPLL_PREDIV 23
>>> +#define CLKID_MPLL0_DIV 24
>>> +#define CLKID_MPLL1_DIV 26
>>> +#define CLKID_MPLL2_DIV 28
>>> +#define CLKID_MPLL3_DIV 30
>>> +
>>> +#define NR_PLL_CLKS 32
>>> +/* include the CLKIDs that have been made part of the DT binding */
>>> +#include <dt-bindings/clock/amlogic,s4-pll-clkc.h>
>>> +
>>> +#endif /* __MESON_S4_PLL_H__ */
>>
>> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-15 6:17 ` Yu Tu
@ 2022-08-29 9:43 ` Jerome Brunet
2022-08-30 6:05 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-29 9:43 UTC (permalink / raw)
To: Yu Tu, Krzysztof Kozlowski, linux-clk, linux-arm-kernel,
linux-amlogic, linux-kernel, devicetree, Rob Herring,
Neil Armstrong, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On Mon 15 Aug 2022 at 14:17, Yu Tu <yu.tu@amlogic.com> wrote:
> Hi Jerome,
>
> On 2022/8/10 21:32, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>> On Fri 05 Aug 2022 at 17:39, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>>> Hi Krzysztof,
>>> Thank you for your reply.
>>>
>>> On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
>>>> [ EXTERNAL EMAIL ]
>>>> On 05/08/2022 10:57, Yu Tu wrote:
>>>>> Added information about the S4 SOC PLL Clock controller in DT.
>>>>>
>>>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>>>> ---
>>>>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>>>>> 1 file changed, 8 insertions(+)
>>>>>
>>>>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>> index ff213618a598..a816b1f7694b 100644
>>>>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>>>>> #size-cells = <2>;
>>>>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>>>> + clkc_pll: pll-clock-controller@8000 {
>>>> Node names should be generic - clock-controller.
>>>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>>>
>>> I will change to clkc_pll: clock-controller@8000, in next version.
>> Same comment applies to the binding doc.
> OKay.
>> Also it would be nice to split this in two series.
>> Bindings and drivers in one, arm64 dt in the other. These changes goes
>> in through different trees.
> At present, Bindings, DTS and drivers are three series. Do you mean to put
> Bindings and drivers together? If so, checkpatch.pl will report a warning.
Yes because patches are not in yet so there is a good reason to ignore
the warning. Warning will never show up on the actual tree if the
patches are correctly ordered.
>
>>
>>>> Best regards,
>>>> Krzysztof
>>>> .
>> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-15 6:34 ` Yu Tu
2022-08-15 13:20 ` Yu Tu
@ 2022-08-29 9:46 ` Jerome Brunet
2022-08-30 6:08 ` Yu Tu
1 sibling, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-29 9:46 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Mon 15 Aug 2022 at 14:34, Yu Tu <yu.tu@amlogic.com> wrote:
> Hi Jerome,
>
> On 2022/8/10 21:47, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>> On Fri 05 Aug 2022 at 16:57, Yu Tu <yu.tu@amlogic.com> wrote:
>> */
[... ]
>>> +#define ANACTRL_FIXPLL_CTRL0 (0x0010 << 2)
>> I already commented on the "<< 2" . Remove them please.
> Sorry, maybe I didn't pay attention to this comment earlier. A little bit
> of a question why is this not okay? I understand isn't it better for the
> compiler to help with this calculation itself?
Because it is aweful to read
Also please trim your replies.
It is a bit annoying to search for your comments in the replies
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-15 13:20 ` Yu Tu
@ 2022-08-29 9:48 ` Jerome Brunet
2022-08-30 6:13 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-29 9:48 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>> +
>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>> + .data = &(struct meson_clk_pll_data){
>>>> + .en = {
>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>> + .shift = 28,
>>>> + .width = 1,
>>>> + },
>>>> + .m = {
>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>> + .shift = 0,
>>>> + .width = 8,
>>>> + },
>>>> + .n = {
>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>> + .shift = 10,
>>>> + .width = 5,
>>>> + },
>>>> + .frac = {
>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>> + .shift = 0,
>>>> + .width = 17,
>>>> + },
>>>> + .l = {
>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>> + .shift = 31,
>>>> + .width = 1,
>>>> + },
>>>> + .rst = {
>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>> + .shift = 29,
>>>> + .width = 1,
>>>> + },
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "hdmi_pll_dco",
>>>> + .ops = &meson_clk_pll_ro_ops,
>>>> + .parent_data = (const struct clk_parent_data []) {
>>>> + { .fw_name = "xtal", }
>>>> + },
>>>> + .num_parents = 1,
>>>> + /*
>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>> + * possible
>>>> + */
>>>
>>> Is it really ?
>>>
>>> Given that HDMI support for the s4 is there yet, the
>>> addresses have changes and the region is no longer a syscon, it is time
>>> for the HDMI driver to get fixed.
> The HDMI PLL is configured in the Uboot phase and does not change the
> frequency in the kernel phase. So we use the NOCACHE flag and
> "ro_ops".
That's no reason to put NOCACHE or ro-ops
If you want the frequencies to be statically assinged, the correct way
would be through assigned-rate in DT I guess.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 6/6] clk: meson: s4: add s4 SoC peripheral clock controller driver
[not found] ` <8f40cb49-fdc5-20cd-343b-8ce50e5d6d97@amlogic.com>
@ 2022-08-29 12:19 ` Jerome Brunet
2022-08-30 8:20 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-29 12:19 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Tue 16 Aug 2022 at 20:00, Yu Tu <yu.tu@amlogic.com> wrote:
Please trim your replies
>>> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
>>> index f4244edc7b28..ec6beb9284d3 100644
>>> --- a/drivers/clk/meson/Kconfig
>>> +++ b/drivers/clk/meson/Kconfig
>>> @@ -127,4 +127,17 @@ config COMMON_CLK_S4_PLL
>>> Support for the pll clock controller on Amlogic S805X2 and S905Y4 devices,
>>> aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and AQ229.
>>> Say Y if you want peripherals and CPU frequency scaling to work.
>>> +
>>> +config COMMON_CLK_S4
>>> + tristate "S4 SoC Peripherals clock controllers support"
>>> + depends on ARM64
>>> + default y
>>> + select COMMON_CLK_MESON_REGMAP
>>> + select COMMON_CLK_MESON_DUALDIV
>>> + select COMMON_CLK_MESON_VID_PLL_DIV
>>> + select COMMON_CLK_S4_PLL
>> Do you really this ? your driver does not even include the related
>> header.
> If the PLL driver is not turned on in DTS, will it not cause an error?
>>
I don't get the question.
Kconfig list compile deps. S4 PLL is not a compile dep of the peripheral
controller.
If you really want to, you may use 'imply'.
>>
>>> +static const struct clk_parent_data sys_ab_clk_parent_data[] = {
>>> + { .fw_name = "xtal" },
>>> + { .fw_name = "fclk_div2" },
>>> + { .fw_name = "fclk_div3" },
>>> + { .fw_name = "fclk_div4" },
>>> + { .fw_name = "fclk_div5" },
>>> + { .fw_name = "fclk_div7" },
>>> + { .hw = &s4_rtc_clk.hw }
>>> +};
>>> +
>>> +static struct clk_regmap s4_sysclk_b_sel = {
>>> + .data = &(struct clk_regmap_mux_data){
>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>> + .mask = 0x7,
>>> + .shift = 26,
>>> + .table = mux_table_sys_ab_clk_sel,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "sysclk_b_sel",
>>> + .ops = &clk_regmap_mux_ro_ops,
>> Why is this using the RO ops ?
> Sys_clk is initialized during the Uboot phase and is fixed at
> 166.666MHz. So I'm going to change it to ro.
That really much depends on the bootloader and is a pretty weak design.
The bootloader deps should be kept as minimal as possible.
I see no reason for RO.
You may cut rate propagation on the user if you need to and continue to
whatever you want in your u-boot
>>
>>> + .parent_data = sys_ab_clk_parent_data,
>>> + .num_parents = ARRAY_SIZE(sys_ab_clk_parent_data),
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_sysclk_b_div = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>> + .shift = 16,
>>> + .width = 10,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "sysclk_b_div",
>>> + .ops = &clk_regmap_divider_ro_ops,
>> Same here and for the rest of the sys part
> Same above.
We can play that game for a while
>>> +
>>> +/* Video Clocks */
>>> +static struct clk_regmap s4_vid_pll_div = {
>>> + .data = &(struct meson_vid_pll_div_data){
>>> + .val = {
>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>> + .shift = 0,
>>> + .width = 15,
>>> + },
>>> + .sel = {
>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>> + .shift = 16,
>>> + .width = 2,
>>> + },
>>> + },
>>> + .hw.init = &(struct clk_init_data) {
>>> + .name = "vid_pll_div",
>>> + .ops = &meson_vid_pll_div_ro_ops,
>> Why RO ? applies to the rest of the video part.
> Because vid_pll_div parent is HDMI_PLL, and HDMI_PLL is a fixed
> frequency. Flags is CLK_SET_RATE_PARENT. So we use RO.
If the HDMI_PLL is fixed somehow, that is not reason for this clock to
be RO
> Can I remove RO and use CLK_SET_RATE_NO_REPARENT instead, which one do you
> think is more reasonable?
Neither. CLK_SET_RATE_NO_REPARENT makes no sense, it is not mux
>
>>
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .fw_name = "hdmi_pll", }
>>> + },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_vid_pll_sel = {
>>> + .data = &(struct clk_regmap_mux_data){
>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>> + .mask = 0x1,
>>> + .shift = 18,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "vid_pll_sel",
>>> + .ops = &clk_regmap_mux_ops,
>>> + /*
>>> + * bit 18 selects from 2 possible parents:
>>> + * vid_pll_div or hdmi_pll
>>> + */
>>> + .parent_data = (const struct clk_parent_data []) {
>>> + { .hw = &s4_vid_pll_div.hw },
>>> + { .fw_name = "hdmi_pll", }
>>> + },
>>> + .num_parents = 2,
>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>> Why ? are you planning to DT assigned clocks to statically set this ?
> Because vid_pll_sel one parent is HDMI_PLL, and HDMI_PLL is a fixed
> frequency. To prevent modification, use CLK_SET_RATE_NO_REPARENT.
Again, this makes no sense.
>>
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_vid_pll = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>> + .bit_idx = 19,
>>> + },
>>> + .hw.init = &(struct clk_init_data) {
>>> + .name = "vid_pll",
>>> + .ops = &clk_regmap_gate_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_vid_pll_sel.hw
>>> + },
>>> + .num_parents = 1,
>>> + .flags = CLK_SET_RATE_PARENT,
>>> + },
>>> +};
>>> +
>>> +static const struct clk_parent_data s4_vclk_parent_data[] = {
>>> + { .hw = &s4_vid_pll.hw },
>>> + { .fw_name = "gp0_pll", },
>>> + { .fw_name = "hifi_pll", },
>>> + { .fw_name = "mpll1", },
>>> + { .fw_name = "fclk_div3", },
>>> + { .fw_name = "fclk_div4", },
>>> + { .fw_name = "fclk_div5", },
>>> + { .fw_name = "fclk_div7", },
>>> +};
>>> +
>>> +static struct clk_regmap s4_vclk_sel = {
>>> + .data = &(struct clk_regmap_mux_data){
>>> + .offset = CLKCTRL_VID_CLK_CTRL,
>>> + .mask = 0x7,
>>> + .shift = 16,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "vclk_sel",
>>> + .ops = &clk_regmap_mux_ops,
>>> + .parent_data = s4_vclk_parent_data,
>>> + .num_parents = ARRAY_SIZE(s4_vclk_parent_data),
>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>> Same
> Since fclk_div* is a fixed frequency value, mplL1 and hifi_pll and gp0_pll
> are used by other specialized modules, vid_pll has CLK_SET_RATE_PARENT. The
> parent of vid_pll is that vid_pll_sel uses CLK_SET_RATE_NO_REPARENT.
Still not good.
You don't have CLK_SET_RATE, propagation is stopped and parent clock
will not changed. The best parent will be picked but not changed.
If one parent MUST NOT be picked, just remove it from the list and add a
explaining why
[...]
>>> +
>>> +static struct clk_regmap s4_ts_clk_div = {
>>> + .data = &(struct clk_regmap_div_data){
>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>> + .shift = 0,
>>> + .width = 8,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "ts_clk_div",
>>> + .ops = &clk_regmap_divider_ops,
>>> + .parent_data = &(const struct clk_parent_data) {
>>> + .fw_name = "xtal",
>>> + },
>>> + .num_parents = 1,
>> propagation stopped ?
> Its parent is xtal, so I should use CLK_SET_RATE_NO_REPARENT.
Still no. You seem to have problem with the meaning of
CLK_SET_RATE_NO_REPARENT.
* CLK_SET_RATE_NO_REPARENT: means the parent will no be changed, even if
selecting another parent would result in a closer rate to the
request. It makes sense only if the clock has several parents
* CLK_SET_RATE_PARENT: means rate change may propagate the parent,
meaning the rate of the parent may change if it help the child achieve
a closer rate to the request
>>
>>> + },
>>> +};
>>> +
>>> +static struct clk_regmap s4_ts_clk_gate = {
>>> + .data = &(struct clk_regmap_gate_data){
>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>> + .bit_idx = 8,
>>> + },
>>> + .hw.init = &(struct clk_init_data){
>>> + .name = "ts_clk",
>>> + .ops = &clk_regmap_gate_ops,
>>> + .parent_hws = (const struct clk_hw *[]) {
>>> + &s4_ts_clk_div.hw
>>> + },
>>> + .num_parents = 1,
>>> + },
>> propagation stopped ?
> I will add CLK_SET_RATE_PARENT.
[...]
>>> +/* EMMC/NAND clock */
>>> +
>>> +static const struct clk_parent_data s4_sd_emmc_clk0_parent_data[] = {
>>> + { .fw_name = "xtal", },
>>> + { .fw_name = "fclk_div2", },
>>> + { .fw_name = "fclk_div3", },
>>> + { .fw_name = "hifi_pll", },
>>> + { .fw_name = "fclk_div2p5", },
>>> + /*
>>> + * Following these parent clocks, we should also have had mpll2, mpll3
>>> + * and gp0_pll but these clocks are too precious to be used here. All
>>> + * the necessary rates for MMC and NAND operation can be acheived using
>>> + * hifi_pll or fclk_div clocks
>>> + */
>> You don't want to list mplls but hifi_pll is fine ? seems dangerous.
> hifi pll is for EMMC and NAND on this SoC.
That deserve a better explanation.
Why can't it use fdiv2 and xtal like the previous SoCs ?
Which PLLs are you using for Audio then ?
Typical operation on these SoCs usually require 3 PLLs to acheive all rates
>>
>>> +/*
>>> + * gen clk is designed for debug/monitor some internal clock quality. Some of the
>>> + * corresponding clock sources are not described in the clock tree, so they are skipped.
>>> + */
>> Still feels a bit light, don't you think ? Among all the clocks, can't
>> you add a bit more parents here ? It would certainly help debug down the road
> [16:12] is gen_clk source select.All is:
> 0: cts_oscin_clk
> 1:cts_rtc_clk
> 2:sys_pll_div16 (internal clock)
> 3:ddr_pll_div32 (internal clock)
> 4: vid_pll
> 5: gp0_pll
> 7: hifi_pll
> 10:adc_dpll_clk_b3 (internal clock for debug)
> 11:adc_dpll_intclk (internal clock for debug)
> 12:clk_msr_src(select from all internal clock except PLLs);
> 16: no used
> 17: sys_cpu_clk_div16 (internal clock)
> 19: fclk_div2
> 20: fclk_div2p5
> 21: fclk_div3
> 22: fclk_div4
> 23: fclk_div5
> 24: fclk_div7
> 25: mpll0
> 26: mpll1
> 27: mpll2
> 28: mpll3
> So i only added the clocks that will actually be used, and some debugging
> clock peripherals will not be used.
you may at least add vid_pll
>>
>>> +static u32 s4_gen_clk_mux_table[] = { 0, 5, 7, 19, 21, 22,
>>> + 23, 24, 25, 26, 27, 28 };
>>> +static const struct clk_parent_data s4_gen_clk_parent_data[] = {
>>> + { .fw_name = "xtal", },
>>> + { .fw_name = "gp0_pll", },
>>> + { .fw_name = "hifi_pll", },
>>> + { .fw_name = "fclk_div2", },
>>> + { .fw_name = "fclk_div3", },
>>> + { .fw_name = "fclk_div4", },
>>> + { .fw_name = "fclk_div5", },
>>> + { .fw_name = "fclk_div7", },
>>> + { .fw_name = "mpll0", },
>>> + { .fw_name = "mpll1", },
>>> + { .fw_name = "mpll2", },
>>> + { .fw_name = "mpll3", },
>>> +};
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-29 9:43 ` Jerome Brunet
@ 2022-08-30 6:05 ` Yu Tu
2022-08-30 6:36 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-30 6:05 UTC (permalink / raw)
To: Jerome Brunet, Krzysztof Kozlowski, linux-clk, linux-arm-kernel,
linux-amlogic, linux-kernel, devicetree, Rob Herring,
Neil Armstrong, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/29 17:43, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Mon 15 Aug 2022 at 14:17, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> Hi Jerome,
>>
>> On 2022/8/10 21:32, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>> On Fri 05 Aug 2022 at 17:39, Yu Tu <yu.tu@amlogic.com> wrote:
>>>
>>>> Hi Krzysztof,
>>>> Thank you for your reply.
>>>>
>>>> On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
>>>>> [ EXTERNAL EMAIL ]
>>>>> On 05/08/2022 10:57, Yu Tu wrote:
>>>>>> Added information about the S4 SOC PLL Clock controller in DT.
>>>>>>
>>>>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>>>>> ---
>>>>>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>>>>>> 1 file changed, 8 insertions(+)
>>>>>>
>>>>>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>> index ff213618a598..a816b1f7694b 100644
>>>>>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>>>>>> #size-cells = <2>;
>>>>>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>>>>> + clkc_pll: pll-clock-controller@8000 {
>>>>> Node names should be generic - clock-controller.
>>>>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>>>>
>>>> I will change to clkc_pll: clock-controller@8000, in next version.
>>> Same comment applies to the binding doc.
>> OKay.
>>> Also it would be nice to split this in two series.
>>> Bindings and drivers in one, arm64 dt in the other. These changes goes
>>> in through different trees.
>> At present, Bindings, DTS and drivers are three series. Do you mean to put
>> Bindings and drivers together? If so, checkpatch.pl will report a warning.
>
> Yes because patches are not in yet so there is a good reason to ignore
> the warning. Warning will never show up on the actual tree if the
> patches are correctly ordered.
I think Binding, DTS and drivers use three series and you said two
series is not a big problem. Three series are recommended for
checkpatch.pl, I think it should be easy for that to separate and merge。
I've sent it to V4. Please look at V4 and give some comments.
>
>>
>>>
>>>>> Best regards,
>>>>> Krzysztof
>>>>> .
>>> .
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-29 9:46 ` Jerome Brunet
@ 2022-08-30 6:08 ` Yu Tu
0 siblings, 0 replies; 41+ messages in thread
From: Yu Tu @ 2022-08-30 6:08 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/29 17:46, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Mon 15 Aug 2022 at 14:34, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> Hi Jerome,
>>
>> On 2022/8/10 21:47, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>> On Fri 05 Aug 2022 at 16:57, Yu Tu <yu.tu@amlogic.com> wrote:
>>> */
>
> [... ]
>
>>>> +#define ANACTRL_FIXPLL_CTRL0 (0x0010 << 2)
>>> I already commented on the "<< 2" . Remove them please.
>> Sorry, maybe I didn't pay attention to this comment earlier. A little bit
>> of a question why is this not okay? I understand isn't it better for the
>> compiler to help with this calculation itself?
>
> Because it is aweful to read
>
> Also please trim your replies.
> It is a bit annoying to search for your comments in the replies
>
Okay. Like this?
>
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-29 9:48 ` Jerome Brunet
@ 2022-08-30 6:13 ` Yu Tu
2022-08-30 6:44 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-30 6:13 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/29 17:48, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>
>>>>> +
>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>> + .data = &(struct meson_clk_pll_data){
>>>>> + .en = {
>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>> + .shift = 28,
>>>>> + .width = 1,
>>>>> + },
>>>>> + .m = {
>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>> + .shift = 0,
>>>>> + .width = 8,
>>>>> + },
>>>>> + .n = {
>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>> + .shift = 10,
>>>>> + .width = 5,
>>>>> + },
>>>>> + .frac = {
>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>> + .shift = 0,
>>>>> + .width = 17,
>>>>> + },
>>>>> + .l = {
>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>> + .shift = 31,
>>>>> + .width = 1,
>>>>> + },
>>>>> + .rst = {
>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>> + .shift = 29,
>>>>> + .width = 1,
>>>>> + },
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "hdmi_pll_dco",
>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>> + { .fw_name = "xtal", }
>>>>> + },
>>>>> + .num_parents = 1,
>>>>> + /*
>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>> + * possible
>>>>> + */
>>>>
>>>> Is it really ?
>>>>
>>>> Given that HDMI support for the s4 is there yet, the
>>>> addresses have changes and the region is no longer a syscon, it is time
>>>> for the HDMI driver to get fixed.
>> The HDMI PLL is configured in the Uboot phase and does not change the
>> frequency in the kernel phase. So we use the NOCACHE flag and
>> "ro_ops".
>
> That's no reason to put NOCACHE or ro-ops
>
> If you want the frequencies to be statically assinged, the correct way
> would be through assigned-rate in DT I guess.
Okay. You're right. However, when registering with OPS, HDMI PLL will be
reset. It takes time for PLL to stabilize the output frequency, which
will lead to the startup screen flashing.
I would like to know how to solve this problem if not using ro_ops.
>
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-30 6:05 ` Yu Tu
@ 2022-08-30 6:36 ` Jerome Brunet
2022-08-30 7:06 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-30 6:36 UTC (permalink / raw)
To: Yu Tu, Krzysztof Kozlowski, linux-clk, linux-arm-kernel,
linux-amlogic, linux-kernel, devicetree, Rob Herring,
Neil Armstrong, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On Tue 30 Aug 2022 at 14:05, Yu Tu <yu.tu@amlogic.com> wrote:
> On 2022/8/29 17:43, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>> On Mon 15 Aug 2022 at 14:17, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>>> Hi Jerome,
>>>
>>> On 2022/8/10 21:32, Jerome Brunet wrote:
>>>> [ EXTERNAL EMAIL ]
>>>> On Fri 05 Aug 2022 at 17:39, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>
>>>>> Hi Krzysztof,
>>>>> Thank you for your reply.
>>>>>
>>>>> On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
>>>>>> [ EXTERNAL EMAIL ]
>>>>>> On 05/08/2022 10:57, Yu Tu wrote:
>>>>>>> Added information about the S4 SOC PLL Clock controller in DT.
>>>>>>>
>>>>>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>>>>>> ---
>>>>>>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>>>>>>> 1 file changed, 8 insertions(+)
>>>>>>>
>>>>>>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>>> index ff213618a598..a816b1f7694b 100644
>>>>>>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>>>>>>> #size-cells = <2>;
>>>>>>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>>>>>> + clkc_pll: pll-clock-controller@8000 {
>>>>>> Node names should be generic - clock-controller.
>>>>>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>>>>>
>>>>> I will change to clkc_pll: clock-controller@8000, in next version.
>>>> Same comment applies to the binding doc.
>>> OKay.
>>>> Also it would be nice to split this in two series.
>>>> Bindings and drivers in one, arm64 dt in the other. These changes goes
>>>> in through different trees.
>>> At present, Bindings, DTS and drivers are three series. Do you mean to put
>>> Bindings and drivers together? If so, checkpatch.pl will report a warning.
>> Yes because patches are not in yet so there is a good reason to ignore
>> the warning. Warning will never show up on the actual tree if the
>> patches are correctly ordered.
>
> I think Binding, DTS and drivers use three series and you said two series
> is not a big problem. Three series are recommended for checkpatch.pl, I
> think it should be easy for that to separate and merge。
No - There is only 2 series. 1 for the bindings and clock drivers and
one for the DT once things are in
>
> I've sent it to V4. Please look at V4 and give some comments.
>
That's not how it works. You sent that before v3 review was done. There
are still comments that needed to be addressed
Given the time it takes to make that review I going to completly skip v4
and I'd like on the comment to addressed before you send another version
>>
>>>
>>>>
>>>>>> Best regards,
>>>>>> Krzysztof
>>>>>> .
>>>> .
>> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-30 6:13 ` Yu Tu
@ 2022-08-30 6:44 ` Jerome Brunet
2022-08-30 7:37 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-08-30 6:44 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
> On 2022/8/29 17:48, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>>>>>> +
>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>> + .en = {
>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>> + .shift = 28,
>>>>>> + .width = 1,
>>>>>> + },
>>>>>> + .m = {
>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>> + .shift = 0,
>>>>>> + .width = 8,
>>>>>> + },
>>>>>> + .n = {
>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>> + .shift = 10,
>>>>>> + .width = 5,
>>>>>> + },
>>>>>> + .frac = {
>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>> + .shift = 0,
>>>>>> + .width = 17,
>>>>>> + },
>>>>>> + .l = {
>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>> + .shift = 31,
>>>>>> + .width = 1,
>>>>>> + },
>>>>>> + .rst = {
>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>> + .shift = 29,
>>>>>> + .width = 1,
>>>>>> + },
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "hdmi_pll_dco",
>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>> + { .fw_name = "xtal", }
>>>>>> + },
>>>>>> + .num_parents = 1,
>>>>>> + /*
>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>> + * possible
>>>>>> + */
>>>>>
>>>>> Is it really ?
>>>>>
>>>>> Given that HDMI support for the s4 is there yet, the
>>>>> addresses have changes and the region is no longer a syscon, it is time
>>>>> for the HDMI driver to get fixed.
>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>> "ro_ops".
>> That's no reason to put NOCACHE or ro-ops
>> If you want the frequencies to be statically assinged, the correct way
>> would be through assigned-rate in DT I guess.
>
> Okay. You're right. However, when registering with OPS, HDMI PLL will be
> reset. It takes time for PLL to stabilize the output frequency, which will
> lead to the startup screen flashing.
>
> I would like to know how to solve this problem if not using ro_ops.
>
>>
You can add new ops or tweak the current init function.
Safest would be to do the following :
* Check if the PLLs is already on.
* Check if the 'pll->init_regs' matches what is already set
- if so, you can skip the reset
- if not, you need to reset as usual
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT
2022-08-30 6:36 ` Jerome Brunet
@ 2022-08-30 7:06 ` Yu Tu
0 siblings, 0 replies; 41+ messages in thread
From: Yu Tu @ 2022-08-30 7:06 UTC (permalink / raw)
To: Jerome Brunet, Krzysztof Kozlowski, linux-clk, linux-arm-kernel,
linux-amlogic, linux-kernel, devicetree, Rob Herring,
Neil Armstrong, Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/30 14:36, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Tue 30 Aug 2022 at 14:05, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> On 2022/8/29 17:43, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>> On Mon 15 Aug 2022 at 14:17, Yu Tu <yu.tu@amlogic.com> wrote:
>>>
>>>> Hi Jerome,
>>>>
>>>> On 2022/8/10 21:32, Jerome Brunet wrote:
>>>>> [ EXTERNAL EMAIL ]
>>>>> On Fri 05 Aug 2022 at 17:39, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>
>>>>>> Hi Krzysztof,
>>>>>> Thank you for your reply.
>>>>>>
>>>>>> On 2022/8/5 17:16, Krzysztof Kozlowski wrote:
>>>>>>> [ EXTERNAL EMAIL ]
>>>>>>> On 05/08/2022 10:57, Yu Tu wrote:
>>>>>>>> Added information about the S4 SOC PLL Clock controller in DT.
>>>>>>>>
>>>>>>>> Signed-off-by: Yu Tu <yu.tu@amlogic.com>
>>>>>>>> ---
>>>>>>>> arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 8 ++++++++
>>>>>>>> 1 file changed, 8 insertions(+)
>>>>>>>>
>>>>>>>> diff --git a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>>>> index ff213618a598..a816b1f7694b 100644
>>>>>>>> --- a/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>>>> +++ b/arch/arm64/boot/dts/amlogic/meson-s4.dtsi
>>>>>>>> @@ -92,6 +92,14 @@ apb4: apb4@fe000000 {
>>>>>>>> #size-cells = <2>;
>>>>>>>> ranges = <0x0 0x0 0x0 0xfe000000 0x0 0x480000>;
>>>>>>>> + clkc_pll: pll-clock-controller@8000 {
>>>>>>> Node names should be generic - clock-controller.
>>>>>>> https://devicetree-specification.readthedocs.io/en/latest/chapter2-devicetree-basics.html#generic-names-recommendation
>>>>>>>
>>>>>> I will change to clkc_pll: clock-controller@8000, in next version.
>>>>> Same comment applies to the binding doc.
>>>> OKay.
>>>>> Also it would be nice to split this in two series.
>>>>> Bindings and drivers in one, arm64 dt in the other. These changes goes
>>>>> in through different trees.
>>>> At present, Bindings, DTS and drivers are three series. Do you mean to put
>>>> Bindings and drivers together? If so, checkpatch.pl will report a warning.
>>> Yes because patches are not in yet so there is a good reason to ignore
>>> the warning. Warning will never show up on the actual tree if the
>>> patches are correctly ordered.
>>
>> I think Binding, DTS and drivers use three series and you said two series
>> is not a big problem. Three series are recommended for checkpatch.pl, I
>> think it should be easy for that to separate and merge。
>
> No - There is only 2 series. 1 for the bindings and clock drivers and
> one for the DT once things are in
All right, we'll do it your way.
>
>>
>> I've sent it to V4. Please look at V4 and give some comments.
>>
>
> That's not how it works. You sent that before v3 review was done. There
> are still comments that needed to be addressed
Yes. But can you reply faster?
>
> Given the time it takes to make that review I going to completly skip v4
> and I'd like on the comment to addressed before you send another version
>
What can I say? It's up to you.
>
>>>
>>>>
>>>>>
>>>>>>> Best regards,
>>>>>>> Krzysztof
>>>>>>> .
>>>>> .
>>> .
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-30 6:44 ` Jerome Brunet
@ 2022-08-30 7:37 ` Yu Tu
2022-09-21 8:40 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-30 7:37 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/30 14:44, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> On 2022/8/29 17:48, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>
>>>>>>> +
>>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>>> + .en = {
>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>> + .shift = 28,
>>>>>>> + .width = 1,
>>>>>>> + },
>>>>>>> + .m = {
>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>> + .shift = 0,
>>>>>>> + .width = 8,
>>>>>>> + },
>>>>>>> + .n = {
>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>> + .shift = 10,
>>>>>>> + .width = 5,
>>>>>>> + },
>>>>>>> + .frac = {
>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>>> + .shift = 0,
>>>>>>> + .width = 17,
>>>>>>> + },
>>>>>>> + .l = {
>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>> + .shift = 31,
>>>>>>> + .width = 1,
>>>>>>> + },
>>>>>>> + .rst = {
>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>> + .shift = 29,
>>>>>>> + .width = 1,
>>>>>>> + },
>>>>>>> + },
>>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>>> + .name = "hdmi_pll_dco",
>>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>>> + { .fw_name = "xtal", }
>>>>>>> + },
>>>>>>> + .num_parents = 1,
>>>>>>> + /*
>>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>>> + * possible
>>>>>>> + */
>>>>>>
>>>>>> Is it really ?
>>>>>>
>>>>>> Given that HDMI support for the s4 is there yet, the
>>>>>> addresses have changes and the region is no longer a syscon, it is time
>>>>>> for the HDMI driver to get fixed.
>>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>>> "ro_ops".
>>> That's no reason to put NOCACHE or ro-ops
>>> If you want the frequencies to be statically assinged, the correct way
>>> would be through assigned-rate in DT I guess.
>>
>> Okay. You're right. However, when registering with OPS, HDMI PLL will be
>> reset. It takes time for PLL to stabilize the output frequency, which will
>> lead to the startup screen flashing.
>>
>> I would like to know how to solve this problem if not using ro_ops.
>>
>>>
>
> You can add new ops or tweak the current init function.
HDMI PLL is not different from other PLLS, so I think adding OPS is weird.
>
> Safest would be to do the following :
> * Check if the PLLs is already on.
> * Check if the 'pll->init_regs' matches what is already set
> - if so, you can skip the reset
> - if not, you need to reset as usual
static int meson_clk_pll_init(struct clk_hw *hw)
{
struct clk_regmap *clk = to_clk_regmap(hw);
struct meson_clk_pll_data *pll = meson_clk_pll_data(clk);
if (pll->init_count) {
meson_parm_write(clk->map, &pll->rst, 1);
regmap_multi_reg_write(clk->map, pll->init_regs,
| pll->init_count);
meson_parm_write(clk->map, &pll->rst, 0);
}
return 0;
}
Because the init function looks like this. Therefore, HDMI PLL
init_count is not given. Can I change it like this?
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 6/6] clk: meson: s4: add s4 SoC peripheral clock controller driver
2022-08-29 12:19 ` [PATCH V3 6/6] clk: meson: s4: add s4 SoC peripheral clock controller driver Jerome Brunet
@ 2022-08-30 8:20 ` Yu Tu
2022-09-21 9:01 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-08-30 8:20 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
On 2022/8/29 20:19, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Tue 16 Aug 2022 at 20:00, Yu Tu <yu.tu@amlogic.com> wrote:
>
> Please trim your replies
>
>>>> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
>>>> index f4244edc7b28..ec6beb9284d3 100644
>>>> --- a/drivers/clk/meson/Kconfig
>>>> +++ b/drivers/clk/meson/Kconfig
>>>> @@ -127,4 +127,17 @@ config COMMON_CLK_S4_PLL
>>>> Support for the pll clock controller on Amlogic S805X2 and S905Y4 devices,
>>>> aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and AQ229.
>>>> Say Y if you want peripherals and CPU frequency scaling to work.
>>>> +
>>>> +config COMMON_CLK_S4
>>>> + tristate "S4 SoC Peripherals clock controllers support"
>>>> + depends on ARM64
>>>> + default y
>>>> + select COMMON_CLK_MESON_REGMAP
>>>> + select COMMON_CLK_MESON_DUALDIV
>>>> + select COMMON_CLK_MESON_VID_PLL_DIV
>>>> + select COMMON_CLK_S4_PLL
>>> Do you really this ? your driver does not even include the related
>>> header.
>> If the PLL driver is not turned on in DTS, will it not cause an error?
>>>
>
> I don't get the question.
> Kconfig list compile deps. S4 PLL is not a compile dep of the peripheral
> controller.
>
> If you really want to, you may use 'imply'.
V4 has been changed as you suggested.
>>>
>>>> +static const struct clk_parent_data sys_ab_clk_parent_data[] = {
>>>> + { .fw_name = "xtal" },
>>>> + { .fw_name = "fclk_div2" },
>>>> + { .fw_name = "fclk_div3" },
>>>> + { .fw_name = "fclk_div4" },
>>>> + { .fw_name = "fclk_div5" },
>>>> + { .fw_name = "fclk_div7" },
>>>> + { .hw = &s4_rtc_clk.hw }
>>>> +};
>>>> +
>>>> +static struct clk_regmap s4_sysclk_b_sel = {
>>>> + .data = &(struct clk_regmap_mux_data){
>>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>>> + .mask = 0x7,
>>>> + .shift = 26,
>>>> + .table = mux_table_sys_ab_clk_sel,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "sysclk_b_sel",
>>>> + .ops = &clk_regmap_mux_ro_ops,
>>> Why is this using the RO ops ?
>> Sys_clk is initialized during the Uboot phase and is fixed at
>> 166.666MHz. So I'm going to change it to ro.
>
> That really much depends on the bootloader and is a pretty weak design.
> The bootloader deps should be kept as minimal as possible.
>
> I see no reason for RO.
>
> You may cut rate propagation on the user if you need to and continue to
> whatever you want in your u-boot
I think I know what you mean. But we let the user be in control and not
set the frequency, which can be risky. If you insist, I will change it
as you suggest.
>
>>>
>>>> + .parent_data = sys_ab_clk_parent_data,
>>>> + .num_parents = ARRAY_SIZE(sys_ab_clk_parent_data),
>>>> + },
>>>> +};
>>>> +
>>>> +static struct clk_regmap s4_sysclk_b_div = {
>>>> + .data = &(struct clk_regmap_div_data){
>>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>>> + .shift = 16,
>>>> + .width = 10,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "sysclk_b_div",
>>>> + .ops = &clk_regmap_divider_ro_ops,
>>> Same here and for the rest of the sys part
>> Same above.
>
> We can play that game for a while
Ah, you're so funny.
>
>>>> +
>>>> +/* Video Clocks */
>>>> +static struct clk_regmap s4_vid_pll_div = {
>>>> + .data = &(struct meson_vid_pll_div_data){
>>>> + .val = {
>>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>>> + .shift = 0,
>>>> + .width = 15,
>>>> + },
>>>> + .sel = {
>>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>>> + .shift = 16,
>>>> + .width = 2,
>>>> + },
>>>> + },
>>>> + .hw.init = &(struct clk_init_data) {
>>>> + .name = "vid_pll_div",
>>>> + .ops = &meson_vid_pll_div_ro_ops,
>>> Why RO ? applies to the rest of the video part.
>> Because vid_pll_div parent is HDMI_PLL, and HDMI_PLL is a fixed
>> frequency. Flags is CLK_SET_RATE_PARENT. So we use RO.
>
> If the HDMI_PLL is fixed somehow, that is not reason for this clock to
> be RO
>
>> Can I remove RO and use CLK_SET_RATE_NO_REPARENT instead, which one do you
>> think is more reasonable?
>
> Neither. CLK_SET_RATE_NO_REPARENT makes no sense, it is not mux
>
"drivers/clk/meson/vid-pll-div.c"
This file only provides ro_ops. Maybe the submission records will give
us the answer.
In fact, our hardware design is the same as the G12 series.
>>
>>>
>>>> + .parent_data = (const struct clk_parent_data []) {
>>>> + { .fw_name = "hdmi_pll", }
>>>> + },
>>>> + .num_parents = 1,
>>>> + .flags = CLK_SET_RATE_PARENT,
>>>> + },
>>>> +};
>>>> +
>>>> +static struct clk_regmap s4_vid_pll_sel = {
>>>> + .data = &(struct clk_regmap_mux_data){
>>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>>> + .mask = 0x1,
>>>> + .shift = 18,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "vid_pll_sel",
>>>> + .ops = &clk_regmap_mux_ops,
>>>> + /*
>>>> + * bit 18 selects from 2 possible parents:
>>>> + * vid_pll_div or hdmi_pll
>>>> + */
>>>> + .parent_data = (const struct clk_parent_data []) {
>>>> + { .hw = &s4_vid_pll_div.hw },
>>>> + { .fw_name = "hdmi_pll", }
>>>> + },
>>>> + .num_parents = 2,
>>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>>> Why ? are you planning to DT assigned clocks to statically set this ?
>> Because vid_pll_sel one parent is HDMI_PLL, and HDMI_PLL is a fixed
>> frequency. To prevent modification, use CLK_SET_RATE_NO_REPARENT.
>
> Again, this makes no sense.
Unfortunately you don't read V4, in fact I have corrected in V4.
".flags = CLK_SET_RATE_PARENT," in V4. Is that okay with you?
>
>>>
>>>> + },
>>>> +};
>>>> +
>>>> +static struct clk_regmap s4_vid_pll = {
>>>> + .data = &(struct clk_regmap_gate_data){
>>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>>> + .bit_idx = 19,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data) {
>>>> + .name = "vid_pll",
>>>> + .ops = &clk_regmap_gate_ops,
>>>> + .parent_hws = (const struct clk_hw *[]) {
>>>> + &s4_vid_pll_sel.hw
>>>> + },
>>>> + .num_parents = 1,
>>>> + .flags = CLK_SET_RATE_PARENT,
>>>> + },
>>>> +};
>>>> +
>>>> +static const struct clk_parent_data s4_vclk_parent_data[] = {
>>>> + { .hw = &s4_vid_pll.hw },
>>>> + { .fw_name = "gp0_pll", },
>>>> + { .fw_name = "hifi_pll", },
>>>> + { .fw_name = "mpll1", },
>>>> + { .fw_name = "fclk_div3", },
>>>> + { .fw_name = "fclk_div4", },
>>>> + { .fw_name = "fclk_div5", },
>>>> + { .fw_name = "fclk_div7", },
>>>> +};
>>>> +
>>>> +static struct clk_regmap s4_vclk_sel = {
>>>> + .data = &(struct clk_regmap_mux_data){
>>>> + .offset = CLKCTRL_VID_CLK_CTRL,
>>>> + .mask = 0x7,
>>>> + .shift = 16,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "vclk_sel",
>>>> + .ops = &clk_regmap_mux_ops,
>>>> + .parent_data = s4_vclk_parent_data,
>>>> + .num_parents = ARRAY_SIZE(s4_vclk_parent_data),
>>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>>> Same
>> Since fclk_div* is a fixed frequency value, mplL1 and hifi_pll and gp0_pll
>> are used by other specialized modules, vid_pll has CLK_SET_RATE_PARENT. The
>> parent of vid_pll is that vid_pll_sel uses CLK_SET_RATE_NO_REPARENT.
>
> Still not good.
>
> You don't have CLK_SET_RATE, propagation is stopped and parent clock
> will not changed. The best parent will be picked but not changed.
>
> If one parent MUST NOT be picked, just remove it from the list and add a
> explaining why
>
> [...]
Okay.
>
>>>> +
>>>> +static struct clk_regmap s4_ts_clk_div = {
>>>> + .data = &(struct clk_regmap_div_data){
>>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>>> + .shift = 0,
>>>> + .width = 8,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "ts_clk_div",
>>>> + .ops = &clk_regmap_divider_ops,
>>>> + .parent_data = &(const struct clk_parent_data) {
>>>> + .fw_name = "xtal",
>>>> + },
>>>> + .num_parents = 1,
>>> propagation stopped ?
>> Its parent is xtal, so I should use CLK_SET_RATE_NO_REPARENT.
>
> Still no. You seem to have problem with the meaning of
> CLK_SET_RATE_NO_REPARENT.
>
> * CLK_SET_RATE_NO_REPARENT: means the parent will no be changed, even if
> selecting another parent would result in a closer rate to the
> request. It makes sense only if the clock has several parents
>
> * CLK_SET_RATE_PARENT: means rate change may propagate the parent,
> meaning the rate of the parent may change if it help the child achieve
> a closer rate to the request
Thank you for explaining.I got it.
>
>>>
>>>> + },
>>>> +};
>>>> +
>>>> +static struct clk_regmap s4_ts_clk_gate = {
>>>> + .data = &(struct clk_regmap_gate_data){
>>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>>> + .bit_idx = 8,
>>>> + },
>>>> + .hw.init = &(struct clk_init_data){
>>>> + .name = "ts_clk",
>>>> + .ops = &clk_regmap_gate_ops,
>>>> + .parent_hws = (const struct clk_hw *[]) {
>>>> + &s4_ts_clk_div.hw
>>>> + },
>>>> + .num_parents = 1,
>>>> + },
>>> propagation stopped ?
>> I will add CLK_SET_RATE_PARENT.
>
> [...]
>
>>>> +/* EMMC/NAND clock */
>>>> +
>>>> +static const struct clk_parent_data s4_sd_emmc_clk0_parent_data[] = {
>>>> + { .fw_name = "xtal", },
>>>> + { .fw_name = "fclk_div2", },
>>>> + { .fw_name = "fclk_div3", },
>>>> + { .fw_name = "hifi_pll", },
>>>> + { .fw_name = "fclk_div2p5", },
>>>> + /*
>>>> + * Following these parent clocks, we should also have had mpll2, mpll3
>>>> + * and gp0_pll but these clocks are too precious to be used here. All
>>>> + * the necessary rates for MMC and NAND operation can be acheived using
>>>> + * hifi_pll or fclk_div clocks
>>>> + */
>>> You don't want to list mplls but hifi_pll is fine ? seems dangerous.
>> hifi pll is for EMMC and NAND on this SoC.
>
> That deserve a better explanation.
> Why can't it use fdiv2 and xtal like the previous SoCs ?
>
> Which PLLs are you using for Audio then ?
> Typical operation on these SoCs usually require 3 PLLs to acheive all rates
>
I'll list all the clocks and let the driver itself select Parent as needed.
>>>
>
>
>>>> +/*
>>>> + * gen clk is designed for debug/monitor some internal clock quality. Some of the
>>>> + * corresponding clock sources are not described in the clock tree, so they are skipped.
>>>> + */
>>> Still feels a bit light, don't you think ? Among all the clocks, can't
>>> you add a bit more parents here ? It would certainly help debug down the road
>> [16:12] is gen_clk source select.All is:
>> 0: cts_oscin_clk
>> 1:cts_rtc_clk
>> 2:sys_pll_div16 (internal clock)
>> 3:ddr_pll_div32 (internal clock)
>> 4: vid_pll
>> 5: gp0_pll
>> 7: hifi_pll
>> 10:adc_dpll_clk_b3 (internal clock for debug)
>> 11:adc_dpll_intclk (internal clock for debug)
>> 12:clk_msr_src(select from all internal clock except PLLs);
>> 16: no used
>> 17: sys_cpu_clk_div16 (internal clock)
>> 19: fclk_div2
>> 20: fclk_div2p5
>> 21: fclk_div3
>> 22: fclk_div4
>> 23: fclk_div5
>> 24: fclk_div7
>> 25: mpll0
>> 26: mpll1
>> 27: mpll2
>> 28: mpll3
>> So i only added the clocks that will actually be used, and some debugging
>> clock peripherals will not be used.
>
> you may at least add vid_pll
Okay.
>
>>>
>>>> +static u32 s4_gen_clk_mux_table[] = { 0, 5, 7, 19, 21, 22,
>>>> + 23, 24, 25, 26, 27, 28 };
>>>> +static const struct clk_parent_data s4_gen_clk_parent_data[] = {
>>>> + { .fw_name = "xtal", },
>>>> + { .fw_name = "gp0_pll", },
>>>> + { .fw_name = "hifi_pll", },
>>>> + { .fw_name = "fclk_div2", },
>>>> + { .fw_name = "fclk_div3", },
>>>> + { .fw_name = "fclk_div4", },
>>>> + { .fw_name = "fclk_div5", },
>>>> + { .fw_name = "fclk_div7", },
>>>> + { .fw_name = "mpll0", },
>>>> + { .fw_name = "mpll1", },
>>>> + { .fw_name = "mpll2", },
>>>> + { .fw_name = "mpll3", },
>>>> +};
>
> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-08-30 7:37 ` Yu Tu
@ 2022-09-21 8:40 ` Yu Tu
2022-09-28 15:27 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-09-21 8:40 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Jerome,
On 2022/8/30 15:37, Yu Tu wrote:
>
>
> On 2022/8/30 14:44, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>>
>>
>> On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>>> On 2022/8/29 17:48, Jerome Brunet wrote:
>>>> [ EXTERNAL EMAIL ]
>>>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>
>>>>>>>> +
>>>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>>>> + .en = {
>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>> + .shift = 28,
>>>>>>>> + .width = 1,
>>>>>>>> + },
>>>>>>>> + .m = {
>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>> + .shift = 0,
>>>>>>>> + .width = 8,
>>>>>>>> + },
>>>>>>>> + .n = {
>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>> + .shift = 10,
>>>>>>>> + .width = 5,
>>>>>>>> + },
>>>>>>>> + .frac = {
>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>>>> + .shift = 0,
>>>>>>>> + .width = 17,
>>>>>>>> + },
>>>>>>>> + .l = {
>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>> + .shift = 31,
>>>>>>>> + .width = 1,
>>>>>>>> + },
>>>>>>>> + .rst = {
>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>> + .shift = 29,
>>>>>>>> + .width = 1,
>>>>>>>> + },
>>>>>>>> + },
>>>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>>>> + .name = "hdmi_pll_dco",
>>>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>>>> + { .fw_name = "xtal", }
>>>>>>>> + },
>>>>>>>> + .num_parents = 1,
>>>>>>>> + /*
>>>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>>>> + * possible
>>>>>>>> + */
>>>>>>>
>>>>>>> Is it really ?
>>>>>>>
>>>>>>> Given that HDMI support for the s4 is there yet, the
>>>>>>> addresses have changes and the region is no longer a syscon, it
>>>>>>> is time
>>>>>>> for the HDMI driver to get fixed.
>>>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>>>> "ro_ops".
>>>> That's no reason to put NOCACHE or ro-ops
>>>> If you want the frequencies to be statically assinged, the correct way
>>>> would be through assigned-rate in DT I guess.
>>>
>>> Okay. You're right. However, when registering with OPS, HDMI PLL will be
>>> reset. It takes time for PLL to stabilize the output frequency, which
>>> will
>>> lead to the startup screen flashing.
>>>
>>> I would like to know how to solve this problem if not using ro_ops.
>>>
>>>>
>>
>> You can add new ops or tweak the current init function.
>
> HDMI PLL is not different from other PLLS, so I think adding OPS is weird.
>
>>
>> Safest would be to do the following :
>> * Check if the PLLs is already on.
>> * Check if the 'pll->init_regs' matches what is already set
>> - if so, you can skip the reset
>> - if not, you need to reset as usual
>
> static int meson_clk_pll_init(struct clk_hw *hw)
> {
> struct clk_regmap *clk = to_clk_regmap(hw);
> struct meson_clk_pll_data *pll = meson_clk_pll_data(clk);
>
>
> if (pll->init_count) {
> meson_parm_write(clk->map, &pll->rst, 1);
> regmap_multi_reg_write(clk->map, pll->init_regs,
> | pll->init_count);
> meson_parm_write(clk->map, &pll->rst, 0);
> }
>
>
> return 0;
> }
>
> Because the init function looks like this. Therefore, HDMI PLL
> init_count is not given. Can I change it like this?
I don't know if this change meets your requirements? Please give us your
valuable advice.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 6/6] clk: meson: s4: add s4 SoC peripheral clock controller driver
2022-08-30 8:20 ` Yu Tu
@ 2022-09-21 9:01 ` Yu Tu
2022-09-28 15:35 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-09-21 9:01 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Jerome,
On 2022/8/30 16:20, Yu Tu wrote:
>
>
> On 2022/8/29 20:19, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>>
>>
>> On Tue 16 Aug 2022 at 20:00, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>> Please trim your replies
>>
>>>>> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
>>>>> index f4244edc7b28..ec6beb9284d3 100644
>>>>> --- a/drivers/clk/meson/Kconfig
>>>>> +++ b/drivers/clk/meson/Kconfig
>>>>> @@ -127,4 +127,17 @@ config COMMON_CLK_S4_PLL
>>>>> Support for the pll clock controller on Amlogic S805X2 and
>>>>> S905Y4 devices,
>>>>> aka s4. Amlogic S805X2 and S905Y4 devices include AQ222
>>>>> and AQ229.
>>>>> Say Y if you want peripherals and CPU frequency scaling to
>>>>> work.
>>>>> +
>>>>> +config COMMON_CLK_S4
>>>>> + tristate "S4 SoC Peripherals clock controllers support"
>>>>> + depends on ARM64
>>>>> + default y
>>>>> + select COMMON_CLK_MESON_REGMAP
>>>>> + select COMMON_CLK_MESON_DUALDIV
>>>>> + select COMMON_CLK_MESON_VID_PLL_DIV
>>>>> + select COMMON_CLK_S4_PLL
>>>> Do you really this ? your driver does not even include the related
>>>> header.
>>> If the PLL driver is not turned on in DTS, will it not cause an error?
>>>>
>>
>> I don't get the question.
>> Kconfig list compile deps. S4 PLL is not a compile dep of the peripheral
>> controller.
>>
>> If you really want to, you may use 'imply'.
>
> V4 has been changed as you suggested.
The next edition is being changed according to your requirements. Please
give us your valuable opinions.
>
>>>>
>>>>> +static const struct clk_parent_data sys_ab_clk_parent_data[] = {
>>>>> + { .fw_name = "xtal" },
>>>>> + { .fw_name = "fclk_div2" },
>>>>> + { .fw_name = "fclk_div3" },
>>>>> + { .fw_name = "fclk_div4" },
>>>>> + { .fw_name = "fclk_div5" },
>>>>> + { .fw_name = "fclk_div7" },
>>>>> + { .hw = &s4_rtc_clk.hw }
>>>>> +};
>>>>> +
>>>>> +static struct clk_regmap s4_sysclk_b_sel = {
>>>>> + .data = &(struct clk_regmap_mux_data){
>>>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>>>> + .mask = 0x7,
>>>>> + .shift = 26,
>>>>> + .table = mux_table_sys_ab_clk_sel,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "sysclk_b_sel",
>>>>> + .ops = &clk_regmap_mux_ro_ops,
>>>> Why is this using the RO ops ?
>>> Sys_clk is initialized during the Uboot phase and is fixed at
>>> 166.666MHz. So I'm going to change it to ro.
>>
>> That really much depends on the bootloader and is a pretty weak design.
>> The bootloader deps should be kept as minimal as possible.
>>
>> I see no reason for RO.
>>
>> You may cut rate propagation on the user if you need to and continue to
>> whatever you want in your u-boot
>
> I think I know what you mean. But we let the user be in control and not
> set the frequency, which can be risky. If you insist, I will change it
> as you suggest.
It has been changed as you requested.
>
>>
>>>>
>>>>> + .parent_data = sys_ab_clk_parent_data,
>>>>> + .num_parents = ARRAY_SIZE(sys_ab_clk_parent_data),
>>>>> + },
>>>>> +};
>>>>> +
>>>>> +static struct clk_regmap s4_sysclk_b_div = {
>>>>> + .data = &(struct clk_regmap_div_data){
>>>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>>>> + .shift = 16,
>>>>> + .width = 10,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "sysclk_b_div",
>>>>> + .ops = &clk_regmap_divider_ro_ops,
>>>> Same here and for the rest of the sys part
>>> Same above.
>>
>> We can play that game for a while
>
> Ah, you're so funny.
>
>>
>>>>> +
>>>>> +/* Video Clocks */
>>>>> +static struct clk_regmap s4_vid_pll_div = {
>>>>> + .data = &(struct meson_vid_pll_div_data){
>>>>> + .val = {
>>>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>>>> + .shift = 0,
>>>>> + .width = 15,
>>>>> + },
>>>>> + .sel = {
>>>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>>>> + .shift = 16,
>>>>> + .width = 2,
>>>>> + },
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data) {
>>>>> + .name = "vid_pll_div",
>>>>> + .ops = &meson_vid_pll_div_ro_ops,
>>>> Why RO ? applies to the rest of the video part.
>>> Because vid_pll_div parent is HDMI_PLL, and HDMI_PLL is a fixed
>>> frequency. Flags is CLK_SET_RATE_PARENT. So we use RO.
>>
>> If the HDMI_PLL is fixed somehow, that is not reason for this clock to
>> be RO
>>
>>> Can I remove RO and use CLK_SET_RATE_NO_REPARENT instead, which one
>>> do you
>>> think is more reasonable?
>>
>> Neither. CLK_SET_RATE_NO_REPARENT makes no sense, it is not mux
>>
>
> "drivers/clk/meson/vid-pll-div.c"
> This file only provides ro_ops. Maybe the submission records will give
> us the answer.
>
> In fact, our hardware design is the same as the G12 series.
I don't know if you checked this commit, but there is only one
"ro_ops"in this place right now.
The S4 SoC is consistent with the G12A/B and GX series.
>
>>>
>>>>
>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>> + { .fw_name = "hdmi_pll", }
>>>>> + },
>>>>> + .num_parents = 1,
>>>>> + .flags = CLK_SET_RATE_PARENT,
>>>>> + },
>>>>> +};
>>>>> +
>>>>> +static struct clk_regmap s4_vid_pll_sel = {
>>>>> + .data = &(struct clk_regmap_mux_data){
>>>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>>>> + .mask = 0x1,
>>>>> + .shift = 18,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "vid_pll_sel",
>>>>> + .ops = &clk_regmap_mux_ops,
>>>>> + /*
>>>>> + * bit 18 selects from 2 possible parents:
>>>>> + * vid_pll_div or hdmi_pll
>>>>> + */
>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>> + { .hw = &s4_vid_pll_div.hw },
>>>>> + { .fw_name = "hdmi_pll", }
>>>>> + },
>>>>> + .num_parents = 2,
>>>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>>>> Why ? are you planning to DT assigned clocks to statically set this ?
>>> Because vid_pll_sel one parent is HDMI_PLL, and HDMI_PLL is a fixed
>>> frequency. To prevent modification, use CLK_SET_RATE_NO_REPARENT.
>>
>> Again, this makes no sense.
>
> Unfortunately you don't read V4, in fact I have corrected in V4.
>
> ".flags = CLK_SET_RATE_PARENT," in V4. Is that okay with you?
I don't know what you think?
>
>>
>>>>
>>>>> + },
>>>>> +};
>>>>> +
>>>>> +static struct clk_regmap s4_vid_pll = {
>>>>> + .data = &(struct clk_regmap_gate_data){
>>>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>>>> + .bit_idx = 19,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data) {
>>>>> + .name = "vid_pll",
>>>>> + .ops = &clk_regmap_gate_ops,
>>>>> + .parent_hws = (const struct clk_hw *[]) {
>>>>> + &s4_vid_pll_sel.hw
>>>>> + },
>>>>> + .num_parents = 1,
>>>>> + .flags = CLK_SET_RATE_PARENT,
>>>>> + },
>>>>> +};
>>>>> +
>>>>> +static const struct clk_parent_data s4_vclk_parent_data[] = {
>>>>> + { .hw = &s4_vid_pll.hw },
>>>>> + { .fw_name = "gp0_pll", },
>>>>> + { .fw_name = "hifi_pll", },
>>>>> + { .fw_name = "mpll1", },
>>>>> + { .fw_name = "fclk_div3", },
>>>>> + { .fw_name = "fclk_div4", },
>>>>> + { .fw_name = "fclk_div5", },
>>>>> + { .fw_name = "fclk_div7", },
>>>>> +};
>>>>> +
>>>>> +static struct clk_regmap s4_vclk_sel = {
>>>>> + .data = &(struct clk_regmap_mux_data){
>>>>> + .offset = CLKCTRL_VID_CLK_CTRL,
>>>>> + .mask = 0x7,
>>>>> + .shift = 16,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "vclk_sel",
>>>>> + .ops = &clk_regmap_mux_ops,
>>>>> + .parent_data = s4_vclk_parent_data,
>>>>> + .num_parents = ARRAY_SIZE(s4_vclk_parent_data),
>>>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>>>> Same
>>> Since fclk_div* is a fixed frequency value, mplL1 and hifi_pll and
>>> gp0_pll
>>> are used by other specialized modules, vid_pll has
>>> CLK_SET_RATE_PARENT. The
>>> parent of vid_pll is that vid_pll_sel uses CLK_SET_RATE_NO_REPARENT.
>>
>> Still not good.
>>
>> You don't have CLK_SET_RATE, propagation is stopped and parent clock
>> will not changed. The best parent will be picked but not changed.
>>
>> If one parent MUST NOT be picked, just remove it from the list and add a
>> explaining why
>>
>> [...]
>
> Okay.
In the next edition I will change it to ".flags = CLK_SET_RATE_PARENT".
>
>>
>>>>> +
>>>>> +static struct clk_regmap s4_ts_clk_div = {
>>>>> + .data = &(struct clk_regmap_div_data){
>>>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>>>> + .shift = 0,
>>>>> + .width = 8,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "ts_clk_div",
>>>>> + .ops = &clk_regmap_divider_ops,
>>>>> + .parent_data = &(const struct clk_parent_data) {
>>>>> + .fw_name = "xtal",
>>>>> + },
>>>>> + .num_parents = 1,
>>>> propagation stopped ?
>>> Its parent is xtal, so I should use CLK_SET_RATE_NO_REPARENT.
>>
>> Still no. You seem to have problem with the meaning of
>> CLK_SET_RATE_NO_REPARENT.
>>
>> * CLK_SET_RATE_NO_REPARENT: means the parent will no be changed, even if
>> selecting another parent would result in a closer rate to the
>> request. It makes sense only if the clock has several parents
>>
>> * CLK_SET_RATE_PARENT: means rate change may propagate the parent,
>> meaning the rate of the parent may change if it help the child achieve
>> a closer rate to the request
>
> Thank you for explaining.I got it.
>
>>
>>>>
>>>>> + },
>>>>> +};
>>>>> +
>>>>> +static struct clk_regmap s4_ts_clk_gate = {
>>>>> + .data = &(struct clk_regmap_gate_data){
>>>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>>>> + .bit_idx = 8,
>>>>> + },
>>>>> + .hw.init = &(struct clk_init_data){
>>>>> + .name = "ts_clk",
>>>>> + .ops = &clk_regmap_gate_ops,
>>>>> + .parent_hws = (const struct clk_hw *[]) {
>>>>> + &s4_ts_clk_div.hw
>>>>> + },
>>>>> + .num_parents = 1,
>>>>> + },
>>>> propagation stopped ?
>>> I will add CLK_SET_RATE_PARENT.
>>
>> [...]
>>
>>>>> +/* EMMC/NAND clock */
>>>>> +
>>>>> +static const struct clk_parent_data s4_sd_emmc_clk0_parent_data[] = {
>>>>> + { .fw_name = "xtal", },
>>>>> + { .fw_name = "fclk_div2", },
>>>>> + { .fw_name = "fclk_div3", },
>>>>> + { .fw_name = "hifi_pll", },
>>>>> + { .fw_name = "fclk_div2p5", },
>>>>> + /*
>>>>> + * Following these parent clocks, we should also have had
>>>>> mpll2, mpll3
>>>>> + * and gp0_pll but these clocks are too precious to be used
>>>>> here. All
>>>>> + * the necessary rates for MMC and NAND operation can be
>>>>> acheived using
>>>>> + * hifi_pll or fclk_div clocks
>>>>> + */
>>>> You don't want to list mplls but hifi_pll is fine ? seems dangerous.
>>> hifi pll is for EMMC and NAND on this SoC.
>>
>> That deserve a better explanation.
>> Why can't it use fdiv2 and xtal like the previous SoCs ?
>>
>> Which PLLs are you using for Audio then ?
>> Typical operation on these SoCs usually require 3 PLLs to acheive all
>> rates
>>
>
> I'll list all the clocks and let the driver itself select Parent as needed.
I don't know what you think?
>
>>>>
>>
>>
>>>>> +/*
>>>>> + * gen clk is designed for debug/monitor some internal clock
>>>>> quality. Some of the
>>>>> + * corresponding clock sources are not described in the clock
>>>>> tree, so they are skipped.
>>>>> + */
>>>> Still feels a bit light, don't you think ? Among all the clocks, can't
>>>> you add a bit more parents here ? It would certainly help debug down
>>>> the road
>>> [16:12] is gen_clk source select.All is:
>>> 0: cts_oscin_clk
>>> 1:cts_rtc_clk
>>> 2:sys_pll_div16 (internal clock)
>>> 3:ddr_pll_div32 (internal clock)
>>> 4: vid_pll
>>> 5: gp0_pll
>>> 7: hifi_pll
>>> 10:adc_dpll_clk_b3 (internal clock for debug)
>>> 11:adc_dpll_intclk (internal clock for debug)
>>> 12:clk_msr_src(select from all internal clock except PLLs);
>>> 16: no used
>>> 17: sys_cpu_clk_div16 (internal clock)
>>> 19: fclk_div2
>>> 20: fclk_div2p5
>>> 21: fclk_div3
>>> 22: fclk_div4
>>> 23: fclk_div5
>>> 24: fclk_div7
>>> 25: mpll0
>>> 26: mpll1
>>> 27: mpll2
>>> 28: mpll3
>>> So i only added the clocks that will actually be used, and some
>>> debugging
>>> clock peripherals will not be used.
>>
>> you may at least add vid_pll
>
> Okay.
It has been changed as you suggested.
>
>>
>>>>
>>>>> +static u32 s4_gen_clk_mux_table[] = { 0, 5, 7, 19, 21, 22,
>>>>> + 23, 24, 25, 26, 27, 28 };
>>>>> +static const struct clk_parent_data s4_gen_clk_parent_data[] = {
>>>>> + { .fw_name = "xtal", },
>>>>> + { .fw_name = "gp0_pll", },
>>>>> + { .fw_name = "hifi_pll", },
>>>>> + { .fw_name = "fclk_div2", },
>>>>> + { .fw_name = "fclk_div3", },
>>>>> + { .fw_name = "fclk_div4", },
>>>>> + { .fw_name = "fclk_div5", },
>>>>> + { .fw_name = "fclk_div7", },
>>>>> + { .fw_name = "mpll0", },
>>>>> + { .fw_name = "mpll1", },
>>>>> + { .fw_name = "mpll2", },
>>>>> + { .fw_name = "mpll3", },
>>>>> +};
>>
>> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-09-21 8:40 ` Yu Tu
@ 2022-09-28 15:27 ` Jerome Brunet
2022-09-29 7:07 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-09-28 15:27 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Wed 21 Sep 2022 at 16:40, Yu Tu <yu.tu@amlogic.com> wrote:
> Hi Jerome,
>
> On 2022/8/30 15:37, Yu Tu wrote:
>> On 2022/8/30 14:44, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>>
>>>
>>> On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
>>>
>>>> On 2022/8/29 17:48, Jerome Brunet wrote:
>>>>> [ EXTERNAL EMAIL ]
>>>>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>
>>>>>>>>> +
>>>>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>>>>> + .en = {
>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>> + .shift = 28,
>>>>>>>>> + .width = 1,
>>>>>>>>> + },
>>>>>>>>> + .m = {
>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>> + .shift = 0,
>>>>>>>>> + .width = 8,
>>>>>>>>> + },
>>>>>>>>> + .n = {
>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>> + .shift = 10,
>>>>>>>>> + .width = 5,
>>>>>>>>> + },
>>>>>>>>> + .frac = {
>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>>>>> + .shift = 0,
>>>>>>>>> + .width = 17,
>>>>>>>>> + },
>>>>>>>>> + .l = {
>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>> + .shift = 31,
>>>>>>>>> + .width = 1,
>>>>>>>>> + },
>>>>>>>>> + .rst = {
>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>> + .shift = 29,
>>>>>>>>> + .width = 1,
>>>>>>>>> + },
>>>>>>>>> + },
>>>>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>>>>> + .name = "hdmi_pll_dco",
>>>>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>>>>> + { .fw_name = "xtal", }
>>>>>>>>> + },
>>>>>>>>> + .num_parents = 1,
>>>>>>>>> + /*
>>>>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>>>>> + * possible
>>>>>>>>> + */
>>>>>>>>
>>>>>>>> Is it really ?
>>>>>>>>
>>>>>>>> Given that HDMI support for the s4 is there yet, the
>>>>>>>> addresses have changes and the region is no longer a syscon, it is
>>>>>>>> time
>>>>>>>> for the HDMI driver to get fixed.
>>>>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>>>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>>>>> "ro_ops".
>>>>> That's no reason to put NOCACHE or ro-ops
>>>>> If you want the frequencies to be statically assinged, the correct way
>>>>> would be through assigned-rate in DT I guess.
>>>>
>>>> Okay. You're right. However, when registering with OPS, HDMI PLL will be
>>>> reset. It takes time for PLL to stabilize the output frequency, which
>>>> will
>>>> lead to the startup screen flashing.
>>>>
>>>> I would like to know how to solve this problem if not using ro_ops.
>>>>
>>>>>
>>>
>>> You can add new ops or tweak the current init function.
>> HDMI PLL is not different from other PLLS, so I think adding OPS is
>> weird.
>>
>>>
>>> Safest would be to do the following :
>>> * Check if the PLLs is already on.
>>> * Check if the 'pll->init_regs' matches what is already set
>>> - if so, you can skip the reset
>>> - if not, you need to reset as usual
>> static int meson_clk_pll_init(struct clk_hw *hw)
>> {
>> struct clk_regmap *clk = to_clk_regmap(hw);
>> struct meson_clk_pll_data *pll = meson_clk_pll_data(clk);
>> if (pll->init_count) {
>> meson_parm_write(clk->map, &pll->rst, 1);
>> regmap_multi_reg_write(clk->map, pll->init_regs,
>> | pll->init_count);
>> meson_parm_write(clk->map, &pll->rst, 0);
>> }
>> return 0;
>> }
>> Because the init function looks like this. Therefore, HDMI PLL init_count
>> is not given.
I don't get the remark. You've got pll->init_count right there.
>> Can I change it like this?
What change ? The function above looks a lot like meson_clk_pll_init()
in the actual source
>
> I don't know if this change meets your requirements? Please give us your
> valuable advice.
What change ?
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 6/6] clk: meson: s4: add s4 SoC peripheral clock controller driver
2022-09-21 9:01 ` Yu Tu
@ 2022-09-28 15:35 ` Jerome Brunet
0 siblings, 0 replies; 41+ messages in thread
From: Jerome Brunet @ 2022-09-28 15:35 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Wed 21 Sep 2022 at 17:01, Yu Tu <yu.tu@amlogic.com> wrote:
> Hi Jerome,
>
> On 2022/8/30 16:20, Yu Tu wrote:
>> On 2022/8/29 20:19, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>>
>>>
>>> On Tue 16 Aug 2022 at 20:00, Yu Tu <yu.tu@amlogic.com> wrote:
>>>
>>> Please trim your replies
>>>
>>>>>> diff --git a/drivers/clk/meson/Kconfig b/drivers/clk/meson/Kconfig
>>>>>> index f4244edc7b28..ec6beb9284d3 100644
>>>>>> --- a/drivers/clk/meson/Kconfig
>>>>>> +++ b/drivers/clk/meson/Kconfig
>>>>>> @@ -127,4 +127,17 @@ config COMMON_CLK_S4_PLL
>>>>>> Support for the pll clock controller on Amlogic S805X2 and
>>>>>> S905Y4 devices,
>>>>>> aka s4. Amlogic S805X2 and S905Y4 devices include AQ222 and
>>>>>> AQ229.
>>>>>> Say Y if you want peripherals and CPU frequency scaling to
>>>>>> work.
>>>>>> +
>>>>>> +config COMMON_CLK_S4
>>>>>> + tristate "S4 SoC Peripherals clock controllers support"
>>>>>> + depends on ARM64
>>>>>> + default y
>>>>>> + select COMMON_CLK_MESON_REGMAP
>>>>>> + select COMMON_CLK_MESON_DUALDIV
>>>>>> + select COMMON_CLK_MESON_VID_PLL_DIV
>>>>>> + select COMMON_CLK_S4_PLL
>>>>> Do you really this ? your driver does not even include the related
>>>>> header.
>>>> If the PLL driver is not turned on in DTS, will it not cause an error?
>>>>>
>>>
>>> I don't get the question.
>>> Kconfig list compile deps. S4 PLL is not a compile dep of the peripheral
>>> controller.
>>>
>>> If you really want to, you may use 'imply'.
>> V4 has been changed as you suggested.
>
> The next edition is being changed according to your requirements. Please
> give us your valuable opinions.
>
>>
>>>>>
>>>>>> +static const struct clk_parent_data sys_ab_clk_parent_data[] = {
>>>>>> + { .fw_name = "xtal" },
>>>>>> + { .fw_name = "fclk_div2" },
>>>>>> + { .fw_name = "fclk_div3" },
>>>>>> + { .fw_name = "fclk_div4" },
>>>>>> + { .fw_name = "fclk_div5" },
>>>>>> + { .fw_name = "fclk_div7" },
>>>>>> + { .hw = &s4_rtc_clk.hw }
>>>>>> +};
>>>>>> +
>>>>>> +static struct clk_regmap s4_sysclk_b_sel = {
>>>>>> + .data = &(struct clk_regmap_mux_data){
>>>>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>>>>> + .mask = 0x7,
>>>>>> + .shift = 26,
>>>>>> + .table = mux_table_sys_ab_clk_sel,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "sysclk_b_sel",
>>>>>> + .ops = &clk_regmap_mux_ro_ops,
>>>>> Why is this using the RO ops ?
>>>> Sys_clk is initialized during the Uboot phase and is fixed at
>>>> 166.666MHz. So I'm going to change it to ro.
>>>
>>> That really much depends on the bootloader and is a pretty weak design.
>>> The bootloader deps should be kept as minimal as possible.
>>>
>>> I see no reason for RO.
>>>
>>> You may cut rate propagation on the user if you need to and continue to
>>> whatever you want in your u-boot
>> I think I know what you mean. But we let the user be in control and not
>> set the frequency, which can be risky. If you insist, I will change it as
>> you suggest.
>
> It has been changed as you requested.
>
>>
>>>
>>>>>
>>>>>> + .parent_data = sys_ab_clk_parent_data,
>>>>>> + .num_parents = ARRAY_SIZE(sys_ab_clk_parent_data),
>>>>>> + },
>>>>>> +};
>>>>>> +
>>>>>> +static struct clk_regmap s4_sysclk_b_div = {
>>>>>> + .data = &(struct clk_regmap_div_data){
>>>>>> + .offset = CLKCTRL_SYS_CLK_CTRL0,
>>>>>> + .shift = 16,
>>>>>> + .width = 10,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "sysclk_b_div",
>>>>>> + .ops = &clk_regmap_divider_ro_ops,
>>>>> Same here and for the rest of the sys part
>>>> Same above.
>>>
>>> We can play that game for a while
>> Ah, you're so funny.
>>
>>>
>>>>>> +
>>>>>> +/* Video Clocks */
>>>>>> +static struct clk_regmap s4_vid_pll_div = {
>>>>>> + .data = &(struct meson_vid_pll_div_data){
>>>>>> + .val = {
>>>>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>>>>> + .shift = 0,
>>>>>> + .width = 15,
>>>>>> + },
>>>>>> + .sel = {
>>>>>> + .reg_off = CLKCTRL_VID_PLL_CLK_DIV,
>>>>>> + .shift = 16,
>>>>>> + .width = 2,
>>>>>> + },
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data) {
>>>>>> + .name = "vid_pll_div",
>>>>>> + .ops = &meson_vid_pll_div_ro_ops,
>>>>> Why RO ? applies to the rest of the video part.
>>>> Because vid_pll_div parent is HDMI_PLL, and HDMI_PLL is a fixed
>>>> frequency. Flags is CLK_SET_RATE_PARENT. So we use RO.
>>>
>>> If the HDMI_PLL is fixed somehow, that is not reason for this clock to
>>> be RO
>>>
>>>> Can I remove RO and use CLK_SET_RATE_NO_REPARENT instead, which one do
>>>> you
>>>> think is more reasonable?
>>>
>>> Neither. CLK_SET_RATE_NO_REPARENT makes no sense, it is not mux
>>>
>> "drivers/clk/meson/vid-pll-div.c"
>> This file only provides ro_ops. Maybe the submission records will give us
>> the answer.
>> In fact, our hardware design is the same as the G12 series.
>
> I don't know if you checked this commit, but there is only one "ro_ops"in
> this place right now.
>
> The S4 SoC is consistent with the G12A/B and GX series.
>
I missed the vid_pll type, Ok then.
>>
>>>>
>>>>>
>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>> + { .fw_name = "hdmi_pll", }
>>>>>> + },
>>>>>> + .num_parents = 1,
>>>>>> + .flags = CLK_SET_RATE_PARENT,
>>>>>> + },
>>>>>> +};
>>>>>> +
>>>>>> +static struct clk_regmap s4_vid_pll_sel = {
>>>>>> + .data = &(struct clk_regmap_mux_data){
>>>>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>>>>> + .mask = 0x1,
>>>>>> + .shift = 18,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "vid_pll_sel",
>>>>>> + .ops = &clk_regmap_mux_ops,
>>>>>> + /*
>>>>>> + * bit 18 selects from 2 possible parents:
>>>>>> + * vid_pll_div or hdmi_pll
>>>>>> + */
>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>> + { .hw = &s4_vid_pll_div.hw },
>>>>>> + { .fw_name = "hdmi_pll", }
>>>>>> + },
>>>>>> + .num_parents = 2,
>>>>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>>>>> Why ? are you planning to DT assigned clocks to statically set this ?
>>>> Because vid_pll_sel one parent is HDMI_PLL, and HDMI_PLL is a fixed
>>>> frequency. To prevent modification, use CLK_SET_RATE_NO_REPARENT.
>>>
>>> Again, this makes no sense.
>> Unfortunately you don't read V4, in fact I have corrected in V4.
>> ".flags = CLK_SET_RATE_PARENT," in V4. Is that okay with you?
>
> I don't know what you think?
>
ok
>>
>>>
>>>>>
>>>>>> + },
>>>>>> +};
>>>>>> +
>>>>>> +static struct clk_regmap s4_vid_pll = {
>>>>>> + .data = &(struct clk_regmap_gate_data){
>>>>>> + .offset = CLKCTRL_VID_PLL_CLK_DIV,
>>>>>> + .bit_idx = 19,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data) {
>>>>>> + .name = "vid_pll",
>>>>>> + .ops = &clk_regmap_gate_ops,
>>>>>> + .parent_hws = (const struct clk_hw *[]) {
>>>>>> + &s4_vid_pll_sel.hw
>>>>>> + },
>>>>>> + .num_parents = 1,
>>>>>> + .flags = CLK_SET_RATE_PARENT,
>>>>>> + },
>>>>>> +};
>>>>>> +
>>>>>> +static const struct clk_parent_data s4_vclk_parent_data[] = {
>>>>>> + { .hw = &s4_vid_pll.hw },
>>>>>> + { .fw_name = "gp0_pll", },
>>>>>> + { .fw_name = "hifi_pll", },
>>>>>> + { .fw_name = "mpll1", },
>>>>>> + { .fw_name = "fclk_div3", },
>>>>>> + { .fw_name = "fclk_div4", },
>>>>>> + { .fw_name = "fclk_div5", },
>>>>>> + { .fw_name = "fclk_div7", },
>>>>>> +};
>>>>>> +
>>>>>> +static struct clk_regmap s4_vclk_sel = {
>>>>>> + .data = &(struct clk_regmap_mux_data){
>>>>>> + .offset = CLKCTRL_VID_CLK_CTRL,
>>>>>> + .mask = 0x7,
>>>>>> + .shift = 16,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "vclk_sel",
>>>>>> + .ops = &clk_regmap_mux_ops,
>>>>>> + .parent_data = s4_vclk_parent_data,
>>>>>> + .num_parents = ARRAY_SIZE(s4_vclk_parent_data),
>>>>>> + .flags = CLK_SET_RATE_NO_REPARENT,
>>>>> Same
>>>> Since fclk_div* is a fixed frequency value, mplL1 and hifi_pll and
>>>> gp0_pll
>>>> are used by other specialized modules, vid_pll has
>>>> CLK_SET_RATE_PARENT. The
>>>> parent of vid_pll is that vid_pll_sel uses CLK_SET_RATE_NO_REPARENT.
>>>
>>> Still not good.
>>>
>>> You don't have CLK_SET_RATE, propagation is stopped and parent clock
>>> will not changed. The best parent will be picked but not changed.
>>>
>>> If one parent MUST NOT be picked, just remove it from the list and add a
>>> explaining why
>>>
>>> [...]
>> Okay.
>
> In the next edition I will change it to ".flags = CLK_SET_RATE_PARENT".
>
If you clock rate propagation is ok, then OK
>>
>>>
>>>>>> +
>>>>>> +static struct clk_regmap s4_ts_clk_div = {
>>>>>> + .data = &(struct clk_regmap_div_data){
>>>>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>>>>> + .shift = 0,
>>>>>> + .width = 8,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "ts_clk_div",
>>>>>> + .ops = &clk_regmap_divider_ops,
>>>>>> + .parent_data = &(const struct clk_parent_data) {
>>>>>> + .fw_name = "xtal",
>>>>>> + },
>>>>>> + .num_parents = 1,
>>>>> propagation stopped ?
>>>> Its parent is xtal, so I should use CLK_SET_RATE_NO_REPARENT.
>>>
>>> Still no. You seem to have problem with the meaning of
>>> CLK_SET_RATE_NO_REPARENT.
>>>
>>> * CLK_SET_RATE_NO_REPARENT: means the parent will no be changed, even if
>>> selecting another parent would result in a closer rate to the
>>> request. It makes sense only if the clock has several parents
>>>
>>> * CLK_SET_RATE_PARENT: means rate change may propagate the parent,
>>> meaning the rate of the parent may change if it help the child achieve
>>> a closer rate to the request
>> Thank you for explaining.I got it.
>>
>>>
>>>>>
>>>>>> + },
>>>>>> +};
>>>>>> +
>>>>>> +static struct clk_regmap s4_ts_clk_gate = {
>>>>>> + .data = &(struct clk_regmap_gate_data){
>>>>>> + .offset = CLKCTRL_TS_CLK_CTRL,
>>>>>> + .bit_idx = 8,
>>>>>> + },
>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>> + .name = "ts_clk",
>>>>>> + .ops = &clk_regmap_gate_ops,
>>>>>> + .parent_hws = (const struct clk_hw *[]) {
>>>>>> + &s4_ts_clk_div.hw
>>>>>> + },
>>>>>> + .num_parents = 1,
>>>>>> + },
>>>>> propagation stopped ?
>>>> I will add CLK_SET_RATE_PARENT.
>>>
>>> [...]
>>>
>>>>>> +/* EMMC/NAND clock */
>>>>>> +
>>>>>> +static const struct clk_parent_data s4_sd_emmc_clk0_parent_data[] = {
>>>>>> + { .fw_name = "xtal", },
>>>>>> + { .fw_name = "fclk_div2", },
>>>>>> + { .fw_name = "fclk_div3", },
>>>>>> + { .fw_name = "hifi_pll", },
>>>>>> + { .fw_name = "fclk_div2p5", },
>>>>>> + /*
>>>>>> + * Following these parent clocks, we should also have had mpll2,
>>>>>> mpll3
>>>>>> + * and gp0_pll but these clocks are too precious to be used
>>>>>> here. All
>>>>>> + * the necessary rates for MMC and NAND operation can be
>>>>>> acheived using
>>>>>> + * hifi_pll or fclk_div clocks
>>>>>> + */
>>>>> You don't want to list mplls but hifi_pll is fine ? seems dangerous.
>>>> hifi pll is for EMMC and NAND on this SoC.
>>>
>>> That deserve a better explanation.
>>> Why can't it use fdiv2 and xtal like the previous SoCs ?
>>>
>>> Which PLLs are you using for Audio then ?
>>> Typical operation on these SoCs usually require 3 PLLs to acheive all
>>> rates
>>>
>> I'll list all the clocks and let the driver itself select Parent as
>> needed.
>
> I don't know what you think?
>
ok
>>
>>>>>
>>>
>>>
>>>>>> +/*
>>>>>> + * gen clk is designed for debug/monitor some internal clock
>>>>>> quality. Some of the
>>>>>> + * corresponding clock sources are not described in the clock tree,
>>>>>> so they are skipped.
>>>>>> + */
>>>>> Still feels a bit light, don't you think ? Among all the clocks, can't
>>>>> you add a bit more parents here ? It would certainly help debug down
>>>>> the road
>>>> [16:12] is gen_clk source select.All is:
>>>> 0: cts_oscin_clk
>>>> 1:cts_rtc_clk
>>>> 2:sys_pll_div16 (internal clock)
>>>> 3:ddr_pll_div32 (internal clock)
>>>> 4: vid_pll
>>>> 5: gp0_pll
>>>> 7: hifi_pll
>>>> 10:adc_dpll_clk_b3 (internal clock for debug)
>>>> 11:adc_dpll_intclk (internal clock for debug)
>>>> 12:clk_msr_src(select from all internal clock except PLLs);
>>>> 16: no used
>>>> 17: sys_cpu_clk_div16 (internal clock)
>>>> 19: fclk_div2
>>>> 20: fclk_div2p5
>>>> 21: fclk_div3
>>>> 22: fclk_div4
>>>> 23: fclk_div5
>>>> 24: fclk_div7
>>>> 25: mpll0
>>>> 26: mpll1
>>>> 27: mpll2
>>>> 28: mpll3
>>>> So i only added the clocks that will actually be used, and some
>>>> debugging
>>>> clock peripherals will not be used.
>>>
>>> you may at least add vid_pll
>> Okay.
>
> It has been changed as you suggested.
>
>>
>>>
>>>>>
>>>>>> +static u32 s4_gen_clk_mux_table[] = { 0, 5, 7, 19, 21, 22,
>>>>>> + 23, 24, 25, 26, 27, 28 };
>>>>>> +static const struct clk_parent_data s4_gen_clk_parent_data[] = {
>>>>>> + { .fw_name = "xtal", },
>>>>>> + { .fw_name = "gp0_pll", },
>>>>>> + { .fw_name = "hifi_pll", },
>>>>>> + { .fw_name = "fclk_div2", },
>>>>>> + { .fw_name = "fclk_div3", },
>>>>>> + { .fw_name = "fclk_div4", },
>>>>>> + { .fw_name = "fclk_div5", },
>>>>>> + { .fw_name = "fclk_div7", },
>>>>>> + { .fw_name = "mpll0", },
>>>>>> + { .fw_name = "mpll1", },
>>>>>> + { .fw_name = "mpll2", },
>>>>>> + { .fw_name = "mpll3", },
>>>>>> +};
>>>
>>> .
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-09-28 15:27 ` Jerome Brunet
@ 2022-09-29 7:07 ` Yu Tu
2022-10-22 12:22 ` Jerome Brunet
0 siblings, 1 reply; 41+ messages in thread
From: Yu Tu @ 2022-09-29 7:07 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Jerome,
Thank you for your reply.
On 2022/9/28 23:27, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Wed 21 Sep 2022 at 16:40, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> Hi Jerome,
>>
>> On 2022/8/30 15:37, Yu Tu wrote:
>>> On 2022/8/30 14:44, Jerome Brunet wrote:
>>>> [ EXTERNAL EMAIL ]
>>>>
>>>>
>>>> On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>
>>>>> On 2022/8/29 17:48, Jerome Brunet wrote:
>>>>>> [ EXTERNAL EMAIL ]
>>>>>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>>
>>>>>>>>>> +
>>>>>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>>>>>> + .en = {
>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>> + .shift = 28,
>>>>>>>>>> + .width = 1,
>>>>>>>>>> + },
>>>>>>>>>> + .m = {
>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>> + .shift = 0,
>>>>>>>>>> + .width = 8,
>>>>>>>>>> + },
>>>>>>>>>> + .n = {
>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>> + .shift = 10,
>>>>>>>>>> + .width = 5,
>>>>>>>>>> + },
>>>>>>>>>> + .frac = {
>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>>>>>> + .shift = 0,
>>>>>>>>>> + .width = 17,
>>>>>>>>>> + },
>>>>>>>>>> + .l = {
>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>> + .shift = 31,
>>>>>>>>>> + .width = 1,
>>>>>>>>>> + },
>>>>>>>>>> + .rst = {
>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>> + .shift = 29,
>>>>>>>>>> + .width = 1,
>>>>>>>>>> + },
>>>>>>>>>> + },
>>>>>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>>>>>> + .name = "hdmi_pll_dco",
>>>>>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>>>>>> + { .fw_name = "xtal", }
>>>>>>>>>> + },
>>>>>>>>>> + .num_parents = 1,
>>>>>>>>>> + /*
>>>>>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>>>>>> + * possible
>>>>>>>>>> + */
>>>>>>>>>
>>>>>>>>> Is it really ?
>>>>>>>>>
>>>>>>>>> Given that HDMI support for the s4 is there yet, the
>>>>>>>>> addresses have changes and the region is no longer a syscon, it is
>>>>>>>>> time
>>>>>>>>> for the HDMI driver to get fixed.
>>>>>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>>>>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>>>>>> "ro_ops".
>>>>>> That's no reason to put NOCACHE or ro-ops
>>>>>> If you want the frequencies to be statically assinged, the correct way
>>>>>> would be through assigned-rate in DT I guess.
>>>>>
>>>>> Okay. You're right. However, when registering with OPS, HDMI PLL will be
>>>>> reset. It takes time for PLL to stabilize the output frequency, which
>>>>> will
>>>>> lead to the startup screen flashing.
>>>>>
>>>>> I would like to know how to solve this problem if not using ro_ops.
>>>>>
>>>>>>
>>>>
>>>> You can add new ops or tweak the current init function.
>>> HDMI PLL is not different from other PLLS, so I think adding OPS is
>>> weird.
>>>
>>>>
>>>> Safest would be to do the following :
>>>> * Check if the PLLs is already on.
>>>> * Check if the 'pll->init_regs' matches what is already set
>>>> - if so, you can skip the reset
>>>> - if not, you need to reset as usual
>>> static int meson_clk_pll_init(struct clk_hw *hw)
>>> {
>>> struct clk_regmap *clk = to_clk_regmap(hw);
>>> struct meson_clk_pll_data *pll = meson_clk_pll_data(clk);
>>> if (pll->init_count) {
>>> meson_parm_write(clk->map, &pll->rst, 1);
>>> regmap_multi_reg_write(clk->map, pll->init_regs,
>>> | pll->init_count);
>>> meson_parm_write(clk->map, &pll->rst, 0);
>>> }
>>> return 0;
>>> }
>>> Because the init function looks like this. Therefore, HDMI PLL init_count
>>> is not given.
>
> I don't get the remark. You've got pll->init_count right there.
>
>>> Can I change it like this?
>
> What change ? The function above looks a lot like meson_clk_pll_init()
> in the actual source
>
>>
>> I don't know if this change meets your requirements? Please give us your
>> valuable advice.
>
> What change ?
static struct clk_regmap s4_hdmi_pll_dco = {
.data = &(struct meson_clk_pll_data){
.en = {
.reg_off = ANACTRL_HDMIPLL_CTRL0,
.shift = 28,
.width = 1,
},
.m = {
.reg_off = ANACTRL_HDMIPLL_CTRL0,
.shift = 0,
.width = 8,
},
.n = {
.reg_off = ANACTRL_HDMIPLL_CTRL0,
.shift = 10,
.width = 5,
},
.frac = {
.reg_off = ANACTRL_HDMIPLL_CTRL1,
.shift = 0,
.width = 17,
},
.l = {
.reg_off = ANACTRL_HDMIPLL_CTRL0,
.shift = 31,
.width = 1,
},
.rst = {
.reg_off = ANACTRL_HDMIPLL_CTRL0,
.shift = 29,
.width = 1,
},
.range = &s4_gp0_pll_mult_range,
},
.hw.init = &(struct clk_init_data){
.name = "hdmi_pll_dco",
.ops = &meson_clk_pll_ops,
.parent_data = (const struct clk_parent_data []) {
{ .fw_name = "xtal", }
},
.num_parents = 1,
},
};
This is my code right now. Because init_count and init_regs are not
defined, HDMI PLL is not reset. In this way, the kernel will not blink
when the Uboot starts. Then in the kernel stage, if we want to change
the HDMI PLL frequency value, we can directly change M, N and OD. In
fact, we will not change the HDMI PLL frequency value later.
I wonder if you accept this change?
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-09-29 7:07 ` Yu Tu
@ 2022-10-22 12:22 ` Jerome Brunet
2022-10-24 11:33 ` Yu Tu
0 siblings, 1 reply; 41+ messages in thread
From: Jerome Brunet @ 2022-10-22 12:22 UTC (permalink / raw)
To: Yu Tu, linux-clk, linux-arm-kernel, linux-amlogic, linux-kernel,
devicetree, Rob Herring, Neil Armstrong, Kevin Hilman,
Michael Turquette, Stephen Boyd, Krzysztof Kozlowski,
Martin Blumenstingl
On Thu 29 Sep 2022 at 15:07, Yu Tu <yu.tu@amlogic.com> wrote:
> Hi Jerome,
> Thank you for your reply.
>
> On 2022/9/28 23:27, Jerome Brunet wrote:
>> [ EXTERNAL EMAIL ]
>> On Wed 21 Sep 2022 at 16:40, Yu Tu <yu.tu@amlogic.com> wrote:
>>
>>> Hi Jerome,
>>>
>>> On 2022/8/30 15:37, Yu Tu wrote:
>>>> On 2022/8/30 14:44, Jerome Brunet wrote:
>>>>> [ EXTERNAL EMAIL ]
>>>>>
>>>>>
>>>>> On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>
>>>>>> On 2022/8/29 17:48, Jerome Brunet wrote:
>>>>>>> [ EXTERNAL EMAIL ]
>>>>>>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>>>
>>>>>>>>>>> +
>>>>>>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>>>>>>> + .en = {
>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>> + .shift = 28,
>>>>>>>>>>> + .width = 1,
>>>>>>>>>>> + },
>>>>>>>>>>> + .m = {
>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>> + .shift = 0,
>>>>>>>>>>> + .width = 8,
>>>>>>>>>>> + },
>>>>>>>>>>> + .n = {
>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>> + .shift = 10,
>>>>>>>>>>> + .width = 5,
>>>>>>>>>>> + },
>>>>>>>>>>> + .frac = {
>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>>>>>>> + .shift = 0,
>>>>>>>>>>> + .width = 17,
>>>>>>>>>>> + },
>>>>>>>>>>> + .l = {
>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>> + .shift = 31,
>>>>>>>>>>> + .width = 1,
>>>>>>>>>>> + },
>>>>>>>>>>> + .rst = {
>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>> + .shift = 29,
>>>>>>>>>>> + .width = 1,
>>>>>>>>>>> + },
>>>>>>>>>>> + },
>>>>>>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>>>>>>> + .name = "hdmi_pll_dco",
>>>>>>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>>>>>>> + { .fw_name = "xtal", }
>>>>>>>>>>> + },
>>>>>>>>>>> + .num_parents = 1,
>>>>>>>>>>> + /*
>>>>>>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>>>>>>> + * possible
>>>>>>>>>>> + */
>>>>>>>>>>
>>>>>>>>>> Is it really ?
>>>>>>>>>>
>>>>>>>>>> Given that HDMI support for the s4 is there yet, the
>>>>>>>>>> addresses have changes and the region is no longer a syscon, it is
>>>>>>>>>> time
>>>>>>>>>> for the HDMI driver to get fixed.
>>>>>>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>>>>>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>>>>>>> "ro_ops".
>>>>>>> That's no reason to put NOCACHE or ro-ops
>>>>>>> If you want the frequencies to be statically assinged, the correct way
>>>>>>> would be through assigned-rate in DT I guess.
>>>>>>
>>>>>> Okay. You're right. However, when registering with OPS, HDMI PLL will be
>>>>>> reset. It takes time for PLL to stabilize the output frequency, which
>>>>>> will
>>>>>> lead to the startup screen flashing.
>>>>>>
>>>>>> I would like to know how to solve this problem if not using ro_ops.
>>>>>>
>>>>>>>
>>>>>
>>>>> You can add new ops or tweak the current init function.
>>>> HDMI PLL is not different from other PLLS, so I think adding OPS is
>>>> weird.
>>>>
>>>>>
>>>>> Safest would be to do the following :
>>>>> * Check if the PLLs is already on.
>>>>> * Check if the 'pll->init_regs' matches what is already set
>>>>> - if so, you can skip the reset
>>>>> - if not, you need to reset as usual
>>>> static int meson_clk_pll_init(struct clk_hw *hw)
>>>> {
>>>> struct clk_regmap *clk = to_clk_regmap(hw);
>>>> struct meson_clk_pll_data *pll = meson_clk_pll_data(clk);
>>>> if (pll->init_count) {
>>>> meson_parm_write(clk->map, &pll->rst, 1);
>>>> regmap_multi_reg_write(clk->map, pll->init_regs,
>>>> | pll->init_count);
>>>> meson_parm_write(clk->map, &pll->rst, 0);
>>>> }
>>>> return 0;
>>>> }
>>>> Because the init function looks like this. Therefore, HDMI PLL init_count
>>>> is not given.
>> I don't get the remark. You've got pll->init_count right there.
>>
>>>> Can I change it like this?
>> What change ? The function above looks a lot like meson_clk_pll_init()
>> in the actual source
>>
>>>
>>> I don't know if this change meets your requirements? Please give us your
>>> valuable advice.
>> What change ?
>
> static struct clk_regmap s4_hdmi_pll_dco = { .data = &(struct
> meson_clk_pll_data){ .en = {
> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 28,
> .width = 1, }, .m = {
> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 0,
> .width = 8, }, .n = {
> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 10,
> .width = 5, }, .frac = {
> .reg_off = ANACTRL_HDMIPLL_CTRL1, .shift = 0,
> .width = 17, }, .l = {
> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 31,
> .width = 1, }, .rst = {
> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 29,
> .width = 1, }, .range =
> &s4_gp0_pll_mult_range, }, .hw.init = &(struct
> clk_init_data){ .name = "hdmi_pll_dco",
> .ops = &meson_clk_pll_ops, .parent_data = (const struct
> clk_parent_data []) { { .fw_name = "xtal", }
> }, .num_parents = 1, }, };
I'm sorry but I can't read this
>
> This is my code right now. Because init_count and init_regs are not
> defined, HDMI PLL is not reset. In this way, the kernel will not blink when
> the Uboot starts. Then in the kernel stage, if we want to change the HDMI
> PLL frequency value, we can directly change M, N and OD. In fact, we will
> not change the HDMI PLL frequency value later.
>
> I wonder if you accept this change?
I think it would be better to separate this topic from your new SoC support
I think you can make another version of the clock controller for the new
SoC, keeping the pll_ops as it is now.
With another series, you may submit something to avoid reseting the PLL if
is already correctly initialized. This can be useful to all the SoC, not
just the last one. It will be easier to discuss on properly formatted
patches
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
* Re: [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver
2022-10-22 12:22 ` Jerome Brunet
@ 2022-10-24 11:33 ` Yu Tu
0 siblings, 0 replies; 41+ messages in thread
From: Yu Tu @ 2022-10-24 11:33 UTC (permalink / raw)
To: Jerome Brunet, linux-clk, linux-arm-kernel, linux-amlogic,
linux-kernel, devicetree, Rob Herring, Neil Armstrong,
Kevin Hilman, Michael Turquette, Stephen Boyd,
Krzysztof Kozlowski, Martin Blumenstingl
Hi Jerome,
Thank you for your reply.
On 2022/10/22 20:22, Jerome Brunet wrote:
> [ EXTERNAL EMAIL ]
>
>
> On Thu 29 Sep 2022 at 15:07, Yu Tu <yu.tu@amlogic.com> wrote:
>
>> Hi Jerome,
>> Thank you for your reply.
>>
>> On 2022/9/28 23:27, Jerome Brunet wrote:
>>> [ EXTERNAL EMAIL ]
>>> On Wed 21 Sep 2022 at 16:40, Yu Tu <yu.tu@amlogic.com> wrote:
>>>
>>>> Hi Jerome,
>>>>
>>>> On 2022/8/30 15:37, Yu Tu wrote:
>>>>> On 2022/8/30 14:44, Jerome Brunet wrote:
>>>>>> [ EXTERNAL EMAIL ]
>>>>>>
>>>>>>
>>>>>> On Tue 30 Aug 2022 at 14:13, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>>
>>>>>>> On 2022/8/29 17:48, Jerome Brunet wrote:
>>>>>>>> [ EXTERNAL EMAIL ]
>>>>>>>> On Mon 15 Aug 2022 at 21:20, Yu Tu <yu.tu@amlogic.com> wrote:
>>>>>>>>
>>>>>>>>>>>> +
>>>>>>>>>>>> +static struct clk_regmap s4_hdmi_pll_dco = {
>>>>>>>>>>>> + .data = &(struct meson_clk_pll_data){
>>>>>>>>>>>> + .en = {
>>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>>> + .shift = 28,
>>>>>>>>>>>> + .width = 1,
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .m = {
>>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>>> + .shift = 0,
>>>>>>>>>>>> + .width = 8,
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .n = {
>>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>>> + .shift = 10,
>>>>>>>>>>>> + .width = 5,
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .frac = {
>>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL1,
>>>>>>>>>>>> + .shift = 0,
>>>>>>>>>>>> + .width = 17,
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .l = {
>>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>>> + .shift = 31,
>>>>>>>>>>>> + .width = 1,
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .rst = {
>>>>>>>>>>>> + .reg_off = ANACTRL_HDMIPLL_CTRL0,
>>>>>>>>>>>> + .shift = 29,
>>>>>>>>>>>> + .width = 1,
>>>>>>>>>>>> + },
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .hw.init = &(struct clk_init_data){
>>>>>>>>>>>> + .name = "hdmi_pll_dco",
>>>>>>>>>>>> + .ops = &meson_clk_pll_ro_ops,
>>>>>>>>>>>> + .parent_data = (const struct clk_parent_data []) {
>>>>>>>>>>>> + { .fw_name = "xtal", }
>>>>>>>>>>>> + },
>>>>>>>>>>>> + .num_parents = 1,
>>>>>>>>>>>> + /*
>>>>>>>>>>>> + * Display directly handle hdmi pll registers ATM, we need
>>>>>>>>>>>> + * NOCACHE to keep our view of the clock as accurate as
>>>>>>>>>>>> + * possible
>>>>>>>>>>>> + */
>>>>>>>>>>>
>>>>>>>>>>> Is it really ?
>>>>>>>>>>>
>>>>>>>>>>> Given that HDMI support for the s4 is there yet, the
>>>>>>>>>>> addresses have changes and the region is no longer a syscon, it is
>>>>>>>>>>> time
>>>>>>>>>>> for the HDMI driver to get fixed.
>>>>>>>>> The HDMI PLL is configured in the Uboot phase and does not change the
>>>>>>>>> frequency in the kernel phase. So we use the NOCACHE flag and
>>>>>>>>> "ro_ops".
>>>>>>>> That's no reason to put NOCACHE or ro-ops
>>>>>>>> If you want the frequencies to be statically assinged, the correct way
>>>>>>>> would be through assigned-rate in DT I guess.
>>>>>>>
>>>>>>> Okay. You're right. However, when registering with OPS, HDMI PLL will be
>>>>>>> reset. It takes time for PLL to stabilize the output frequency, which
>>>>>>> will
>>>>>>> lead to the startup screen flashing.
>>>>>>>
>>>>>>> I would like to know how to solve this problem if not using ro_ops.
>>>>>>>
>>>>>>>>
>>>>>>
>>>>>> You can add new ops or tweak the current init function.
>>>>> HDMI PLL is not different from other PLLS, so I think adding OPS is
>>>>> weird.
>>>>>
>>>>>>
>>>>>> Safest would be to do the following :
>>>>>> * Check if the PLLs is already on.
>>>>>> * Check if the 'pll->init_regs' matches what is already set
>>>>>> - if so, you can skip the reset
>>>>>> - if not, you need to reset as usual
>>>>> static int meson_clk_pll_init(struct clk_hw *hw)
>>>>> {
>>>>> struct clk_regmap *clk = to_clk_regmap(hw);
>>>>> struct meson_clk_pll_data *pll = meson_clk_pll_data(clk);
>>>>> if (pll->init_count) {
>>>>> meson_parm_write(clk->map, &pll->rst, 1);
>>>>> regmap_multi_reg_write(clk->map, pll->init_regs,
>>>>> | pll->init_count);
>>>>> meson_parm_write(clk->map, &pll->rst, 0);
>>>>> }
>>>>> return 0;
>>>>> }
>>>>> Because the init function looks like this. Therefore, HDMI PLL init_count
>>>>> is not given.
>>> I don't get the remark. You've got pll->init_count right there.
>>>
>>>>> Can I change it like this?
>>> What change ? The function above looks a lot like meson_clk_pll_init()
>>> in the actual source
>>>
>>>>
>>>> I don't know if this change meets your requirements? Please give us your
>>>> valuable advice.
>>> What change ?
>>
>> static struct clk_regmap s4_hdmi_pll_dco = { .data = &(struct
>> meson_clk_pll_data){ .en = {
>> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 28,
>> .width = 1, }, .m = {
>> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 0,
>> .width = 8, }, .n = {
>> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 10,
>> .width = 5, }, .frac = {
>> .reg_off = ANACTRL_HDMIPLL_CTRL1, .shift = 0,
>> .width = 17, }, .l = {
>> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 31,
>> .width = 1, }, .rst = {
>> .reg_off = ANACTRL_HDMIPLL_CTRL0, .shift = 29,
>> .width = 1, }, .range =
>> &s4_gp0_pll_mult_range, }, .hw.init = &(struct
>> clk_init_data){ .name = "hdmi_pll_dco",
>> .ops = &meson_clk_pll_ops, .parent_data = (const struct
>> clk_parent_data []) { { .fw_name = "xtal", }
>> }, .num_parents = 1, }, };
>
> I'm sorry but I can't read this
>
>>
>> This is my code right now. Because init_count and init_regs are not
>> defined, HDMI PLL is not reset. In this way, the kernel will not blink when
>> the Uboot starts. Then in the kernel stage, if we want to change the HDMI
>> PLL frequency value, we can directly change M, N and OD. In fact, we will
>> not change the HDMI PLL frequency value later.
>>
>> I wonder if you accept this change?
Sorry, the code is not formatted properly.
>
> I think it would be better to separate this topic from your new SoC support
>
> I think you can make another version of the clock controller for the new
> SoC, keeping the pll_ops as it is now.
>
> With another series, you may submit something to avoid reseting the PLL if
> is already correctly initialized. This can be useful to all the SoC, not
> just the last one. It will be easier to discuss on properly formatted
> patches
Do you mean that hdmi_pll, I will submit the next version of code
according to pll_ro_ops first, and then we will discuss how to reject
the patch of reset pll in another topic?
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 41+ messages in thread
end of thread, other threads:[~2022-10-24 11:45 UTC | newest]
Thread overview: 41+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-08-05 8:57 [PATCH V3 0/6] Add S4 SoC PLL and Peripheral clock controller Yu Tu
2022-08-05 8:57 ` [PATCH V3 1/6] dt-bindings: clock: meson: add S4 SoC PLL clock controller bindings Yu Tu
2022-08-05 9:13 ` Krzysztof Kozlowski
2022-08-05 8:57 ` [PATCH V3 2/6] arm64: dts: meson: add S4 Soc PLL clock controller in DT Yu Tu
2022-08-05 9:16 ` Krzysztof Kozlowski
2022-08-05 9:39 ` Yu Tu
2022-08-10 13:32 ` Jerome Brunet
2022-08-15 6:17 ` Yu Tu
2022-08-29 9:43 ` Jerome Brunet
2022-08-30 6:05 ` Yu Tu
2022-08-30 6:36 ` Jerome Brunet
2022-08-30 7:06 ` Yu Tu
2022-08-05 8:57 ` [PATCH V3 3/6] clk: meson: S4: add support for Amlogic S4 SoC PLL clock driver Yu Tu
2022-08-10 13:47 ` Jerome Brunet
2022-08-15 6:34 ` Yu Tu
2022-08-15 13:20 ` Yu Tu
2022-08-29 9:48 ` Jerome Brunet
2022-08-30 6:13 ` Yu Tu
2022-08-30 6:44 ` Jerome Brunet
2022-08-30 7:37 ` Yu Tu
2022-09-21 8:40 ` Yu Tu
2022-09-28 15:27 ` Jerome Brunet
2022-09-29 7:07 ` Yu Tu
2022-10-22 12:22 ` Jerome Brunet
2022-10-24 11:33 ` Yu Tu
2022-08-29 9:46 ` Jerome Brunet
2022-08-30 6:08 ` Yu Tu
2022-08-05 8:57 ` [PATCH V3 4/6] dt-bindings: clk: meson: add S4 SoC peripheral clock controller bindings Yu Tu
2022-08-05 9:15 ` Krzysztof Kozlowski
2022-08-05 9:33 ` Yu Tu
2022-08-08 6:16 ` Krzysztof Kozlowski
2022-08-08 10:00 ` Yu Tu
2022-08-05 8:57 ` [PATCH V3 5/6] arm64: dts: meson: add S4 Soc Peripheral clock controller in DT Yu Tu
2022-08-05 9:16 ` Krzysztof Kozlowski
2022-08-05 9:36 ` Yu Tu
2022-08-08 6:17 ` Krzysztof Kozlowski
2022-08-08 10:02 ` Yu Tu
[not found] ` <20220805085716.5635-7-yu.tu@amlogic.com>
[not found] ` <1jedxlzxyz.fsf@starbuckisacylon.baylibre.com>
[not found] ` <8f40cb49-fdc5-20cd-343b-8ce50e5d6d97@amlogic.com>
2022-08-29 12:19 ` [PATCH V3 6/6] clk: meson: s4: add s4 SoC peripheral clock controller driver Jerome Brunet
2022-08-30 8:20 ` Yu Tu
2022-09-21 9:01 ` Yu Tu
2022-09-28 15:35 ` Jerome Brunet
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).