clk: sunxi: gates support
Subsystems (5)
Emilio López
maintainercover clk: sunxi: gates support
Hi,
This patchset adds support for the main sunxi gates; namely, those found on AXI, AHB, APB0 and APB1 clocks. The series depends on Maxime's UART rework and on the basic sunxi clock driver by me.
Patches 1 and 2 implement the actual gate support, and patch 3 switches the UARTs to use the correct gates now that they are available.
Patch 4 and 5 add clock support to our pinctrl driver, as we need to keep the pio clock running for gpio to work.
Patch 6 drops the ignore flags; now that we have actual clock users that will keep the important clocks running.
Thanks,
Emilio
Emilio L?pez (6):
clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
arm: sunxi: Add clock definitions for AXI, AHB, APB0, APB1 gates
arm: sunxi: use the right clock phandles for UARTs
pinctrl: sunxi: add clock support
arm: sunxi: Add clock to pinctrl node
clk: sunxi: drop CLK_IGNORE_UNUSED
Documentation/devicetree/bindings/clock/sunxi.txt | 108 +++++++++++++++++++++- arch/arm/boot/dts/sun4i-a10.dtsi | 13 +-- arch/arm/boot/dts/sun5i-a13.dtsi | 1 + arch/arm/boot/dts/sunxi.dtsi | 53 ++++++++++- drivers/clk/sunxi/clk-sunxi.c | 98 +++++++++++++++++++- drivers/pinctrl/pinctrl-sunxi.c | 8 ++ 6 files changed, 267 insertions(+), 14 deletions(-)
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH 1/6 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
This patchset adds DT support for all the AXI, AHB, APB0 and APB1 gates present on sunxi SoCs.
--- Documentation/devicetree/bindings/clock/sunxi.txt | 108 +++++++++++++++++++++- drivers/clk/sunxi/clk-sunxi.c | 90 +++++++++++++++++- 2 files changed, 196 insertions(+), 2 deletions(-)
Documentation/devicetree/bindings/clock/sunxi.txt +107 -1
@@ -10,15 +10,23 @@ Required properties: "allwinner,sunxi-pll1-clk" - for the main PLL clock "allwinner,sunxi-cpu-clk" - for the CPU multiplexer clock "allwinner,sunxi-axi-clk" - for the sunxi AXI clock + "allwinner,sunxi-axi-gates-clk" - for the AXI gates "allwinner,sunxi-ahb-clk" - for the sunxi AHB clock + "allwinner,sunxi-ahb-gates-clk" - for the AHB gates "allwinner,sunxi-apb0-clk" - for the sunxi APB0 clock + "allwinner,sunxi-apb0-gates-clk" - for the APB0 gates "allwinner,sunxi-apb1-clk" - for the sunxi APB1 clock "allwinner,sunxi-apb1-mux-clk" - for the sunxi APB1 clock muxing + "allwinner,sunxi-apb1-gates-clk" - for the APB1 gates Required properties for all clocks: - reg : shall be the control register address for the clock. - clocks : shall be the input parent clock(s) phandle for the clock -- #clock-cells : from common clock binding; shall be set to 0. +- #clock-cells : from common clock binding; shall be set to 0 except for + "allwinner,sunxi-*-gates-clk" where it shall be set to 1 + +Additionally, "allwinner,sunxi-*-gates-clk" clocks require: +- clock-output-names : the corresponding gate names that the clock controls For example:
@@ -42,3 +50,101 @@ cpu: cpu at 01c20054 { reg = <0x01c20054 0x4>; clocks = <&osc32k>, <&osc24M>, <&pll1>; }; + + + +Gate clock outputs + +The "allwinner,sunxi-*-gates-clk" clocks provide several gatable outputs; +their corresponding offsets are listed below: + + * AXI gates ("allwinner,sunxi-axi-gates-clk") + + DRAM 0 + + * AHB gates ("allwinner,sunxi-ahb-gates-clk") + + USB0 0 + EHCI0 1 + OHCI0 2* + EHCI1 3 + OHCI1 4* + SS 5 + DMA 6 + BIST 7 + MMC0 8 + MMC1 9 + MMC2 10 + MMC3 11 + MS 12** + NAND 13 + SDRAM 14 + + ACE 16 + EMAC 17 + TS 18 + + SPI0 20 + SPI1 21 + SPI2 22 + SPI3 23 + PATA 24 + SATA 25** + GPS 26* + + VE 32 + TVD 33 + TVE0 34 + TVE1 35 + LCD0 36 + LCD1 37 + + CSI0 40 + CSI1 41 + + HDMI 43 + DE_BE0 44 + DE_BE1 45 + DE_FE0 46 + DE_FE1 47 + + MP 50 + + MALI400 52 + + * APB0 gates ("allwinner,sunxi-apb0-gates-clk") + + CODEC 0 + SPDIF 1* + AC97 2 + IIS 3 + + PIO 5 + IR0 6 + IR1 7 + + KEYPAD 10 + + * APB1 gates ("allwinner,sunxi-apb1-gates-clk") + + TWI0 0 + TWI1 1 + TWI2 2 + + CAN 4 + SCR 5 + PS20 6 + PS21 7 + + UART0 16 + UART1 17 + UART2 18 + UART3 19 + UART4 20 + UART5 21 + UART6 22 + UART7 23 + +Notation: + [*]: The datasheet didn't mention these, but they are present on AW code + [**]: The datasheet had this marked as "NC" but they are used on AW code
drivers/clk/sunxi/clk-sunxi.c +88 -0
104 lines not on this page · raw message
Signature · 1 line
1.8.2
PATCH v2 1/6 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Hi Emilio,
I have a few comments below,
Le 22/03/2013 15:20, Emilio L?pez a ?crit :
Quoted 15 lines
This patchset adds DT support for all the AXI, AHB, APB0 and APB1 gates present on sunxi SoCs.
Signed-off-by Emilio L?pez <[email protected]>--- Documentation/devicetree/bindings/clock/sunxi.txt | 108 +++++++++++++++++++++- drivers/clk/sunxi/clk-sunxi.c | 90 +++++++++++++++++- 2 files changed, 196 insertions(+), 2 deletions(-)Documentation/devicetree/bindings/clock/sunxi.txt +0
-0@@ -10,15 +10,23 @@ Required properties: "allwinner,sunxi-pll1-clk" - for the main PLL clock
Documentation/devicetree/bindings/clock/sunxi.txt Required properties: · line 20
Quoted 10 lines
Documentation/devicetree/bindings/clock/sunxi.txt +4
-0"allwinner,sunxi-cpu-clk" - for the CPU multiplexer clock "allwinner,sunxi-axi-clk" - for the sunxi AXI clock + "allwinner,sunxi-axi-gates-clk" - for the AXI gates "allwinner,sunxi-ahb-clk" - for the sunxi AHB clock + "allwinner,sunxi-ahb-gates-clk" - for the AHB gates "allwinner,sunxi-apb0-clk" - for the sunxi APB0 clock + "allwinner,sunxi-apb0-gates-clk" - for the APB0 gates "allwinner,sunxi-apb1-clk" - for the sunxi APB1 clock "allwinner,sunxi-apb1-mux-clk" - for the sunxi APB1 clock muxing + "allwinner,sunxi-apb1-gates-clk" - for the APB1 gates
It has been pointed to me that the usual way of dealing with compatible IPs in several generation of SoC in device tree is usually to use the oldest SoC name to support it, and not some generic term like we use.
I'm currently struggling while doing some work on the A31, where some IPs differ in a significant way from the A10/A13, and we can't just use sunxi anymore.
I have a patchset that renames every other IP we have doing this, but could you rename your compatible strings to allwinner,sun4i* as a first patch for the already existing clocks, and use that prefix in the following patches as well? Moreover, since it has been mostly tested on an A10, if the IP came to differ to some aspect in the A13, it will make our life easier as well.
Quoted 39 lines
Documentation/devicetree/bindings/clock/sunxi.txt +27
-1Required properties for all clocks: - reg : shall be the control register address for the clock. - clocks : shall be the input parent clock(s) phandle for the clock -- #clock-cells : from common clock binding; shall be set to 0. +- #clock-cells : from common clock binding; shall be set to 0 except for + "allwinner,sunxi-*-gates-clk" where it shall be set to 1 + +Additionally, "allwinner,sunxi-*-gates-clk" clocks require: +- clock-output-names : the corresponding gate names that the clock controls For example:@@ -42,3 +50,101 @@ cpu: cpu at 01c20054 { reg = <0x01c20054 0x4>; clocks = <&osc32k>, <&osc24M>, <&pll1>; }; + + + +Gate clock outputs + +The "allwinner,sunxi-*-gates-clk" clocks provide several gatable outputs; +their corresponding offsets are listed below: + + * AXI gates ("allwinner,sunxi-axi-gates-clk") + + DRAM 0 + + * AHB gates ("allwinner,sunxi-ahb-gates-clk") + + USB0 0 + EHCI0 1 + OHCI0 2* + EHCI1 3 + OHCI1 4* + SS 5 + DMA 6 + BIST 7
Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 84
Quoted 10 lines
Documentation/devicetree/bindings/clock/sunxi.txt +10
-0+ MMC0 8 + MMC1 9 + MMC2 10 + MMC3 11 + MS 12** + NAND 13 + SDRAM 14 + + ACE 16 + EMAC 17
Hmmm, this makes me thinking that I used wemac in the pinctrl driver, and that it's not really consistent with the datasheet, what you have here, and what I did in the EMAC driver. I'll need to change that...
Quoted 38 lines
Documentation/devicetree/bindings/clock/sunxi.txt +38
-0+ TS 18 + + SPI0 20 + SPI1 21 + SPI2 22 + SPI3 23 + PATA 24 + SATA 25** + GPS 26* + + VE 32 + TVD 33 + TVE0 34 + TVE1 35 + LCD0 36 + LCD1 37 + + CSI0 40 + CSI1 41 + + HDMI 43 + DE_BE0 44 + DE_BE1 45 + DE_FE0 46 + DE_FE1 47 + + MP 50 + + MALI400 52 + + * APB0 gates ("allwinner,sunxi-apb0-gates-clk") + + CODEC 0 + SPDIF 1* + AC97 2 + IIS 3 + + PIO 5
Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 132
Quoted 10 lines
Documentation/devicetree/bindings/clock/sunxi.txt +10
-0+ IR0 6 + IR1 7 + + KEYPAD 10 + + * APB1 gates ("allwinner,sunxi-apb1-gates-clk") + + TWI0 0 + TWI1 1 + TWI2 2
I'd rather see here I2C. I know that it's called TWI in the datasheet, but the term I2C is more commonly used in the kernel.
Quoted 7 lines
Documentation/devicetree/bindings/clock/sunxi.txt +7
-0+ CAN 4 + SCR 5 + PS20 6 + PS21 7 + + UART0 16 + UART1 17
Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 150
Quoted 10 lines
Documentation/devicetree/bindings/clock/sunxi.txt +10
-0+ UART2 18 + UART3 19 + UART4 20 + UART5 21 + UART6 22 + UART7 23 + +Notation: + [*]: The datasheet didn't mention these, but they are present on AW code + [**]: The datasheet had this marked as "NC" but they are used on AW code
I'm really happy with the general documentation, thanks for this. However, all of this is true only for the A10, could you mention it somewhere?
Thanks,
Maxime
Signature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
Emilio López
maintainerPATCH 1/6 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Hi Maxime,
El 25/03/13 06:43, Maxime Ripard escribi?:
Quoted 39 lines
Hi Emilio,
I have a few comments below,
Le 22/03/2013 15:20, Emilio L?pez a ?crit :
Quoted 15 lines
This patchset adds DT support for all the AXI, AHB, APB0 and APB1 gates present on sunxi SoCs.
Signed-off-by Emilio L?pez <[email protected]>--- Documentation/devicetree/bindings/clock/sunxi.txt | 108 +++++++++++++++++++++- drivers/clk/sunxi/clk-sunxi.c | 90 +++++++++++++++++- 2 files changed, 196 insertions(+), 2 deletions(-)Documentation/devicetree/bindings/clock/sunxi.txt +0
-0@@ -10,15 +10,23 @@ Required properties: "allwinner,sunxi-pll1-clk" - for the main PLL clockDocumentation/devicetree/bindings/clock/sunxi.txt Required properties: · line 20
Quoted 10 lines
Documentation/devicetree/bindings/clock/sunxi.txt +4
-0"allwinner,sunxi-cpu-clk" - for the CPU multiplexer clock "allwinner,sunxi-axi-clk" - for the sunxi AXI clock + "allwinner,sunxi-axi-gates-clk" - for the AXI gates "allwinner,sunxi-ahb-clk" - for the sunxi AHB clock + "allwinner,sunxi-ahb-gates-clk" - for the AHB gates "allwinner,sunxi-apb0-clk" - for the sunxi APB0 clock + "allwinner,sunxi-apb0-gates-clk" - for the APB0 gates "allwinner,sunxi-apb1-clk" - for the sunxi APB1 clock "allwinner,sunxi-apb1-mux-clk" - for the sunxi APB1 clock muxing + "allwinner,sunxi-apb1-gates-clk" - for the APB1 gatesIt has been pointed to me that the usual way of dealing with compatible IPs in several generation of SoC in device tree is usually to use the oldest SoC name to support it, and not some generic term like we use.
I'm currently struggling while doing some work on the A31, where some IPs differ in a significant way from the A10/A13, and we can't just use sunxi anymore.
Quoted 6 lines
I have a patchset that renames every other IP we have doing this, but could you rename your compatible strings to allwinner,sun4i* as a first patch for the already existing clocks, and use that prefix in the following patches as well? Moreover, since it has been mostly tested on an A10, if the IP came to differ to some aspect in the A13, it will make our life easier as well.
Ok, I'll do so.
Quoted 98 lines
Quoted 40 lines
Documentation/devicetree/bindings/clock/sunxi.txt +27
-1Required properties for all clocks: - reg : shall be the control register address for the clock. - clocks : shall be the input parent clock(s) phandle for the clock -- #clock-cells : from common clock binding; shall be set to 0. +- #clock-cells : from common clock binding; shall be set to 0 except for + "allwinner,sunxi-*-gates-clk" where it shall be set to 1 + +Additionally, "allwinner,sunxi-*-gates-clk" clocks require: +- clock-output-names : the corresponding gate names that the clock controls For example:@@ -42,3 +50,101 @@ cpu: cpu at 01c20054 { reg = <0x01c20054 0x4>; clocks = <&osc32k>, <&osc24M>, <&pll1>; }; + + + +Gate clock outputs + +The "allwinner,sunxi-*-gates-clk" clocks provide several gatable outputs; +their corresponding offsets are listed below: + + * AXI gates ("allwinner,sunxi-axi-gates-clk") + + DRAM 0 + + * AHB gates ("allwinner,sunxi-ahb-gates-clk") + + USB0 0 + EHCI0 1 + OHCI0 2* + EHCI1 3 + OHCI1 4* + SS 5 + DMA 6 + BIST 7Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 84
Quoted 10 lines
Documentation/devicetree/bindings/clock/sunxi.txt +10
-0+ MMC0 8 + MMC1 9 + MMC2 10 + MMC3 11 + MS 12** + NAND 13 + SDRAM 14 + + ACE 16 + EMAC 17Hmmm, this makes me thinking that I used wemac in the pinctrl driver, and that it's not really consistent with the datasheet, what you have here, and what I did in the EMAC driver. I'll need to change that...
Quoted 43 lines
Documentation/devicetree/bindings/clock/sunxi.txt +43
-0+ TS 18 + + SPI0 20 + SPI1 21 + SPI2 22 + SPI3 23 + PATA 24 + SATA 25** + GPS 26* + + VE 32 + TVD 33 + TVE0 34 + TVE1 35 + LCD0 36 + LCD1 37 + + CSI0 40 + CSI1 41 + + HDMI 43 + DE_BE0 44 + DE_BE1 45 + DE_FE0 46 + DE_FE1 47 + + MP 50 + + MALI400 52 + + * APB0 gates ("allwinner,sunxi-apb0-gates-clk") + + CODEC 0 + SPDIF 1* + AC97 2 + IIS 3 + + PIO 5 + IR0 6 + IR1 7 + + KEYPAD 10 +
Quoted 8 lines
Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 132
Quoted 5 lines
Documentation/devicetree/bindings/clock/sunxi.txt +5
-0+ * APB1 gates ("allwinner,sunxi-apb1-gates-clk") + + TWI0 0 + TWI1 1 + TWI2 2I'd rather see here I2C. I know that it's called TWI in the datasheet, but the term I2C is more commonly used in the kernel.
Ok, I'll change it. Do you want me to rename the gate names on the DT too to match? ("apb1_twi0" -> "apb1_i2c0")
Quoted 14 lines
Quoted 14 lines
Documentation/devicetree/bindings/clock/sunxi.txt +13
-0+ CAN 4 + SCR 5 + PS20 6 + PS21 7 + + UART0 16 + UART1 17 + UART2 18 + UART3 19 + UART4 20 + UART5 21 + UART6 22 + UART7 23
Quoted 8 lines
Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 150
Quoted 4 lines
Documentation/devicetree/bindings/clock/sunxi.txt +4
-0+ +Notation: + [*]: The datasheet didn't mention these, but they are present on AW code + [**]: The datasheet had this marked as "NC" but they are used on AW codeI'm really happy with the general documentation, thanks for this. However, all of this is true only for the A10, could you mention it somewhere?
I could add a note saying so. We could also rename this file to sun4i.txt and then and add a sun5i.txt with the A13 gate descriptions too; what would you prefer?
Thanks for the review,
Emilio
PATCH v2 1/6 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Le 25/03/2013 11:17, Emilio L?pez a ?crit :
Quoted 7 lines
El 25/03/13 06:43, Maxime Ripard escribi?:
Quoted 6 lines
Quoted 5 lines
Documentation/devicetree/bindings/clock/sunxi.txt +5
-0+ * APB1 gates ("allwinner,sunxi-apb1-gates-clk") + + TWI0 0 + TWI1 1 + TWI2 2
Quoted 5 lines
Quoted 2 lines
I'd rather see here I2C. I know that it's called TWI in the datasheet, but the term I2C is more commonly used in the kernel.
Ok, I'll change it. Do you want me to rename the gate names on the DT too to match? ("
apb1_twi0" -> "apb1_i2c0")
Yes, it would be nice.
Quoted 22 lines
Quoted 13 lines
Quoted 13 lines
Documentation/devicetree/bindings/clock/sunxi.txt +13
-0+ CAN 4 + SCR 5 + PS20 6 + PS21 7 + + UART0 16 + UART1 17 + UART2 18 + UART3 19 + UART4 20 + UART5 21 + UART6 22 + UART7 23Quoted 8 lines
Documentation/devicetree/bindings/clock/sunxi.txt cpu: cpu at 01c20054 { · line 150
Quoted 4 lines
Documentation/devicetree/bindings/clock/sunxi.txt +4
-0+ +Notation: + [*]: The datasheet didn't mention these, but they are present on AW code + [**]: The datasheet had this marked as "NC" but they are used on AW codeI'm really happy with the general documentation, thanks for this. However, all of this is true only for the A10, could you mention it somewhere?
Quoted 5 lines
I could add a note saying so. We could also rename this file to sun4i.txt and then and add a sun5i.txt with the A13 gate descriptions too; what would you prefer?
From what we've seen so far, A10 and A13 will share a lot except the
gates in the documentation, so maybe you can keep this sunxi.txt file with the generic comments, and add a note at the bottom saying to refer to either sun4i.txt or sun5i.txt files for details.
Maxime
Signature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
Emilio López
maintainerPATCH 2/6 arm: sunxi: Add clock definitions for AXI, AHB, APB0, APB1 gates
In reply to Emilio LópezThis commit adds the corresponding DT bindings for all the AXI, AHB, APB0 and APB1 gates.
--- arch/arm/boot/dts/sunxi.dtsi | 49 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+)
arch/arm/boot/dts/sunxi.dtsi +49 -0
@@ -80,6 +80,14 @@ clocks = <&cpu>; }; + axi_gates: axi_gates at 01c2005c { + #clock-cells = <1>; + compatible = "allwinner,sunxi-axi-gates-clk"; + reg = <0x01c2005c 0x4>; + clocks = <&axi>; + clock-output-names = "axi_dram"; + }; + ahb: ahb at 01c20054 { #clock-cells = <0>; compatible = "allwinner,sunxi-ahb-clk";
@@ -87,6 +95,24 @@ clocks = <&axi>; }; + ahb_gates: ahb_gates at 01c20060 { + #clock-cells = <1>; + compatible = "allwinner,sunxi-ahb-gates-clk"; + reg = <0x01c20060 0x8>; + clocks = <&ahb>; + clock-output-names = "ahb_usb0", "ahb_ehci0", + "ahb_ohci0", "ahb_ehci1", "ahb_ohci1", "ahb_ss", + "ahb_dma", "ahb_bist", "ahb_mmc0", "ahb_mmc1", + "ahb_mmc2", "ahb_mmc3", "ahb_ms", "ahb_nand", + "ahb_sdram", "ahb_ace", "ahb_emac", "ahb_ts", + "ahb_spi0", "ahb_spi1", "ahb_spi2", "ahb_spi3", + "ahb_pata", "ahb_sata", "ahb_gps", "ahb_ve", + "ahb_tvd", "ahb_tve0", "ahb_tve1", "ahb_lcd0", + "ahb_lcd1", "ahb_csi0", "ahb_csi1", "ahb_hdmi", + "ahb_de_be0", "ahb_de_be1", "ahb_de_fe0", + "ahb_de_fe1", "ahb_mp", "ahb_mali400"; + }; + apb0: apb0 at 01c20054 { #clock-cells = <0>; compatible = "allwinner,sunxi-apb0-clk";
@@ -94,6 +120,16 @@ clocks = <&ahb>; }; + apb0_gates: apb0_gates at 01c20068 { + #clock-cells = <1>; + compatible = "allwinner,sunxi-apb0-gates-clk"; + reg = <0x01c20068 0x4>; + clocks = <&apb0>; + clock-output-names = "apb0_codec", "apb0_spdif", + "apb0_ac97", "apb0_iis", "apb0_pio", "apb0_ir0", + "apb0_ir1", "apb0_keypad"; + }; + /* dummy is pll62 */ apb1_mux: apb1_mux at 01c20058 { #clock-cells = <0>;
@@ -108,6 +144,19 @@ reg = <0x01c20058 0x4>; clocks = <&apb1_mux>; }; + + apb1_gates: apb1_gates at 01c2006c { + #clock-cells = <1>; + compatible = "allwinner,sunxi-apb1-gates-clk"; + reg = <0x01c2006c 0x4>; + clocks = <&apb1>; + clock-output-names = "apb1_twi0", "apb1_twi1", + "apb1_twi2", "apb1_can", "apb1_scr", + "apb1_ps20", "apb1_ps21", "apb1_uart0", + "apb1_uart1", "apb1_uart2", "apb1_uart3", + "apb1_uart4", "apb1_uart5", "apb1_uart6", + "apb1_uart7"; + }; }; soc {
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH 3/6 arm: sunxi: use the right clock phandles for UARTs
In reply to Emilio LópezAll the UARTs are connected to clock gates; now that our clock driver is able to handle them, make the switch.
--- arch/arm/boot/dts/sun4i-a10.dtsi | 12 ++++++------ arch/arm/boot/dts/sunxi.dtsi | 4 ++-- 2 files changed, 8 insertions(+), 8 deletions(-)
arch/arm/boot/dts/sun4i-a10.dtsi +6 -6
@@ -54,7 +54,7 @@ interrupts = <1>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 16>; status = "disabled"; };
@@ -64,7 +64,7 @@ interrupts = <3>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 18>; status = "disabled"; };
@@ -74,7 +74,7 @@ interrupts = <17>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 20>; status = "disabled"; };
@@ -84,7 +84,7 @@ interrupts = <18>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 21>; status = "disabled"; };
@@ -94,7 +94,7 @@ interrupts = <19>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 22>; status = "disabled"; };
@@ -104,7 +104,7 @@ interrupts = <20>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 23>; status = "disabled"; }; };
arch/arm/boot/dts/sunxi.dtsi +2 -2
@@ -191,7 +191,7 @@ interrupts = <2>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 17>; status = "disabled"; };
@@ -201,7 +201,7 @@ interrupts = <4>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 19>; status = "disabled"; }; };
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH 4/6 pinctrl: sunxi: add clock support
In reply to Emilio LópezFor the port controller to work, we need to enable the apb0_pio gate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.
--- drivers/pinctrl/pinctrl-sunxi.c | 8 ++++++++ 1 file changed, 8 insertions(+)
drivers/pinctrl/pinctrl-sunxi.c +8 -0
@@ -11,6 +11,7 @@*/ #include <linux/io.h> +#include <linux/clk.h> #include <linux/gpio.h> #include <linux/module.h> #include <linux/of.h>
@@ -1917,6 +1918,7 @@ static int sunxi_pinctrl_probe(struct platform_device *pdev)struct pinctrl_pin_desc *pins;struct sunxi_pinctrl *pctl;int i, ret, last_pin; +struct clk *clk; pctl =devm_kzalloc (&pdev->dev,sizeof (*pctl),GFP_KERNEL );if (!pctl)
@@ -1987,6 +1989,12 @@ static int sunxi_pinctrl_probe(struct platform_device *pdev) goto gpiochip_error; } + clk =devm_clk_get (&pdev->dev,NULL ); +if (IS_ERR (clk)) + goto gpiochip_error; + +clk_prepare_enable (clk); +dev_info (&pdev->dev,"initialized sunXi PIO driver\n" );return 0 ;
Signature · 1 line
1.8.2
PATCH 4/6 pinctrl: sunxi: add clock support
Quoted 5 lines · Emilio L?pez
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Signed-off-by Emilio L?pez <[email protected]>
I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
Yours,
Linus Walleij
PATCH v2 4/6 pinctrl: sunxi: add clock support
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :
Quoted 6 lines · Emilio L?pez
Quoted 5 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.
Quoted 6 lines
Quoted 1 line
Signed-off-by Emilio L?pez <[email protected]>Acked-by Linus Walleij <[email protected]>I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
Maxime
Signature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
PATCH 4/6 pinctrl: sunxi: add clock support
Quoted 10 lines · Maxime Ripard
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :Quoted 8 lines · Emilio L?pez
Quoted 6 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Signed-off-by Emilio L?pez <[email protected]>
Quoted 8 lines
Quoted 4 lines
Acked-by Linus Walleij <[email protected]>I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
OK then, patch applied to the pinctrl tree, thanks!
Yours,
Linus Walleij
PATCH v2 4/6 pinctrl: sunxi: add clock support
Hi Linus,
Le 03/04/2013 13:59, Linus Walleij a ?crit :
Quoted 15 lines
Quoted 12 lines · Maxime Ripard
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :Quoted 10 lines · Emilio L?pez
Quoted 6 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Signed-off-by Emilio L?pez <[email protected]>Acked-by Linus Walleij <[email protected]>
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
OK then, patch applied to the pinctrl tree, thanks!
Ouch, the previous patches aren't applied yet by Mike, and we didn't have some comments from him about these patches, so I can't really say if the previous patches will get merge by 3.10 or not... So we will maybe end up with a broken pinctrl stuff.
Mike, do you have some comments on Emilio's patches?
If you do, we can still fix things up later by applying a slightly modified version of patches 5/6 with a clock that is already there, for it to load properly.
Thanks,
Maxime
Signature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
PATCH v2 4/6 pinctrl: sunxi: add clock support
Quoting Maxime Ripard (2013-04-03 14:20:37)
Quoted 32 lines
Hi Linus,
Le 03/04/2013 13:59, Linus Walleij a ?crit :
Quoted 15 lines
Quoted 12 lines · Maxime Ripard
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :Quoted 10 lines · Emilio L?pez
Quoted 6 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Signed-off-by Emilio L?pez <[email protected]>Acked-by Linus Walleij <[email protected]>Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
OK then, patch applied to the pinctrl tree, thanks!
Ouch, the previous patches aren't applied yet by Mike, and we didn't have some comments from him about these patches, so I can't really say if the previous patches will get merge by 3.10 or not... So we will maybe end up with a broken pinctrl stuff.
Quoted 6 lines
Mike, do you have some comments on Emilio's patches?
If you do, we can still fix things up later by applying a slightly modified version of patches 5/6 with a clock that is already there, for it to load properly.
Hi Maxime,
I did leave some comments a few hours ago. It would be better if the fixes could be rolled in now, but if that is somehow impossible then I don't mind taking them as-is with fixes at a later date. Just let me know.
Regards,
Mike
Quoted 7 lines
Thanks,
MaximeSignature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
PATCH 4/6 pinctrl: sunxi: add clock support
In reply to Maxime RipardQuoted 23 lines · Maxime Ripard
Le 03/04/2013 13:59, Linus Walleij a ?crit :
Quoted 22 lines
Quoted 18 lines · Maxime Ripard
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :Quoted 6 lines · Emilio L?pez
Quoted 5 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Quoted 6 lines
Quoted 1 line
Signed-off-by Emilio L?pez <[email protected]>Acked-by Linus Walleij <[email protected]>I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
Quoted 6 lines
Quoted 1 line
OK then, patch applied to the pinctrl tree, thanks!
Ouch, the previous patches aren't applied yet by Mike, and we didn't have some comments from him about these patches, so I can't really say if the previous patches will get merge by 3.10 or not... So we will maybe end up with a broken pinctrl stuff.
This is usually why we keep patches together and merge whole series into one tree, such as ARM SoC.
Well, maybe I'll drop it then ...
Yours,
Linus Walleij
PATCH v2 4/6 pinctrl: sunxi: add clock support
Quoting Linus Walleij (2013-04-04 04:37:23)
Quoted 32 lines
Quoted 23 lines · Maxime Ripard
Le 03/04/2013 13:59, Linus Walleij a ?crit :
Quoted 22 lines
Quoted 18 lines · Maxime Ripard
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :Quoted 6 lines · Emilio L?pez
Quoted 5 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Quoted 6 lines
Quoted 1 line
Signed-off-by Emilio L?pez <[email protected]>Acked-by Linus Walleij <[email protected]>I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
Quoted 6 lines
Quoted 1 line
OK then, patch applied to the pinctrl tree, thanks!
Ouch, the previous patches aren't applied yet by Mike, and we didn't have some comments from him about these patches, so I can't really say if the previous patches will get merge by 3.10 or not... So we will maybe end up with a broken pinctrl stuff.
Quoted 5 lines
This is usually why we keep patches together and merge whole series into one tree, such as ARM SoC.
Well, maybe I'll drop it then ...
I plan on taking in the three clk patches after getting final comments from Gregory. So maybe you do not need to drop the pinctrl patches?
Regards,
Mike
Quoted 2 lines
Yours,
Linus Walleij
PATCH 4/6 pinctrl: sunxi: add clock support
Quoted 33 lines · Mike Turquette
Quoting Linus Walleij (2013-04-04 04:37:23)
Quoted 32 lines
Quoted 29 lines · Maxime Ripard
Le 03/04/2013 13:59, Linus Walleij a ?crit :
Quoted 15 lines
Quoted 12 lines · Maxime Ripard
Hi Linus,
Le 27/03/2013 14:14, Linus Walleij a ?crit :Quoted 10 lines · Emilio L?pez
Quoted 6 lines
For the port controller to work, we need to enable the
apb0_piogate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.Signed-off-by Emilio L?pez <[email protected]>Acked-by Linus Walleij <[email protected]>Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
I guess you'll want to merge these as one series to some platform-specific tree, tell me if you want it merged in the pinctrl tree.
The easiest would probably be to merge this patch through your patch, with Mike taking the clock driver patches and me taking the dt ones, once we agree on the other patches.
OK then, patch applied to the pinctrl tree, thanks!
Ouch, the previous patches aren't applied yet by Mike, and we didn't have some comments from him about these patches, so I can't really say if the previous patches will get merge by 3.10 or not... So we will maybe end up with a broken pinctrl stuff.
Quoted 8 lines
Quoted 5 lines
This is usually why we keep patches together and merge whole series into one tree, such as ARM SoC.
Well, maybe I'll drop it then ...
I plan on taking in the three clk patches after getting final comments from Gregory. So maybe you do not need to drop the pinctrl patches?
No, I'm keeping it around, no problem.
Thanks Mike,
Linus Walleij
Emilio López
maintainerPATCH 5/6 arm: sunxi: Add clock to pinctrl node
In reply to Emilio LópezThe port controller needs the apb0_pio clock enabled to be able to work. This commit declares that on the device tree.
--- arch/arm/boot/dts/sun4i-a10.dtsi | 1 + arch/arm/boot/dts/sun5i-a13.dtsi | 1 + 2 files changed, 2 insertions(+)
arch/arm/boot/dts/sun4i-a10.dtsi +1 -0
@@ -21,6 +21,7 @@ pio: pinctrl at 01c20800 { compatible = "allwinner,sun4i-a10-pinctrl"; reg = <0x01c20800 0x400>; + clocks = <&apb0_gates 5>; gpio-controller; #address-cells = <1>; #size-cells = <0>;
arch/arm/boot/dts/sun5i-a13.dtsi +1 -0
@@ -22,6 +22,7 @@ pio: pinctrl at 01c20800 { compatible = "allwinner,sun5i-a13-pinctrl"; reg = <0x01c20800 0x400>; + clocks = <&apb0_gates 5>; gpio-controller; #address-cells = <1>; #size-cells = <0>;
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH 6/6 clk: sunxi: drop CLK_IGNORE_UNUSED
In reply to Emilio LópezThis flag was in place to prevent important clocks from getting gated while they had no users. Now that the UART driver supports clocks properly, we can drop this.
--- drivers/clk/sunxi/clk-sunxi.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/clk/sunxi/clk-sunxi.c +4 -4
@@ -41,8 +41,8 @@ static void __init sunxi_osc_clk_setup(struct device_node *node) parent =of_clk_get_parent_name (node,0 ); - clk =clk_register_gate (NULL , clk_name, parent,CLK_IGNORE_UNUSED , - reg,SUNXI_OSC24M_GATE ,0 , &clk_lock); + clk =clk_register_gate (NULL , clk_name, parent,0 , reg, +SUNXI_OSC24M_GATE ,0 , &clk_lock);if (clk) {of_clk_add_provider (node, of_clk_src_simple_get, clk);
@@ -198,8 +198,8 @@ static void __init sunxi_factors_clk_setup(struct device_node *node, parent =of_clk_get_parent_name (node,0 ); - clk =clk_register_factors (NULL , clk_name, parent,CLK_IGNORE_UNUSED , - reg, data->table, data->getter, &clk_lock); + clk =clk_register_factors (NULL , clk_name, parent,0 , reg, + data->table, data->getter, &clk_lock);if (clk) {of_clk_add_provider (node, of_clk_src_simple_get, clk);
Signature · 1 line
1.8.2
Emilio López
maintainerv2 cover clk: sunxi: gates support
In reply to Emilio LópezHi,
This patchset adds support for the main sunxi gates; namely, those found on AXI, AHB, APB0 and APB1 clocks. The series depends on Maxime's UART rework and on the basic sunxi clock driver by me.
Patches 1 and 2 implement the actual gate support, and patch 3 switches the UARTs to use the correct gates now that they are available.
Patch 4 and 5 add clock support to our pinctrl driver, as we need to keep the pio clock running for gpio to work.
Patch 6 drops the ignore flags; now that we have actual clock users that will keep the important clocks running.
Patch 7 is an unrelated oneliner to remove an unnecesary kmalloc call.
The main changes from v1 are renamed compatible strings and a clarification on the documentation, see each individual patch for more details.
Thanks,
Emilio
Emilio L?pez (7):
clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
arm: sunxi: Add clock definitions for AXI, AHB, APB0, APB1 gates
arm: sunxi: use the right clock phandles for UARTs
pinctrl: sunxi: add clock support
arm: sunxi: Add clock to pinctrl node
clk: sunxi: drop CLK_IGNORE_UNUSED
clk: sunxi: drop an unnecesary kmalloc
Emilio López
maintainerPATCH v2 1/7 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
This patchset adds DT support for all the AXI, AHB, APB0 and APB1 gates present on sunxi SoCs.
---
Changes from v1:
- renamed compatibles to sun4i
- renamed TWI -> I2C
- added note about gate availability on sun5i
Documentation/devicetree/bindings/clock/sunxi.txt | 109 +++++++++++++++++++++- drivers/clk/sunxi/clk-sunxi.c | 88 +++++++++++++++++ 2 files changed, 196 insertions(+), 1 deletion(-)
Documentation/devicetree/bindings/clock/sunxi.txt +108 -1
@@ -10,15 +10,23 @@ Required properties: "allwinner,sun4i-pll1-clk" - for the main PLL clock "allwinner,sun4i-cpu-clk" - for the CPU multiplexer clock "allwinner,sun4i-axi-clk" - for the AXI clock + "allwinner,sun4i-axi-gates-clk" - for the AXI gates "allwinner,sun4i-ahb-clk" - for the AHB clock + "allwinner,sun4i-ahb-gates-clk" - for the AHB gates "allwinner,sun4i-apb0-clk" - for the APB0 clock + "allwinner,sun4i-apb0-gates-clk" - for the APB0 gates "allwinner,sun4i-apb1-clk" - for the APB1 clock "allwinner,sun4i-apb1-mux-clk" - for the APB1 clock muxing + "allwinner,sun4i-apb1-gates-clk" - for the APB1 gates Required properties for all clocks: - reg : shall be the control register address for the clock. - clocks : shall be the input parent clock(s) phandle for the clock -- #clock-cells : from common clock binding; shall be set to 0. +- #clock-cells : from common clock binding; shall be set to 0 except for + "allwinner,sun4i-*-gates-clk" where it shall be set to 1 + +Additionally, "allwinner,sun4i-*-gates-clk" clocks require: +- clock-output-names : the corresponding gate names that the clock controls For example:
@@ -42,3 +50,102 @@ cpu: cpu at 01c20054 { reg = <0x01c20054 0x4>; clocks = <&osc32k>, <&osc24M>, <&pll1>; }; + + + +Gate clock outputs + +The "allwinner,sun4i-*-gates-clk" clocks provide several gatable outputs; +their corresponding offsets as present on sun4i are listed below. Note that +some of these gates are not present on sun5i. + + * AXI gates ("allwinner,sun4i-axi-gates-clk") + + DRAM 0 + + * AHB gates ("allwinner,sun4i-ahb-gates-clk") + + USB0 0 + EHCI0 1 + OHCI0 2* + EHCI1 3 + OHCI1 4* + SS 5 + DMA 6 + BIST 7 + MMC0 8 + MMC1 9 + MMC2 10 + MMC3 11 + MS 12** + NAND 13 + SDRAM 14 + + ACE 16 + EMAC 17 + TS 18 + + SPI0 20 + SPI1 21 + SPI2 22 + SPI3 23 + PATA 24 + SATA 25** + GPS 26* + + VE 32 + TVD 33 + TVE0 34 + TVE1 35 + LCD0 36 + LCD1 37 + + CSI0 40 + CSI1 41 + + HDMI 43 + DE_BE0 44 + DE_BE1 45 + DE_FE0 46 + DE_FE1 47 + + MP 50 + + MALI400 52 + + * APB0 gates ("allwinner,sun4i-apb0-gates-clk") + + CODEC 0 + SPDIF 1* + AC97 2 + IIS 3 + + PIO 5 + IR0 6 + IR1 7 + + KEYPAD 10 + + * APB1 gates ("allwinner,sun4i-apb1-gates-clk") + + I2C0 0 + I2C1 1 + I2C2 2 + + CAN 4 + SCR 5 + PS20 6 + PS21 7 + + UART0 16 + UART1 17 + UART2 18 + UART3 19 + UART4 20 + UART5 21 + UART6 22 + UART7 23 + +Notation: + [*]: The datasheet didn't mention these, but they are present on AW code + [**]: The datasheet had this marked as "NC" but they are used on AW code
drivers/clk/sunxi/clk-sunxi.c +88 -0
104 lines not on this page · raw message
Signature · 1 line
1.8.2
PATCH v2 1/7 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Quoting Emilio L?pez (2013-03-27 14:20:37)
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 305
Quoted 9 lines
drivers/clk/sunxi/clk-sunxi.c +1
-0@@ -302,6 +302,82 @@ static void __init sunxi_divider_clk_setup(struct device_node *node, } +
A lot of white space between these functions.
Quoted 48 lines
drivers/clk/sunxi/clk-sunxi.c +48
-0+/** +* sunxi_gates_clk_setup() - Setup function for leaf gates on clocks +*/ + +#define SUNXI_GATES_MAX_SIZE 64 + +struct gates_data { + DECLARE_BITMAP(mask, SUNXI_GATES_MAX_SIZE); +}; + +static const __initconst struct gates_data axi_gates_data = { + .mask = {1 }, +}; + +static const __initconst struct gates_data ahb_gates_data = { + .mask = {0x7F77FFF ,0x14FB3F }, +}; + +static const __initconst struct gates_data apb0_gates_data = { + .mask = {0x4EF }, +}; + +static const __initconst struct gates_data apb1_gates_data = { + .mask = {0xFF00F7 }, +}; + +static void __init sunxi_gates_clk_setup (struct device_node *node, +struct gates_data *data) +{ +struct clk_onecell_data *clk_data; +const char *clk_parent; +const char *clk_name; +void *reg; +int qty; +int i =0 ; +int j =0 ; +int ignore; + + reg =of_iomap (node,0 ); + + clk_parent =of_clk_get_parent_name (node,0 ); + +/* Worst-case size approximation and memory allocation */ + qty =find_last_bit (data->mask,SUNXI_GATES_MAX_SIZE ); + clk_data =kmalloc (sizeof (struct clk_onecell_data ),GFP_KERNEL ); +if (!clk_data) +return ; + clk_data->clks =kzalloc ((qty+1 ) *sizeof (struct clk *),GFP_KERNEL );
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 363
Quoted 10 lines
drivers/clk/sunxi/clk-sunxi.c +10
-0+if (!clk_data->clks) { +kfree (clk_data); +return ; + } + +for_each_set_bit (i, data->mask,SUNXI_GATES_MAX_SIZE ) { +of_property_read_string_index (node,"clock-output-names" , + j, &clk_name); + +/* No driver claims this clock, but it should remain gated */
Should the comment read, "ungated" instead of "gated"?
Quoted 37 lines
drivers/clk/sunxi/clk-sunxi.c +26
-0+ ignore = !strcmp ("ahb_sdram" , clk_name) ?CLK_IGNORE_UNUSED :0 ; + + clk_data->clks[i] =clk_register_gate (NULL , clk_name, + clk_parent, ignore, + reg +4 * (i/32 ), i %32 , +0 , &clk_lock); +WARN_ON (IS_ERR (clk_data->clks[i])); + + j++; + } + +/* Adjust to the real max */ + clk_data->clk_num = i; + +of_clk_add_provider (node, of_clk_src_onecell_get, clk_data); +} +/* Matches for of_clk_init */ static const __initconst struct of_device_id clk_match[] = { {.compatible ="fixed-clock" , .data = of_fixed_clk_setup,},@@ -331,6 +407,15 @@ static const __initconst struct of_device_id clk_mux_match[] = { {} }; +/* Matches for gate clocks */ +static const __initconst struct of_device_id clk_gates_match[] = { + {.compatible ="allwinner,sun4i-axi-gates-clk" , .data = &axi_gates_data,}, + {.compatible ="allwinner,sun4i-ahb-gates-clk" , .data = &ahb_gates_data,}, + {.compatible ="allwinner,sun4i-apb0-gates-clk" , .data = &apb0_gates_data,}, + {.compatible ="allwinner,sun4i-apb1-gates-clk" , .data = &apb1_gates_data,}, + {} +}; +static void __init of_sunxi_table_clock_setup (const struct of_device_id *clk_match,void *function) {
drivers/clk/sunxi/clk-sunxi.c void __init sunxi_init_clocks(void) · line 449
Quoted 6 lines
drivers/clk/sunxi/clk-sunxi.c +3
-0@@ -359,4 +444,7 @@ void __init sunxi_init_clocks(void)/* Register mux clocks */ of_sunxi_table_clock_setup (clk_mux_match, sunxi_mux_clk_setup); + +/* Register gate clocks */ +of_sunxi_table_clock_setup (clk_gates_match, sunxi_gates_clk_setup);
I'm still a device tree noob, so this may be a dumb question. Can the above be converted to of_clk_init?
Regards,
Mike
Quoted 3 lines
drivers/clk/sunxi/clk-sunxi.c +0
-0}Signature · 1 line
1.8.2
Emilio López
maintainerPATCH v2 1/7 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Hi Mike,
El 03/04/13 18:48, Mike Turquette escribi?:
Quoted 9 lines
Quoting Emilio L?pez (2013-03-27 14:20:37)
Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c +0
-0@@ -302,6 +302,82 @@ static void __init sunxi_divider_clk_setup(struct device_node *node, }
Quoted 3 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 305
Quoted 1 line
drivers/clk/sunxi/clk-sunxi.c +1
-0+A lot of white space between these functions.
All of the function blocks are separated with three spaces on the file; I thought it looked more readable that way, but I don't have any strong opinion on separation. Is there any standard for this used on the kernel?
In any case, and to keep consistency, can we handle this on a follow-up patch?
Quoted 52 lines
Quoted 52 lines
drivers/clk/sunxi/clk-sunxi.c +52
-0+/** +* sunxi_gates_clk_setup() - Setup function for leaf gates on clocks +*/ + +#define SUNXI_GATES_MAX_SIZE 64 + +struct gates_data { + DECLARE_BITMAP(mask, SUNXI_GATES_MAX_SIZE); +}; + +static const __initconst struct gates_data axi_gates_data = { + .mask = {1 }, +}; + +static const __initconst struct gates_data ahb_gates_data = { + .mask = {0x7F77FFF ,0x14FB3F }, +}; + +static const __initconst struct gates_data apb0_gates_data = { + .mask = {0x4EF }, +}; + +static const __initconst struct gates_data apb1_gates_data = { + .mask = {0xFF00F7 }, +}; + +static void __init sunxi_gates_clk_setup (struct device_node *node, +struct gates_data *data) +{ +struct clk_onecell_data *clk_data; +const char *clk_parent; +const char *clk_name; +void *reg; +int qty; +int i =0 ; +int j =0 ; +int ignore; + + reg =of_iomap (node,0 ); + + clk_parent =of_clk_get_parent_name (node,0 ); + +/* Worst-case size approximation and memory allocation */ + qty =find_last_bit (data->mask,SUNXI_GATES_MAX_SIZE ); + clk_data =kmalloc (sizeof (struct clk_onecell_data ),GFP_KERNEL ); +if (!clk_data) +return ; + clk_data->clks =kzalloc ((qty+1 ) *sizeof (struct clk *),GFP_KERNEL ); +if (!clk_data->clks) { +kfree (clk_data); +return ; + }
Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 363
Quoted 6 lines
drivers/clk/sunxi/clk-sunxi.c +6
-0+ +for_each_set_bit (i, data->mask,SUNXI_GATES_MAX_SIZE ) { +of_property_read_string_index (node,"clock-output-names" , + j, &clk_name); + +/* No driver claims this clock, but it should remain gated */ Should the comment read, "ungated" instead of "gated"?
Indeed, good catch. Do you want me to resend the series, or can you amend this when picking the patches?
Quoted 38 lines
Quoted 38 lines
drivers/clk/sunxi/clk-sunxi.c +26
-0+ ignore = !strcmp ("ahb_sdram" , clk_name) ?CLK_IGNORE_UNUSED :0 ; + + clk_data->clks[i] =clk_register_gate (NULL , clk_name, + clk_parent, ignore, + reg +4 * (i/32 ), i %32 , +0 , &clk_lock); +WARN_ON (IS_ERR (clk_data->clks[i])); + + j++; + } + +/* Adjust to the real max */ + clk_data->clk_num = i; + +of_clk_add_provider (node, of_clk_src_onecell_get, clk_data); +} +/* Matches for of_clk_init */ static const __initconst struct of_device_id clk_match[] = { {.compatible ="fixed-clock" , .data = of_fixed_clk_setup,},@@ -331,6 +407,15 @@ static const __initconst struct of_device_id clk_mux_match[] = { {} }; +/* Matches for gate clocks */ +static const __initconst struct of_device_id clk_gates_match[] = { + {.compatible ="allwinner,sun4i-axi-gates-clk" , .data = &axi_gates_data,}, + {.compatible ="allwinner,sun4i-ahb-gates-clk" , .data = &ahb_gates_data,}, + {.compatible ="allwinner,sun4i-apb0-gates-clk" , .data = &apb0_gates_data,}, + {.compatible ="allwinner,sun4i-apb1-gates-clk" , .data = &apb1_gates_data,}, + {} +}; +static void __init of_sunxi_table_clock_setup (const struct of_device_id *clk_match,void *function) {@@ -359,4 +444,7 @@ void __init sunxi_init_clocks(void)
Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c void __init sunxi_init_clocks(void) · line 449
Quoted 5 lines
drivers/clk/sunxi/clk-sunxi.c +3
-0/* Register mux clocks */ of_sunxi_table_clock_setup (clk_mux_match, sunxi_mux_clk_setup); + +/* Register gate clocks */ +of_sunxi_table_clock_setup (clk_gates_match, sunxi_gates_clk_setup);I'm still a device tree noob, so this may be a dumb question. Can the above be converted to
of_clk_init?
As far as I know, you can't, because of_clk_init doesn't allow for custom data to be passed to the functions. If we were to use of_clk_init we would need one function per clock, and it would be mostly duplicated code / wrappers. I've added Gregory on Cc, please correct me if this is not the case.
Thanks for the review,
Emilio
PATCH v2 1/7 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Quoting Emilio L?pez (2013-04-03 18:19:22)
Quoted 16 lines
Hi Mike,
El 03/04/13 18:48, Mike Turquette escribi?:
Quoted 9 lines
Quoting Emilio L?pez (2013-03-27 14:20:37)
Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c +0
-0@@ -302,6 +302,82 @@ static void __init sunxi_divider_clk_setup(struct device_node *node, }Quoted 3 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 305
Quoted 1 line
drivers/clk/sunxi/clk-sunxi.c +1
-0+A lot of white space between these functions.
Quoted 7 lines
All of the function blocks are separated with three spaces on the file; I thought it looked more readable that way, but I don't have any strong opinion on separation. Is there any standard for this used on the kernel?
In any case, and to keep consistency, can we handle this on a follow-up patch?
If it's consistent throughout the file then go ahead and keep it.
Quoted 61 lines
Quoted 52 lines
Quoted 52 lines
drivers/clk/sunxi/clk-sunxi.c +52
-0+/** +* sunxi_gates_clk_setup() - Setup function for leaf gates on clocks +*/ + +#define SUNXI_GATES_MAX_SIZE 64 + +struct gates_data { + DECLARE_BITMAP(mask, SUNXI_GATES_MAX_SIZE); +}; + +static const __initconst struct gates_data axi_gates_data = { + .mask = {1 }, +}; + +static const __initconst struct gates_data ahb_gates_data = { + .mask = {0x7F77FFF ,0x14FB3F }, +}; + +static const __initconst struct gates_data apb0_gates_data = { + .mask = {0x4EF }, +}; + +static const __initconst struct gates_data apb1_gates_data = { + .mask = {0xFF00F7 }, +}; + +static void __init sunxi_gates_clk_setup (struct device_node *node, +struct gates_data *data) +{ +struct clk_onecell_data *clk_data; +const char *clk_parent; +const char *clk_name; +void *reg; +int qty; +int i =0 ; +int j =0 ; +int ignore; + + reg =of_iomap (node,0 ); + + clk_parent =of_clk_get_parent_name (node,0 ); + +/* Worst-case size approximation and memory allocation */ + qty =find_last_bit (data->mask,SUNXI_GATES_MAX_SIZE ); + clk_data =kmalloc (sizeof (struct clk_onecell_data ),GFP_KERNEL ); +if (!clk_data) +return ; + clk_data->clks =kzalloc ((qty+1 ) *sizeof (struct clk *),GFP_KERNEL ); +if (!clk_data->clks) { +kfree (clk_data); +return ; + }Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 363
Quoted 6 lines
drivers/clk/sunxi/clk-sunxi.c +6
-0+ +for_each_set_bit (i, data->mask,SUNXI_GATES_MAX_SIZE ) { +of_property_read_string_index (node,"clock-output-names" , + j, &clk_name); + +/* No driver claims this clock, but it should remain gated */ Should the comment read, "ungated" instead of "gated"?
Quoted 3 lines
Indeed, good catch. Do you want me to resend the series, or can you amend this when picking the patches?
I can amend it, but I don't like to get into the habit of doing that too often.
I'll wait on Gregory's response on the of_clk_init stuff before I do so.
Regards,
Mike
Quoted 56 lines
Quoted 38 lines
Quoted 38 lines
drivers/clk/sunxi/clk-sunxi.c +26
-0+ ignore = !strcmp ("ahb_sdram" , clk_name) ?CLK_IGNORE_UNUSED :0 ; + + clk_data->clks[i] =clk_register_gate (NULL , clk_name, + clk_parent, ignore, + reg +4 * (i/32 ), i %32 , +0 , &clk_lock); +WARN_ON (IS_ERR (clk_data->clks[i])); + + j++; + } + +/* Adjust to the real max */ + clk_data->clk_num = i; + +of_clk_add_provider (node, of_clk_src_onecell_get, clk_data); +} +/* Matches for of_clk_init */ static const __initconst struct of_device_id clk_match[] = { {.compatible ="fixed-clock" , .data = of_fixed_clk_setup,},@@ -331,6 +407,15 @@ static const __initconst struct of_device_id clk_mux_match[] = { {} }; +/* Matches for gate clocks */ +static const __initconst struct of_device_id clk_gates_match[] = { + {.compatible ="allwinner,sun4i-axi-gates-clk" , .data = &axi_gates_data,}, + {.compatible ="allwinner,sun4i-ahb-gates-clk" , .data = &ahb_gates_data,}, + {.compatible ="allwinner,sun4i-apb0-gates-clk" , .data = &apb0_gates_data,}, + {.compatible ="allwinner,sun4i-apb1-gates-clk" , .data = &apb1_gates_data,}, + {} +}; +static void __init of_sunxi_table_clock_setup (const struct of_device_id *clk_match,void *function) {@@ -359,4 +444,7 @@ void __init sunxi_init_clocks(void)Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c void __init sunxi_init_clocks(void) · line 449
Quoted 5 lines
drivers/clk/sunxi/clk-sunxi.c +3
-0/* Register mux clocks */ of_sunxi_table_clock_setup (clk_mux_match, sunxi_mux_clk_setup); + +/* Register gate clocks */ +of_sunxi_table_clock_setup (clk_gates_match, sunxi_gates_clk_setup);I'm still a device tree noob, so this may be a dumb question. Can the above be converted to
of_clk_init?As far as I know, you can't, because
of_clk_initdoesn't allow for custom data to be passed to the functions. If we were to useof_clk_initwe would need one function per clock, and it would be mostly duplicated code / wrappers. I've added Gregory on Cc, please correct me if this is not the case.Thanks for the review,
Emilio
PATCH v2 1/7 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Quoted 92 lines · Mike Turquette
Quoting Emilio L??pez (2013-04-03 18:19:22)
Quoted 16 lines
Hi Mike,
El 03/04/13 18:48, Mike Turquette escribi??:
Quoted 9 lines
Quoting Emilio L??pez (2013-03-27 14:20:37)
Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c +0
-0@@ -302,6 +302,82 @@ static void __init sunxi_divider_clk_setup(struct device_node *node, }Quoted 3 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 305
Quoted 1 line
drivers/clk/sunxi/clk-sunxi.c +1
-0+A lot of white space between these functions.
Quoted 7 lines
All of the function blocks are separated with three spaces on the file; I thought it looked more readable that way, but I don't have any strong opinion on separation. Is there any standard for this used on the kernel?
In any case, and to keep consistency, can we handle this on a follow-up patch?
If it's consistent throughout the file then go ahead and keep it.
Quoted 61 lines
Quoted 52 lines
Quoted 52 lines
drivers/clk/sunxi/clk-sunxi.c +52
-0+/** +* sunxi_gates_clk_setup() - Setup function for leaf gates on clocks +*/ + +#define SUNXI_GATES_MAX_SIZE 64 + +struct gates_data { + DECLARE_BITMAP(mask, SUNXI_GATES_MAX_SIZE); +}; + +static const __initconst struct gates_data axi_gates_data = { + .mask = {1 }, +}; + +static const __initconst struct gates_data ahb_gates_data = { + .mask = {0x7F77FFF ,0x14FB3F }, +}; + +static const __initconst struct gates_data apb0_gates_data = { + .mask = {0x4EF }, +}; + +static const __initconst struct gates_data apb1_gates_data = { + .mask = {0xFF00F7 }, +}; + +static void __init sunxi_gates_clk_setup (struct device_node *node, +struct gates_data *data) +{ +struct clk_onecell_data *clk_data; +const char *clk_parent; +const char *clk_name; +void *reg; +int qty; +int i =0 ; +int j =0 ; +int ignore; + + reg =of_iomap (node,0 ); + + clk_parent =of_clk_get_parent_name (node,0 ); + +/* Worst-case size approximation and memory allocation */ + qty =find_last_bit (data->mask,SUNXI_GATES_MAX_SIZE ); + clk_data =kmalloc (sizeof (struct clk_onecell_data ),GFP_KERNEL ); +if (!clk_data) +return ; + clk_data->clks =kzalloc ((qty+1 ) *sizeof (struct clk *),GFP_KERNEL ); +if (!clk_data->clks) { +kfree (clk_data); +return ; + }Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 363
Quoted 6 lines
drivers/clk/sunxi/clk-sunxi.c +6
-0+ +for_each_set_bit (i, data->mask,SUNXI_GATES_MAX_SIZE ) { +of_property_read_string_index (node,"clock-output-names" , + j, &clk_name); + +/* No driver claims this clock, but it should remain gated */ Should the comment read, "ungated" instead of "gated"?
Quoted 3 lines
Indeed, good catch. Do you want me to resend the series, or can you amend this when picking the patches?
Quoted 60 lines
I can amend it, but I don't like to get into the habit of doing that too often.
I'll wait on Gregory's response on the
of_clk_initstuff before I do so.Regards,
MikeQuoted 52 lines
Quoted 38 lines
Quoted 38 lines
drivers/clk/sunxi/clk-sunxi.c +26
-0+ ignore = !strcmp ("ahb_sdram" , clk_name) ?CLK_IGNORE_UNUSED :0 ; + + clk_data->clks[i] =clk_register_gate (NULL , clk_name, + clk_parent, ignore, + reg +4 * (i/32 ), i %32 , +0 , &clk_lock); +WARN_ON (IS_ERR (clk_data->clks[i])); + + j++; + } + +/* Adjust to the real max */ + clk_data->clk_num = i; + +of_clk_add_provider (node, of_clk_src_onecell_get, clk_data); +} +/* Matches for of_clk_init */ static const __initconst struct of_device_id clk_match[] = { {.compatible ="fixed-clock" , .data = of_fixed_clk_setup,},@@ -331,6 +407,15 @@ static const __initconst struct of_device_id clk_mux_match[] = { {} }; +/* Matches for gate clocks */ +static const __initconst struct of_device_id clk_gates_match[] = { + {.compatible ="allwinner,sun4i-axi-gates-clk" , .data = &axi_gates_data,}, + {.compatible ="allwinner,sun4i-ahb-gates-clk" , .data = &ahb_gates_data,}, + {.compatible ="allwinner,sun4i-apb0-gates-clk" , .data = &apb0_gates_data,}, + {.compatible ="allwinner,sun4i-apb1-gates-clk" , .data = &apb1_gates_data,}, + {} +}; +static void __init of_sunxi_table_clock_setup (const struct of_device_id *clk_match,void *function) {@@ -359,4 +444,7 @@ void __init sunxi_init_clocks(void)Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c void __init sunxi_init_clocks(void) · line 449
Quoted 5 lines
drivers/clk/sunxi/clk-sunxi.c +3
-0/* Register mux clocks */ of_sunxi_table_clock_setup (clk_mux_match, sunxi_mux_clk_setup); + +/* Register gate clocks */ +of_sunxi_table_clock_setup (clk_gates_match, sunxi_gates_clk_setup);I'm still a device tree noob, so this may be a dumb question. Can the above be converted to
of_clk_init?As far as I know, you can't, because
of_clk_initdoesn't allow for custom data to be passed to the functions. If we were to useof_clk_initwe would need one function per clock, and it would be mostly duplicated code / wrappers. I've added Gregory on Cc, please correct me if this is not the case.
I confirm that the current implementation of of_clk_init only take setup functions. That was also the reason why we didn't use it in mvebu/clk-core.c for example.
Maybe it should be a good improvement to allow of_clk_init to receive a function _and_ data for a given node. Something like that;
typedef void (*of_clk_init_cb_t)(struct device_node *, void *data); struct clk_of_setup {of_clk_init_cb_t clk_init_cb;
void* data;
}
void __init of_clk_init(const struct of_device_id *matches)
{
struct device_node *np;
if (!matches)
matches = __clk_of_table;
for_each_matching_node(np, matches) {
const struct of_device_id *match = of_match_node(matches, np);
match->clk_init_cb(np, match->data);
}
}I have just writen this code in this email I didn't even try to compile this code. This was just a rough idea which could be use as a base for future patch for 3.11. With a good coccinelle script it should be not too complicated.
But about this current patch, I am fine with it, and you can add my
Quoted 4 lines
Quoted 4 lines
Thanks for the review,
Emilio
Signature · 4 lines
Gregory Clement, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
PATCH v2 1/7 clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
Quoting Gregory CLEMENT (2013-04-04 13:45:24)
Quoted 180 lines · Mike Turquette
Quoted 92 lines
Quoting Emilio L??pez (2013-04-03 18:19:22)
Quoted 16 lines
Hi Mike,
El 03/04/13 18:48, Mike Turquette escribi??:
Quoted 9 lines
Quoting Emilio L??pez (2013-03-27 14:20:37)
Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c +0
-0@@ -302,6 +302,82 @@ static void __init sunxi_divider_clk_setup(struct device_node *node, }Quoted 3 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 305
Quoted 1 line
drivers/clk/sunxi/clk-sunxi.c +1
-0+A lot of white space between these functions.
Quoted 7 lines
All of the function blocks are separated with three spaces on the file; I thought it looked more readable that way, but I don't have any strong opinion on separation. Is there any standard for this used on the kernel?
In any case, and to keep consistency, can we handle this on a follow-up patch?
If it's consistent throughout the file then go ahead and keep it.
Quoted 61 lines
Quoted 52 lines
Quoted 52 lines
drivers/clk/sunxi/clk-sunxi.c +52
-0+/** +* sunxi_gates_clk_setup() - Setup function for leaf gates on clocks +*/ + +#define SUNXI_GATES_MAX_SIZE 64 + +struct gates_data { + DECLARE_BITMAP(mask, SUNXI_GATES_MAX_SIZE); +}; + +static const __initconst struct gates_data axi_gates_data = { + .mask = {1 }, +}; + +static const __initconst struct gates_data ahb_gates_data = { + .mask = {0x7F77FFF ,0x14FB3F }, +}; + +static const __initconst struct gates_data apb0_gates_data = { + .mask = {0x4EF }, +}; + +static const __initconst struct gates_data apb1_gates_data = { + .mask = {0xFF00F7 }, +}; + +static void __init sunxi_gates_clk_setup (struct device_node *node, +struct gates_data *data) +{ +struct clk_onecell_data *clk_data; +const char *clk_parent; +const char *clk_name; +void *reg; +int qty; +int i =0 ; +int j =0 ; +int ignore; + + reg =of_iomap (node,0 ); + + clk_parent =of_clk_get_parent_name (node,0 ); + +/* Worst-case size approximation and memory allocation */ + qty =find_last_bit (data->mask,SUNXI_GATES_MAX_SIZE ); + clk_data =kmalloc (sizeof (struct clk_onecell_data ),GFP_KERNEL ); +if (!clk_data) +return ; + clk_data->clks =kzalloc ((qty+1 ) *sizeof (struct clk *),GFP_KERNEL ); +if (!clk_data->clks) { +kfree (clk_data); +return ; + }Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c static void __init sunxi_divider_clk_setup(struct device_node *node, · line 363
Quoted 6 lines
drivers/clk/sunxi/clk-sunxi.c +6
-0+ +for_each_set_bit (i, data->mask,SUNXI_GATES_MAX_SIZE ) { +of_property_read_string_index (node,"clock-output-names" , + j, &clk_name); + +/* No driver claims this clock, but it should remain gated */ Should the comment read, "ungated" instead of "gated"?
Quoted 3 lines
Indeed, good catch. Do you want me to resend the series, or can you amend this when picking the patches?
Quoted 60 lines
I can amend it, but I don't like to get into the habit of doing that too often.
I'll wait on Gregory's response on the
of_clk_initstuff before I do so.Regards,
MikeQuoted 52 lines
Quoted 38 lines
Quoted 38 lines
drivers/clk/sunxi/clk-sunxi.c +26
-0+ ignore = !strcmp ("ahb_sdram" , clk_name) ?CLK_IGNORE_UNUSED :0 ; + + clk_data->clks[i] =clk_register_gate (NULL , clk_name, + clk_parent, ignore, + reg +4 * (i/32 ), i %32 , +0 , &clk_lock); +WARN_ON (IS_ERR (clk_data->clks[i])); + + j++; + } + +/* Adjust to the real max */ + clk_data->clk_num = i; + +of_clk_add_provider (node, of_clk_src_onecell_get, clk_data); +} +/* Matches for of_clk_init */ static const __initconst struct of_device_id clk_match[] = { {.compatible ="fixed-clock" , .data = of_fixed_clk_setup,},@@ -331,6 +407,15 @@ static const __initconst struct of_device_id clk_mux_match[] = { {} }; +/* Matches for gate clocks */ +static const __initconst struct of_device_id clk_gates_match[] = { + {.compatible ="allwinner,sun4i-axi-gates-clk" , .data = &axi_gates_data,}, + {.compatible ="allwinner,sun4i-ahb-gates-clk" , .data = &ahb_gates_data,}, + {.compatible ="allwinner,sun4i-apb0-gates-clk" , .data = &apb0_gates_data,}, + {.compatible ="allwinner,sun4i-apb1-gates-clk" , .data = &apb1_gates_data,}, + {} +}; +static void __init of_sunxi_table_clock_setup (const struct of_device_id *clk_match,void *function) {@@ -359,4 +444,7 @@ void __init sunxi_init_clocks(void)Quoted 8 lines
drivers/clk/sunxi/clk-sunxi.c void __init sunxi_init_clocks(void) · line 449
Quoted 5 lines
drivers/clk/sunxi/clk-sunxi.c +3
-0/* Register mux clocks */ of_sunxi_table_clock_setup (clk_mux_match, sunxi_mux_clk_setup); + +/* Register gate clocks */ +of_sunxi_table_clock_setup (clk_gates_match, sunxi_gates_clk_setup);I'm still a device tree noob, so this may be a dumb question. Can the above be converted to
of_clk_init?As far as I know, you can't, because
of_clk_initdoesn't allow for custom data to be passed to the functions. If we were to useof_clk_initwe would need one function per clock, and it would be mostly duplicated code / wrappers. I've added Gregory on Cc, please correct me if this is not the case.I confirm that the current implementation of
of_clk_initonly take setup functions. That was also the reason why we didn't use it inmvebu/clk-core.cfor example.Maybe it should be a good improvement to allow
of_clk_initto receive a function _and_ data for a given node. Something like that;typedef void (*of_clk_init_cb_t)(struct
device_node*, void *data); structclk_of_setup{of_clk_init_cb_tclk_init_cb;
void* data;
}void
__initof_clk_init(const struct of_device_id *matches){ struct device_node *np; if (!matches) matches = __clk_of_table; for_each_matching_node(np, matches) { const struct of_device_id *match = of_match_node(matches, np); match->clk_init_cb(np, match->data); } }
Quoted 4 lines
I have just writen this code in this email I didn't even try to compile this code. This was just a rough idea which could be use as a base for future patch for 3.11. With a good coccinelle script it should be not too complicated.
Sure, maybe some solution can be found for 3.11.
Quoted 4 lines
But about this current patch, I am fine with it, and you can add my
Reviewed-by Gregory CLEMENT <[email protected]>
Thanks. The three clock patches have been pushed to clk-next.
Regards,
Mike
Quoted 12 lines
Quoted 5 lines
Quoted 5 lines
Thanks for the review,
Emilio
Signature · 4 lines
Gregory Clement, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
Emilio López
maintainerPATCH v2 2/7 arm: sunxi: Add clock definitions for AXI, AHB, APB0, APB1 gates
In reply to Emilio LópezThis commit adds the corresponding DT bindings for all the AXI, AHB, APB0 and APB1 gates.
---
Changes from v1:
- renamed compatible strings to sun4i
- rename twi gates to i2c
arch/arm/boot/dts/sunxi.dtsi | 49 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+)
arch/arm/boot/dts/sunxi.dtsi +49 -0
@@ -80,6 +80,14 @@ clocks = <&cpu>; }; + axi_gates: axi_gates at 01c2005c { + #clock-cells = <1>; + compatible = "allwinner,sun4i-axi-gates-clk"; + reg = <0x01c2005c 0x4>; + clocks = <&axi>; + clock-output-names = "axi_dram"; + }; + ahb: ahb at 01c20054 { #clock-cells = <0>; compatible = "allwinner,sun4i-ahb-clk";
@@ -87,6 +95,24 @@ clocks = <&axi>; }; + ahb_gates: ahb_gates at 01c20060 { + #clock-cells = <1>; + compatible = "allwinner,sun4i-ahb-gates-clk"; + reg = <0x01c20060 0x8>; + clocks = <&ahb>; + clock-output-names = "ahb_usb0", "ahb_ehci0", + "ahb_ohci0", "ahb_ehci1", "ahb_ohci1", "ahb_ss", + "ahb_dma", "ahb_bist", "ahb_mmc0", "ahb_mmc1", + "ahb_mmc2", "ahb_mmc3", "ahb_ms", "ahb_nand", + "ahb_sdram", "ahb_ace", "ahb_emac", "ahb_ts", + "ahb_spi0", "ahb_spi1", "ahb_spi2", "ahb_spi3", + "ahb_pata", "ahb_sata", "ahb_gps", "ahb_ve", + "ahb_tvd", "ahb_tve0", "ahb_tve1", "ahb_lcd0", + "ahb_lcd1", "ahb_csi0", "ahb_csi1", "ahb_hdmi", + "ahb_de_be0", "ahb_de_be1", "ahb_de_fe0", + "ahb_de_fe1", "ahb_mp", "ahb_mali400"; + }; + apb0: apb0 at 01c20054 { #clock-cells = <0>; compatible = "allwinner,sun4i-apb0-clk";
@@ -94,6 +120,16 @@ clocks = <&ahb>; }; + apb0_gates: apb0_gates at 01c20068 { + #clock-cells = <1>; + compatible = "allwinner,sun4i-apb0-gates-clk"; + reg = <0x01c20068 0x4>; + clocks = <&apb0>; + clock-output-names = "apb0_codec", "apb0_spdif", + "apb0_ac97", "apb0_iis", "apb0_pio", "apb0_ir0", + "apb0_ir1", "apb0_keypad"; + }; + /* dummy is pll62 */ apb1_mux: apb1_mux at 01c20058 { #clock-cells = <0>;
@@ -108,6 +144,19 @@ reg = <0x01c20058 0x4>; clocks = <&apb1_mux>; }; + + apb1_gates: apb1_gates at 01c2006c { + #clock-cells = <1>; + compatible = "allwinner,sun4i-apb1-gates-clk"; + reg = <0x01c2006c 0x4>; + clocks = <&apb1>; + clock-output-names = "apb1_i2c0", "apb1_i2c1", + "apb1_i2c2", "apb1_can", "apb1_scr", + "apb1_ps20", "apb1_ps21", "apb1_uart0", + "apb1_uart1", "apb1_uart2", "apb1_uart3", + "apb1_uart4", "apb1_uart5", "apb1_uart6", + "apb1_uart7"; + }; }; soc {
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH v2 3/7 arm: sunxi: use the right clock phandles for UARTs
In reply to Emilio LópezAll the UARTs are connected to clock gates; now that our clock driver is able to handle them, make the switch.
---
Changes from v1:
- no changes
arch/arm/boot/dts/sun4i-a10.dtsi | 12 ++++++------ arch/arm/boot/dts/sunxi.dtsi | 4 ++-- 2 files changed, 8 insertions(+), 8 deletions(-)
arch/arm/boot/dts/sun4i-a10.dtsi +6 -6
@@ -54,7 +54,7 @@ interrupts = <1>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 16>; status = "disabled"; };
@@ -64,7 +64,7 @@ interrupts = <3>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 18>; status = "disabled"; };
@@ -74,7 +74,7 @@ interrupts = <17>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 20>; status = "disabled"; };
@@ -84,7 +84,7 @@ interrupts = <18>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 21>; status = "disabled"; };
@@ -94,7 +94,7 @@ interrupts = <19>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 22>; status = "disabled"; };
@@ -104,7 +104,7 @@ interrupts = <20>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 23>; status = "disabled"; }; };
arch/arm/boot/dts/sunxi.dtsi +2 -2
@@ -191,7 +191,7 @@ interrupts = <2>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 17>; status = "disabled"; };
@@ -201,7 +201,7 @@ interrupts = <4>; reg-shift = <2>; reg-io-width = <4>; - clocks = <&osc>; + clocks = <&apb1_gates 19>; status = "disabled"; }; };
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH v2 4/7 pinctrl: sunxi: add clock support
In reply to Emilio LópezFor the port controller to work, we need to enable the apb0_pio gate. This commit adds the ability to enable one clock specified on the device tree to the pinctrl driver.
---
Changes from v1:
- no changes
drivers/pinctrl/pinctrl-sunxi.c | 8 ++++++++ 1 file changed, 8 insertions(+)
drivers/pinctrl/pinctrl-sunxi.c +8 -0
@@ -11,6 +11,7 @@*/ #include <linux/io.h> +#include <linux/clk.h> #include <linux/gpio.h> #include <linux/module.h> #include <linux/of.h>
@@ -1917,6 +1918,7 @@ static int sunxi_pinctrl_probe(struct platform_device *pdev)struct pinctrl_pin_desc *pins;struct sunxi_pinctrl *pctl;int i, ret, last_pin; +struct clk *clk; pctl =devm_kzalloc (&pdev->dev,sizeof (*pctl),GFP_KERNEL );if (!pctl)
@@ -1987,6 +1989,12 @@ static int sunxi_pinctrl_probe(struct platform_device *pdev) goto gpiochip_error; } + clk =devm_clk_get (&pdev->dev,NULL ); +if (IS_ERR (clk)) + goto gpiochip_error; + +clk_prepare_enable (clk); +dev_info (&pdev->dev,"initialized sunXi PIO driver\n" );return 0 ;
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH v2 5/7 arm: sunxi: Add clock to pinctrl node
In reply to Emilio LópezThe port controller needs the apb0_pio clock enabled to be able to work. This commit declares that on the device tree.
---
Changes from v1:
- no changes
arch/arm/boot/dts/sun4i-a10.dtsi | 1 + arch/arm/boot/dts/sun5i-a13.dtsi | 1 + 2 files changed, 2 insertions(+)
arch/arm/boot/dts/sun4i-a10.dtsi +1 -0
@@ -21,6 +21,7 @@ pio: pinctrl at 01c20800 { compatible = "allwinner,sun4i-a10-pinctrl"; reg = <0x01c20800 0x400>; + clocks = <&apb0_gates 5>; gpio-controller; #address-cells = <1>; #size-cells = <0>;
arch/arm/boot/dts/sun5i-a13.dtsi +1 -0
@@ -22,6 +22,7 @@ pio: pinctrl at 01c20800 { compatible = "allwinner,sun5i-a13-pinctrl"; reg = <0x01c20800 0x400>; + clocks = <&apb0_gates 5>; gpio-controller; #address-cells = <1>; #size-cells = <0>;
Signature · 1 line
1.8.2
Emilio López
maintainerPATCH v2 6/7 clk: sunxi: drop CLK_IGNORE_UNUSED
In reply to Emilio LópezThis flag was in place to prevent important clocks from getting gated while they had no users. Now that the UART driver supports clocks properly, we can drop this.
---
Changes from v1:
- no changes
drivers/clk/sunxi/clk-sunxi.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/clk/sunxi/clk-sunxi.c +4 -4
@@ -41,8 +41,8 @@ static void __init sunxi_osc_clk_setup(struct device_node *node) parent =of_clk_get_parent_name (node,0 ); - clk =clk_register_gate (NULL , clk_name, parent,CLK_IGNORE_UNUSED , - reg,SUNXI_OSC24M_GATE ,0 , &clk_lock); + clk =clk_register_gate (NULL , clk_name, parent,0 , reg, +SUNXI_OSC24M_GATE ,0 , &clk_lock);if (clk) {of_clk_add_provider (node, of_clk_src_simple_get, clk);
@@ -198,8 +198,8 @@ static void __init sunxi_factors_clk_setup(struct device_node *node, parent =of_clk_get_parent_name (node,0 ); - clk =clk_register_factors (NULL , clk_name, parent,CLK_IGNORE_UNUSED , - reg, data->table, data->getter, &clk_lock); + clk =clk_register_factors (NULL , clk_name, parent,0 , reg, + data->table, data->getter, &clk_lock);if (clk) {of_clk_add_provider (node, of_clk_src_simple_get, clk);
Signature · 1 line
1.8.2
PATCH v2 6/7 clk: sunxi: drop CLK_IGNORE_UNUSED
Quoting Emilio L?pez (2013-03-27 14:20:42)
Quoted 5 lines
This flag was in place to prevent important clocks from getting gated while they had no users. Now that the UART driver supports clocks properly, we can drop this.
Signed-off-by Emilio L?pez <[email protected]>
This change looks good to me.
Regards,
Mike
Quoted 36 lines
---
Changes from v1:
- no changesdrivers/clk/sunxi/clk-sunxi.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)drivers/clk/sunxi/clk-sunxi.c +4
-4@@ -41,8 +41,8 @@ static void __init sunxi_osc_clk_setup(struct device_node *node) parent =of_clk_get_parent_name (node,0 ); - clk =clk_register_gate (NULL , clk_name, parent,CLK_IGNORE_UNUSED , - reg,SUNXI_OSC24M_GATE ,0 , &clk_lock); + clk =clk_register_gate (NULL , clk_name, parent,0 , reg, +SUNXI_OSC24M_GATE ,0 , &clk_lock);if (clk) {of_clk_add_provider (node, of_clk_src_simple_get, clk);@@ -198,8 +198,8 @@ static void __init sunxi_factors_clk_setup(struct device_node *node, parent =of_clk_get_parent_name (node,0 ); - clk =clk_register_factors (NULL , clk_name, parent,CLK_IGNORE_UNUSED , - reg, data->table, data->getter, &clk_lock); + clk =clk_register_factors (NULL , clk_name, parent,0 , reg, + data->table, data->getter, &clk_lock);if (clk) {of_clk_add_provider (node, of_clk_src_simple_get, clk);Signature · 1 line
1.8.2
Emilio López
maintainerPATCH v2 7/7 clk: sunxi: drop an unnecesary kmalloc
In reply to Emilio Lópezclk_register will copy this information, so we can just use a normal array and do one less dynamic allocation.
---
Changes from v1:
- this patch was not present on v1
drivers/clk/sunxi/clk-sunxi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/clk/sunxi/clk-sunxi.c +1 -1
@@ -232,7 +232,7 @@ static void __init sunxi_mux_clk_setup(struct device_node *node, {struct clk *clk;const char *clk_name = node->name; -const char **parents =kmalloc (sizeof (char *) *5 ,GFP_KERNEL ); +const char *parents[5 ];void *reg;int i =0 ;
Signature · 1 line
1.8.2
PATCH v2 7/7 clk: sunxi: drop an unnecesary kmalloc
Quoting Emilio L?pez (2013-03-27 14:20:43)
Quoted 4 lines
clk_registerwill copy this information, so we can just use a normal array and do one less dynamic allocation.Signed-off-by Emilio L?pez <[email protected]>
Looks good to me.
Regards,
Mike
Quoted 23 lines
---
Changes from v1:
- this patch was not present on v1drivers/clk/sunxi/clk-sunxi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/clk/sunxi/clk-sunxi.c +1
-1@@ -232,7 +232,7 @@ static void __init sunxi_mux_clk_setup(struct device_node *node, {struct clk *clk;const char *clk_name = node->name; -const char **parents =kmalloc (sizeof (char *) *5 ,GFP_KERNEL ); +const char *parents[5 ];void *reg;int i =0 ;Signature · 1 line
1.8.2
v2 cover clk: sunxi: gates support
In reply to Emilio LópezQuoting Emilio L?pez (2013-03-27 14:20:36)
Quoted 12 lines
Hi,
This patchset adds support for the main sunxi gates; namely, those found on AXI, AHB, APB0 and APB1 clocks. The series depends on Maxime's UART rework and on the basic sunxi clock driver by me.
Patches 1 and 2 implement the actual gate support, and patch 3 switches the UARTs to use the correct gates now that they are available.
Patch 4 and 5 add clock support to our pinctrl driver, as we need to keep the pio clock running for gpio to work.
Quoted 8 lines
Patch 6 drops the ignore flags; now that we have actual clock users that will keep the important clocks running.
Patch 7 is an unrelated oneliner to remove an unnecesary kmalloc call.
The main changes from v1 are renamed compatible strings and a clarification on the documentation, see each individual patch for more details.
Hi Emilio,
Which tree(s) did you want these patches to go through?
Regards,
Mike
Quoted 12 lines
Thanks,
Emilio
Emilio L?pez (7):
clk: sunxi: Add support for AXI, AHB, APB0 and APB1 gates
arm: sunxi: Add clock definitions for AXI, AHB, APB0, APB1 gates
arm: sunxi: use the right clock phandles for UARTs
pinctrl: sunxi: add clock support
arm: sunxi: Add clock to pinctrl node
clk: sunxi: dropCLK_IGNORE_UNUSED
clk: sunxi: drop an unnecesary kmalloc
Emilio López
maintainerv2 cover clk: sunxi: gates support
Hi Mike,
El 03/04/13 18:45, Mike Turquette escribi?:
Quoted 18 lines
Quoting Emilio L?pez (2013-03-27 14:20:36)
Quoted 17 lines
Hi,
This patchset adds support for the main sunxi gates; namely, those found on AXI, AHB, APB0 and APB1 clocks. The series depends on Maxime's UART rework and on the basic sunxi clock driver by me.
Patches 1 and 2 implement the actual gate support, and patch 3 switches the UARTs to use the correct gates now that they are available.
Patch 4 and 5 add clock support to our pinctrl driver, as we need to keep the pio clock running for gpio to work.
Patch 6 drops the ignore flags; now that we have actual clock users that will keep the important clocks running.
Patch 7 is an unrelated oneliner to remove an unnecesary kmalloc call.
Quoted 7 lines
Quoted 3 lines
The main changes from v1 are renamed compatible strings and a clarification on the documentation, see each individual patch for more details.
Hi Emilio,
Which tree(s) did you want these patches to go through?
The rest of the clk driver is on your tree, so I think 1, 6 and 7 should go through there too, unless Maxime has any objection.
Thanks,
Emilio
v2 cover clk: sunxi: gates support
In reply to Mike TurquetteHi Mike
Le 03/04/2013 23:45, Mike Turquette a ?crit :
Quoted 18 lines
Quoting Emilio L?pez (2013-03-27 14:20:36)
Quoted 17 lines
Hi,
This patchset adds support for the main sunxi gates; namely, those found on AXI, AHB, APB0 and APB1 clocks. The series depends on Maxime's UART rework and on the basic sunxi clock driver by me.
Patches 1 and 2 implement the actual gate support, and patch 3 switches the UARTs to use the correct gates now that they are available.
Patch 4 and 5 add clock support to our pinctrl driver, as we need to keep the pio clock running for gpio to work.
Patch 6 drops the ignore flags; now that we have actual clock users that will keep the important clocks running.
Patch 7 is an unrelated oneliner to remove an unnecesary kmalloc call.
Quoted 7 lines
Quoted 3 lines
The main changes from v1 are renamed compatible strings and a clarification on the documentation, see each individual patch for more details.
Hi Emilio,
Which tree(s) did you want these patches to go through?
1, 6 and 7 are changes on the clock driver in itself that depends on the patches you have in your tree, so I guess you should take them through your tree.
2,3 and 5 are dt patches, so I'll take them, and patch 4 has already been applied by LinusW.
Thanks,
Maxime
Signature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
Maxime Ripard
Appliedv2 cover clk: sunxi: gates support
In reply to Emilio LópezHi,
Le 27/03/2013 22:20, Emilio L?pez a ?crit :
Quoted 12 lines
Hi,
This patchset adds support for the main sunxi gates; namely, those found on AXI, AHB, APB0 and APB1 clocks. The series depends on Maxime's UART rework and on the basic sunxi clock driver by me.
Patches 1 and 2 implement the actual gate support, and patch 3 switches the UARTs to use the correct gates now that they are available.
Patch 4 and 5 add clock support to our pinctrl driver, as we need to keep the pio clock running for gpio to work.
Quoted 7 lines
Patch 6 drops the ignore flags; now that we have actual clock users that will keep the important clocks running.
Patch 7 is an unrelated oneliner to remove an unnecesary kmalloc call.
The main changes from v1 are renamed compatible strings and a clarification on the documentation, see each individual patch for more details.
I merged the patches 2, 3 and 5 into my sunxi/dt-for-3.10 branch.
Thanks,
Maxime
Signature · 3 lines
Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com