[PATCH v7 0/5] arm: Base support for Renesas RZN1D-DB Board

[PATCH v7 0/5] arm: Base support for Renesas RZN1D-DB Board

[Michel Pollet]

This series adds the plain basic support for booting a bare
kernel on the RZ/N1D-DB Board. It's been trimmed to the strict
minimum as a 'base', further patches that could add the
rest of the support.

Note on the clock driver: Current usage of the clocks on Linux
involves Linux 'claiming' all of them, disabling the one it doesn't
need and so on.
On *this* architecture it can't be done, there is at least one other
OS running on the CM3 core that claims it's own clock; Linux can claim
some others but definitely not start disabling stuff it isn't supposed to.

Thanks for the comments on the previous versions!

v7:
 + Removed mention of 'rz[/]n1' from everywhere.
 + Removed unwanted documentation.
 + Renamed clock node back to sysctrl.
 + Renamed rzn1-clocks.h to r9a06g032-sysctrl.h to match
 + Made the clock driver claim the sysctrl node.
 + Fixed a couple of 'sparse' warning in the clock driver.
v6:
 + Fix for suggestion by Geert Uytterhoeven
 + Removed "renesas,rzn1" from the board bindings
 + Removed patches already merged.
 + Removed reboot driver
 + Added a whole clock infrastructure.
 + Rebased on next-20180517
v5:
 + Given the problems I have with getting in some structure around the
   sysctrl block, I've removed the MFD, I've now attached the reboot
   driver on it's  own pair of registers.
 + Rebased on next-20180417
v4:
 + Fixes for suggestions by Simon Horman
 + Fixes for suggestions by Jacopo Mondi
 + Fixes for suggestions by Geert Uytterhoeven
 + Renamed the r9a06g0xx.dtsi file, given up on trying to get a family
   common file in, so dropped potential RZ/N1S support and now only
   focus on RZ/N1D for this patchset.
 + Added 'always-on' to the architected timer node, because it is.
 + Added ARCH_R9A06G032, to match others patterns like RCAR
 + Sorted the .dts files, added empty lines as required.
 + Fixed patch prefixes to match git-log for bindings&dts
 + Merged board .dts & Makefile changes together
 + Rebased on next-20180410
v3:
 + Fixes for suggestions by Geert Uytterhoeven
 + Removed SoC Specific renesas,r9a06g032-xxx, as it's not needed for now.
 + Kept renesas,rzn1 as a family/generic for this family.
 + Fixed a couple of the commit messages.
 + Added Geert's Reviewed-By where appropriate.
v2:
 + Fixes for suggestions by Simon Horman
 + Fixes for suggestions by Rob Herring
 + Fixes for suggestions by Geert Uytterhoeven
 + Removed the mach file
 + Added a MFD base for the sysctrl block
 + Added a regmap based sub driver for the reboot handler
 + Renamed the files to match shmobile conventions
 + Adapted the compatible= strings to reflect 'family' vs 'part'
   distinction.
 + Removed the sysctrl.h file entirelly.
 + Fixed every warnings from the DTC compiler on W=12 mode.
 + Split the device-tree patches from the code.

Michel Pollet (5):
  dt-bindings: Add the r9a06g032-sysctrl.h file
  dt-bindings: clock: renesas,r9a06g032-sysctrl: documentation
  ARM: dts: Renesas R9A06G032 base device tree file
  ARM: dts: Renesas RZN1D-DB Board base file
  clk: renesas: Renesas R9A06G032 clock driver

 .../bindings/clock/renesas,r9a06g032-sysctrl.txt   |  32 +
 arch/arm/boot/dts/Makefile                         |   1 +
 arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts        |  29 +
 arch/arm/boot/dts/r9a06g032.dtsi                   |  86 +++
 drivers/clk/renesas/Kconfig                        |   6 +
 drivers/clk/renesas/Makefile                       |   1 +
 drivers/clk/renesas/r9a06g032-clocks.c             | 812 +++++++++++++++++++++
 include/dt-bindings/clock/r9a06g032-sysctrl.h      | 187 +++++
 8 files changed, 1154 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt
 create mode 100644 arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts
 create mode 100644 arch/arm/boot/dts/r9a06g032.dtsi
 create mode 100644 drivers/clk/renesas/r9a06g032-clocks.c
 create mode 100644 include/dt-bindings/clock/r9a06g032-sysctrl.h

-- 
2.7.4

[PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Michel Pollet]

This adds the constants necessary to use the renesas,r9a06g032-sysctrl node.

Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
Reviewed-by: Rob Herring <robh@kernel.org>
---
 include/dt-bindings/clock/r9a06g032-sysctrl.h | 187 ++++++++++++++++++++++++++
 1 file changed, 187 insertions(+)
 create mode 100644 include/dt-bindings/clock/r9a06g032-sysctrl.h

diff --git a/include/dt-bindings/clock/r9a06g032-sysctrl.h b/include/dt-bindings/clock/r9a06g032-sysctrl.h
new file mode 100644
index 0000000..e3fe69e
--- /dev/null
+++ b/include/dt-bindings/clock/r9a06g032-sysctrl.h
@@ -0,0 +1,187 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * R9A06G032 sysctrl IDs
+ *
+ * Copyright (C) 2018 Renesas Electronics Europe Limited
+ *
+ * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
+ * Derived from zx-reboot.c
+ */
+
+#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
+#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
+
+#define R9A06G032_CLKOUT			0
+#define R9A06G032_CLK_PLL_USB			1
+#define R9A06G032_CLK_48			1 /* AKA CLK_PLL_USB */
+#define R9A06G032_CLKOUT_D10			2
+#define R9A06G032_CLKOUT_D16			3
+#define R9A06G032_MSEBIS_CLK			3 /* AKA CLKOUT_D16 */
+#define R9A06G032_MSEBIM_CLK			3 /* AKA CLKOUT_D16 */
+#define R9A06G032_CLKOUT_D160			4
+#define R9A06G032_CLKOUT_D1OR2			5
+#define R9A06G032_CLK_DDRPHY_PLLCLK		5 /* AKA CLKOUT_D1OR2 */
+#define R9A06G032_CLKOUT_D20			6
+#define R9A06G032_CLK50				6 /* AKA CLKOUT_D20 */
+#define R9A06G032_CLKOUT_D40			7
+#define R9A06G032_CLK25				7 /* AKA CLKOUT_D40 */
+#define R9A06G032_CLKOUT_D5			8
+#define R9A06G032_CLKOUT_D8			9
+#define R9A06G032_CLK125			9 /* AKA CLKOUT_D8 */
+#define R9A06G032_DIV_ADC			10
+#define R9A06G032_DIV_I2C			11
+#define R9A06G032_DIV_NAND			12
+#define R9A06G032_DIV_P1_PG			13
+#define R9A06G032_DIV_P2_PG			14
+#define R9A06G032_DIV_P3_PG			15
+#define R9A06G032_DIV_P4_PG			16
+#define R9A06G032_DIV_P5_PG			17
+#define R9A06G032_CLK_P5_PG1			17 /* AKA DIV_P5_PG */
+#define R9A06G032_DIV_P6_PG			18
+#define R9A06G032_DIV_QSPI0			19
+#define R9A06G032_DIV_QSPI1			20
+#define R9A06G032_DIV_REF_SYNC			21
+#define R9A06G032_CLK_REF_SYNC			21 /* AKA DIV_REF_SYNC */
+#define R9A06G032_DIV_SDIO0			22
+#define R9A06G032_DIV_SDIO1			23
+#define R9A06G032_DIV_SWITCH			24
+#define R9A06G032_DIV_UART			25
+#define R9A06G032_CLK_25_PG4			26
+#define R9A06G032_CLK_25_PG5			27
+#define R9A06G032_CLK_25_PG6			28
+#define R9A06G032_CLK_25_PG7			29
+#define R9A06G032_CLK_25_PG8			30
+#define R9A06G032_CLK_ADC			31
+#define R9A06G032_CLK_ECAT100			32
+#define R9A06G032_CLK_HSR100			33
+#define R9A06G032_CLK_I2C0			34
+#define R9A06G032_CLK_I2C1			35
+#define R9A06G032_CLK_MII_REF			36
+#define R9A06G032_CLK_NAND			37
+#define R9A06G032_CLK_NOUSBP2_PG6		38
+#define R9A06G032_CLK_P1_PG2			39
+#define R9A06G032_CLK_P1_PG3			40
+#define R9A06G032_CLK_P1_PG4			41
+#define R9A06G032_CLK_P4_PG3			42
+#define R9A06G032_CLK_P4_PG4			43
+#define R9A06G032_CLK_P6_PG1			44
+#define R9A06G032_CLK_P6_PG2			45
+#define R9A06G032_CLK_P6_PG3			46
+#define R9A06G032_CLK_P6_PG4			47
+#define R9A06G032_CLK_PCI_USB			48
+#define R9A06G032_CLK_QSPI0			49
+#define R9A06G032_CLK_QSPI1			50
+#define R9A06G032_CLK_RGMII_REF			51
+#define R9A06G032_CLK_RMII_REF			52
+#define R9A06G032_CLK_SDIO0			53
+#define R9A06G032_CLK_SDIO1			54
+#define R9A06G032_CLK_SERCOS100			55
+#define R9A06G032_CLK_SLCD			56
+#define R9A06G032_CLK_SPI0			57
+#define R9A06G032_CLK_SPI1			58
+#define R9A06G032_CLK_SPI2			59
+#define R9A06G032_CLK_SPI3			60
+#define R9A06G032_CLK_SPI4			61
+#define R9A06G032_CLK_SPI5			62
+#define R9A06G032_CLK_SWITCH			63
+#define R9A06G032_DIV_MOTOR			64
+#define R9A06G032_HCLK_ECAT125			65
+#define R9A06G032_HCLK_PINCONFIG		66
+#define R9A06G032_HCLK_SERCOS			67
+#define R9A06G032_HCLK_SGPIO2			68
+#define R9A06G032_HCLK_SGPIO3			69
+#define R9A06G032_HCLK_SGPIO4			70
+#define R9A06G032_HCLK_TIMER0			71
+#define R9A06G032_HCLK_TIMER1			72
+#define R9A06G032_HCLK_USBF			73
+#define R9A06G032_HCLK_USBH			74
+#define R9A06G032_HCLK_USBPM			75
+#define R9A06G032_CLK_48_PG_F			76
+#define R9A06G032_CLK_48_PG4			77
+#define R9A06G032_CLK_DDRPHY_PLLCLK_D4		78
+#define R9A06G032_CLK_ECAT100_D4		79
+#define R9A06G032_CLK_HSR100_D2			80
+#define R9A06G032_CLK_REF_SYNC_D4		81
+#define R9A06G032_CLK_DDRPHY_PCLK		81 /* AKA CLK_REF_SYNC_D4 */
+#define R9A06G032_CLK_FW			81 /* AKA CLK_REF_SYNC_D4 */
+#define R9A06G032_CLK_CRYPTO			81 /* AKA CLK_REF_SYNC_D4 */
+#define R9A06G032_CLK_REF_SYNC_D8		82
+#define R9A06G032_CLK_SERCOS100_D2		83
+#define R9A06G032_DIV_CA7			84
+#define R9A06G032_CLK_A7MP			84 /* AKA DIV_CA7 */
+#define R9A06G032_HCLK_CAN0			85
+#define R9A06G032_HCLK_CAN1			86
+#define R9A06G032_HCLK_DELTASIGMA		87
+#define R9A06G032_HCLK_PWMPTO			88
+#define R9A06G032_HCLK_RSV			89
+#define R9A06G032_HCLK_SGPIO0			90
+#define R9A06G032_HCLK_SGPIO1			91
+#define R9A06G032_RTOS_MDC			92
+#define R9A06G032_CLK_CM3			93
+#define R9A06G032_CLK_DDRC			94
+#define R9A06G032_CLK_ECAT25			95
+#define R9A06G032_CLK_HSR50			96
+#define R9A06G032_CLK_HW_RTOS			97
+#define R9A06G032_CLK_SERCOS50			98
+#define R9A06G032_HCLK_ADC			99
+#define R9A06G032_HCLK_CM3			100
+#define R9A06G032_HCLK_CRYPTO_EIP150		101
+#define R9A06G032_HCLK_CRYPTO_EIP93		102
+#define R9A06G032_HCLK_DDRC			103
+#define R9A06G032_HCLK_DMA0			104
+#define R9A06G032_HCLK_DMA1			105
+#define R9A06G032_HCLK_GMAC0			106
+#define R9A06G032_HCLK_GMAC1			107
+#define R9A06G032_HCLK_GPIO0			108
+#define R9A06G032_HCLK_GPIO1			109
+#define R9A06G032_HCLK_GPIO2			110
+#define R9A06G032_HCLK_HSR			111
+#define R9A06G032_HCLK_I2C0			112
+#define R9A06G032_HCLK_I2C1			113
+#define R9A06G032_HCLK_LCD			114
+#define R9A06G032_HCLK_MSEBI_M			115
+#define R9A06G032_HCLK_MSEBI_S			116
+#define R9A06G032_HCLK_NAND			117
+#define R9A06G032_HCLK_PG_I			118
+#define R9A06G032_HCLK_PG19			119
+#define R9A06G032_HCLK_PG20			120
+#define R9A06G032_HCLK_PG3			121
+#define R9A06G032_HCLK_PG4			122
+#define R9A06G032_HCLK_QSPI0			123
+#define R9A06G032_HCLK_QSPI1			124
+#define R9A06G032_HCLK_ROM			125
+#define R9A06G032_HCLK_RTC			126
+#define R9A06G032_HCLK_SDIO0			127
+#define R9A06G032_HCLK_SDIO1			128
+#define R9A06G032_HCLK_SEMAP			129
+#define R9A06G032_HCLK_SPI0			130
+#define R9A06G032_HCLK_SPI1			131
+#define R9A06G032_HCLK_SPI2			132
+#define R9A06G032_HCLK_SPI3			133
+#define R9A06G032_HCLK_SPI4			134
+#define R9A06G032_HCLK_SPI5			135
+#define R9A06G032_HCLK_SWITCH			136
+#define R9A06G032_HCLK_SWITCH_RG		137
+#define R9A06G032_HCLK_UART0			138
+#define R9A06G032_HCLK_UART1			139
+#define R9A06G032_HCLK_UART2			140
+#define R9A06G032_HCLK_UART3			141
+#define R9A06G032_HCLK_UART4			142
+#define R9A06G032_HCLK_UART5			143
+#define R9A06G032_HCLK_UART6			144
+#define R9A06G032_HCLK_UART7			145
+#define R9A06G032_CLK_UART0			146
+#define R9A06G032_CLK_UART1			147
+#define R9A06G032_CLK_UART2			148
+#define R9A06G032_CLK_UART3			149
+#define R9A06G032_CLK_UART4			150
+#define R9A06G032_CLK_UART5			151
+#define R9A06G032_CLK_UART6			152
+#define R9A06G032_CLK_UART7			153
+
+#define R9A06G032_UART_GROUP_012		154
+#define R9A06G032_UART_GROUP_34567		155
+
+#define R9A06G032_CLOCK_COUNT		(R9A06G032_UART_GROUP_34567 + 1)
+
+#endif /* __DT_BINDINGS_R9A06G032_SYSCTRL_H__ */
-- 
2.7.4

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
<michel.pollet@bp.renesas.com> wrote:
> This adds the constants necessary to use the renesas,r9a06g032-sysctrl node.
>
> Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

Thanks for your patch!

> --- /dev/null
> +++ b/include/dt-bindings/clock/r9a06g032-sysctrl.h

You can still call this file r9a06g032-clocks.h, if you want, as it
contains clock definitions only.

> @@ -0,0 +1,187 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * R9A06G032 sysctrl IDs
> + *
> + * Copyright (C) 2018 Renesas Electronics Europe Limited
> + *
> + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
> + * Derived from zx-reboot.c
> + */
> +
> +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> +
> +#define R9A06G032_CLKOUT                       0
> +#define R9A06G032_CLK_PLL_USB                  1
> +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
> +#define R9A06G032_CLKOUT_D10                   2
> +#define R9A06G032_CLKOUT_D16                   3
> +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
> +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
> +#define R9A06G032_CLKOUT_D160                  4
> +#define R9A06G032_CLKOUT_D1OR2                 5
> +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 */

[...]

I have 3 comments:

  1. I had expected this list to match (both name- and order-wise) Appendix
     C ("Clock Tree Structure") in the RZ/N1[DSL] User’s Manual: System
     Introduction, Multiplexing, Electrical and Mechanical Information.
     That would make it easier to review.

  2. These definitions seems to be a mix of:
       1. Internal core clocks,
       2. Other core clocks,
       3. Module clocks.

     The internal clocks do not need to be referenced from DT, so I would
     not make them part of the DT bindings.

  3. It looks like the module clocks can be referred to by register offset
     and bit position, which is similar to how this is handled on R-Car
     SoCs.
     Perhaps you can just drop the definitions for these from the header
     file, and refer to them by (combined) register offset and bit position
     instead?
     Or am I missing something?
     I believe this can also be done for the module resets (later).

Thanks again!

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
<michel.pollet@bp.renesas.com> wrote:
> This adds the constants necessary to use the renesas,r9a06g032-sysctrl no=
de.
>
> Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

Thanks for your patch!

> --- /dev/null
> +++ b/include/dt-bindings/clock/r9a06g032-sysctrl.h

You can still call this file r9a06g032-clocks.h, if you want, as it
contains clock definitions only.

> @@ -0,0 +1,187 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * R9A06G032 sysctrl IDs
> + *
> + * Copyright (C) 2018 Renesas Electronics Europe Limited
> + *
> + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
> + * Derived from zx-reboot.c
> + */
> +
> +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> +
> +#define R9A06G032_CLKOUT                       0
> +#define R9A06G032_CLK_PLL_USB                  1
> +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
> +#define R9A06G032_CLKOUT_D10                   2
> +#define R9A06G032_CLKOUT_D16                   3
> +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
> +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
> +#define R9A06G032_CLKOUT_D160                  4
> +#define R9A06G032_CLKOUT_D1OR2                 5
> +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 */

[...]

I have 3 comments:

  1. I had expected this list to match (both name- and order-wise) Appendix
     C ("Clock Tree Structure") in the RZ/N1[DSL] User=E2=80=99s Manual: Sy=
stem
     Introduction, Multiplexing, Electrical and Mechanical Information.
     That would make it easier to review.

  2. These definitions seems to be a mix of:
       1. Internal core clocks,
       2. Other core clocks,
       3. Module clocks.

     The internal clocks do not need to be referenced from DT, so I would
     not make them part of the DT bindings.

  3. It looks like the module clocks can be referred to by register offset
     and bit position, which is similar to how this is handled on R-Car
     SoCs.
     Perhaps you can just drop the definitions for these from the header
     file, and refer to them by (combined) register offset and bit position
     instead?
     Or am I missing something?
     I believe this can also be done for the module resets (later).

Thanks again!

Gr{oetje,eeting}s,

                        Geert

--=20
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k=
.org

In personal conversations with technical people, I call myself a hacker. Bu=
t
when I'm talking to journalists I just say "programmer" or something like t=
hat.
                                -- Linus Torvalds

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[M P]

On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
wrote:

> Hi Michel,

Hi Geert,


> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
> <michel.pollet@bp.renesas.com> wrote:
> > This adds the constants necessary to use the renesas,r9a06g032-sysctrl
node.
> >
> > Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

> Thanks for your patch!

> > --- /dev/null
> > +++ b/include/dt-bindings/clock/r9a06g032-sysctrl.h

> You can still call this file r9a06g032-clocks.h, if you want, as it
> contains clock definitions only.

Well, having renamed the node, I thought it made sense to keep the naming
consistent. Any further #define could be added to here instead of having to
add another file...


> > @@ -0,0 +1,187 @@
> > +/* SPDX-License-Identifier: GPL-2.0 */
> > +/*
> > + * R9A06G032 sysctrl IDs
> > + *
> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
> > + *
> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
> > + * Derived from zx-reboot.c
> > + */
> > +
> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> > +
> > +#define R9A06G032_CLKOUT                       0
> > +#define R9A06G032_CLK_PLL_USB                  1
> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
> > +#define R9A06G032_CLKOUT_D10                   2
> > +#define R9A06G032_CLKOUT_D16                   3
> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
> > +#define R9A06G032_CLKOUT_D160                  4
> > +#define R9A06G032_CLKOUT_D1OR2                 5
> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 */

> [...]

> I have 3 comments:

>    1. I had expected this list to match (both name- and order-wise)
Appendix
>       C ("Clock Tree Structure") in the RZ/N1[DSL] User’s Manual: System
>       Introduction, Multiplexing, Electrical and Mechanical Information.
>       That would make it easier to review.

Well, that document was made a *long* time after the internal documentation
we used to generate the clock lists. There are a few things we had to do:

* Renumber peripherals. We decided a long time ago that u-boot and linux
would stick to zero based peripherals, and not one based numbers. It's next
to impossible to keep mixing number based across software base, so we use
UART0 while the hardware manual mentions UART1 -- It *is* documented
extensively with out SDK, and makes life using linux a lot easier. It's
across all our SDK, u-boot, webapps readmes, howtos etc.

* Rename some peripherals. Plenty of peripherals names made little sense
and had zero consistency, in fact, many names were different depending on
the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
"QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match linux
conventions.

* Other internal reasons I can't document here

Also, the value here are made up anyway -- so I've decided to sort them to
make sure any clock that has a parent is numbered *after* the parent...

>    2. These definitions seems to be a mix of:
>         1. Internal core clocks,
>         2. Other core clocks,
>         3. Module clocks.

>       The internal clocks do not need to be referenced from DT, so I would
>       not make them part of the DT bindings.

Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
'something' needing them? Why would I decide NOT to include them,
as they are there? I 'factored' a few of them to the same number when
applicable.

This is all done automatically BTW -- a script takes the original clocking
spreadsheet, and converts it into a table, correcting 'human input' as it
goes along.

>    3. It looks like the module clocks can be referred to by register offset
>       and bit position, which is similar to how this is handled on R-Car
>       SoCs.
>       Perhaps you can just drop the definitions for these from the header
>       file, and refer to them by (combined) register offset and bit
position
>       instead?
>       Or am I missing something?
>       I believe this can also be done for the module resets (later).

If you look in the .c file, you'll see that most gate have not just one
register/bit pair associated with them -- there are several, spread
across registers.. Also, there are clocks in there with two gates,
or none. Given that, I've decided a separate numbering
made sense.
As mentioned, it's arbitrary, with 'root' clocks numbered lowered than
sub-clocks.

Thank you!

Michel

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[M P]

On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
wrote:

> Hi Michel,

Hi Geert,


> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
> <michel.pollet@bp.renesas.com> wrote:
> > This adds the constants necessary to use the renesas,r9a06g032-sysctrl
node.
> >
> > Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

> Thanks for your patch!

> > --- /dev/null
> > +++ b/include/dt-bindings/clock/r9a06g032-sysctrl.h

> You can still call this file r9a06g032-clocks.h, if you want, as it
> contains clock definitions only.

Well, having renamed the node, I thought it made sense to keep the naming
consistent. Any further #define could be added to here instead of having to
add another file...


> > @@ -0,0 +1,187 @@
> > +/* SPDX-License-Identifier: GPL-2.0 */
> > +/*
> > + * R9A06G032 sysctrl IDs
> > + *
> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
> > + *
> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
> > + * Derived from zx-reboot.c
> > + */
> > +
> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> > +
> > +#define R9A06G032_CLKOUT                       0
> > +#define R9A06G032_CLK_PLL_USB                  1
> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
> > +#define R9A06G032_CLKOUT_D10                   2
> > +#define R9A06G032_CLKOUT_D16                   3
> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
> > +#define R9A06G032_CLKOUT_D160                  4
> > +#define R9A06G032_CLKOUT_D1OR2                 5
> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 *=
/

> [...]

> I have 3 comments:

>    1. I had expected this list to match (both name- and order-wise)
Appendix
>       C ("Clock Tree Structure") in the RZ/N1[DSL] User=E2=80=99s Manual:=
 System
>       Introduction, Multiplexing, Electrical and Mechanical Information.
>       That would make it easier to review.

Well, that document was made a *long* time after the internal documentation
we used to generate the clock lists. There are a few things we had to do:

* Renumber peripherals. We decided a long time ago that u-boot and linux
would stick to zero based peripherals, and not one based numbers. It's next
to impossible to keep mixing number based across software base, so we use
UART0 while the hardware manual mentions UART1 -- It *is* documented
extensively with out SDK, and makes life using linux a lot easier. It's
across all our SDK, u-boot, webapps readmes, howtos etc.

* Rename some peripherals. Plenty of peripherals names made little sense
and had zero consistency, in fact, many names were different depending on
the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
"QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match linux
conventions.

* Other internal reasons I can't document here

Also, the value here are made up anyway -- so I've decided to sort them to
make sure any clock that has a parent is numbered *after* the parent...

>    2. These definitions seems to be a mix of:
>         1. Internal core clocks,
>         2. Other core clocks,
>         3. Module clocks.

>       The internal clocks do not need to be referenced from DT, so I woul=
d
>       not make them part of the DT bindings.

Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
'something' needing them? Why would I decide NOT to include them,
as they are there? I 'factored' a few of them to the same number when
applicable.

This is all done automatically BTW -- a script takes the original clocking
spreadsheet, and converts it into a table, correcting 'human input' as it
goes along.

>    3. It looks like the module clocks can be referred to by register offs=
et
>       and bit position, which is similar to how this is handled on R-Car
>       SoCs.
>       Perhaps you can just drop the definitions for these from the header
>       file, and refer to them by (combined) register offset and bit
position
>       instead?
>       Or am I missing something?
>       I believe this can also be done for the module resets (later).

If you look in the .c file, you'll see that most gate have not just one
register/bit pair associated with them -- there are several, spread
across registers.. Also, there are clocks in there with two gates,
or none. Given that, I've decided a separate numbering
made sense.
As mentioned, it's arbitrary, with 'root' clocks numbered lowered than
sub-clocks.

Thank you!

Michel

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 31, 2018 at 11:11 AM, M P <buserror@gmail.com> wrote:
> On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
> wrote:
>> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
>> <michel.pollet@bp.renesas.com> wrote:
>> > This adds the constants necessary to use the renesas,r9a06g032-sysctrl
> node.

>> > @@ -0,0 +1,187 @@
>> > +/* SPDX-License-Identifier: GPL-2.0 */
>> > +/*
>> > + * R9A06G032 sysctrl IDs
>> > + *
>> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
>> > + *
>> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
>> > + * Derived from zx-reboot.c
>> > + */
>> > +
>> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> > +
>> > +#define R9A06G032_CLKOUT                       0
>> > +#define R9A06G032_CLK_PLL_USB                  1
>> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
>> > +#define R9A06G032_CLKOUT_D10                   2
>> > +#define R9A06G032_CLKOUT_D16                   3
>> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
>> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
>> > +#define R9A06G032_CLKOUT_D160                  4
>> > +#define R9A06G032_CLKOUT_D1OR2                 5
>> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 */
>
>> [...]
>
>> I have 3 comments:
>
>>    1. I had expected this list to match (both name- and order-wise)
> Appendix
>>       C ("Clock Tree Structure") in the RZ/N1[DSL] User’s Manual: System
>>       Introduction, Multiplexing, Electrical and Mechanical Information.
>>       That would make it easier to review.
>
> Well, that document was made a *long* time after the internal documentation
> we used to generate the clock lists. There are a few things we had to do:
>
> * Renumber peripherals. We decided a long time ago that u-boot and linux
> would stick to zero based peripherals, and not one based numbers. It's next
> to impossible to keep mixing number based across software base, so we use
> UART0 while the hardware manual mentions UART1 -- It *is* documented
> extensively with out SDK, and makes life using linux a lot easier. It's
> across all our SDK, u-boot, webapps readmes, howtos etc.
>
> * Rename some peripherals. Plenty of peripherals names made little sense
> and had zero consistency, in fact, many names were different depending on
> the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
> "QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match linux
> conventions.
>
> * Other internal reasons I can't document here
>
> Also, the value here are made up anyway -- so I've decided to sort them to
> make sure any clock that has a parent is numbered *after* the parent...

Well, all of that combines makes it very hard for us to review the list.

>>    2. These definitions seems to be a mix of:
>>         1. Internal core clocks,
>>         2. Other core clocks,
>>         3. Module clocks.
>
>>       The internal clocks do not need to be referenced from DT, so I would
>>       not make them part of the DT bindings.
>
> Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
> 'something' needing them? Why would I decide NOT to include them,
> as they are there? I 'factored' a few of them to the same number when

Just general safety w.r.t. unchangeable DT definitions: anything that is
not listed here cannot be declared wrong later.

> applicable.

You're 100% sure they're the same?

> This is all done automatically BTW -- a script takes the original clocking
> spreadsheet, and converts it into a table, correcting 'human input' as it
> goes along.

So the internal spreadsheet doesn't match the public documentation neither?

>>    3. It looks like the module clocks can be referred to by register offset
>>       and bit position, which is similar to how this is handled on R-Car
>>       SoCs.
>>       Perhaps you can just drop the definitions for these from the header
>>       file, and refer to them by (combined) register offset and bit
> position
>>       instead?
>>       Or am I missing something?
>>       I believe this can also be done for the module resets (later).
>
> If you look in the .c file, you'll see that most gate have not just one
> register/bit pair associated with them -- there are several, spread
> across registers.. Also, there are clocks in there with two gates,
> or none. Given that, I've decided a separate numbering
> made sense.

OK, fair enough.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 31, 2018 at 11:11 AM, M P <buserror@gmail.com> wrote:
> On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
> wrote:
>> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
>> <michel.pollet@bp.renesas.com> wrote:
>> > This adds the constants necessary to use the renesas,r9a06g032-sysctrl
> node.

>> > @@ -0,0 +1,187 @@
>> > +/* SPDX-License-Identifier: GPL-2.0 */
>> > +/*
>> > + * R9A06G032 sysctrl IDs
>> > + *
>> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
>> > + *
>> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
>> > + * Derived from zx-reboot.c
>> > + */
>> > +
>> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> > +
>> > +#define R9A06G032_CLKOUT                       0
>> > +#define R9A06G032_CLK_PLL_USB                  1
>> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB *=
/
>> > +#define R9A06G032_CLKOUT_D10                   2
>> > +#define R9A06G032_CLKOUT_D16                   3
>> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
>> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
>> > +#define R9A06G032_CLKOUT_D160                  4
>> > +#define R9A06G032_CLKOUT_D1OR2                 5
>> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 =
*/
>
>> [...]
>
>> I have 3 comments:
>
>>    1. I had expected this list to match (both name- and order-wise)
> Appendix
>>       C ("Clock Tree Structure") in the RZ/N1[DSL] User=E2=80=99s Manual=
: System
>>       Introduction, Multiplexing, Electrical and Mechanical Information.
>>       That would make it easier to review.
>
> Well, that document was made a *long* time after the internal documentati=
on
> we used to generate the clock lists. There are a few things we had to do:
>
> * Renumber peripherals. We decided a long time ago that u-boot and linux
> would stick to zero based peripherals, and not one based numbers. It's ne=
xt
> to impossible to keep mixing number based across software base, so we use
> UART0 while the hardware manual mentions UART1 -- It *is* documented
> extensively with out SDK, and makes life using linux a lot easier. It's
> across all our SDK, u-boot, webapps readmes, howtos etc.
>
> * Rename some peripherals. Plenty of peripherals names made little sense
> and had zero consistency, in fact, many names were different depending on
> the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
> "QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match linu=
x
> conventions.
>
> * Other internal reasons I can't document here
>
> Also, the value here are made up anyway -- so I've decided to sort them t=
o
> make sure any clock that has a parent is numbered *after* the parent...

Well, all of that combines makes it very hard for us to review the list.

>>    2. These definitions seems to be a mix of:
>>         1. Internal core clocks,
>>         2. Other core clocks,
>>         3. Module clocks.
>
>>       The internal clocks do not need to be referenced from DT, so I wou=
ld
>>       not make them part of the DT bindings.
>
> Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
> 'something' needing them? Why would I decide NOT to include them,
> as they are there? I 'factored' a few of them to the same number when

Just general safety w.r.t. unchangeable DT definitions: anything that is
not listed here cannot be declared wrong later.

> applicable.

You're 100% sure they're the same?

> This is all done automatically BTW -- a script takes the original clockin=
g
> spreadsheet, and converts it into a table, correcting 'human input' as it
> goes along.

So the internal spreadsheet doesn't match the public documentation neither?

>>    3. It looks like the module clocks can be referred to by register off=
set
>>       and bit position, which is similar to how this is handled on R-Car
>>       SoCs.
>>       Perhaps you can just drop the definitions for these from the heade=
r
>>       file, and refer to them by (combined) register offset and bit
> position
>>       instead?
>>       Or am I missing something?
>>       I believe this can also be done for the module resets (later).
>
> If you look in the .c file, you'll see that most gate have not just one
> register/bit pair associated with them -- there are several, spread
> across registers.. Also, there are clocks in there with two gates,
> or none. Given that, I've decided a separate numbering
> made sense.

OK, fair enough.

Gr{oetje,eeting}s,

                        Geert

--=20
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k=
.org

In personal conversations with technical people, I call myself a hacker. Bu=
t
when I'm talking to journalists I just say "programmer" or something like t=
hat.
                                -- Linus Torvalds

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[M P]

Hi Geert,

On Thu, 31 May 2018 at 10:32, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> Hi Michel,
>
> On Thu, May 31, 2018 at 11:11 AM, M P <buserror@gmail.com> wrote:
> > On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
> > wrote:
> >> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
> >> <michel.pollet@bp.renesas.com> wrote:
> >> > This adds the constants necessary to use the renesas,r9a06g032-sysctrl
> > node.
>
> >> > @@ -0,0 +1,187 @@
> >> > +/* SPDX-License-Identifier: GPL-2.0 */
> >> > +/*
> >> > + * R9A06G032 sysctrl IDs
> >> > + *
> >> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
> >> > + *
> >> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
> >> > + * Derived from zx-reboot.c
> >> > + */
> >> > +
> >> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> >> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> >> > +
> >> > +#define R9A06G032_CLKOUT                       0
> >> > +#define R9A06G032_CLK_PLL_USB                  1
> >> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
> >> > +#define R9A06G032_CLKOUT_D10                   2
> >> > +#define R9A06G032_CLKOUT_D16                   3
> >> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
> >> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
> >> > +#define R9A06G032_CLKOUT_D160                  4
> >> > +#define R9A06G032_CLKOUT_D1OR2                 5
> >> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 */
> >
> >> [...]
> >
> >> I have 3 comments:
> >
> >>    1. I had expected this list to match (both name- and order-wise)
> > Appendix
> >>       C ("Clock Tree Structure") in the RZ/N1[DSL] User’s Manual: System
> >>       Introduction, Multiplexing, Electrical and Mechanical Information.
> >>       That would make it easier to review.
> >
> > Well, that document was made a *long* time after the internal documentation
> > we used to generate the clock lists. There are a few things we had to do:
> >
> > * Renumber peripherals. We decided a long time ago that u-boot and linux
> > would stick to zero based peripherals, and not one based numbers. It's next
> > to impossible to keep mixing number based across software base, so we use
> > UART0 while the hardware manual mentions UART1 -- It *is* documented
> > extensively with out SDK, and makes life using linux a lot easier. It's
> > across all our SDK, u-boot, webapps readmes, howtos etc.
> >
> > * Rename some peripherals. Plenty of peripherals names made little sense
> > and had zero consistency, in fact, many names were different depending on
> > the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
> > "QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match linux
> > conventions.
> >
> > * Other internal reasons I can't document here
> >
> > Also, the value here are made up anyway -- so I've decided to sort them to
> > make sure any clock that has a parent is numbered *after* the parent...
>
> Well, all of that combines makes it very hard for us to review the list.

I understand -- the previous format was a lot more readable, here I had to
work to make the table as small as I could, and quite a few bits of readability
were lost in the process. It's already a huge file.

I'm not sure if that would help, but here is the link to the old table
format I used
https://github.com/renesas-rz/rzn1_linux/blob/rzn1-stable/drivers/clk/rzn1/rzn1-clkctrl-tables.h

I had to throw that away to make up the new composite table that now contains
what was in the device tree file before. However the names are the same.

Perhaps I could change how I encode the register/bit pairs to use 8+8 bits to
make the hex constants more readable?

>
> >>    2. These definitions seems to be a mix of:
> >>         1. Internal core clocks,
> >>         2. Other core clocks,
> >>         3. Module clocks.
> >
> >>       The internal clocks do not need to be referenced from DT, so I would
> >>       not make them part of the DT bindings.
> >
> > Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
> > 'something' needing them? Why would I decide NOT to include them,
> > as they are there? I 'factored' a few of them to the same number when
>
> Just general safety w.r.t. unchangeable DT definitions: anything that is
> not listed here cannot be declared wrong later.

Well, I need these #define *somewhere* as I use them in my driver. So what
do you want me to do?
+ Remove the header, move the constants into the driver?
+ Use hard coded numbers in the DT and remove the header entirely?
+ Duplicate the header, keep the full one with the driver, and only list
the CA7+UART0 one in the dt-binding and duplicate the ones I need as
I go along?

>
> > applicable.
>
> You're 100% sure they're the same?

Yes, the script logic hasn't changed in 2 years, and we've been using that
hierarchy extensively since.
Basically the logic is that if a clock doesn't have a gate and isn't a divisor
of some sort, it's factored with it's parent clock... it used to be
done with a DT
alias of the same name.

>
> > This is all done automatically BTW -- a script takes the original clocking
> > spreadsheet, and converts it into a table, correcting 'human input' as it
> > goes along.
>
> So the internal spreadsheet doesn't match the public documentation neither?

Nope, as usual, people who wrote the documentation went their own way at
some point, and didn't backport any changes they made.

> Gr{oetje,eeting}s,
>
>                         Geert

Cheers,
Michel

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[M P]

Hi Geert,

On Thu, 31 May 2018 at 10:32, Geert Uytterhoeven <geert@linux-m68k.org> wro=
te:
>
> Hi Michel,
>
> On Thu, May 31, 2018 at 11:11 AM, M P <buserror@gmail.com> wrote:
> > On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
> > wrote:
> >> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
> >> <michel.pollet@bp.renesas.com> wrote:
> >> > This adds the constants necessary to use the renesas,r9a06g032-sysct=
rl
> > node.
>
> >> > @@ -0,0 +1,187 @@
> >> > +/* SPDX-License-Identifier: GPL-2.0 */
> >> > +/*
> >> > + * R9A06G032 sysctrl IDs
> >> > + *
> >> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
> >> > + *
> >> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.co=
m>
> >> > + * Derived from zx-reboot.c
> >> > + */
> >> > +
> >> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> >> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
> >> > +
> >> > +#define R9A06G032_CLKOUT                       0
> >> > +#define R9A06G032_CLK_PLL_USB                  1
> >> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB=
 */
> >> > +#define R9A06G032_CLKOUT_D10                   2
> >> > +#define R9A06G032_CLKOUT_D16                   3
> >> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 =
*/
> >> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 =
*/
> >> > +#define R9A06G032_CLKOUT_D160                  4
> >> > +#define R9A06G032_CLKOUT_D1OR2                 5
> >> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR=
2 */
> >
> >> [...]
> >
> >> I have 3 comments:
> >
> >>    1. I had expected this list to match (both name- and order-wise)
> > Appendix
> >>       C ("Clock Tree Structure") in the RZ/N1[DSL] User=E2=80=99s Manu=
al: System
> >>       Introduction, Multiplexing, Electrical and Mechanical Informatio=
n.
> >>       That would make it easier to review.
> >
> > Well, that document was made a *long* time after the internal documenta=
tion
> > we used to generate the clock lists. There are a few things we had to d=
o:
> >
> > * Renumber peripherals. We decided a long time ago that u-boot and linu=
x
> > would stick to zero based peripherals, and not one based numbers. It's =
next
> > to impossible to keep mixing number based across software base, so we u=
se
> > UART0 while the hardware manual mentions UART1 -- It *is* documented
> > extensively with out SDK, and makes life using linux a lot easier. It's
> > across all our SDK, u-boot, webapps readmes, howtos etc.
> >
> > * Rename some peripherals. Plenty of peripherals names made little sens=
e
> > and had zero consistency, in fact, many names were different depending =
on
> > the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
> > "QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match li=
nux
> > conventions.
> >
> > * Other internal reasons I can't document here
> >
> > Also, the value here are made up anyway -- so I've decided to sort them=
 to
> > make sure any clock that has a parent is numbered *after* the parent...
>
> Well, all of that combines makes it very hard for us to review the list.

I understand -- the previous format was a lot more readable, here I had to
work to make the table as small as I could, and quite a few bits of readabi=
lity
were lost in the process. It's already a huge file.

I'm not sure if that would help, but here is the link to the old table
format I used
https://github.com/renesas-rz/rzn1_linux/blob/rzn1-stable/drivers/clk/rzn1/=
rzn1-clkctrl-tables.h

I had to throw that away to make up the new composite table that now contai=
ns
what was in the device tree file before. However the names are the same.

Perhaps I could change how I encode the register/bit pairs to use 8+8 bits =
to
make the hex constants more readable?

>
> >>    2. These definitions seems to be a mix of:
> >>         1. Internal core clocks,
> >>         2. Other core clocks,
> >>         3. Module clocks.
> >
> >>       The internal clocks do not need to be referenced from DT, so I w=
ould
> >>       not make them part of the DT bindings.
> >
> > Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
> > 'something' needing them? Why would I decide NOT to include them,
> > as they are there? I 'factored' a few of them to the same number when
>
> Just general safety w.r.t. unchangeable DT definitions: anything that is
> not listed here cannot be declared wrong later.

Well, I need these #define *somewhere* as I use them in my driver. So what
do you want me to do?
+ Remove the header, move the constants into the driver?
+ Use hard coded numbers in the DT and remove the header entirely?
+ Duplicate the header, keep the full one with the driver, and only list
the CA7+UART0 one in the dt-binding and duplicate the ones I need as
I go along?

>
> > applicable.
>
> You're 100% sure they're the same?

Yes, the script logic hasn't changed in 2 years, and we've been using that
hierarchy extensively since.
Basically the logic is that if a clock doesn't have a gate and isn't a divi=
sor
of some sort, it's factored with it's parent clock... it used to be
done with a DT
alias of the same name.

>
> > This is all done automatically BTW -- a script takes the original clock=
ing
> > spreadsheet, and converts it into a table, correcting 'human input' as =
it
> > goes along.
>
> So the internal spreadsheet doesn't match the public documentation neithe=
r?

Nope, as usual, people who wrote the documentation went their own way at
some point, and didn't backport any changes they made.

> Gr{oetje,eeting}s,
>
>                         Geert

Cheers,
Michel

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 31, 2018 at 12:01 PM, M P <buserror@gmail.com> wrote:
> On Thu, 31 May 2018 at 10:32, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>> On Thu, May 31, 2018 at 11:11 AM, M P <buserror@gmail.com> wrote:
>> > On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org>
>> > wrote:
>> >> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
>> >> <michel.pollet@bp.renesas.com> wrote:
>> >> > This adds the constants necessary to use the renesas,r9a06g032-sysctrl
>> > node.
>>
>> >> > @@ -0,0 +1,187 @@
>> >> > +/* SPDX-License-Identifier: GPL-2.0 */
>> >> > +/*
>> >> > + * R9A06G032 sysctrl IDs
>> >> > + *
>> >> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
>> >> > + *
>> >> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
>> >> > + * Derived from zx-reboot.c
>> >> > + */
>> >> > +
>> >> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> >> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> >> > +
>> >> > +#define R9A06G032_CLKOUT                       0
>> >> > +#define R9A06G032_CLK_PLL_USB                  1
>> >> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_USB */
>> >> > +#define R9A06G032_CLKOUT_D10                   2
>> >> > +#define R9A06G032_CLKOUT_D16                   3
>> >> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16 */
>> >> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16 */
>> >> > +#define R9A06G032_CLKOUT_D160                  4
>> >> > +#define R9A06G032_CLKOUT_D1OR2                 5
>> >> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1OR2 */
>> >
>> >> [...]
>> >
>> >> I have 3 comments:
>> >
>> >>    1. I had expected this list to match (both name- and order-wise)
>> > Appendix
>> >>       C ("Clock Tree Structure") in the RZ/N1[DSL] User’s Manual: System
>> >>       Introduction, Multiplexing, Electrical and Mechanical Information.
>> >>       That would make it easier to review.
>> >
>> > Well, that document was made a *long* time after the internal documentation
>> > we used to generate the clock lists. There are a few things we had to do:
>> >
>> > * Renumber peripherals. We decided a long time ago that u-boot and linux
>> > would stick to zero based peripherals, and not one based numbers. It's next
>> > to impossible to keep mixing number based across software base, so we use
>> > UART0 while the hardware manual mentions UART1 -- It *is* documented
>> > extensively with out SDK, and makes life using linux a lot easier. It's
>> > across all our SDK, u-boot, webapps readmes, howtos etc.
>> >
>> > * Rename some peripherals. Plenty of peripherals names made little sense
>> > and had zero consistency, in fact, many names were different depending on
>> > the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
>> > "QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match linux
>> > conventions.
>> >
>> > * Other internal reasons I can't document here
>> >
>> > Also, the value here are made up anyway -- so I've decided to sort them to
>> > make sure any clock that has a parent is numbered *after* the parent...
>>
>> Well, all of that combines makes it very hard for us to review the list.
>
> I understand -- the previous format was a lot more readable, here I had to
> work to make the table as small as I could, and quite a few bits of readability
> were lost in the process. It's already a huge file.
>
> I'm not sure if that would help, but here is the link to the old table
> format I used
> https://github.com/renesas-rz/rzn1_linux/blob/rzn1-stable/drivers/clk/rzn1/rzn1-clkctrl-tables.h
>
> I had to throw that away to make up the new composite table that now contains
> what was in the device tree file before. However the names are the same.
>
> Perhaps I could change how I encode the register/bit pairs to use 8+8 bits to
> make the hex constants more readable?

Alternatively, you can use a macro to pack them, cfr. MOD_CLK_PACK()
in renesas-cpg-mssr.

>> >>    2. These definitions seems to be a mix of:
>> >>         1. Internal core clocks,
>> >>         2. Other core clocks,
>> >>         3. Module clocks.
>> >
>> >>       The internal clocks do not need to be referenced from DT, so I would
>> >>       not make them part of the DT bindings.
>> >
>> > Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
>> > 'something' needing them? Why would I decide NOT to include them,
>> > as they are there? I 'factored' a few of them to the same number when
>>
>> Just general safety w.r.t. unchangeable DT definitions: anything that is
>> not listed here cannot be declared wrong later.
>
> Well, I need these #define *somewhere* as I use them in my driver. So what
> do you want me to do?
> + Remove the header, move the constants into the driver?
> + Use hard coded numbers in the DT and remove the header entirely?
> + Duplicate the header, keep the full one with the driver, and only list
> the CA7+UART0 one in the dt-binding and duplicate the ones I need as
> I go along?

None of these. You can just move the #defines that will never be needed
from DT into the driver source file.

Cfr. "Internal Core Clocks" in drivers/clk/renesas/r8a7795-cpg-mssr.c.

>> > applicable.
>>
>> You're 100% sure they're the same?
>
> Yes, the script logic hasn't changed in 2 years, and we've been using that
> hierarchy extensively since.
> Basically the logic is that if a clock doesn't have a gate and isn't a divisor
> of some sort, it's factored with it's parent clock... it used to be
> done with a DT
> alias of the same name.

OK.

>> > This is all done automatically BTW -- a script takes the original clocking
>> > spreadsheet, and converts it into a table, correcting 'human input' as it
>> > goes along.
>>
>> So the internal spreadsheet doesn't match the public documentation neither?
>
> Nope, as usual, people who wrote the documentation went their own way at
> some point, and didn't backport any changes they made.

Well, I guess we'll have to live with that...

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v7 1/5] dt-bindings: Add the r9a06g032-sysctrl.h file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 31, 2018 at 12:01 PM, M P <buserror@gmail.com> wrote:
> On Thu, 31 May 2018 at 10:32, Geert Uytterhoeven <geert@linux-m68k.org> w=
rote:
>> On Thu, May 31, 2018 at 11:11 AM, M P <buserror@gmail.com> wrote:
>> > On Fri, 25 May 2018 at 11:31, Geert Uytterhoeven <geert@linux-m68k.org=
>
>> > wrote:
>> >> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
>> >> <michel.pollet@bp.renesas.com> wrote:
>> >> > This adds the constants necessary to use the renesas,r9a06g032-sysc=
trl
>> > node.
>>
>> >> > @@ -0,0 +1,187 @@
>> >> > +/* SPDX-License-Identifier: GPL-2.0 */
>> >> > +/*
>> >> > + * R9A06G032 sysctrl IDs
>> >> > + *
>> >> > + * Copyright (C) 2018 Renesas Electronics Europe Limited
>> >> > + *
>> >> > + * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.c=
om>
>> >> > + * Derived from zx-reboot.c
>> >> > + */
>> >> > +
>> >> > +#ifndef __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> >> > +#define __DT_BINDINGS_R9A06G032_SYSCTRL_H__
>> >> > +
>> >> > +#define R9A06G032_CLKOUT                       0
>> >> > +#define R9A06G032_CLK_PLL_USB                  1
>> >> > +#define R9A06G032_CLK_48                       1 /* AKA CLK_PLL_US=
B */
>> >> > +#define R9A06G032_CLKOUT_D10                   2
>> >> > +#define R9A06G032_CLKOUT_D16                   3
>> >> > +#define R9A06G032_MSEBIS_CLK                   3 /* AKA CLKOUT_D16=
 */
>> >> > +#define R9A06G032_MSEBIM_CLK                   3 /* AKA CLKOUT_D16=
 */
>> >> > +#define R9A06G032_CLKOUT_D160                  4
>> >> > +#define R9A06G032_CLKOUT_D1OR2                 5
>> >> > +#define R9A06G032_CLK_DDRPHY_PLLCLK            5 /* AKA CLKOUT_D1O=
R2 */
>> >
>> >> [...]
>> >
>> >> I have 3 comments:
>> >
>> >>    1. I had expected this list to match (both name- and order-wise)
>> > Appendix
>> >>       C ("Clock Tree Structure") in the RZ/N1[DSL] User=E2=80=99s Man=
ual: System
>> >>       Introduction, Multiplexing, Electrical and Mechanical Informati=
on.
>> >>       That would make it easier to review.
>> >
>> > Well, that document was made a *long* time after the internal document=
ation
>> > we used to generate the clock lists. There are a few things we had to =
do:
>> >
>> > * Renumber peripherals. We decided a long time ago that u-boot and lin=
ux
>> > would stick to zero based peripherals, and not one based numbers. It's=
 next
>> > to impossible to keep mixing number based across software base, so we =
use
>> > UART0 while the hardware manual mentions UART1 -- It *is* documented
>> > extensively with out SDK, and makes life using linux a lot easier. It'=
s
>> > across all our SDK, u-boot, webapps readmes, howtos etc.
>> >
>> > * Rename some peripherals. Plenty of peripherals names made little sen=
se
>> > and had zero consistency, in fact, many names were different depending=
 on
>> > the place they were used! -- "flashnand"+"nand_flash"+"FNAND" etc,
>> > "QUADSPI"+"QSPI" etc etc so we also re-normalized the names to match l=
inux
>> > conventions.
>> >
>> > * Other internal reasons I can't document here
>> >
>> > Also, the value here are made up anyway -- so I've decided to sort the=
m to
>> > make sure any clock that has a parent is numbered *after* the parent..=
.
>>
>> Well, all of that combines makes it very hard for us to review the list.
>
> I understand -- the previous format was a lot more readable, here I had t=
o
> work to make the table as small as I could, and quite a few bits of reada=
bility
> were lost in the process. It's already a huge file.
>
> I'm not sure if that would help, but here is the link to the old table
> format I used
> https://github.com/renesas-rz/rzn1_linux/blob/rzn1-stable/drivers/clk/rzn=
1/rzn1-clkctrl-tables.h
>
> I had to throw that away to make up the new composite table that now cont=
ains
> what was in the device tree file before. However the names are the same.
>
> Perhaps I could change how I encode the register/bit pairs to use 8+8 bit=
s to
> make the hex constants more readable?

Alternatively, you can use a macro to pack them, cfr. MOD_CLK_PACK()
in renesas-cpg-mssr.

>> >>    2. These definitions seems to be a mix of:
>> >>         1. Internal core clocks,
>> >>         2. Other core clocks,
>> >>         3. Module clocks.
>> >
>> >>       The internal clocks do not need to be referenced from DT, so I =
would
>> >>       not make them part of the DT bindings.
>> >
>> > Why? I'm told that "Bindings aren't for linux" -- why can't I imagine
>> > 'something' needing them? Why would I decide NOT to include them,
>> > as they are there? I 'factored' a few of them to the same number when
>>
>> Just general safety w.r.t. unchangeable DT definitions: anything that is
>> not listed here cannot be declared wrong later.
>
> Well, I need these #define *somewhere* as I use them in my driver. So wha=
t
> do you want me to do?
> + Remove the header, move the constants into the driver?
> + Use hard coded numbers in the DT and remove the header entirely?
> + Duplicate the header, keep the full one with the driver, and only list
> the CA7+UART0 one in the dt-binding and duplicate the ones I need as
> I go along?

None of these. You can just move the #defines that will never be needed
from DT into the driver source file.

Cfr. "Internal Core Clocks" in drivers/clk/renesas/r8a7795-cpg-mssr.c.

>> > applicable.
>>
>> You're 100% sure they're the same?
>
> Yes, the script logic hasn't changed in 2 years, and we've been using tha=
t
> hierarchy extensively since.
> Basically the logic is that if a clock doesn't have a gate and isn't a di=
visor
> of some sort, it's factored with it's parent clock... it used to be
> done with a DT
> alias of the same name.

OK.

>> > This is all done automatically BTW -- a script takes the original cloc=
king
>> > spreadsheet, and converts it into a table, correcting 'human input' as=
 it
>> > goes along.
>>
>> So the internal spreadsheet doesn't match the public documentation neith=
er?
>
> Nope, as usual, people who wrote the documentation went their own way at
> some point, and didn't backport any changes they made.

Well, I guess we'll have to live with that...

Gr{oetje,eeting}s,

                        Geert

--=20
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k=
.org

In personal conversations with technical people, I call myself a hacker. Bu=
t
when I'm talking to journalists I just say "programmer" or something like t=
hat.
                                -- Linus Torvalds

[PATCH v7 2/5] dt-bindings: clock: renesas,r9a06g032-sysctrl: documentation

[Michel Pollet]

The Renesas R9A06G032 SYSCTRL node description.

Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
---
 .../bindings/clock/renesas,r9a06g032-sysctrl.txt   | 32 ++++++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt

diff --git a/Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt b/Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt
new file mode 100644
index 0000000..63eb66cd
--- /dev/null
+++ b/Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt
@@ -0,0 +1,32 @@
+* Renesas R9A06G032 SYSCTRL
+
+Required Properties:
+
+  - compatible: Must be:
+    - "renesas,r9a06g032-sysctrl"
+  - reg: Base address and length of the SYSCTRL IO block.
+  - #clock-cells: Must be 1
+
+Examples
+--------
+
+  - SYSCTRL node:
+
+	sysctrl: sysctrl@4000c000 {
+		compatible = "renesas,r9a06g032-sysctrl";
+		reg = <0x4000c000 0x1000>;
+		#clock-cells = <1>;
+	};
+
+  - Other nodes can use the clocks provided by SYSCTRL as in:
+
+	#include <dt-bindings/clock/r9a06g032-sysctrl.h>
+	uart0: serial@40060000 {
+		compatible = "snps,dw-apb-uart";
+		reg = <0x40060000 0x400>;
+		interrupts = <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>;
+		reg-shift = <2>;
+		reg-io-width = <4>;
+		clocks = <&sysctrl R9A06G032_CLK_UART0>;
+		clock-names = "baudclk";
+	};
-- 
2.7.4

Re: [PATCH v7 2/5] dt-bindings: clock: renesas,r9a06g032-sysctrl: documentation

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
<michel.pollet@bp.renesas.com> wrote:
> The Renesas R9A06G032 SYSCTRL node description.
>
> Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

Thanks for your patch!

> --- /dev/null
> +++ b/Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt
> @@ -0,0 +1,32 @@
> +* Renesas R9A06G032 SYSCTRL
> +
> +Required Properties:
> +
> +  - compatible: Must be:
> +    - "renesas,r9a06g032-sysctrl"
> +  - reg: Base address and length of the SYSCTRL IO block.
> +  - #clock-cells: Must be 1

No clocks/clock-names for the external clock inputs?

"RZ/N1 has 3 clock sources, 1 reference clock inputs for RGMII, and 2
 reference clock outputs for RMII/MII."

Given the documentation explicitly mentions the module clocks are to be
used for power-management, you may want to add #power-domain-cells as well,
and let the driver register clock domain. But that can be added later
(although it will break backwards compatibility with old DTBs).

As PWRCTRL_* registers allow to reset individual modules, #reset-cells is
another thing to add later.  It's good to start thinking early about how to
reference resets, though.
E.g. on other Renesas-SoCs, module resets uses the same numerical
references as module clocks.


> +
> +Examples
> +--------
> +
> +  - SYSCTRL node:
> +
> +       sysctrl: sysctrl@4000c000 {

system-controller@

> +               compatible = "renesas,r9a06g032-sysctrl";
> +               reg = <0x4000c000 0x1000>;
> +               #clock-cells = <1>;
> +       };

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v7 2/5] dt-bindings: clock: renesas,r9a06g032-sysctrl: documentation

[M P]

Hi Geert,

On Fri, 25 May 2018 at 10:23, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> Hi Michel,
>
> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
> <michel.pollet@bp.renesas.com> wrote:
> > The Renesas R9A06G032 SYSCTRL node description.
> >
> > Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
>
> Thanks for your patch!
>
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt
> > @@ -0,0 +1,32 @@
> > +* Renesas R9A06G032 SYSCTRL
> > +
> > +Required Properties:
> > +
> > +  - compatible: Must be:
> > +    - "renesas,r9a06g032-sysctrl"
> > +  - reg: Base address and length of the SYSCTRL IO block.
> > +  - #clock-cells: Must be 1
>
> No clocks/clock-names for the external clock inputs?
>
> "RZ/N1 has 3 clock sources, 1 reference clock inputs for RGMII, and 2
>  reference clock outputs for RMII/MII."
>
> Given the documentation explicitly mentions the module clocks are to be
> used for power-management, you may want to add #power-domain-cells as well,
> and let the driver register clock domain. But that can be added later
> (although it will break backwards compatibility with old DTBs).
>
> As PWRCTRL_* registers allow to reset individual modules, #reset-cells is
> another thing to add later.  It's good to start thinking early about how to
> reference resets, though.
> E.g. on other Renesas-SoCs, module resets uses the same numerical
> references as module clocks.

As you said, could we add all that later, as appropriate? Here I tried
to trim it
down to the the bare minimum -- my previous version of the driver had
separate reset  descriptors, but this one has been all compacted to do just
what it's supposed to do: clocks.

Or, so you want to add another DT index to refer to other reset indexes etc?
ie not use the of_clk_src_onecell_get provider? That COULD work and yes, the
indexes would stay the same, I'd just have to get the reset descriptor from the
clock object. We haven't had a use for individual resets so far.

> Gr{oetje,eeting}s,
>
>                         Geert

Cheers,
Michel

Re: [PATCH v7 2/5] dt-bindings: clock: renesas,r9a06g032-sysctrl: documentation

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 31, 2018 at 12:16 PM, M P <buserror@gmail.com> wrote:
> On Fri, 25 May 2018 at 10:23, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
>> <michel.pollet@bp.renesas.com> wrote:
>> > The Renesas R9A06G032 SYSCTRL node description.
>> >
>> > Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
>>
>> Thanks for your patch!
>>
>> > --- /dev/null
>> > +++ b/Documentation/devicetree/bindings/clock/renesas,r9a06g032-sysctrl.txt
>> > @@ -0,0 +1,32 @@
>> > +* Renesas R9A06G032 SYSCTRL
>> > +
>> > +Required Properties:
>> > +
>> > +  - compatible: Must be:
>> > +    - "renesas,r9a06g032-sysctrl"
>> > +  - reg: Base address and length of the SYSCTRL IO block.
>> > +  - #clock-cells: Must be 1
>>
>> No clocks/clock-names for the external clock inputs?
>>
>> "RZ/N1 has 3 clock sources, 1 reference clock inputs for RGMII, and 2
>>  reference clock outputs for RMII/MII."
>>
>> Given the documentation explicitly mentions the module clocks are to be
>> used for power-management, you may want to add #power-domain-cells as well,
>> and let the driver register clock domain. But that can be added later
>> (although it will break backwards compatibility with old DTBs).
>>
>> As PWRCTRL_* registers allow to reset individual modules, #reset-cells is
>> another thing to add later.  It's good to start thinking early about how to
>> reference resets, though.
>> E.g. on other Renesas-SoCs, module resets uses the same numerical
>> references as module clocks.
>
> As you said, could we add all that later, as appropriate? Here I tried
> to trim it
> down to the the bare minimum -- my previous version of the driver had
> separate reset  descriptors, but this one has been all compacted to do just
> what it's supposed to do: clocks.

Yes, it can be added later.
I just wanted to mention it, so you could already think about it, and to avoid a
possible "but we could have nicely integrated reset and clock support if we
did ..." later.

> Or, so you want to add another DT index to refer to other reset indexes etc?
> ie not use the of_clk_src_onecell_get provider? That COULD work and yes, the
> indexes would stay the same, I'd just have to get the reset descriptor from the
> clock object. We haven't had a use for individual resets so far.

Reset indices come from DT, too, but the reset framework doesn't use an
xlate function itself, but just passes the index to the reset driver.
How you obtain the register and bits to reset the device is competlely up
to to the reset driver.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[PATCH v7 3/5] ARM: dts: Renesas R9A06G032 base device tree file

[Michel Pollet]

This adds the Renesas R9A06G032 bare bone support.

This currently only handles generic parts (gic, architected timer)
and a UART.

Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
---
 arch/arm/boot/dts/r9a06g032.dtsi | 86 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 arch/arm/boot/dts/r9a06g032.dtsi

diff --git a/arch/arm/boot/dts/r9a06g032.dtsi b/arch/arm/boot/dts/r9a06g032.dtsi
new file mode 100644
index 0000000..9534f1b
--- /dev/null
+++ b/arch/arm/boot/dts/r9a06g032.dtsi
@@ -0,0 +1,86 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Base Device Tree Source for the Renesas RZ/N1D (R9A06G032)
+ *
+ * Copyright (C) 2018 Renesas Electronics Europe Limited
+ *
+ */
+
+#include <dt-bindings/interrupt-controller/arm-gic.h>
+#include <dt-bindings/clock/r9a06g032-sysctrl.h>
+
+/ {
+	compatible = "renesas,r9a06g032";
+	#address-cells = <1>;
+	#size-cells = <1>;
+
+	cpus {
+		#address-cells = <1>;
+		#size-cells = <0>;
+
+		cpu@0 {
+			device_type = "cpu";
+			compatible = "arm,cortex-a7";
+			reg = <0>;
+			clocks = <&sysctrl R9A06G032_DIV_CA7>;
+		};
+
+		cpu@1 {
+			device_type = "cpu";
+			compatible = "arm,cortex-a7";
+			reg = <1>;
+			clocks = <&sysctrl R9A06G032_DIV_CA7>;
+		};
+	};
+
+	soc {
+		compatible = "simple-bus";
+		#address-cells = <1>;
+		#size-cells = <1>;
+		interrupt-parent = <&gic>;
+		ranges;
+
+		sysctrl: sysctrl@4000c000 {
+			compatible = "renesas,r9a06g032-sysctrl";
+			reg = <0x4000c000 0x1000>;
+			status = "okay";
+			#clock-cells = <1>;
+		};
+
+		uart0: serial@40060000 {
+			compatible = "snps,dw-apb-uart";
+			reg = <0x40060000 0x400>;
+			interrupts = <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>;
+			reg-shift = <2>;
+			reg-io-width = <4>;
+			clocks = <&sysctrl R9A06G032_CLK_UART0>;
+			clock-names = "baudclk";
+			status = "disabled";
+		};
+
+		gic: gic@44101000 {
+			compatible = "arm,cortex-a7-gic", "arm,gic-400";
+			interrupt-controller;
+			#interrupt-cells = <3>;
+			reg = <0x44101000 0x1000>, /* Distributer */
+			      <0x44102000 0x2000>, /* CPU interface */
+			      <0x44104000 0x2000>, /* Virt interface control */
+			      <0x44106000 0x2000>; /* Virt CPU interface */
+			interrupts =
+				<GIC_PPI 9 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_HIGH)>;
+		};
+	};
+
+	timer {
+		compatible = "arm,cortex-a7-timer",
+			     "arm,armv7-timer";
+		interrupt-parent = <&gic>;
+		arm,cpu-registers-not-fw-configured;
+		always-on;
+		interrupts =
+			<GIC_PPI 13 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_LOW)>,
+			<GIC_PPI 14 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_LOW)>,
+			<GIC_PPI 11 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_LOW)>,
+			<GIC_PPI 10 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_LOW)>;
+	};
+};
-- 
2.7.4

Re: [PATCH v7 3/5] ARM: dts: Renesas R9A06G032 base device tree file

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
<michel.pollet@bp.renesas.com> wrote:
> This adds the Renesas R9A06G032 bare bone support.
>
> This currently only handles generic parts (gic, architected timer)
> and a UART.
>
> Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

Thanks for your patch!

> --- /dev/null
> +++ b/arch/arm/boot/dts/r9a06g032.dtsi

> +       soc {
> +               compatible = "simple-bus";
> +               #address-cells = <1>;
> +               #size-cells = <1>;
> +               interrupt-parent = <&gic>;
> +               ranges;
> +
> +               sysctrl: sysctrl@4000c000 {

system-controller@

> +                       compatible = "renesas,r9a06g032-sysctrl";
> +                       reg = <0x4000c000 0x1000>;
> +                       status = "okay";
> +                       #clock-cells = <1>;
> +               };

With the minor nit above resolved, and pending acceptance of the
sysctrl bindings:

Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v7 3/5] ARM: dts: Renesas R9A06G032 base device tree file

[Simon Horman]

On Fri, May 25, 2018 at 11:27:46AM +0200, Geert Uytterhoeven wrote:
> Hi Michel,
> 
> On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
> <michel.pollet@bp.renesas.com> wrote:
> > This adds the Renesas R9A06G032 bare bone support.
> >
> > This currently only handles generic parts (gic, architected timer)
> > and a UART.
> >
> > Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
> 
> Thanks for your patch!
> 
> > --- /dev/null
> > +++ b/arch/arm/boot/dts/r9a06g032.dtsi
> 
> > +       soc {
> > +               compatible = "simple-bus";
> > +               #address-cells = <1>;
> > +               #size-cells = <1>;
> > +               interrupt-parent = <&gic>;
> > +               ranges;
> > +
> > +               sysctrl: sysctrl@4000c000 {
> 
> system-controller@
> 
> > +                       compatible = "renesas,r9a06g032-sysctrl";
> > +                       reg = <0x4000c000 0x1000>;
> > +                       status = "okay";
> > +                       #clock-cells = <1>;
> > +               };
> 
> With the minor nit above resolved, and pending acceptance of the
> sysctrl bindings:
> 
> Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Thanks. I have marked this patch, and the following one, as deferred
pending acceptance of those bindings. Please repost or otherwise ping me
once that has happened.

[PATCH v7 4/5] ARM: dts: Renesas RZN1D-DB Board base file

[Michel Pollet]

This adds a base device tree file for the RZN1-DB board, with only the
basic support allowing the system to boot to a prompt. Only one UART is
used, with only a single CPU running.

Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
---
 arch/arm/boot/dts/Makefile                  |  1 +
 arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts | 29 +++++++++++++++++++++++++++++
 2 files changed, 30 insertions(+)
 create mode 100644 arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts

diff --git a/arch/arm/boot/dts/Makefile b/arch/arm/boot/dts/Makefile
index f4753b0..6157897 100644
--- a/arch/arm/boot/dts/Makefile
+++ b/arch/arm/boot/dts/Makefile
@@ -814,6 +814,7 @@ dtb-$(CONFIG_ARCH_RENESAS) += \
 	r8a7793-gose.dtb \
 	r8a7794-alt.dtb \
 	r8a7794-silk.dtb \
+	r9a06g032-rzn1d400-db.dtb \
 	sh73a0-kzm9g.dtb
 dtb-$(CONFIG_ARCH_ROCKCHIP) += \
 	rv1108-evb.dtb \
diff --git a/arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts b/arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts
new file mode 100644
index 0000000..292af58
--- /dev/null
+++ b/arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts
@@ -0,0 +1,29 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Device Tree Source for the RZN1D-DB Board
+ *
+ * Copyright (C) 2018 Renesas Electronics Europe Limited
+ *
+ */
+
+/dts-v1/;
+
+#include "r9a06g032.dtsi"
+
+/ {
+	model = "RZN1D-DB Board";
+	compatible = "renesas,rzn1d400-db",
+			"renesas,r9a06g032";
+
+	chosen {
+		stdout-path = "serial0:115200n8";
+	};
+
+	aliases {
+		serial0 = &uart0;
+	};
+};
+
+&uart0 {
+	status = "okay";
+};
-- 
2.7.4

Re: [PATCH v7 4/5] ARM: dts: Renesas RZN1D-DB Board base file

[Geert Uytterhoeven]

On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
<michel.pollet@bp.renesas.com> wrote:
> This adds a base device tree file for the RZN1-DB board, with only the
> basic support allowing the system to boot to a prompt. Only one UART is
> used, with only a single CPU running.
>
> Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

> --- /dev/null
> +++ b/arch/arm/boot/dts/r9a06g032-rzn1d400-db.dts
> @@ -0,0 +1,29 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Device Tree Source for the RZN1D-DB Board
> + *
> + * Copyright (C) 2018 Renesas Electronics Europe Limited
> + *
> + */
> +
> +/dts-v1/;
> +
> +#include "r9a06g032.dtsi"
> +
> +/ {
> +       model = "RZN1D-DB Board";
> +       compatible = "renesas,rzn1d400-db",
> +                       "renesas,r9a06g032";

Nit: This could fit on a single line.

> +
> +       chosen {
> +               stdout-path = "serial0:115200n8";
> +       };
> +
> +       aliases {
> +               serial0 = &uart0;
> +       };
> +};

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[PATCH v7 5/5] clk: renesas: Renesas R9A06G032 clock driver

[Michel Pollet]

This provides a clock driver for the Renesas R09A06G032.
This uses a structure derived from both the RCAR gen2 driver as well as
the renesas-cpg-mssr driver.

Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>
---
 drivers/clk/renesas/Kconfig            |   6 +
 drivers/clk/renesas/Makefile           |   1 +
 drivers/clk/renesas/r9a06g032-clocks.c | 812 +++++++++++++++++++++++++++++++++
 3 files changed, 819 insertions(+)
 create mode 100644 drivers/clk/renesas/r9a06g032-clocks.c

diff --git a/drivers/clk/renesas/Kconfig b/drivers/clk/renesas/Kconfig
index f9ba71311..9022bbe 100644
--- a/drivers/clk/renesas/Kconfig
+++ b/drivers/clk/renesas/Kconfig
@@ -21,6 +21,7 @@ config CLK_RENESAS
 	select CLK_R8A77980 if ARCH_R8A77980
 	select CLK_R8A77990 if ARCH_R8A77990
 	select CLK_R8A77995 if ARCH_R8A77995
+	select CLK_R9A06G032 if ARCH_R9A06G032
 	select CLK_SH73A0 if ARCH_SH73A0
 
 if CLK_RENESAS
@@ -125,6 +126,11 @@ config CLK_R8A77995
 	bool "R-Car D3 clock support" if COMPILE_TEST
 	select CLK_RCAR_GEN3_CPG
 
+config CLK_R9A06G032
+	bool "Renesas R9A06G032 clock driver"
+	help
+	  This is a driver for R9A06G032 clocks
+
 config CLK_SH73A0
 	bool "SH-Mobile AG5 clock support" if COMPILE_TEST
 	select CLK_RENESAS_CPG_MSTP
diff --git a/drivers/clk/renesas/Makefile b/drivers/clk/renesas/Makefile
index fe5bac9..e4aa3d6 100644
--- a/drivers/clk/renesas/Makefile
+++ b/drivers/clk/renesas/Makefile
@@ -20,6 +20,7 @@ obj-$(CONFIG_CLK_R8A77970)		+= r8a77970-cpg-mssr.o
 obj-$(CONFIG_CLK_R8A77980)		+= r8a77980-cpg-mssr.o
 obj-$(CONFIG_CLK_R8A77990)		+= r8a77990-cpg-mssr.o
 obj-$(CONFIG_CLK_R8A77995)		+= r8a77995-cpg-mssr.o
+obj-$(CONFIG_CLK_R9A06G032)		+= r9a06g032-clocks.o
 obj-$(CONFIG_CLK_SH73A0)		+= clk-sh73a0.o
 
 # Family
diff --git a/drivers/clk/renesas/r9a06g032-clocks.c b/drivers/clk/renesas/r9a06g032-clocks.c
new file mode 100644
index 0000000..ad61cda
--- /dev/null
+++ b/drivers/clk/renesas/r9a06g032-clocks.c
@@ -0,0 +1,812 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * R9A09G032 clock driver
+ *
+ * Copyright (C) 2018 Renesas Electronics Europe Limited
+ *
+ * Michel Pollet <michel.pollet@bp.renesas.com>, <buserror@gmail.com>
+ */
+
+#include <linux/clk.h>
+#include <linux/clk-provider.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/math64.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <dt-bindings/clock/r9a06g032-sysctrl.h>
+
+struct r9a06g032_gate {
+	uint16_t gate, reset, ready, midle,
+		scon, mirack, mistat;
+};
+
+/* This is used to describe a clock for instantiation */
+struct r9a06g032_clkdesc {
+	const char *name;
+	uint32_t type: 3;
+	uint32_t index: 8;
+	uint32_t source : 8; /* source index + 1 (0 == none) */
+	/* these are used to populate the bitsel struct */
+	union {
+		struct r9a06g032_gate gate;
+		/* for dividers */
+		struct {
+			unsigned int div_min : 10, div_max : 10, reg: 10;
+			uint16_t div_table[4];
+		};
+		/* For fixed-factor ones */
+		uint16_t div;
+		unsigned int factor;
+		unsigned int frequency;
+		/* for dual gate */
+		struct {
+			uint16_t group : 1, index: 3;
+			uint16_t sel, g1, r1, g2, r2;
+		} dual;
+	};
+} __packed;
+
+#define I_GATE(_clk, _rst, _rdy, _midle, _scon, _mirack, _mistat) \
+	{ .gate = _clk, .reset = _rst, \
+		.ready = _rdy, .midle = _midle, \
+		.scon = _scon, .mirack = _mirack, .mistat = _mistat }
+#define D_GATE(_idx, _n, _src, ...) \
+	{ .type = K_GATE, .index = R9A06G032_##_idx, \
+		.source = 1 + R9A06G032_##_src, .name = _n, \
+		.gate = I_GATE(__VA_ARGS__), }
+#define D_FC(_idx, _n, _freq) \
+	{ .type = K_FC, .index = R9A06G032_##_idx, .name = _n, .frequency = _freq, }
+#define D_FFC(_idx, _n, _src, _div) \
+	{ .type = K_FFC, .index = R9A06G032_##_idx, \
+		.source = 1 + R9A06G032_##_src, .name = _n, \
+		.div = _div, }
+#define D_DIV(_idx, _n, _src, _reg, _min, _max, ...) \
+	{ .type = K_DIV, .index = R9A06G032_##_idx, \
+		.source = 1 + R9A06G032_##_src, .name = _n, \
+		.reg = _reg, .div_min = _min, .div_max = _max, \
+		.div_table = { __VA_ARGS__ } }
+#define D_UGATE(_idx, _n, _src, _g, _gi, _g1, _r1, _g2, _r2) \
+	{ .type = K_DUALGATE, .index = R9A06G032_##_idx, \
+		.source = 1 + R9A06G032_##_src, .name = _n, \
+		.dual = { .group = _g, .index = _gi, \
+			.g1 = _g1, .r1 = _r1, .g2 = _g2, .r2 = _r2 }, }
+
+enum { K_GATE = 0, K_FFC, K_FC, K_DIV, K_BITSEL, K_DUALGATE };
+
+static const struct r9a06g032_clkdesc r9a06g032_clocks[] __initconst = {
+	D_FC(CLKOUT, "clkout", 1000000000),
+	D_FC(CLK_PLL_USB, "clk_pll_usb", 48000000),
+	D_FFC(CLKOUT_D10, "clkout_d10", CLKOUT, 10),
+	D_FFC(CLKOUT_D16, "clkout_d16", CLKOUT, 16),
+	D_FFC(CLKOUT_D160, "clkout_d160", CLKOUT, 160),
+	D_DIV(CLKOUT_D1OR2, "clkout_d1or2", CLKOUT, 0, 1, 2),
+	D_FFC(CLKOUT_D20, "clkout_d20", CLKOUT, 20),
+	D_FFC(CLKOUT_D40, "clkout_d40", CLKOUT, 40),
+	D_FFC(CLKOUT_D5, "clkout_d5", CLKOUT, 5),
+	D_FFC(CLKOUT_D8, "clkout_d8", CLKOUT, 8),
+	D_DIV(DIV_ADC, "div_adc", CLKOUT, 77, 50, 250),
+	D_DIV(DIV_I2C, "div_i2c", CLKOUT, 78, 12, 16),
+	D_DIV(DIV_NAND, "div_nand", CLKOUT, 82, 12, 32),
+	D_DIV(DIV_P1_PG, "div_p1_pg", CLKOUT, 68, 12, 200),
+	D_DIV(DIV_P2_PG, "div_p2_pg", CLKOUT, 62, 12, 128),
+	D_DIV(DIV_P3_PG, "div_p3_pg", CLKOUT, 64, 8, 128),
+	D_DIV(DIV_P4_PG, "div_p4_pg", CLKOUT, 66, 8, 128),
+	D_DIV(DIV_P5_PG, "div_p5_pg", CLKOUT, 71, 10, 40),
+	D_DIV(DIV_P6_PG, "div_p6_pg", CLKOUT, 18, 12, 64),
+	D_DIV(DIV_QSPI0, "div_qspi0", CLKOUT, 73, 3, 7),
+	D_DIV(DIV_QSPI1, "div_qspi1", CLKOUT, 25, 3, 7),
+	D_DIV(DIV_REF_SYNC, "div_ref_sync", CLKOUT, 56, 2, 16, 2, 4, 8, 16),
+	D_DIV(DIV_SDIO0, "div_sdio0", CLKOUT, 74, 20, 128),
+	D_DIV(DIV_SDIO1, "div_sdio1", CLKOUT, 75, 20, 128),
+	D_DIV(DIV_SWITCH, "div_switch", CLKOUT, 37, 5, 40),
+	D_DIV(DIV_UART, "div_uart", CLKOUT, 79, 12, 128),
+	D_GATE(CLK_25_PG4, "clk_25_pg4", CLKOUT_D40, 0x749, 0x74a, 0x74b, 0, 0xae3, 0, 0),
+	D_GATE(CLK_25_PG5, "clk_25_pg5", CLKOUT_D40, 0x74c, 0x74d, 0x74e, 0, 0xae4, 0, 0),
+	D_GATE(CLK_25_PG6, "clk_25_pg6", CLKOUT_D40, 0x74f, 0x750, 0x751, 0, 0xae5, 0, 0),
+	D_GATE(CLK_25_PG7, "clk_25_pg7", CLKOUT_D40, 0x752, 0x753, 0x754, 0, 0xae6, 0, 0),
+	D_GATE(CLK_25_PG8, "clk_25_pg8", CLKOUT_D40, 0x755, 0x756, 0x757, 0, 0xae7, 0, 0),
+	D_GATE(CLK_ADC, "clk_adc", DIV_ADC, 0x1ea, 0x1eb, 0, 0, 0, 0, 0),
+	D_GATE(CLK_ECAT100, "clk_ecat100", CLKOUT_D10, 0x405, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_HSR100, "clk_hsr100", CLKOUT_D10, 0x483, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_I2C0, "clk_i2c0", DIV_I2C, 0x1e6, 0x1e7, 0, 0, 0, 0, 0),
+	D_GATE(CLK_I2C1, "clk_i2c1", DIV_I2C, 0x1e8, 0x1e9, 0, 0, 0, 0, 0),
+	D_GATE(CLK_MII_REF, "clk_mii_ref", CLKOUT_D40, 0x342, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_NAND, "clk_nand", DIV_NAND, 0x284, 0x285, 0, 0, 0, 0, 0),
+	D_GATE(CLK_NOUSBP2_PG6, "clk_nousbp2_pg6", DIV_P2_PG, 0x774, 0x775, 0, 0, 0, 0, 0),
+	D_GATE(CLK_P1_PG2, "clk_p1_pg2", DIV_P1_PG, 0x862, 0x863, 0, 0, 0, 0, 0),
+	D_GATE(CLK_P1_PG3, "clk_p1_pg3", DIV_P1_PG, 0x864, 0x865, 0, 0, 0, 0, 0),
+	D_GATE(CLK_P1_PG4, "clk_p1_pg4", DIV_P1_PG, 0x866, 0x867, 0, 0, 0, 0, 0),
+	D_GATE(CLK_P4_PG3, "clk_p4_pg3", DIV_P4_PG, 0x824, 0x825, 0, 0, 0, 0, 0),
+	D_GATE(CLK_P4_PG4, "clk_p4_pg4", DIV_P4_PG, 0x826, 0x827, 0, 0, 0, 0, 0),
+	D_GATE(CLK_P6_PG1, "clk_p6_pg1", DIV_P6_PG, 0x8a0, 0x8a1, 0x8a2, 0, 0xb60, 0, 0),
+	D_GATE(CLK_P6_PG2, "clk_p6_pg2", DIV_P6_PG, 0x8a3, 0x8a4, 0x8a5, 0, 0xb61, 0, 0),
+	D_GATE(CLK_P6_PG3, "clk_p6_pg3", DIV_P6_PG, 0x8a6, 0x8a7, 0x8a8, 0, 0xb62, 0, 0),
+	D_GATE(CLK_P6_PG4, "clk_p6_pg4", DIV_P6_PG, 0x8a9, 0x8aa, 0x8ab, 0, 0xb63, 0, 0),
+	D_GATE(CLK_QSPI0, "clk_qspi0", DIV_QSPI0, 0x2a4, 0x2a5, 0, 0, 0, 0, 0),
+	D_GATE(CLK_QSPI1, "clk_qspi1", DIV_QSPI1, 0x484, 0x485, 0, 0, 0, 0, 0),
+	D_GATE(CLK_RGMII_REF, "clk_rgmii_ref", CLKOUT_D8, 0x340, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_RMII_REF, "clk_rmii_ref", CLKOUT_D20, 0x341, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SDIO0, "clk_sdio0", DIV_SDIO0, 0x64, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SDIO1, "clk_sdio1", DIV_SDIO1, 0x644, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SERCOS100, "clk_sercos100", CLKOUT_D10, 0x425, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SLCD, "clk_slcd", DIV_P1_PG, 0x860, 0x861, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SPI0, "clk_spi0", DIV_P3_PG, 0x7e0, 0x7e1, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SPI1, "clk_spi1", DIV_P3_PG, 0x7e2, 0x7e3, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SPI2, "clk_spi2", DIV_P3_PG, 0x7e4, 0x7e5, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SPI3, "clk_spi3", DIV_P3_PG, 0x7e6, 0x7e7, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SPI4, "clk_spi4", DIV_P4_PG, 0x820, 0x821, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SPI5, "clk_spi5", DIV_P4_PG, 0x822, 0x823, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SWITCH, "clk_switch", DIV_SWITCH, 0x982, 0x983, 0, 0, 0, 0, 0),
+	D_DIV(DIV_MOTOR, "div_motor", CLKOUT_D5, 84, 2, 8),
+	D_GATE(HCLK_ECAT125, "hclk_ecat125", CLKOUT_D8, 0x400, 0x401, 0, 0x402, 0, 0x440, 0x441),
+	D_GATE(HCLK_PINCONFIG, "hclk_pinconfig", CLKOUT_D40, 0x740, 0x741, 0x742, 0, 0xae0, 0, 0),
+	D_GATE(HCLK_SERCOS, "hclk_sercos", CLKOUT_D10, 0x420, 0x422, 0, 0x421, 0, 0x460, 0x461),
+	D_GATE(HCLK_SGPIO2, "hclk_sgpio2", DIV_P5_PG, 0x8c3, 0x8c4, 0x8c5, 0, 0xb41, 0, 0),
+	D_GATE(HCLK_SGPIO3, "hclk_sgpio3", DIV_P5_PG, 0x8c6, 0x8c7, 0x8c8, 0, 0xb42, 0, 0),
+	D_GATE(HCLK_SGPIO4, "hclk_sgpio4", DIV_P5_PG, 0x8c9, 0x8ca, 0x8cb, 0, 0xb43, 0, 0),
+	D_GATE(HCLK_TIMER0, "hclk_timer0", CLKOUT_D40, 0x743, 0x744, 0x745, 0, 0xae1, 0, 0),
+	D_GATE(HCLK_TIMER1, "hclk_timer1", CLKOUT_D40, 0x746, 0x747, 0x748, 0, 0xae2, 0, 0),
+	D_GATE(HCLK_USBF, "hclk_usbf", CLKOUT_D8, 0xe3, 0, 0, 0xe4, 0, 0x102, 0x103),
+	D_GATE(HCLK_USBH, "hclk_usbh", CLKOUT_D8, 0xe0, 0xe1, 0, 0xe2, 0, 0x100, 0x101),
+	D_GATE(HCLK_USBPM, "hclk_usbpm", CLKOUT_D8, 0xe5, 0, 0, 0, 0, 0, 0),
+	D_GATE(CLK_48_PG_F, "clk_48_pg_f", CLK_48, 0x78c, 0x78d, 0, 0x78e, 0, 0xb04, 0xb05),
+	D_GATE(CLK_48_PG4, "clk_48_pg4", CLK_48, 0x789, 0x78a, 0x78b, 0, 0xb03, 0, 0),
+	D_FFC(CLK_DDRPHY_PLLCLK_D4, "clk_ddrphy_pllclk_d4", CLK_DDRPHY_PLLCLK, 4),
+	D_FFC(CLK_ECAT100_D4, "clk_ecat100_d4", CLK_ECAT100, 4),
+	D_FFC(CLK_HSR100_D2, "clk_hsr100_d2", CLK_HSR100, 2),
+	D_FFC(CLK_REF_SYNC_D4, "clk_ref_sync_d4", CLK_REF_SYNC, 4),
+	D_FFC(CLK_REF_SYNC_D8, "clk_ref_sync_d8", CLK_REF_SYNC, 8),
+	D_FFC(CLK_SERCOS100_D2, "clk_sercos100_d2", CLK_SERCOS100, 2),
+	D_DIV(DIV_CA7, "div_ca7", CLK_REF_SYNC, 57, 1, 4, 1, 2, 4),
+	D_GATE(HCLK_CAN0, "hclk_can0", CLK_48, 0x783, 0x784, 0x785, 0, 0xb01, 0, 0),
+	D_GATE(HCLK_CAN1, "hclk_can1", CLK_48, 0x786, 0x787, 0x788, 0, 0xb02, 0, 0),
+	D_GATE(HCLK_DELTASIGMA, "hclk_deltasigma", DIV_MOTOR, 0x1ef, 0x1f0, 0x1f1, 0, 0, 0, 0),
+	D_GATE(HCLK_PWMPTO, "hclk_pwmpto", DIV_MOTOR, 0x1ec, 0x1ed, 0x1ee, 0, 0, 0, 0),
+	D_GATE(HCLK_RSV, "hclk_rsv", CLK_48, 0x780, 0x781, 0x782, 0, 0xb00, 0, 0),
+	D_GATE(HCLK_SGPIO0, "hclk_sgpio0", DIV_MOTOR, 0x1e0, 0x1e1, 0x1e2, 0, 0, 0, 0),
+	D_GATE(HCLK_SGPIO1, "hclk_sgpio1", DIV_MOTOR, 0x1e3, 0x1e4, 0x1e5, 0, 0, 0, 0),
+	D_DIV(RTOS_MDC, "rtos_mdc", CLK_REF_SYNC, 100, 80, 640, 80, 160, 320, 640),
+	D_GATE(CLK_CM3, "clk_cm3", CLK_REF_SYNC_D4, 0xba0, 0xba1, 0, 0xba2, 0, 0xbc0, 0xbc1),
+	D_GATE(CLK_DDRC, "clk_ddrc", CLK_DDRPHY_PLLCLK_D4, 0x323, 0x324, 0, 0, 0, 0, 0),
+	D_GATE(CLK_ECAT25, "clk_ecat25", CLK_ECAT100_D4, 0x403, 0x404, 0, 0, 0, 0, 0),
+	D_GATE(CLK_HSR50, "clk_hsr50", CLK_HSR100_D2, 0x484, 0x485, 0, 0, 0, 0, 0),
+	D_GATE(CLK_HW_RTOS, "clk_hw_rtos", CLK_REF_SYNC_D4, 0xc60, 0xc61, 0, 0, 0, 0, 0),
+	D_GATE(CLK_SERCOS50, "clk_sercos50", CLK_SERCOS100_D2, 0x424, 0x423, 0, 0, 0, 0, 0),
+	D_GATE(HCLK_ADC, "hclk_adc", CLK_REF_SYNC_D8, 0x1af, 0x1b0, 0x1b1, 0, 0, 0, 0),
+	D_GATE(HCLK_CM3, "hclk_cm3", CLK_REF_SYNC_D4, 0xc20, 0xc21, 0xc22, 0, 0, 0, 0),
+	D_GATE(HCLK_CRYPTO_EIP150, "hclk_crypto_eip150", CLK_REF_SYNC_D4, 0x123, 0x124, 0x125, 0, 0x142, 0, 0),
+	D_GATE(HCLK_CRYPTO_EIP93, "hclk_crypto_eip93", CLK_REF_SYNC_D4, 0x120, 0x121, 0, 0x122, 0, 0x140, 0x141),
+	D_GATE(HCLK_DDRC, "hclk_ddrc", CLK_REF_SYNC_D4, 0x320, 0x322, 0, 0x321, 0, 0x3a0, 0x3a1),
+	D_GATE(HCLK_DMA0, "hclk_dma0", CLK_REF_SYNC_D4, 0x260, 0x261, 0x262, 0x263, 0x2c0, 0x2c1, 0x2c2),
+	D_GATE(HCLK_DMA1, "hclk_dma1", CLK_REF_SYNC_D4, 0x264, 0x265, 0x266, 0x267, 0x2c3, 0x2c4, 0x2c5),
+	D_GATE(HCLK_GMAC0, "hclk_gmac0", CLK_REF_SYNC_D4, 0x360, 0x361, 0x362, 0x363, 0x3c0, 0x3c1, 0x3c2),
+	D_GATE(HCLK_GMAC1, "hclk_gmac1", CLK_REF_SYNC_D4, 0x380, 0x381, 0x382, 0x383, 0x3e0, 0x3e1, 0x3e2),
+	D_GATE(HCLK_GPIO0, "hclk_gpio0", CLK_REF_SYNC_D4, 0x212, 0x213, 0x214, 0, 0, 0, 0),
+	D_GATE(HCLK_GPIO1, "hclk_gpio1", CLK_REF_SYNC_D4, 0x215, 0x216, 0x217, 0, 0, 0, 0),
+	D_GATE(HCLK_GPIO2, "hclk_gpio2", CLK_REF_SYNC_D4, 0x229, 0x22a, 0x22b, 0, 0, 0, 0),
+	D_GATE(HCLK_HSR, "hclk_hsr", CLK_HSR100_D2, 0x480, 0x482, 0, 0x481, 0, 0x4c0, 0x4c1),
+	D_GATE(HCLK_I2C0, "hclk_i2c0", CLK_REF_SYNC_D8, 0x1a9, 0x1aa, 0x1ab, 0, 0, 0, 0),
+	D_GATE(HCLK_I2C1, "hclk_i2c1", CLK_REF_SYNC_D8, 0x1ac, 0x1ad, 0x1ae, 0, 0, 0, 0),
+	D_GATE(HCLK_LCD, "hclk_lcd", CLK_REF_SYNC_D4, 0x7a0, 0x7a1, 0x7a2, 0, 0xb20, 0, 0),
+	D_GATE(HCLK_MSEBI_M, "hclk_msebi_m", CLK_REF_SYNC_D4, 0x164, 0x165, 0x166, 0, 0x183, 0, 0),
+	D_GATE(HCLK_MSEBI_S, "hclk_msebi_s", CLK_REF_SYNC_D4, 0x160, 0x161, 0x162, 0x163, 0x180, 0x181, 0x182),
+	D_GATE(HCLK_NAND, "hclk_nand", CLK_REF_SYNC_D4, 0x280, 0x281, 0x282, 0x283, 0x2e0, 0x2e1, 0x2e2),
+	D_GATE(HCLK_PG_I, "hclk_pg_i", CLK_REF_SYNC_D4, 0x7ac, 0x7ad, 0, 0x7ae, 0, 0xb24, 0xb25),
+	D_GATE(HCLK_PG19, "hclk_pg19", CLK_REF_SYNC_D4, 0x22c, 0x22d, 0x22e, 0, 0, 0, 0),
+	D_GATE(HCLK_PG20, "hclk_pg20", CLK_REF_SYNC_D4, 0x22f, 0x230, 0x231, 0, 0, 0, 0),
+	D_GATE(HCLK_PG3, "hclk_pg3", CLK_REF_SYNC_D4, 0x7a6, 0x7a7, 0x7a8, 0, 0xb22, 0, 0),
+	D_GATE(HCLK_PG4, "hclk_pg4", CLK_REF_SYNC_D4, 0x7a9, 0x7aa, 0x7ab, 0, 0xb23, 0, 0),
+	D_GATE(HCLK_QSPI0, "hclk_qspi0", CLK_REF_SYNC_D4, 0x2a0, 0x2a1, 0x2a2, 0x2a3, 0x300, 0x301, 0x302),
+	D_GATE(HCLK_QSPI1, "hclk_qspi1", CLK_REF_SYNC_D4, 0x480, 0x481, 0x482, 0x483, 0x4c0, 0x4c1, 0x4c2),
+	D_GATE(HCLK_ROM, "hclk_rom", CLK_REF_SYNC_D4, 0xaa0, 0xaa1, 0xaa2, 0, 0xb80, 0, 0),
+	D_GATE(HCLK_RTC, "hclk_rtc", CLK_REF_SYNC_D8, 0xa00, 0, 0, 0, 0, 0, 0),
+	D_GATE(HCLK_SDIO0, "hclk_sdio0", CLK_REF_SYNC_D4, 0x60, 0x61, 0x62, 0x63, 0x80, 0x81, 0x82),
+	D_GATE(HCLK_SDIO1, "hclk_sdio1", CLK_REF_SYNC_D4, 0x640, 0x641, 0x642, 0x643, 0x660, 0x661, 0x662),
+	D_GATE(HCLK_SEMAP, "hclk_semap", CLK_REF_SYNC_D4, 0x7a3, 0x7a4, 0x7a5, 0, 0xb21, 0, 0),
+	D_GATE(HCLK_SPI0, "hclk_spi0", CLK_REF_SYNC_D4, 0x200, 0x201, 0x202, 0, 0, 0, 0),
+	D_GATE(HCLK_SPI1, "hclk_spi1", CLK_REF_SYNC_D4, 0x203, 0x204, 0x205, 0, 0, 0, 0),
+	D_GATE(HCLK_SPI2, "hclk_spi2", CLK_REF_SYNC_D4, 0x206, 0x207, 0x208, 0, 0, 0, 0),
+	D_GATE(HCLK_SPI3, "hclk_spi3", CLK_REF_SYNC_D4, 0x209, 0x20a, 0x20b, 0, 0, 0, 0),
+	D_GATE(HCLK_SPI4, "hclk_spi4", CLK_REF_SYNC_D4, 0x20c, 0x20d, 0x20e, 0, 0, 0, 0),
+	D_GATE(HCLK_SPI5, "hclk_spi5", CLK_REF_SYNC_D4, 0x20f, 0x210, 0x211, 0, 0, 0, 0),
+	D_GATE(HCLK_SWITCH, "hclk_switch", CLK_REF_SYNC_D4, 0x980, 0, 0x981, 0, 0, 0, 0),
+	D_GATE(HCLK_SWITCH_RG, "hclk_switch_rg", CLK_REF_SYNC_D4, 0xc40, 0xc41, 0xc42, 0, 0, 0, 0),
+	D_GATE(HCLK_UART0, "hclk_uart0", CLK_REF_SYNC_D8, 0x1a0, 0x1a1, 0x1a2, 0, 0, 0, 0),
+	D_GATE(HCLK_UART1, "hclk_uart1", CLK_REF_SYNC_D8, 0x1a3, 0x1a4, 0x1a5, 0, 0, 0, 0),
+	D_GATE(HCLK_UART2, "hclk_uart2", CLK_REF_SYNC_D8, 0x1a6, 0x1a7, 0x1a8, 0, 0, 0, 0),
+	D_GATE(HCLK_UART3, "hclk_uart3", CLK_REF_SYNC_D4, 0x218, 0x219, 0x21a, 0, 0, 0, 0),
+	D_GATE(HCLK_UART4, "hclk_uart4", CLK_REF_SYNC_D4, 0x21b, 0x21c, 0x21d, 0, 0, 0, 0),
+	D_GATE(HCLK_UART5, "hclk_uart5", CLK_REF_SYNC_D4, 0x220, 0x221, 0x222, 0, 0, 0, 0),
+	D_GATE(HCLK_UART6, "hclk_uart6", CLK_REF_SYNC_D4, 0x223, 0x224, 0x225, 0, 0, 0, 0),
+	D_GATE(HCLK_UART7, "hclk_uart7", CLK_REF_SYNC_D4, 0x226, 0x227, 0x228, 0, 0, 0, 0),
+	/*
+	 * These are not hardware clocks, but are needed to handle the special
+	 * case where we have a 'selector bit' that doesn't just change the
+	 * parent for a clock, but also the gate it's suposed to use.
+	 */
+	{
+		.index = R9A06G032_UART_GROUP_012,
+		.name = "uart_group_012",
+		.type = K_BITSEL,
+		.source = 1 + R9A06G032_DIV_UART,
+		/* R9A06G032_SYSCTRL_REG_PWRCTRL_PG1_PR2 */
+		.dual.sel = ((0xec / 4) << 5) | 24,
+		.dual.group = 0,
+	},
+	{
+		.index = R9A06G032_UART_GROUP_34567,
+		.name = "uart_group_34567",
+		.type = K_BITSEL,
+		.source = 1 + R9A06G032_DIV_P2_PG,
+		/* R9A06G032_SYSCTRL_REG_PWRCTRL_PG0_0 */
+		.dual.sel = ((0x34 / 4) << 5) | 30,
+		.dual.group = 1,
+	},
+	D_UGATE(CLK_UART0, "clk_uart0", UART_GROUP_012, 0, 0, 0x1b2, 0x1b3, 0x1b4, 0x1b5),
+	D_UGATE(CLK_UART1, "clk_uart1", UART_GROUP_012, 0, 1, 0x1b6, 0x1b7, 0x1b8, 0x1b9),
+	D_UGATE(CLK_UART2, "clk_uart2", UART_GROUP_012, 0, 2, 0x1ba, 0x1bb, 0x1bc, 0x1bd),
+	D_UGATE(CLK_UART3, "clk_uart3", UART_GROUP_34567, 1, 0, 0x760, 0x761, 0x762, 0x763),
+	D_UGATE(CLK_UART4, "clk_uart4", UART_GROUP_34567, 1, 1, 0x764, 0x765, 0x766, 0x767),
+	D_UGATE(CLK_UART5, "clk_uart5", UART_GROUP_34567, 1, 2, 0x768, 0x769, 0x76a, 0x76b),
+	D_UGATE(CLK_UART6, "clk_uart6", UART_GROUP_34567, 1, 3, 0x76c, 0x76d, 0x76e, 0x76f),
+	D_UGATE(CLK_UART7, "clk_uart7", UART_GROUP_34567, 1, 4, 0x770, 0x771, 0x772, 0x773),
+};
+
+struct r9a06g032_priv {
+	struct clk_onecell_data data;
+	spinlock_t lock;
+	void __iomem *reg;
+};
+
+/* register/bit pairs are encoded as an uint16_t */
+static void clk_rdesc_set(
+	struct r9a06g032_priv *clocks,
+	uint16_t one, unsigned int on)
+{
+	u32 __iomem *reg = ((u32 __iomem *)clocks->reg) + (one >> 5);
+	u32 val = clk_readl(reg);
+
+	val = (val & ~(1U << (one & 0x1f))) | ((!!on) << (one & 0x1f));
+	clk_writel(val, reg);
+}
+
+static int clk_rdesc_get(
+	struct r9a06g032_priv *clocks,
+	uint16_t one)
+{
+	u32 __iomem *reg = ((u32 __iomem *)clocks->reg) + (one >> 5);
+	u32 val = clk_readl(reg);
+
+	return !!(val & (1U << (one & 0x1f)));
+}
+
+/*
+ * This implements the R9A09G032 clock gate 'driver'. We cannot use the system's
+ * clock gate framework as the gates on the R9A09G032 have a special enabling
+ * sequence, therefore we use this little proxy.
+ */
+struct r9a06g032_clk_gate {
+	struct clk_hw hw;
+	struct r9a06g032_priv *clocks;
+	u16 index;
+	unsigned read_only : 1;
+
+	struct r9a06g032_gate gate;
+};
+
+#define to_r9a06g032_gate(_hw) container_of(_hw, struct r9a06g032_clk_gate, hw)
+
+static void r9a06g032_clk_gate_set(
+	struct r9a06g032_priv *clocks,
+	struct r9a06g032_gate *g, int on)
+{
+	unsigned long flags;
+
+	WARN_ON(!g->gate);
+
+	spin_lock_irqsave(&clocks->lock, flags);
+	clk_rdesc_set(clocks, g->gate, on);
+	/* De-assert reset */
+	if (g->reset)
+		clk_rdesc_set(clocks, g->reset, 1);
+	spin_unlock_irqrestore(&clocks->lock, flags);
+
+	/* Hardware manual recommends 5us delay after enabling clock & reset */
+	udelay(5);
+
+	/* If the peripheral is memory mapped (i.e. an AXI slave), there is an
+	 * associated SLVRDY bit in the System Controller that needs to be set
+	 * so that the FlexWAY bus fabric passes on the read/write requests.
+	 */
+	if (g->ready || g->midle) {
+		spin_lock_irqsave(&clocks->lock, flags);
+		if (g->ready)
+			clk_rdesc_set(clocks, g->ready, on);
+		/* Clear 'Master Idle Request' bit */
+		if (g->midle)
+			clk_rdesc_set(clocks, g->midle, !on);
+		spin_unlock_irqrestore(&clocks->lock, flags);
+	}
+	/* Note: We don't wait for FlexWAY Socket Connection signal */
+}
+
+static int r9a06g032_clk_gate_enable(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_gate *g = to_r9a06g032_gate(hw);
+
+	r9a06g032_clk_gate_set(g->clocks, &g->gate, 1);
+	return 0;
+}
+
+static void r9a06g032_clk_gate_disable(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_gate *g = to_r9a06g032_gate(hw);
+
+	if (!g->read_only)
+		r9a06g032_clk_gate_set(g->clocks, &g->gate, 0);
+	else
+		pr_debug("%s %s: disallowed\n", __func__,
+			__clk_get_name(hw->clk));
+}
+
+static int r9a06g032_clk_gate_is_enabled(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_gate *g = to_r9a06g032_gate(hw);
+
+	return clk_rdesc_get(g->clocks, g->gate.gate);
+}
+
+static const struct clk_ops r9a06g032_clk_gate_ops = {
+	.enable = r9a06g032_clk_gate_enable,
+	.disable = r9a06g032_clk_gate_disable,
+	.is_enabled = r9a06g032_clk_gate_is_enabled,
+};
+
+static struct clk *r9a06g032_register_gate(
+	struct r9a06g032_priv *clocks,
+	const char *parent_name,
+	const struct r9a06g032_clkdesc *desc)
+{
+	struct clk *clk;
+	struct r9a06g032_clk_gate *g;
+	struct clk_init_data init;
+
+	g = kzalloc(sizeof(struct r9a06g032_clk_gate), GFP_KERNEL);
+	if (!g)
+		return NULL;
+
+	init.name = desc->name;
+	init.ops = &r9a06g032_clk_gate_ops;
+	init.flags = CLK_IS_BASIC | CLK_SET_RATE_PARENT;
+	init.parent_names = parent_name ? &parent_name : NULL;
+	init.num_parents = parent_name ? 1 : 0;
+
+	g->clocks = clocks;
+	g->index = desc->index;
+	g->gate = desc->gate;
+	g->hw.init = &init;
+	g->read_only = 0;
+
+	clk = clk_register(NULL, &g->hw);
+	if (IS_ERR(clk)) {
+		kfree(g);
+		return NULL;
+	}
+	/*
+	 * important here, some clocks are already in use by the CM3, we
+	 * have to assume they are not Linux's to play with and try to disable
+	 * at the end of the boot!
+	 * Therefore we increase the clock usage count by arbitrarily enabling
+	 * the clock, allowing it to stay untouched at the end of the boot.
+	 */
+	g->read_only = r9a06g032_clk_gate_is_enabled(&g->hw);
+	if (g->read_only)
+		pr_debug("%s was enabled, making read-only\n", desc->name);
+	return clk;
+}
+
+struct r9a06g032_clk_div {
+	struct clk_hw hw;
+	struct r9a06g032_priv *clocks;
+	u16 index;
+	u16 reg;
+	u16 min, max;
+	uint8_t table_size;
+	u16 table[8];	/* we know there are no more than 8 */
+};
+
+#define to_r9a06g032_divider(_hw) \
+		container_of(_hw, struct r9a06g032_clk_div, hw)
+
+static unsigned long r9a06g032_divider_recalc_rate(
+	struct clk_hw *hw,
+	unsigned long parent_rate)
+{
+	struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
+	u32 *reg = ((u32 *)clk->clocks->reg) + clk->reg;
+	long div = clk_readl(reg);
+
+	if (div < clk->min)
+		div = clk->min;
+	else if (div > clk->max)
+		div = clk->max;
+	return DIV_ROUND_UP(parent_rate, div);
+}
+
+/*
+ * Attempts to find a value that is in range of min,max,
+ * and if a table of set dividers was specified for this
+ * register, try to find the fixed divider that is the closest
+ * to the target frequency
+ */
+static long r9a06g032_divider_clamp_div(
+	struct r9a06g032_clk_div *clk,
+	unsigned long rate, unsigned long prate)
+{
+	/* + 1 to cope with rates that have the remainder dropped */
+	long div = DIV_ROUND_UP(prate, rate + 1);
+	int i;
+
+	if (div <= clk->min)
+		return clk->min;
+	if (div >= clk->max)
+		return clk->max;
+
+	for (i = 0; clk->table_size && i < clk->table_size - 1; i++) {
+		if (div >= clk->table[i] && div <= clk->table[i+1]) {
+			unsigned long m = rate -
+				DIV_ROUND_UP(prate, clk->table[i]);
+			unsigned long p =
+				DIV_ROUND_UP(prate, clk->table[i + 1]) -
+				rate;
+			/*
+			 * select the divider that generates
+			 * the value closest to the ideal frequency
+			 */
+			div = p >= m ? clk->table[i] : clk->table[i + 1];
+			return div;
+		}
+	}
+	return div;
+}
+
+static long r9a06g032_divider_round_rate(
+	struct clk_hw *hw, unsigned long rate,
+	unsigned long *prate)
+{
+	struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
+	long div = DIV_ROUND_UP(*prate, rate);
+
+	pr_devel("%s %pC %ld (prate %ld) (wanted div %ld)\n", __func__,
+		hw->clk, rate, *prate, div);
+	pr_devel("   min %d (%ld) max %d (%ld)\n",
+		clk->min, DIV_ROUND_UP(*prate, clk->min),
+		clk->max, DIV_ROUND_UP(*prate, clk->max));
+
+	div = r9a06g032_divider_clamp_div(clk, rate, *prate);
+	/*
+	 * this is a hack. Currently the serial driver asks for a clock rate
+	 * that is 16 times the baud rate -- and that is wildly outside the
+	 * range of the UART divider, somehow there is no provision for that
+	 * case of 'let the divider as is if outside range'.
+	 * The serial driver *shouldn't* play with these clocks anyway, there's
+	 * several uarts attached to this divider, and changing this impacts
+	 * everyone.
+	 */
+	if (clk->index == R9A06G032_DIV_UART) {
+		pr_devel("%s div uart hack!\n", __func__);
+		return clk_get_rate(hw->clk);
+	}
+	pr_devel("%s %pC %ld / %ld = %ld\n", __func__, hw->clk,
+		*prate, div, DIV_ROUND_UP(*prate, div));
+	return DIV_ROUND_UP(*prate, div);
+}
+
+static int r9a06g032_divider_set_rate(
+	struct clk_hw *hw, unsigned long rate,
+	unsigned long parent_rate)
+{
+	struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
+	/* + 1 to cope with rates that have the remainder dropped */
+	u32 div = DIV_ROUND_UP(parent_rate, rate + 1);
+	u32 *reg = ((u32 *)clk->clocks->reg) + clk->reg;
+
+	pr_devel("%s %pC rate %ld parent %ld div %d\n", __func__, hw->clk,
+		rate, parent_rate, div);
+
+	/*
+	 * Need to write the bit 31 with the divider value to
+	 * latch it. Technically we should wait until it has been
+	 * cleared too.
+	 * TODO: Find whether this callback is sleepable, in case
+	 * the hardware /does/ require some sort of spinloop here.
+	 */
+	clk_writel(div | (1U << 31), reg);
+
+	return 0;
+}
+
+static const struct clk_ops r9a06g032_clk_div_ops = {
+	.recalc_rate = r9a06g032_divider_recalc_rate,
+	.round_rate = r9a06g032_divider_round_rate,
+	.set_rate = r9a06g032_divider_set_rate,
+};
+
+static struct clk *r9a06g032_register_divider(
+	struct r9a06g032_priv *clocks,
+	const char *parent_name,
+	const struct r9a06g032_clkdesc *desc)
+{
+	struct r9a06g032_clk_div *div;
+	struct clk *clk;
+	struct clk_init_data init;
+	int i;
+
+	div = kzalloc(sizeof(struct r9a06g032_clk_div), GFP_KERNEL);
+	if (!div)
+		return NULL;
+
+	init.name = desc->name;
+	init.ops = &r9a06g032_clk_div_ops;
+	init.flags = CLK_IS_BASIC | CLK_SET_RATE_PARENT;
+	init.parent_names = parent_name ? &parent_name : NULL;
+	init.num_parents = parent_name ? 1 : 0;
+
+	div->clocks = clocks;
+	div->index = desc->index;
+	div->reg = desc->reg;
+	div->hw.init = &init;
+	div->min = desc->div_min;
+	div->max = desc->div_max;
+	/* populate (optional) divider table fixed values */
+	for (i = 0; i < ARRAY_SIZE(div->table) &&
+			i < ARRAY_SIZE(desc->div_table) &&
+			desc->div_table[i]; i++) {
+		div->table[div->table_size++] = desc->div_table[i];
+	}
+
+	clk = clk_register(NULL, &div->hw);
+	if (IS_ERR(clk)) {
+		kfree(div);
+		return NULL;
+	}
+	return clk;
+}
+
+/*
+ * This clock provider handles the case of the R9A06G032 where you have
+ * peripherals that have two potential clock source and two gates, one for
+ * each of the clock source - the used clock source (for all sub clocks)
+ * is selected by a single bit.
+ * That single bit affects all sub-clocks, and therefore needs to change the
+ * active gate (and turn the others off) and force a recalculation of the rates.
+ *
+ * This implements two clock providers, one 'bitselect' that
+ * handles the switch between both parents, and another 'dualgate'
+ * that knows which gate to poke at, depending on the parent's bit position.
+ */
+struct r9a06g032_clk_bitsel {
+	struct clk_hw	hw;
+	struct r9a06g032_priv *clocks;
+	u16 index;
+	u16 selector;		/* selector register + bit */
+};
+
+#define to_clk_bitselect(_hw) \
+		container_of(_hw, struct r9a06g032_clk_bitsel, hw)
+
+static u8 r9a06g032_clk_mux_get_parent(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_bitsel *set = to_clk_bitselect(hw);
+
+	return clk_rdesc_get(set->clocks, set->selector);
+}
+
+static int r9a06g032_clk_mux_set_parent(struct clk_hw *hw, u8 index)
+{
+	struct r9a06g032_clk_bitsel *set = to_clk_bitselect(hw);
+
+	/* a single bit in the register selects one of two parent clocks */
+	clk_rdesc_set(set->clocks, set->selector, !!index);
+
+	return 0;
+}
+
+static const struct clk_ops clk_bitselect_ops = {
+	.get_parent = r9a06g032_clk_mux_get_parent,
+	.set_parent = r9a06g032_clk_mux_set_parent,
+};
+
+static struct clk *r9a06g032_register_bitsel(
+	struct r9a06g032_priv *clocks,
+	const char *parent_name,
+	const struct r9a06g032_clkdesc *desc)
+{
+	struct clk *clk;
+	struct r9a06g032_clk_bitsel *g;
+	struct clk_init_data init;
+	const char *names[2];
+
+	/* allocate the gate */
+	g = kzalloc(sizeof(struct r9a06g032_clk_bitsel), GFP_KERNEL);
+	if (!g)
+		return NULL;
+
+	names[0] = parent_name;
+	names[1] = "clk_pll_usb";
+
+	init.name = desc->name;
+	init.ops = &clk_bitselect_ops;
+	init.flags = CLK_IS_BASIC | CLK_SET_RATE_PARENT;
+	init.parent_names = names;
+	init.num_parents = 2;
+
+	g->clocks = clocks;
+	g->index = desc->index;
+	g->selector = desc->dual.sel;
+	g->hw.init = &init;
+
+	clk = clk_register(NULL, &g->hw);
+	if (IS_ERR(clk)) {
+		kfree(g);
+		return NULL;
+	}
+	return clk;
+}
+
+struct r9a06g032_clk_dualgate {
+	struct clk_hw	hw;
+	struct r9a06g032_priv *clocks;
+	u16 index;
+	u16 selector;		/* selector register + bit */
+	struct r9a06g032_gate gate[2];
+};
+#define to_clk_dualgate(_hw) \
+		container_of(_hw, struct r9a06g032_clk_dualgate, hw)
+
+static int r9a06g032_clk_dualgate_setenable(
+	struct r9a06g032_clk_dualgate *g, int enable)
+{
+	uint8_t sel_bit = clk_rdesc_get(g->clocks, g->selector);
+
+	/* we always turn off the 'other' gate, regardless */
+	r9a06g032_clk_gate_set(g->clocks, &g->gate[!sel_bit], 0);
+	r9a06g032_clk_gate_set(g->clocks, &g->gate[sel_bit], enable);
+
+	return 0;
+}
+
+static int r9a06g032_clk_dualgate_enable(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_dualgate *gate = to_clk_dualgate(hw);
+
+	r9a06g032_clk_dualgate_setenable(gate, 1);
+
+	return 0;
+}
+
+static void r9a06g032_clk_dualgate_disable(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_dualgate *gate = to_clk_dualgate(hw);
+
+	r9a06g032_clk_dualgate_setenable(gate, 0);
+}
+
+static int r9a06g032_clk_dualgate_is_enabled(struct clk_hw *hw)
+{
+	struct r9a06g032_clk_dualgate *g = to_clk_dualgate(hw);
+	uint8_t sel_bit = clk_rdesc_get(g->clocks, g->selector);
+
+	return clk_rdesc_get(g->clocks, g->gate[sel_bit].gate);
+}
+
+static const struct clk_ops r9a06g032_clk_dualgate_ops = {
+	.enable = r9a06g032_clk_dualgate_enable,
+	.disable = r9a06g032_clk_dualgate_disable,
+	.is_enabled = r9a06g032_clk_dualgate_is_enabled,
+};
+
+static struct clk *r9a06g032_register_dualgate(
+	struct r9a06g032_priv *clocks,
+	const char *parent_name,
+	const struct r9a06g032_clkdesc *desc,
+	uint16_t sel)
+{
+	struct r9a06g032_clk_dualgate *g;
+	struct clk *clk;
+	struct clk_init_data init;
+
+	/* allocate the gate */
+	g = kzalloc(sizeof(struct r9a06g032_clk_dualgate), GFP_KERNEL);
+	if (!g)
+		return NULL;
+	g->clocks = clocks;
+	g->index = desc->index;
+	g->selector = sel;
+	g->gate[0].gate = desc->dual.g1;
+	g->gate[0].reset = desc->dual.r1;
+	g->gate[1].gate = desc->dual.g2;
+	g->gate[1].reset = desc->dual.r2;
+
+	init.name = desc->name;
+	init.ops = &r9a06g032_clk_dualgate_ops;
+	init.flags = CLK_IS_BASIC | CLK_SET_RATE_PARENT;
+	init.parent_names = &parent_name;
+	init.num_parents = 1;
+
+	g->hw.init = &init;
+
+	clk = clk_register(NULL, &g->hw);
+	if (IS_ERR(clk)) {
+		kfree(g);
+		return NULL;
+	}
+	return clk;
+}
+
+static void __init r9a06g032_clocks_init(struct device_node *np)
+{
+	struct r9a06g032_priv *clocks;
+	struct clk **clks;
+	unsigned int i;
+	uint16_t uart_group_sel[2];
+
+	clocks = kzalloc(sizeof(*clocks), GFP_KERNEL);
+	clks = kzalloc(R9A06G032_CLOCK_COUNT * sizeof(struct clk *),
+			GFP_KERNEL);
+	if (clocks == NULL || clks == NULL) {
+		/* We're leaking memory on purpose, there's no point in cleaning
+		 * up as the system won't boot anyway.
+		 */
+		return;
+	}
+	spin_lock_init(&clocks->lock);
+
+	clocks->data.clks = clks;
+	clocks->data.clk_num = R9A06G032_CLOCK_COUNT;
+
+	clocks->reg = of_iomap(np, 0);
+	if (WARN_ON(clocks->reg == NULL))
+		return;
+	for (i = 0; i < ARRAY_SIZE(r9a06g032_clocks); ++i) {
+		const struct r9a06g032_clkdesc *d = &r9a06g032_clocks[i];
+		const char *parent_name = d->source ?
+			__clk_get_name(clocks->data.clks[d->source - 1]) : NULL;
+		struct clk *clk = NULL;
+
+		switch (d->type) {
+		case K_FC:
+			clk = clk_register_fixed_rate(NULL, d->name,
+						parent_name, 0, d->frequency);
+			break;
+		case K_FFC:
+			clk = clk_register_fixed_factor(NULL, d->name,
+						parent_name, 0, 1, d->div);
+			break;
+		case K_GATE:
+			clk = r9a06g032_register_gate(clocks, parent_name, d);
+			break;
+		case K_DIV:
+			clk = r9a06g032_register_divider(clocks,
+					parent_name, d);
+			break;
+		case K_BITSEL:
+			/* keep that selector register around */
+			uart_group_sel[d->dual.group] = d->dual.sel;
+			clk = r9a06g032_register_bitsel(clocks, parent_name, d);
+			break;
+		case K_DUALGATE:
+			clk = r9a06g032_register_dualgate(clocks, parent_name,
+					d, uart_group_sel[d->dual.group]);
+			break;
+		}
+		clocks->data.clks[d->index] = clk;
+	}
+	of_clk_add_provider(np, of_clk_src_onecell_get, &clocks->data);
+}
+CLK_OF_DECLARE(r9a06g032_clks, "renesas,r9a06g032-sysctrl",
+	       r9a06g032_clocks_init);
-- 
2.7.4

Re: [PATCH v7 5/5] clk: renesas: Renesas R9A06G032 clock driver

[kbuild test robot]

Hi Michel,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on renesas-drivers/clk-renesas]
[also build test WARNING on v4.17-rc6]
[cannot apply to robh/for-next]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Michel-Pollet/dt-bindings-Add-the-r9a06g032-sysctrl-h-file/20180526-154235
base:   https://git.kernel.org/pub/scm/linux/kernel/git/geert/renesas-drivers.git clk-renesas
reproduce:
        # apt-get install sparse
        make ARCH=x86_64 allmodconfig
        make C=1 CF=-D__CHECK_ENDIAN__


sparse warnings: (new ones prefixed by >>)

>> drivers/clk/renesas/r9a06g032-clocks.c:430:22: sparse: cast removes address space of expression
>> drivers/clk/renesas/r9a06g032-clocks.c:431:30: sparse: incorrect type in argument 1 (different address spaces) @@    expected unsigned int [noderef] [usertype] <asn:2>*reg @@    got eref] [usertype] <asn:2>*reg @@
   drivers/clk/renesas/r9a06g032-clocks.c:431:30:    expected unsigned int [noderef] [usertype] <asn:2>*reg
   drivers/clk/renesas/r9a06g032-clocks.c:431:30:    got unsigned int [usertype] *reg
   drivers/clk/renesas/r9a06g032-clocks.c:516:22: sparse: cast removes address space of expression
   drivers/clk/renesas/r9a06g032-clocks.c:528:38: sparse: incorrect type in argument 2 (different address spaces) @@    expected unsigned int [noderef] [usertype] <asn:2>*reg @@    got eref] [usertype] <asn:2>*reg @@
   drivers/clk/renesas/r9a06g032-clocks.c:528:38:    expected unsigned int [noderef] [usertype] <asn:2>*reg
   drivers/clk/renesas/r9a06g032-clocks.c:528:38:    got unsigned int [usertype] *reg

vim +430 drivers/clk/renesas/r9a06g032-clocks.c

   421	
   422	#define to_r9a06g032_divider(_hw) \
   423			container_of(_hw, struct r9a06g032_clk_div, hw)
   424	
   425	static unsigned long r9a06g032_divider_recalc_rate(
   426		struct clk_hw *hw,
   427		unsigned long parent_rate)
   428	{
   429		struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
 > 430		u32 *reg = ((u32 *)clk->clocks->reg) + clk->reg;
 > 431		long div = clk_readl(reg);
   432	
   433		if (div < clk->min)
   434			div = clk->min;
   435		else if (div > clk->max)
   436			div = clk->max;
   437		return DIV_ROUND_UP(parent_rate, div);
   438	}
   439	

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

Re: [PATCH v7 5/5] clk: renesas: Renesas R9A06G032 clock driver

[kbuild test robot]

Hi Michel,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on renesas-drivers/clk-renesas]
[also build test WARNING on v4.17-rc6]
[cannot apply to robh/for-next]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Michel-Pollet/dt-bindings-Add-the-r9a06g032-sysctrl-h-file/20180526-154235
base:   https://git.kernel.org/pub/scm/linux/kernel/git/geert/renesas-drivers.git clk-renesas
reproduce:
        # apt-get install sparse
        make ARCH=x86_64 allmodconfig
        make C=1 CF=-D__CHECK_ENDIAN__


sparse warnings: (new ones prefixed by >>)

>> drivers/clk/renesas/r9a06g032-clocks.c:430:22: sparse: cast removes address space of expression
>> drivers/clk/renesas/r9a06g032-clocks.c:431:30: sparse: incorrect type in argument 1 (different address spaces) @@    expected unsigned int [noderef] [usertype] <asn:2>*reg @@    got eref] [usertype] <asn:2>*reg @@
   drivers/clk/renesas/r9a06g032-clocks.c:431:30:    expected unsigned int [noderef] [usertype] <asn:2>*reg
   drivers/clk/renesas/r9a06g032-clocks.c:431:30:    got unsigned int [usertype] *reg
   drivers/clk/renesas/r9a06g032-clocks.c:516:22: sparse: cast removes address space of expression
   drivers/clk/renesas/r9a06g032-clocks.c:528:38: sparse: incorrect type in argument 2 (different address spaces) @@    expected unsigned int [noderef] [usertype] <asn:2>*reg @@    got eref] [usertype] <asn:2>*reg @@
   drivers/clk/renesas/r9a06g032-clocks.c:528:38:    expected unsigned int [noderef] [usertype] <asn:2>*reg
   drivers/clk/renesas/r9a06g032-clocks.c:528:38:    got unsigned int [usertype] *reg

vim +430 drivers/clk/renesas/r9a06g032-clocks.c

   421	
   422	#define to_r9a06g032_divider(_hw) \
   423			container_of(_hw, struct r9a06g032_clk_div, hw)
   424	
   425	static unsigned long r9a06g032_divider_recalc_rate(
   426		struct clk_hw *hw,
   427		unsigned long parent_rate)
   428	{
   429		struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
 > 430		u32 *reg = ((u32 *)clk->clocks->reg) + clk->reg;
 > 431		long div = clk_readl(reg);
   432	
   433		if (div < clk->min)
   434			div = clk->min;
   435		else if (div > clk->max)
   436			div = clk->max;
   437		return DIV_ROUND_UP(parent_rate, div);
   438	}
   439	

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

Re: [PATCH v7 5/5] clk: renesas: Renesas R9A06G032 clock driver

[Geert Uytterhoeven]

Hi Michel,

On Thu, May 24, 2018 at 11:28 AM, Michel Pollet
<michel.pollet@bp.renesas.com> wrote:
> This provides a clock driver for the Renesas R09A06G032.
> This uses a structure derived from both the RCAR gen2 driver as well as
> the renesas-cpg-mssr driver.
>
> Signed-off-by: Michel Pollet <michel.pollet@bp.renesas.com>

Thanks for your patch!

> --- /dev/null
> +++ b/drivers/clk/renesas/r9a06g032-clocks.c

> +/* register/bit pairs are encoded as an uint16_t */
> +static void clk_rdesc_set(
> +       struct r9a06g032_priv *clocks,
> +       uint16_t one, unsigned int on)
> +{
> +       u32 __iomem *reg = ((u32 __iomem *)clocks->reg) + (one >> 5);

Do you need the cast? Gcc does support void pointer arithmetic, and treats
that like byte pointers.

> +       u32 val = clk_readl(reg);
> +
> +       val = (val & ~(1U << (one & 0x1f))) | ((!!on) << (one & 0x1f));
> +       clk_writel(val, reg);

Hence

        clk_writel(val, clocks->reg + 4 * (one >> 5));

Actually clk_{read,write}l() are deprecated, please use {read,write}l() instead.

> +static void r9a06g032_clk_gate_disable(struct clk_hw *hw)
> +{
> +       struct r9a06g032_clk_gate *g = to_r9a06g032_gate(hw);
> +
> +       if (!g->read_only)
> +               r9a06g032_clk_gate_set(g->clocks, &g->gate, 0);
> +       else
> +               pr_debug("%s %s: disallowed\n", __func__,
> +                       __clk_get_name(hw->clk));

You can print the name of a clock using %pC:

    pr_debug("%s %pC: disallowed\n", __func__, hw->clk);

But I don't think you need the check, cfr. below.

> +static struct clk *r9a06g032_register_gate(
> +       struct r9a06g032_priv *clocks,
> +       const char *parent_name,
> +       const struct r9a06g032_clkdesc *desc)
> +{
> +       struct clk *clk;
> +       struct r9a06g032_clk_gate *g;
> +       struct clk_init_data init;
> +
> +       g = kzalloc(sizeof(struct r9a06g032_clk_gate), GFP_KERNEL);
> +       if (!g)
> +               return NULL;
> +
> +       init.name = desc->name;
> +       init.ops = &r9a06g032_clk_gate_ops;
> +       init.flags = CLK_IS_BASIC | CLK_SET_RATE_PARENT;
> +       init.parent_names = parent_name ? &parent_name : NULL;
> +       init.num_parents = parent_name ? 1 : 0;
> +
> +       g->clocks = clocks;
> +       g->index = desc->index;
> +       g->gate = desc->gate;
> +       g->hw.init = &init;
> +       g->read_only = 0;
> +
> +       clk = clk_register(NULL, &g->hw);
> +       if (IS_ERR(clk)) {
> +               kfree(g);
> +               return NULL;
> +       }
> +       /*
> +        * important here, some clocks are already in use by the CM3, we
> +        * have to assume they are not Linux's to play with and try to disable
> +        * at the end of the boot!
> +        * Therefore we increase the clock usage count by arbitrarily enabling
> +        * the clock, allowing it to stay untouched at the end of the boot.
> +        */
> +       g->read_only = r9a06g032_clk_gate_is_enabled(&g->hw);

Is checking if the clock is enabled the recommended way to find out if a
clock is used by the Cortex M3? No need for a table?

> +       if (g->read_only)
> +               pr_debug("%s was enabled, making read-only\n", desc->name);

You can set init.flags |= CLK_IS_CRITICAL instead of using your own flag.

> +static unsigned long r9a06g032_divider_recalc_rate(
> +       struct clk_hw *hw,
> +       unsigned long parent_rate)
> +{
> +       struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
> +       u32 *reg = ((u32 *)clk->clocks->reg) + clk->reg;

Fishy operations on __iomem pointers

> +       long div = clk_readl(reg);

u32 div?

> +
> +       if (div < clk->min)
> +               div = clk->min;
> +       else if (div > clk->max)
> +               div = clk->max;
> +       return DIV_ROUND_UP(parent_rate, div);
> +}
> +
> +/*
> + * Attempts to find a value that is in range of min,max,
> + * and if a table of set dividers was specified for this
> + * register, try to find the fixed divider that is the closest
> + * to the target frequency
> + */
> +static long r9a06g032_divider_clamp_div(
> +       struct r9a06g032_clk_div *clk,
> +       unsigned long rate, unsigned long prate)
> +{
> +       /* + 1 to cope with rates that have the remainder dropped */
> +       long div = DIV_ROUND_UP(prate, rate + 1);
> +       int i;

unsigned int i

> +
> +       if (div <= clk->min)
> +               return clk->min;
> +       if (div >= clk->max)
> +               return clk->max;
> +
> +       for (i = 0; clk->table_size && i < clk->table_size - 1; i++) {

> +static long r9a06g032_divider_round_rate(
> +       struct clk_hw *hw, unsigned long rate,
> +       unsigned long *prate)
> +{
> +       struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
> +       long div = DIV_ROUND_UP(*prate, rate);
> +
> +       pr_devel("%s %pC %ld (prate %ld) (wanted div %ld)\n", __func__,

Ah, you do know about %pC ;-)

> +               hw->clk, rate, *prate, div);
> +       pr_devel("   min %d (%ld) max %d (%ld)\n",
> +               clk->min, DIV_ROUND_UP(*prate, clk->min),
> +               clk->max, DIV_ROUND_UP(*prate, clk->max));
> +
> +       div = r9a06g032_divider_clamp_div(clk, rate, *prate);
> +       /*
> +        * this is a hack. Currently the serial driver asks for a clock rate
> +        * that is 16 times the baud rate -- and that is wildly outside the
> +        * range of the UART divider, somehow there is no provision for that
> +        * case of 'let the divider as is if outside range'.
> +        * The serial driver *shouldn't* play with these clocks anyway, there's
> +        * several uarts attached to this divider, and changing this impacts
> +        * everyone.

Huh?

> +        */
> +       if (clk->index == R9A06G032_DIV_UART) {
> +               pr_devel("%s div uart hack!\n", __func__);
> +               return clk_get_rate(hw->clk);
> +       }
> +       pr_devel("%s %pC %ld / %ld = %ld\n", __func__, hw->clk,
> +               *prate, div, DIV_ROUND_UP(*prate, div));
> +       return DIV_ROUND_UP(*prate, div);
> +}
> +
> +static int r9a06g032_divider_set_rate(
> +       struct clk_hw *hw, unsigned long rate,
> +       unsigned long parent_rate)
> +{
> +       struct r9a06g032_clk_div *clk = to_r9a06g032_divider(hw);
> +       /* + 1 to cope with rates that have the remainder dropped */
> +       u32 div = DIV_ROUND_UP(parent_rate, rate + 1);
> +       u32 *reg = ((u32 *)clk->clocks->reg) + clk->reg;
> +
> +       pr_devel("%s %pC rate %ld parent %ld div %d\n", __func__, hw->clk,
> +               rate, parent_rate, div);
> +
> +       /*
> +        * Need to write the bit 31 with the divider value to
> +        * latch it. Technically we should wait until it has been
> +        * cleared too.
> +        * TODO: Find whether this callback is sleepable, in case
> +        * the hardware /does/ require some sort of spinloop here.
> +        */
> +       clk_writel(div | (1U << 31), reg);

BIT(31)?

> +
> +       return 0;
> +}

> +static struct clk *r9a06g032_register_divider(
> +       struct r9a06g032_priv *clocks,
> +       const char *parent_name,
> +       const struct r9a06g032_clkdesc *desc)
> +{
> +       struct r9a06g032_clk_div *div;
> +       struct clk *clk;
> +       struct clk_init_data init;
> +       int i;

unsigned int i;


> +static void __init r9a06g032_clocks_init(struct device_node *np)
> +{
> +       struct r9a06g032_priv *clocks;
> +       struct clk **clks;
> +       unsigned int i;
> +       uint16_t uart_group_sel[2];
> +
> +       clocks = kzalloc(sizeof(*clocks), GFP_KERNEL);
> +       clks = kzalloc(R9A06G032_CLOCK_COUNT * sizeof(struct clk *),
> +                       GFP_KERNEL);

kcalloc()?

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds