[0/6] clk: thead: Add support for TH1520 AP_SUBSYS clock controller

Message

Drew Fustini June 16, 2024, 1:54 a.m. UTC

This series adds support for the AP sub-system clock controller in the
T-Head TH1520 [1]. Yangtao Li originally submitted this series in May
2023 [2]. Jisheng made additional improvements and then passed on the
work in progress to me.

Changes in v1:
 - Split th1520_clks into groups for gate, mux, div, pll
 - Convert gate clocks to devm_clk_hw_register_gate_parent_data()
 - Convert mux clocks to devm_clk_hw_register_mux_parent_data_table()
 - Split the PLL recalculation into th1520_pll_vco_recalc_rate() and
   th1520_pll_postdiv_recalc_rate(). Based on Emil's comments in v1,
   add logic to handle the fractional portion of feedback divide value
   when the delta-sigma modulator (DSM) is active.
 - Drop clock-names from the binding per Stephan to avoid relying on
   anything other than the cell index when describing clk_parent_data
   (note: I dropped Rob's Rb because I changed the binding patch)
 - Rename reg/reg2 to cfg0/cfg1 to match the SoC documentation
 - Rename struct for pll clocks from ccu_mdiv to ccu_pll
 - Rebase on v6.10-rc3, remove the dts node reordering patch from v3
   now that it is in mainline

Changes in RFC v3 [3]:
 - Drop redundant new line and unused clk label from the dts example in
   the DT binding which I failed to fix in v2.
 - Add patch [4] from Thomas Bonnefille that fixes dts node ordering in
   th1520.dtsi. Conor has already merged it into riscv-dt-for-next so
   the dts patches in this series are based on top of that.
 - Remove fixed uart clock and converted uart DT nodes to use clocks
   from the clock controller.
 - Remove fixed apb clock and converted the dma controller and timer DT
   nodes to use a clock from the clock controller.
 - Made ccu_disable_helper() and ccu_enable_helper() to static functions
 - Follow the advice from Stephen Boyd in Yangtao's original series to
   not use strings for clk tree topology. Created clk_parent_data arrays
   to be used with CLK_HW_INIT_PARENTS_DATA instead of parent strings.
 - Rebase on top of v6.9-rc7

Changes in RFC v2 [5]:
 - squash the header file patch into the DT schema patch
 - describe the changes I made to original series in the cover letter
   instead of the individual patches
 - fix my typo in my email address

Changes in RFC v1 [6] from the original series:
 - corrected the npu_clk enable bit
 - deduplicated CLK_NPU and CLK_NPU_AXI number in header
 - fixed c910_i0_clk reg typo
 - fixed checkpatch and dt_binding_check warnings
 - rebased on v6.9-rc5
 - revised commit descriptions

Thank you,
Drew

[1] https://openbeagle.org/beaglev-ahead/beaglev-ahead/-/blob/main/docs/TH1520%20System%20User%20Manual.pdf
[2] https://lore.kernel.org/all/[e-mail]/
[3] https://lore.kernel.org/all/[e-mail]/
[4] https://lore.kernel.org/all/[e-mail]/
[5] https://lore.kernel.org/all/[e-mail]/
[6] https://lore.kernel.org/all/[e-mail]/

To: Jisheng Zhang <[e-mail]>
To: Guo Ren <[e-mail]>
To: Fu Wei <[e-mail]>
To: Yangtao Li <[e-mail]>
To: Thomas Bonnefille <[e-mail]>
To: Emil Renner Berthing <[e-mail]>
To: Michael Turquette <[e-mail]>
To: Stephen Boyd <[e-mail]>
To: Rob Herring <[e-mail]>
To: Krzysztof Kozlowski <[e-mail]>
To: Conor Dooley <[e-mail]>
To: Paul Walmsley <[e-mail]>
To: Palmer Dabbelt <[e-mail]>
To: Albert Ou <[e-mail]>
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]
Cc: [e-mail]

Signed-off-by: Drew Fustini <[e-mail]>
---
Drew Fustini (6):
      dt-bindings: clock: Document T-Head TH1520 AP_SUBSYS controller
      clk: thead: Add support for T-Head TH1520 AP_SUBSYS clocks
      riscv: dts: thead: Add TH1520 AP_SUBSYS clock controller
      riscv: dts: thead: change TH1520 uart nodes to use clock controller
      riscv: dts: thead: change TH1520 mmc nodes to use clock controller
      riscv: dts: thead: update TH1520 dma and timer nodes to use clock controller

 .../bindings/clock/thead,th1520-clk-ap.yaml        |   58 ++
 MAINTAINERS                                        |    3 +
 arch/riscv/boot/dts/thead/th1520-beaglev-ahead.dts |   12 -
 .../boot/dts/thead/th1520-lichee-module-4a.dtsi    |   12 -
 arch/riscv/boot/dts/thead/th1520.dtsi              |   69 +-
 drivers/clk/Kconfig                                |    1 +
 drivers/clk/Makefile                               |    1 +
 drivers/clk/thead/Kconfig                          |   12 +
 drivers/clk/thead/Makefile                         |    2 +
 drivers/clk/thead/clk-th1520-ap.c                  | 1086 ++++++++++++++++++++
 include/dt-bindings/clock/thead,th1520-clk-ap.h    |   96 ++
 11 files changed, 1291 insertions(+), 61 deletions(-)
---
base-commit: 83a7eefedc9b56fe7bfeff13b6c7356688ffa670
change-id: 20240614-th1520-clk-cd585a850d6c

Best regards,

Comments

Conor Dooley June 19, 2024, 11:12 a.m. UTC | #1

On Sat, Jun 15, 2024 at 06:54:29PM -0700, Drew Fustini wrote:
> This series adds support for the AP sub-system clock controller in the
> T-Head TH1520 [1]. Yangtao Li originally submitted this series in May
> 2023 [2]. Jisheng made additional improvements and then passed on the
> work in progress to me.

One thing I noticed on the dts side is that the GPIO controllers have no
clocks provided. Does the AP sub-system clock controller provide their
clocks too?

Drew Fustini June 19, 2024, 6:15 p.m. UTC | #2

On Wed, Jun 19, 2024 at 12:12:30PM +0100, Conor Dooley wrote:
> On Sat, Jun 15, 2024 at 06:54:29PM -0700, Drew Fustini wrote:
> > This series adds support for the AP sub-system clock controller in the
> > T-Head TH1520 [1]. Yangtao Li originally submitted this series in May
> > 2023 [2]. Jisheng made additional improvements and then passed on the
> > work in progress to me.
> 
> One thing I noticed on the dts side is that the GPIO controllers have no
> clocks provided. Does the AP sub-system clock controller provide their
> clocks too?

Good question. I see that dwapb_get_clks() in drivers/gpio/gpio-dwapb.c
does call devm_clk_bulk_get_optional() for "bus" and "db". There doesn't
seem to be to many in-tree examples of clocks being defined for gpio
controllers with compatible "snps,dw-apb-gpio", but I do see that
k210.dtsi defines K210_CLK_APB0 for "bus" and K210_CLK_GPIO for "db".

From the TH1520 System User Manual, I do see the gpio related clocks in
Section 4.4.2.2 AP_SUBSYS. The peripheral clock gate control register
(PERI_CLK_CFG) has:

Bit 20: GPIO3_CLK_EN
Bit  8: GPIO0_CLK_EN
Bit  7: GPIO1_CLK_EN
Bit  6: GPIO2_CLK_EN

I will add these gates to the clk-th1520-ap.c and reference them from
the gpio controller nodes.

Since each gpio controller will only have one clock, do you think I can
omit the clock-names property?

Thanks,
Drew

Link: https://git.beagleboard.org/beaglev-ahead/beaglev-ahead/-/tree/main/docs

Conor Dooley June 20, 2024, 4:30 p.m. UTC | #3

On Wed, Jun 19, 2024 at 11:15:06AM -0700, Drew Fustini wrote:
> On Wed, Jun 19, 2024 at 12:12:30PM +0100, Conor Dooley wrote:
> > On Sat, Jun 15, 2024 at 06:54:29PM -0700, Drew Fustini wrote:
> > > This series adds support for the AP sub-system clock controller in the
> > > T-Head TH1520 [1]. Yangtao Li originally submitted this series in May
> > > 2023 [2]. Jisheng made additional improvements and then passed on the
> > > work in progress to me.
> > 
> > One thing I noticed on the dts side is that the GPIO controllers have no
> > clocks provided. Does the AP sub-system clock controller provide their
> > clocks too?
> 
> Good question. I see that dwapb_get_clks() in drivers/gpio/gpio-dwapb.c
> does call devm_clk_bulk_get_optional() for "bus" and "db". There doesn't
> seem to be to many in-tree examples of clocks being defined for gpio
> controllers with compatible "snps,dw-apb-gpio", but I do see that
> k210.dtsi defines K210_CLK_APB0 for "bus" and K210_CLK_GPIO for "db".
> 
> From the TH1520 System User Manual, I do see the gpio related clocks in
> Section 4.4.2.2 AP_SUBSYS. The peripheral clock gate control register
> (PERI_CLK_CFG) has:
> 
> Bit 20: GPIO3_CLK_EN
> Bit  8: GPIO0_CLK_EN
> Bit  7: GPIO1_CLK_EN
> Bit  6: GPIO2_CLK_EN
> 
> I will add these gates to the clk-th1520-ap.c and reference them from
> the gpio controller nodes.
> 
> Since each gpio controller will only have one clock, do you think I can
> omit the clock-names property?

Sure, thanks for looking into this.

> 
> Thanks,
> Drew
> 
> Link: https://git.beagleboard.org/beaglev-ahead/beaglev-ahead/-/tree/main/docs