Not all u-boot versions initialize the HHI_GP_PLL_CNTL[2-5] registers.
In that case all HHI_GPLL_PLL_CNTL[1-5] registers are 0x0 and when
booting Linux the PLL fails to lock.
The initialization sequence from u-boot is:
- put the PLL into reset
- write 0x59C88000 to HHI_GP_PLL_CNTL2
- write 0xCA463823 to HHI_GP_PLL_CNTL3
- write 0x0286A027 to HHI_GP_PLL_CNTL4
- write 0x00003000 to HHI_GP_PLL_CNTL5
- set M, N, OD and the enable bit
- take the PLL out of reset
- check if it has locked
- disable the PLL
In Linux we already initialize M, N, OD, the enable and the reset bits.
Also the HHI_GP_PLL_CNTL[2-5] registers with these magic values (the
exact meaning is unknown) so the PLL can lock when the vendor u-boot did
not initialize these registers yet.
Fixes: b882964b37 ("clk: meson: meson8b: add support for the GP_PLL clock on Meson8m2")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200501215717.735393-1-martin.blumenstingl@googlemail.com
The "vpu_0" or "vpu_1" clock trees should not be updated while the
clock is running. Enforce this by setting CLK_SET_RATE_GATE on the
"vpu_0" and "vpu_1" gates. This makes the CCF switch to the "vpu_1"
tree when "vpu_0" is currently active and vice versa, which is exactly
what the vendor driver does when updating the frequency of the VPU
clock.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200417184127.1319871-5-martin.blumenstingl@googlemail.com
The DIV{1,2,4,6,12}_EN bits are actually located in HHI_VID_CLK_CNTL
register:
- HHI_VID_CLK_CNTL[0] = DIV1_EN
- HHI_VID_CLK_CNTL[1] = DIV2_EN
- HHI_VID_CLK_CNTL[2] = DIV4_EN
- HHI_VID_CLK_CNTL[3] = DIV6_EN
- HHI_VID_CLK_CNTL[4] = DIV12_EN
Update the bits accordingly so we will enable the bits in the correct
register once we switch these clocks to be mutable.
Fixes: 6cb57c678b ("clk: meson: meson8b: add the read-only video clock trees")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200417184127.1319871-4-martin.blumenstingl@googlemail.com
CLKC_RESET_VID_DIVIDER_CNTL_RESET_N_POST and
CLKC_RESET_VID_DIVIDER_CNTL_RESET_N_PRE are active low. This means:
- asserting them requires setting the register value to 0
- de-asserting them requires setting the register value to 1
Set the register value accordingly for these two reset lines by setting
the inverted the register value compared to all other reset lines.
Fixes: 189621726b ("clk: meson: meson8b: register the built-in reset controller")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200417184127.1319871-3-martin.blumenstingl@googlemail.com
Use hdmi_pll_lvds_out as parent of the vid_pll_in_sel clock. It's not
easy to see that the vendor kernel does the same, but it actually does.
meson_clk_pll_ops in mainline still cannot fully recalculate all rates
from the HDMI PLL registers because some register bits (at the time of
writing it's unknown which bits are used for this) double the HDMI PLL
output rate (compared to simply considering M, N and FRAC) for some (but
not all) PLL settings.
Update the vid_pll_in_sel parent so our clock calculation works for
simple clock settings like the CVBS output (where no rate doubling is
going on). The PLL ops need to be fixed later on for more complex clock
settings (all HDMI rates).
Fixes: 6cb57c678b ("clk: meson: meson8b: add the read-only video clock trees")
Suggested-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200417184127.1319871-2-martin.blumenstingl@googlemail.com
The "mali_0" or "mali_1" clock trees should not be updated while the
clock is running. Enforce this by setting CLK_SET_RATE_GATE on the
"mali_0" and "mali_1" gates. This makes the CCF switch to the "mali_1"
tree when "mali_0" is currently active and vice versa, which is exactly
what the vendor driver does when updating the frequency of the mali
clock.
Also propagate set_rate requests from the gate to the divider and from
the divider to the the mux so the GPU clock frequency can be updated at
runtime (which will be required for GPU DVFS). Don't propagate rate
changes to the mux parents because we don't want to change the MPLL
clocks (these are reserved for audio).
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://lore.kernel.org/r/20200414195031.224021-3-martin.blumenstingl@googlemail.com
The "mali_0" or "mali_1" clock trees should not be updated while the
clock is running. Enforce this by setting CLK_SET_RATE_GATE on the
"mali_0" and "mali_1" gates. This makes the CCF switch to the "mali_1"
tree when "mali_0" is currently active and vice versa, which is exactly
what the vendor driver does when updating the frequency of the mali
clock.
Also propagate set_rate requests from the gate to the divider and from
the divider to the the mux so the GPU clock frequency can be updated at
runtime (which will be required for GPU DVFS). Don't propagate rate
changes to the mux parents because we don't want to change the MPLL
clocks (these are reserved for audio).
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://lore.kernel.org/r/20200414195031.224021-2-martin.blumenstingl@googlemail.com
The HDMI TX controller requires the hdmi_sys clock to be enabled. Allow
changing the whole clock tree now that we know that one of our drivers
requires this.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200330234535.3327513-3-martin.blumenstingl@googlemail.com
Export the HDMI system clock (used by the HDMI transmitter) so it can be
used in the dt-bindings.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Acked-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20200330234535.3327513-2-martin.blumenstingl@googlemail.com
The aiu devices peripheral clocks needs the aiu and aiu_glue clocks to
operate. Reflect this hierarchy in the clock tree.
Fixes: e31a1900c1 ("meson: clk: Add support for clock gates")
Suggested-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
This adds the clocks used for the Amlogic G12A and compatible SoCs SPICC
controller to provide a more complete range of frequencies instead of the
SPICC internal divider over Xtal.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The aiu devices peripheral clocks needs the aiu and aiu_glue clocks to
operate. Reflect this hierarchy in the gxbb clock tree.
Fixes: 738f66d321 ("clk: gxbb: add AmLogic GXBB clk controller driver")
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Add the ACODEC clock gate to the gxl clk controller driver
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
- Support dangerous debugfs actions on clks with dead code
- Convert gpio, fixed-factor, mux, gate, divider basic clks to hw based APIs
* clk-debugfs-danger:
clk: Add support for setting clk_rate via debugfs
* clk-basic-hw:
clk: divider: Add support for specifying parents via DT/pointers
clk: gate: Add support for specifying parents via DT/pointers
clk: mux: Add support for specifying parents via DT/pointers
clk: asm9260: Use parent accuracy in fixed rate clk
clk: fixed-rate: Document that accuracy isn't a rate
clk: fixed-rate: Add clk flags for parent accuracy
clk: fixed-rate: Add support for specifying parents via DT/pointers
clk: fixed-rate: Document accuracy member
clk: fixed-rate: Move to_clk_fixed_rate() to C file
clk: fixed-rate: Remove clk_register_fixed_rate_with_accuracy()
clk: fixed-rate: Convert to clk_hw based APIs
clk: gpio: Use DT way of specifying parents
* clk-renesas:
clk: renesas: Prepare for split of R-Car H3 config symbol
dt-bindings: clock: renesas: cpg-mssr: Fix r8a774b1 typo
clk: renesas: r7s9210: Add SPIBSC clock
clk: renesas: rcar-gen3: Allow changing the RPC[D2] clocks
clk: renesas: Remove use of ARCH_R8A7796
clk: renesas: rcar-gen2: Change multipliers and dividers to u8
* clk-amlogic:
clk: clarify that clk_set_rate() does updates from top to bottom
clk: meson: meson8b: make the CCF use the glitch-free mali mux
clk: meson: pll: Fix by 0 division in __pll_params_to_rate()
clk: meson: g12a: fix missing uart2 in regmap table
clk: meson: meson8b: use of_clk_hw_register to register the clocks
clk: meson: meson8b: don't register the XTAL clock when provided via OF
clk: meson: meson8b: change references to the XTAL clock to use [fw_]name
clk: meson: meson8b: use clk_hw_set_parent in the CPU clock notifier
clk: meson: add a driver for the Meson8/8b/8m2 DDR clock controller
dt-bindings: clock: meson8b: add the clock inputs
dt-bindings: clock: add the Amlogic Meson8 DDR clock controller binding
* clk-allwinner:
clk: sunxi: a23/a33: Export the MIPI PLL
clk: sunxi: a31: Export the MIPI PLL
clk: sunxi-ng: a64: export CLK_CPUX clock for DVFS
clk: sunxi-ng: add mux and pll notifiers for A64 CPU clock
clk: sunxi-ng: r40: Export MBUS clock
clk: sunxi: use of_device_get_match_data
The "mali_0" or "mali_1" clock trees should not be updated while the
clock is running. Enforce this by setting CLK_SET_RATE_GATE on the
"mali_0" and "mali_1" gates. This makes the CCF switch to the "mali_1"
tree when "mali_0" is currently active and vice versa, which is exactly
what the vendor driver does when updating the frequency of the mali
clock.
This fixes a potential hang when changing the GPU frequency at runtime.
Fixes: 74e1f2521f ("clk: meson: meson8b: add the GPU clock tree")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
If the init callback is allowed to request resources, it needs a return
value to report the outcome of such a request.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lkml.kernel.org/r/20190924123954.31561-3-jbrunet@baylibre.com
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Some meson pll registers can be initialized with 0 as N value, introducing
the following division by 0 when computing rate :
UBSAN: Undefined behaviour in drivers/clk/meson/clk-pll.c:75:9
division by zero
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.4.0-rc3-608075-g86c9af8630e1-dirty #400
Call trace:
dump_backtrace+0x0/0x1c0
show_stack+0x14/0x20
dump_stack+0xc4/0x100
ubsan_epilogue+0x14/0x68
__ubsan_handle_divrem_overflow+0x98/0xb8
__pll_params_to_rate+0xdc/0x140
meson_clk_pll_recalc_rate+0x278/0x3a0
__clk_register+0x7c8/0xbb0
devm_clk_hw_register+0x54/0xc0
meson_eeclkc_probe+0xf4/0x1a0
platform_drv_probe+0x54/0xd8
really_probe+0x16c/0x438
driver_probe_device+0xb0/0xf0
device_driver_attach+0x94/0xa0
__driver_attach+0x70/0x108
bus_for_each_dev+0xd8/0x128
driver_attach+0x30/0x40
bus_add_driver+0x1b0/0x2d8
driver_register+0xbc/0x1d0
__platform_driver_register+0x78/0x88
axg_driver_init+0x18/0x20
do_one_initcall+0xc8/0x24c
kernel_init_freeable+0x2b0/0x344
kernel_init+0x10/0x128
ret_from_fork+0x10/0x18
This checks if N is null before doing the division.
Fixes: 7a29a86943 ("clk: meson: Add support for Meson clock controller")
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Remi Pommarel <repk@triplefau.lt>
[jbrunet@baylibre.com: update the comment in above the fix]
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
UART2 peripheral is missing from the regmap fixup table of the g12a family
clock controller. As it is, any access to this clock would Oops, which is
not great.
Add the clock to the table to fix the problem.
Fixes: 085a4ea93d ("clk: meson: g12a: add peripheral clock controller")
Reported-by: Dmitry Shmidt <dimitrysh@google.com>
Tested-by: Dmitry Shmidt <dimitrysh@google.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Switch from clk_hw_register to of_clk_hw_register so we can use
clk_parent_data.fw_name. This will be used to get the "xtal", "ddr_pll"
and possibly others from the .dtb.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The XTAL clock is an actual crystal on the PCB. Thus the meson8b clock
driver should not register the XTAL clock - instead it should be
provided via .dts and then passed to the clock controller.
Skip the registration of the XTAL clock if a parent clock is provided
via OF. Fall back to registering the XTAL clock if this is not the case
to keep support for old .dtbs.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The XTAL clock is an actual crystal which is mounted on the PCB. Thus
the meson8b clock controller driver should not provide the XTAL clock.
The meson8b clock controller driver must not use references to
the meson8b_xtal clock anymore before we can provide the XTAL clock
via OF. Replace the references to the meson8b_xtal.hw by using
clk_parent_data's .fw_name and .name = "xtal" (along with index = -1).
This makes the common clock framework use the clock provided via OF and
if that's not available it falls back to getting the clock by it's name
(which is then the clk_fixed_rate which we register in our driver).
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Switch from clk_set_parent() to clk_hw_set_parent() now that we have a
way to configure a mux clock based on clk_hw pointers. This simplifies
the meson8b_cpu_clk_notifier_cb logic. No functional changes.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The Meson8/Meson8b/Meson8m2 SoCs embed a DDR clock controller in the
MMCBUS registers. There is no public documentation, but the u-boot GPL
sources from the Amlogic BSP show that the DDR clock controller is
identical on all three SoCs:
#define CFG_DDR_CLK 792
#define CFG_PLL_M (((CFG_DDR_CLK/12)*12)/24)
#define CFG_PLL_N 1
#define CFG_PLL_OD 1
// from set_ddr_clock:
t_ddr_pll_cntl= (CFG_PLL_OD << 16)|(CFG_PLL_N<<9)|(CFG_PLL_M<<0)
writel(timing_reg->t_ddr_pll_cntl|(1<<29),AM_DDR_PLL_CNTL);
writel(readl(AM_DDR_PLL_CNTL) & (~(1<<29)),AM_DDR_PLL_CNTL);
// from hx_ddr_power_down_enter: shut down DDR PLL
writel(readl(AM_DDR_PLL_CNTL)|(1<<30),AM_DDR_PLL_CNTL);
do { ... } while((readl(AM_DDR_PLL_CNTL)&(1<<31))==0)
This translates to:
- AM_DDR_PLL_CNTL[29] is the reset bit
- AM_DDR_PLL_CNTL[30] is the enable bit
- AM_DDR_PLL_CNTL[31] is the lock bit
- AM_DDR_PLL_CNTL[8:0] is the m value (assuming the width is 9 bits
based on the start of the n value)
- AM_DDR_PLL_CNTL[13:9] is the n value (assuming the width is 5 bits
based on the start of the od)
- AM_DDR_PLL_CNTL[17:16] is the od (assuming the width is 2 bits based
on other PLLs on this SoC)
Add a driver for this PLL setup because it's used as one of the inputs
of the audio clocks. There may be more clocks inside that clock
controller - those can be added in subsequent patches.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Use devm_platform_ioremap_resource() to simplify the code a bit.
This is detected by coccinelle.
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Add sm1 support the axg audio clock controllers. This new version is
indeed derived from the previous generation, as always, adding a few
new clocks to the mix.
The number of gates now exceeds 32 and do not fit in a single register.
Unfortunately, designers chose to introduce the new gate register
immediately after the original one, at the beginning of the register
space, shifting all the master clock register offsets.
The sm1 also introduce a few mux and divider on the top clock path,
possibly to lower the peripheral clocks of the audio blocks if
necessary.
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The peripheral clock on the sm1 goes through some muxes
and dividers before reaching the audio gates. To model that,
without repeating our self too much, the "top" clock signal
is introduced and will serve as a the parent of the gates.
On the axg and g12a, the top clock is just a pass-through to
the audio peripheral clock provided by the main controller.
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Rearrange the macro definition of the clocks of the axg-audio
controller. Also, the sm1 variant will feature tiny modification
of different blocks in this controller. Because of that, we need
to fallback to the old way of defining parent for some of the
clocks, using signal name.
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Since the addition of the g12a, the last register is
AUDIO_CLK_SPDIFOUT_B_CTRL.
Fixes: 075001385c ("clk: meson: axg-audio: add g12a support")
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Defining the number of each input type is no longer necessary since
we are not using the clk-input hack anymore
Fixes: 282420eed2 ("clk: meson: axg-audio: migrate to the new parent description method")
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
When setting the 100MHz, 500MHz, 666MHz and 1GHz rate for CPU clocks,
CCF will use the SYS_PLL to handle these frequencies, but:
- using FIXED_PLL derived FCLK_DIV2/DIV3 clocks is more precise
- the Amlogic G12A/G12B/SM1 Suspend handling in firmware doesn't
handle entering suspend using SYS_PLL for these frequencies
Adding CLK_MUX_ROUND_CLOSEST on all the muxes of the non-SYS_PLL
cpu clock tree helps CCF always selecting the FCLK_DIV2/DIV3 as source
for these frequencies.
Fixes: ffae8475b9 ("clk: meson: g12a: add notifiers to handle cpu clock change")
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
CLK_SET_RATE_NO_REPARENT is wrongly set on the g12a cpu premux0 clocks
flags, and CLK_SET_RATE_PARENT is required for the g12a cpu premux0 clock
and the g12b cpub premux0 clock, otherwise CCF always selects the SYS_PLL
clock to feed the cpu cluster.
Fixes: ffae8475b9 ("clk: meson: g12a: add notifiers to handle cpu clock change")
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The meson-saradc driver manually sets the input clock for
sar_adc_clk_sel. Update the GXBB clock driver (which is used on GXBB,
GXL and GXM) so the rate settings on sar_adc_clk_div are propagated up
to sar_adc_clk_sel which will let the common clock framework select the
best matching parent clock if we want that.
This makes sar_adc_clk_div consistent with the axg-aoclk and g12a-aoclk
drivers, which both also specify CLK_SET_RATE_PARENT.
Fixes: 33d0fcdfe0 ("clk: gxbb: add the SAR ADC clocks and expose them")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The Amlogic SM1 can set a dedicated clock frequency for each CPU core by
having a dedicate tree for each core similar to the CPU0 tree.
Like the DSU tree, a supplementaty mux has been added to use the CPU0
frequency instead.
But since the cluster only has a single power rail and shares a single PLL,
it's not worth adding 3 unsused clock tree, so we add only the mux to
select the CPU0 clock frequency for each CPU1, CPU2 and CPU3 cores.
They are set read-only because the early boot stages sets them to select
the CPU0 input clock.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The Amlogic SM1 DynamIQ Shared Unit has a dedicated clock tree similar to
the CPU clock tree with a supplementaty mux to select the CPU0 clock
instead.
Leave this as read-only since it's set up by the early boot stages.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Add the new GP1 PLL for the Amlogic SM1 SoC, used to feed the new
DynamIQ Shared Unit of the ARM Cores Complex.
This also adds a dedicated set of clock and compatible for SM1.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
On the g12a, the register space dedicated to the audio clock also
provides some resets. Let the clock controller register a reset
provider as well for this SoC family.
the axg SoC family does not appear to provide this feature.
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
A future patch is going to change semantics of clk_register() so that
clk_hw::init is guaranteed to be NULL after a clk is registered. Avoid
referencing this member here so that we don't run into NULL pointer
exceptions.
Cc: Neil Armstrong <narmstrong@baylibre.com>
Cc: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Link: https://lkml.kernel.org/r/20190731193517.237136-4-sboyd@kernel.org
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Expose the CPUB clock id to add DVFS to the second CPU cluster of
the Amlogic G12B SoC.
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
In order to implement clock switching for the CLKID_CPU_CLK and
CLKID_CPUB_CLK, notifiers are added on specific points of the
clock tree :
cpu_clk / cpub_clk
| \- cpu_clk_dyn
| | \- cpu_clk_premux0
| | |- cpu_clk_postmux0
| | | |- cpu_clk_dyn0_div
| | | \- xtal/fclk_div2/fclk_div3
| | \- xtal/fclk_div2/fclk_div3
| \- cpu_clk_premux1
| |- cpu_clk_postmux1
| | |- cpu_clk_dyn1_div
| | \- xtal/fclk_div2/fclk_div3
| \- xtal/fclk_div2/fclk_div3
\ sys_pll / sys1_pll
This for each cluster, a single one for G12A, two for G12B.
Each cpu_clk_premux1 tree is marked as read-only and CLK_SET_RATE_NO_REPARENT,
to be used as "parking" clock in a safe clock frequency.
A notifier is added on each cpu_clk_premux0 to detech when CCF want to
change the frequency of the cpu_clk_dyn tree.
In this notifier, the cpu_clk_premux1 tree is configured to use the xtal
clock and then the cpu_clk_dyn is switch to cpu_clk_premux1 while CCF
updates the cpu_clk_premux0 tree.
A notifier is added on each sys_pll/sys1_pll to detect when CCF wants to
change the PLL clock source of the cpu_clk.
In this notifier, the cpu_clk is switched to cpu_clk_dyn while CCF
updates the sys_pll/sys1_pll frequency.
A third small notifier is added on each cpu_clk / cpub_clk and cpu_clk_dyn,
add a small delay at PRE_RATE_CHANGE/POST_RATE_CHANGE to let the other
notofiers change propagate before changing the cpu_clk_premux0 and sys_pll
clock trees.
This notifier set permits switching the cpu_clk / cpub_clk without any
glitches and using a safe parking clock while switching between sub-GHz
clocks using the cpu_clk_dyn tree.
This setup has been tested and validated on the Amlogic G12A and G12B
SoCs running the arm64 cpuburn at [1] and cycling between all the possible
cpufreq translations of each cluster and checking the final frequency using
the clock-measurer, script at [2].
[1] https://github.com/ssvb/cpuburn-arm/blob/master/cpuburn-a53.S
[2] https://gist.github.com/superna9999/d4de964dbc0f84b7d527e1df2ddea25f
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Add a clock driver for the cpu dynamic divider, this divider needs
to have a flag set before setting the divider value then removed
while writing the new value to the register.
This drivers implements this behavior and will be used essentially
on the Amlogic G12A and G12B SoCs for cpu clock trees.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
The clk input function which allows clock controllers to register a bypass
clock from a clock producer is no longer needed anymore since meson clock
controllers have migrated to a new parent allocation method.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
During probe, bypass clocks (i.e. ee-in-xtal) are made from device-tree
inputs to provide input clocks which can be access through global name.
The cons of this method are the duplicated clocks, means more string
comparison.
Specify parent directly with device-tree clock name.
Remove the bypass clock registration from the ee probe function.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
This clock controller use the string comparison method to describe parent
relation between the clocks, which is not optimized.
Migrate to the new way by using .parent_hws where possible (ie. when
all clocks are local to the controller) and use .parent_data otherwise.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
This clock controller use the string comparison method to describe parent
relation between the clocks, which is not optimized.
Migrate to the new way by using .parent_hws where possible (ie. when
all clocks are local to the controller) and use .parent_data otherwise.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Tested-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
This clock controller use the string comparison method to describe parent
relation between the clocks, which is not optimized.
Migrate to the new way by using .parent_hws where possible (ie. when
all clocks are local to the controller) and use .parent_data otherwise.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
This clock controller use the string comparison method to describe parent
relation between the clocks, which is not optimized.
Migrate to the new way by using .parent_hws where possible (ie. when
all clocks are local to the controller) and use .parent_data otherwise.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
This clock controller use the string comparison method to describe parent
relation between the clocks, which is not optimized.
Migrate to the new way by using .parent_hws where possible (ie. when
all clocks are local to the controller) and use .parent_data otherwise.
Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
