In commit 4167ac8a65 ("clk: sunxi-ng: sun50i: h6: Modify GPU clock
configuration to support DFS") divider M0 was forced to be 1 in order to
support DFS. However, that left N as it is, at high value of 36. On
boards without devfreq enabled (all of them in kernel 6.0), this
effectively sets GPU frequency to 864 MHz. This is about 100 MHz above
maximum supported frequency.
In order to fix this, let's set N to 18 (register value 17). That way
default frequency of 432 MHz is preserved.
Fixes: 4167ac8a65 ("clk: sunxi-ng: sun50i: h6: Modify GPU clock configuration to support DFS")
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://lore.kernel.org/r/20220928200122.3963509-1-jernej.skrabec@gmail.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Using simple bash script it was discovered that not all CCU registers
can be safely used for DFS, e.g.:
while true
do
devmem 0x3001030 4 0xb0003e02
devmem 0x3001030 4 0xb0001e02
done
Script above changes the GPU_PLL multiplier register value. While the
script is running, the user should interact with the user interface.
Using this method the following results were obtained:
| Register | Name | Bits | Values | Result |
| -- | -- | -- | -- | -- |
| 0x3001030 | GPU_PLL.MULT | 15..8 | 20-62 | OK |
| 0x3001030 | GPU_PLL.INDIV | 1 | 0-1 | OK |
| 0x3001030 | GPU_PLL.OUTDIV | 0 | 0-1 | FAIL |
| 0x3001670 | GPU_CLK.DIV | 3..0 | ANY | FAIL |
DVFS started to work seamlessly once dividers which caused the
glitches were set to fixed values.
Signed-off-by: Roman Stratiienko <r.stratiienko@gmail.com>
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://lore.kernel.org/r/20220705075226.359475-1-r.stratiienko@gmail.com
While it is useful to build all of the CCU drivers at once, only 1-3 of
them will be loaded at a time, or possibly none of them if the kernel is
booted on a non-sunxi platform. These CCU drivers are relatively large;
32-bit drivers have 30-50k of data each, while the 64-bit ones are
50-75k due to the increased pointer overhead. About half of that data
comes from relocations. Let's allow the user to build these drivers as
modules so only the necessary data is loaded.
As a first step, convert the CCUs that are already platform drivers.
When the drivers are built as modules, normally the file name becomes
the module name. However, the current file names are inconsistent with
the <platform>-<peripheral> name used everywhere else: the devicetree
bindings, the platform driver names, and the Kconfig symbols. Use
Makfile logic to rename the modules so they follow the usual pattern.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20211119033338.25486-3-samuel@sholland.org
Use the devm_platform_ioremap_resource() helper instead of
calling platform_get_resource() and devm_ioremap_resource()
separately
Signed-off-by: Cai Huoqing <caihuoqing@baidu.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20210907085206.4611-1-caihuoqing@baidu.com
The CCU drivers are not really designed to be unbound. Unbinding a SoC's
main CCU is especially pointless, as very few of the peripherals on the
SoC will work without it. Let's avoid any potential problems by removing
the bind/unbind attributes from sysfs for these drivers.
This change is not applied to the "secondary" CCUs (DE, USB) as those
could reasonably be unbound without making the system useless.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20210901050526.45673-3-samuel@sholland.org
Currently, unbinding a CCU driver unmaps the device's MMIO region, while
leaving its clocks/resets and their providers registered. This can cause
a page fault later when some clock operation tries to perform MMIO. Fix
this by separating the CCU initialization from the memory allocation,
and then using a devres callback to unregister the clocks and resets.
This also fixes a memory leak of the `struct ccu_reset`, and uses the
correct owner (the specific platform driver) for the clocks and resets.
Early OF clock providers are never unregistered, and limited error
handling is possible, so they are mostly unchanged. The error reporting
is made more consistent by moving the message inside of_sunxi_ccu_probe.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20210901050526.45673-2-samuel@sholland.org
While comparing clocks between the H6 and H616, some of the M factor
ranges were found to be wrong: the manual says they are only covering
two bits [1:0], but our code had "5" in the number-of-bits field.
By writing 0xff into that register in U-Boot and via FEL, it could be
confirmed that bits [4:2] are indeed masked off, so the manual is right.
Change to number of bits in the affected clock's description.
Fixes: 524353ea48 ("clk: sunxi-ng: add support for the Allwinner H6 CCU")
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20210118000912.28116-1-andre.przywara@arm.com
The CEC clock on the H6 SoC is a bit special, since it uses a fixed
pre-dividier for one source clock (the PLL), but conveys the other clock
(32K OSC) directly.
We are using a fixed predivider array for that, but fail to use the right
flag to actually activate that.
Fixes: 524353ea48 ("clk: sunxi-ng: add support for the Allwinner H6 CCU")
Reported-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20210106143246.11255-1-andre.przywara@arm.com
GPU PLL was designed with dynamic frequency switching in mind so driver
can adjust rate based on the GPU load.
Allow GPU clock to change parent rate (GPU PLL is the only possible
parent of GPU clock).
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Audio devices needs exact clock rates in order to correctly reproduce
the sound. Until now, only integer factors were used to configure H6
audio PLL which resulted in inexact rates. Fix that by adding support
for fractional factors using sigma-delta modulation look-up table. It
contains values for two most commonly used audio base frequencies.
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Acked-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
I2S doesn't work if parent rate couldn't be change. Difference between
wanted and actual rate is too big.
Fix this by adding CLK_SET_RATE_PARENT flag to I2S clocks.
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Marcus Cooper <codekipper@gmail.com>
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
With the new clk parenting code and CLK_FIXED_FACTOR_{HW,FW_NAME}
macros, we can reference parents locally via pointers to struct clk_hw
or DT clock-names.
Convert existing CLK_FIXED_FACTOR definitions to either the _HW or
_FW_NAME variant based on whether the parent clock is internal or
external to the CCU.
A forward declaration for struct clk_fixed_factor pll_periph0_4x_clk
is added as the definitions of the fixed factor clocks appear much later
in the file. The position of fixed factor clock definitions will be
moved for all drivers at a later time, before the conversion of all
other clock types.
Acked-by: Maxime Ripard <maxime.ripard@bootlin.com>
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Video related clocks need to set rate as close as possible to the
requested one, so they should be able to change parent clock rate.
When processing 4K video, VPU clock has to be set to higher rate than it
is default parent rate. Because of that, VPU clock should be able to
change parent clock rate.
Add CLK_SET_RATE_PARENT flag to tcon-lcd0, tcon-tv0 and ve.
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
H6 manual and BSP clock driver both states that hdmi-cec clock has two
possible parents, osc32k and pll-periph0-2x with 36621 predivider.
Because pll-periph0-2x is always 1.2 GHz, both parents give same
hdmi-cec rate - 32768 Hz, which is exactly the rate needed for HDMI CEC
controller to operate correctly.
However, for some reason, HDMI CEC controller doesn't work if default
parent (osc32k) is used. BSP HDMI driver also always use pll-periph0-2x
as hdmi-cec clock parent.
In order to solve the issue, preset hdmi-cec clock parent to
pll-periph0-2x.
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
DE2/DE3 mixers have to run at specific frequency in order to work
optimally. This wasn't actually possible for some SoCs because "de"
clock wasn't allowed to adjust parent rate.
Add CLK_SET_RATE_PARENT flag to all "de" clocks which didn't have it
yet.
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
Video PLL factors can be set in a way that final PLL rate is outside
stable range. H6 user manual specifically says that N factor should not
be below 12. While it doesn't says anything about maximum stable rate, it
is clear that PLL doesn't work at 6.096 GHz (254 * 24 MHz).
Set minimum allowed PLL video rate to 288 MHz (12 * 24 MHz) and maximum
to 2.4 GHz, which is maximum in BSP driver.
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
MUX bits for MMC clock register range are 25:24 where 24 is shift
and 2 is width So fix the width number from 3 to 2.
Fixes: 524353ea48 ("clk: sunxi-ng: add support for the Allwinner H6 CCU")
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
On the H6, the MMC module clocks are fixed in the new timing mode,
i.e. they do not have a bit to select the mode. These clocks have
a 2x divider somewhere between the clock and the MMC module.
To be consistent with other SoCs supporting the new timing mode,
we model the 2x divider as a fixed post-divider on the MMC module
clocks.
This patch adds the post-dividers to the MMC clocks, following the
approach on A64.
Fixes: 524353ea48 ("clk: sunxi-ng: add support for the Allwinner H6 CCU")
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
Currently the register offset of the PWM bus gate in Allwinner H6 clock
driver is wrong.
Fix this issue.
Fixes: 542353ea ("clk: sunxi-ng: add support for the Allwinner H6 CCU")
Signed-off-by: Rongyi Chen <chenyi@tt-cool.com>
[Icenowy: refactor commit message]
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
The bus clocks (AHB/APB) on Allwinner H6 have their second divider start
at bit 8, according to the user manual and the BSP code. However,
currently the divider offset is incorrectly set to 16, thus the divider
is not correctly read and the clock frequency is not correctly calculated.
Fix this bit offset on all affected bus clocks in ccu-sun50i-h6.
Cc: stable@vger.kernel.org # v4.17.y
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
The Allwinner H6 CCU has a "HDMI Slow Clock", which is currently missing
in the ccu-sun50i-h6 driver.
Add this missing clock to the driver.
Fixes: 542353ea ("clk: sunxi-ng: add support for the Allwinner H6 CCU")
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
The Allwinner H6 SoC has a CCU which has been largely rearranged.
Add support for it in the sunxi-ng CCU framework.
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Acked-by: Maxime Ripard <maxime.ripard@bootlin.com>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
