STM32h7 pllout frequency calculation overflows. In the
worst case pllsrc_freq can be 50Mhz and plln_mul 512 which will cause
an overflow of the intermediate result which leads to wrong frequency
returned. As no intermediate result can be bigger than 960MHz only the
order of operations is changed.
Signed-off-by: Benjamin Bigler <benjamin.bigler@securiton.ch>
The regulator voltage should be set before the clocks are enabled.
This is especially the case when the MSIS at 48MHz is selected as
SYSCLK.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
Add the "zephyr/" prefix to various #include statements that are
preventing the CI form running with LEGACY_INCLUDE_PATH=n.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
MSIS frequency at boot time can be different from the one we intent to
set from device tree configuration.
In order to avoid issues, read MSIS configuration from registers to get
the actual freq rather than the devicetree one which may be not yet
configured (which is the case at startup).
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
When configuring PLL, we should first make sure we're not running on PLL,
and if running on PLL, first switch to a fixed clock before proceeding
with PLL configuration.
Current code is doing the switch systematically which is not useful as
default startup case is to use MSI as sysclk source.
So add a test before doing this switch.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
Unlike stated in 57df225b396e43358aac4cc998ed2e99fdb57780, RM0456.pdf
reference manual mentions about PLL1R that "Only division by 1 and even
division factors are allowed."
Though, in reference manual, there is one issue on PLL1R values
description, which should actually be:
0000000: pll1_r_ck = vco1_ck
0000001: pll1_r_ck = vco1_ck / 2 (default after reset)
0000010: Not allowed
0000011: pll1_r_ck = vco1_ck / 4
...
This description will be fixed.
Reflect this in binding and driver.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
This power down bit of eSPI module is loacted in the bit 7 of PWDWN_CTL6
register rather than the bit 4. This commit fixs the incorrect setting.
Signed-off-by: Jun Lin <CHLin56@nuvoton.com>
AST10x0 series SOCs provide the clock controller through the syscon
hardware block. The current driver supports the clock gating capability
for the hardware IPs embedded in the SOC. Each clock source has a
clock ID that can simply map to a bit in syscon registers CLK_STOP_CTRL0
(group 0) or CLK_STOP_CTRL1 (group 1). There are some clock sources
that don't have associated clock gating control, which are always on,
are grouped to into group 2.
Signed-off-by: Dylan Hung <dylan_hung@aspeedtech.com>
Some Series were calculating the pll output frequency from an
clock source index instead of the clock source frequency.
This commit resolves this issue for l0, l1.
get_pllout_frequency() is only used for PLLCLK, therefore remove it.
F2, F4, and F7 have several pll dividers and might decide to implement
these as clock sources won't need PLLCLK.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
Update clock_control drivers to use DT_HAS_<compat>_ENABLED Kconfig
symbol to expose the driver and enable it by default based on
devicetree.
We remove 'depend on' Kconfig for symbols that would be implied by
the devicetree node existing.
Signed-off-by: Kumar Gala <galak@kernel.org>
The KW40Z/KW41Z don't have a Fixed Freq MCG clock so the code
associated with that in get_rate fails to build. ifdef around
the code with enum kCLOCK_McgFixedFreqClk so things work
correctly.
Signed-off-by: Kumar Gala <galak@kernel.org>
The ESP32 clock control for ESP32C3 was unified into a single
driver a while back. However the files associated with the
ESP32C3 didn't get removed than. Remove them now.
Signed-off-by: Kumar Gala <galak@kernel.org>
This commit adds required code to setup pll2 and pll3 as defined
in dts. Also these plls can now be used as alternate clock sources.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
This commit changes the range for stm32u5 pll divider values
to allow divider value of 1.
- DIVQ is allowed to beconfigured 1 for all PLL instances
- DIVR can be 1 for PLL2 and PLL3, but is not valid for PLl1.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
Any project with Kconfig option CONFIG_LEGACY_INCLUDE_PATH set to n
couldn't be built because some files were missing zephyr/ prefix in
includes
Re-run the migrate_includes.py script to fix all legacy include paths
Signed-off-by: Tomislav Milkovic <milkovic@byte-lab.com>
irq_lock() returns an unsigned integer key.
Generated by spatch using semantic patch
scripts/coccinelle/irq_lock.cocci
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Add simple clock control node in devicetree for beetle to handle
relationship between drivers (uart, timers, gpio) and clock controller
device.
Signed-off-by: Kumar Gala <galak@kernel.org>
Some stm32 devices, like stm32F4, do not have
a PLL Enable bit on the PLLP nor PLLQ divider
in their PLL config register (PLLCFGR).
The result is a empty function.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
The function get_pllsrc_frequency is added
with possible source HSI or HSE.
With STM32_PLL_P_ENABLED or STM32_PLL_Q_ENABLED,
this function is used by the clock control driver.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
Following zephyr's style guideline, all if statements, including single
line statements shall have braces.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
After replacing STM32_SRC_PLLCLK by the STM32_SRC_PLL_x sources
this function is no longer needed and are therefore removed.
Also, those functions returned a wrong frequency.
They should have used get_pllsrc_frequency() instead of get_pll_source()
to get the input frequency.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
The SOC specific implementations of the clock_stm32_ll_common driver
included the PLL specific functions only when PLL was selected as sysclock.
This commit changes the condition from "STM32_SYSCLK_SRC_PLL"
to "defined(STM32_PLL_ENABLED)".
As a result the pll could also be used as peripheral clock source
in case it is not the sysclock.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
This commit adds support to select pll outputs as peripheral clock
sources to the stm32 common driver.
With this commit they are only available on
STM32G0, STM32G4, STM32L4, STM32L5, STM32WB, and STM32WL.
Support for STM32F2, and STM32F4, which also have p,q,r dividers,
is not enabled in this commit.
Also, stm32_clock_control_get_subsys_rate is extended to return
the configured frequency in case they are enabled, otherwise 0.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
This commit configures the PLL_P divider for SOCs compatible to the
stm32_ll_common driver in case a value was defined via a dts property.
Additionally, in case the divider value is defined in the device
tree, the respective pll output is enabled during initialization
in set_up_plls().
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
In the stm32_ll_common driver the PLL_Q divider should not
be directly written to RCC->PLLCFGR->PLLQ, but should be
translated to the matching register value.
i.e. shifting the value to the correct position of the register
is not enough.
This commit makes sure the divider value taken from device tree, is
correctly translated ot the matching register register value by
converting it to a STM32CUBE LL definition LL_RCC_PLLQ_DIV_xx.
Typcial divider to register value mapping:
G0, WB, WL:
Dividers 2-8 mapped to register values 1-7(0 reserved).
G4, L4, L5:
Dividers 2,4,6,8 mapped to reg values 0,1,2,3.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
All these series share the same defines, and while they are not used
by all socs of the common-driver, this is not exptectd to lead to
any conflict.
By moving the defines they can also be used in clock_stm32_ll_common.c
Additionally, the stm32g0 pll_div define was renamed to pllm
in order to match the other series.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
The STM32_SRC_CLOCK_MIN and STM32_SRC_CLOCK_MAX defines
are not really needed because non valid clock sources are already
filtered out by the precompiler.
Only STM32_SRC_CLOCK_MIN was used once in code and can be replaced.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
This commit rewrite renesas R-Car clock driver in order
to be able to support any new SoC easier.
This work is so creating a clock driver per soc alongside a
common driver for all reneasas r-car boars.
- drivers: create a driver per soc
- create a common driver
- create a common header used by soc & common driver
- create a soc specific driver calling for common driver
- dts: use new compatible
- use old yaml as common yaml
- create a new "child" yaml to define the new compatible field
- change compatible in device tree
As in Linux, the driver can support both r8a77951 and r8a77950
SoC's so we decided to name the new driver as in Linux with Zephyr
prefix : "clock_control_r8a7795_cpg_mssr.c".
Signed-off-by: Aymeric Aillet <aymeric.aillet@iot.bzh>
This commit is reworking clock definition in
order to match linux filetree and definition
as much as possible.
- dt-bindings: rework renesas clocks dt-bindings
- regroup renesas related dt-bindings in a folder
- rename renesas rcar common dt-binding to match linux name
- add soc specific dt-binding matching linux name
- soc dt-bindings are defining clocks matching linux names
- dts: use new clocks names
- move clocks definitions in SoC layer for each core clock entry
- driver: use new clocks names
As seen in this commit, we are declaring clocks for "R8A7795" SoC
to match linux names.
Linux is not declaring "R8A77951" SoC specific files because
its also supporting the first H3 SoC version numbered "R8A77950".
Signed-off-by: Aymeric Aillet <aymeric.aillet@iot.bzh>
STM32_DT_CLOCKS was designed to take a device tree node label name as
argument: STM32_DT_CLOCKS(uart1)
Change its implementation to take a node identifier instead:
STM32_DT_CLOCKS(DT_NODELABEL(uart1)).
This make its usage more flexible since the argument can now be extracted
from other DT macros such as DT_PARENT. Then, the following can be done:
STM32_DT_CLOCKS(DT_PARENT(child_node_label)).
Since it is now possible implement STM32_DT_INST_CLOCKS using
STM32_DT_CLOCKS.
Finally, update existing STM32_DT_CLOCKS users and convert
STM32_INST_CLOCK_INFO users to STM32_CLOCK_INFO.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
Simple driver that allows one to choose the clock speed of xtensa cores.
It's basically a shim layer on top of SOC level driver.
Also, a really simple test case was added, mainly to ensure things are
build and are sane.
Signed-off-by: Ederson de Souza <ederson.desouza@intel.com>
This commit adds a check for the validity of the "sub_system" param in
clock_control_on.
While for clock_control_off there was a check if the clock_subsytem
parameter is in range of STM32_PERIPH_BUS_MIN and STM32_PERIPH_BUS_MAX
this check was missing for clock_control_on.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
-Kept getting an error on STM32g0xx chips that didn't have q-divisor.
-Changed to set prescaler only if it's being used.
Signed-off-by: Dan Higginbotham <daniel@dedesignworks.com>
<soc.h> has been traditionally been used as a proxy to HAL headers,
register definitions, etc. Nowadays, <soc.h> is anarchy. It serves a
different purpose depending on the SoC. In some cases it includes HALs,
in some others it works as a header sink/proxy (for no good reason), as
a register definition when there's no HAL... To make things worse, it is
being included in code that is, in theory, non-SoC specific.
This patch is part of a series intended to improve the situation by
removing <soc.h> usage when not needed, and by eventually removing it.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Adds few missing zephyr/ prefixes to leftover #include statements that
either got added recently or were using double quote format.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
Checking that provided pclken->bus fits bus range in
clock_control_get_rate() is wrong as it could actually be a source
clock and hence not belong to this range.
Also, this kind of check is just not needed since this function should
not be used before call to clock_control_on() or clock_control_configure()
which do the required verification.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
This PR Fixes the Audio PLL Rate Calculation (there was an additional
divide / 8 which is not necessary and does not appear in similar
calculations in example code from the SDK).
Additionally, it adjusts the SAI .dtsi to more correctly configure the
mclk rate, and adds comments specifying what the regististers mean.
Signed-off-by: Nickolas Lapp <nickolaslapp@gmail.com>
