* changes:
perf(psci): pass my_core_pos around instead of calling it repeatedly
refactor(psci): move timestamp collection to psci_pwrdown_cpu
refactor(psci): factor common code out of the standby finisher
refactor(psci): don't use PSCI_INVALID_PWR_LVL to signal OFF state
docs(psci): drop outdated cache maintenance comment
Initializing all early clocks before the MMU is enabled can impact boot
time. Therefore, splitting the setup into A53 clocks and peripheral
clocks can be beneficial, with the peripheral clocks configured after
fully initializing the MMU.
Change-Id: I19644227b66effab8e2c43e64e057ea0c8625ebc
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
Add all clock modules as entries in MMU using dynamic regions.
Change-Id: I56f724ced4bd024554c7b38afd14ea420de80cc6
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
On some platforms plat_my_core_pos is a nontrivial function that takes a
bit of time and the compiler really doesn't like to inline. In the PSCI
library, at least, we have no need to keep repeatedly calling it and we
can instead pass it around as an argument. This saves on a lot of
redundant calls, speeding the library up a bit.
Change-Id: I137f69bea80d7cac90d7a20ffe98e1ba8d77246f
Signed-off-by: Boyan Karatotev <boyan.karatotev@arm.com>
In the SMMU root port programming model, changing both
SMMU_ROOT_CR0.GPCEN and ACCESSEN bits in the same MMIO write operation
is permitted by the architecture but left to the SMMU IP implementation
to determine the order of completing one or the other operation.
Enforce more determinism by setting CR0.GPCEN, wait for CR0ACK.GPCEN
completion, then setting CR0.ACCESSEN and wait for CR0ACK.ACCESSEN
completion.
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com>
Change-Id: I36ba5fbc13d06c6243226008d18a2d57477b0d28
Change the name of these confs to be version agnostic,
we will later use these configs to enforce the mbedtls
minimum version
Change-Id: I1f665c2471877ecc833270c511749ff845046f10
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
The A/B loader [1] meant to be used for convenient CI testing.
The tool is installed into the same location as SA0, where it
conveniently fits due to its size, and where it makes use of
non-volatile PMIC registers to alternate between loading and
starting A or B copy of the BL2. The PMIC registers are used
because CPU registers are lost across reset.
In case the B copy is loaded, it is loaded from 8 MiB offset
from start of HF. In case the B copy fails to boot, a simple
reset of the system will switch back to booting previously
known working A copy and allow recovery.
The A/B loader sets MFIS bit MFISBTSTSR_BOOT_PARTITION to
pass the information which A/B copy is currently booting on
to TFA, which then loads the follow up components from 0 MiB
or 8 MiB offset, depending on whether the A or B copy is being
booted.
The MFISBTSTSR_BOOT_PARTITION interferes with regular A/B
switching during boot from eMMC as the boot media, where
the BootROM also sets MFISBTSTSR_BOOT_PARTITION bit in case
the system boots from SECOND eMMC HW BOOT partition.
Since the A/B loader is meant as a development and CI tool,
isolate the A/B loader use to RPC HF only and furthermore
isolate it behind new RCAR_RPC_HYPERFLASH_ABLOADER option
which is disabled by default.
[1] https://github.com/marex/abloader
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Change-Id: I04ecd50fa1405b78e1ba3949d54029034d4f22d8
Make this function return values from crypto_ret_value.
The previous method of returning the mbedtls error code
on failure meant that the authentication module couldn't
correctly parse failures from this function.
Change-Id: I9fe6eba1fc79e8f81004f8cd202781aea907e963
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
The buffer size allocated for the raw ECDSA signatures
was too small for P-384 signatures. This led to mbedtls
aborting the generation of the raw sig.
Fix this by increasing the buffer size to the required
value.
Change-Id: I06a9cfe1f4cb2603c5fbe945714e90460c24edb8
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
The skipddc_dat variable is only use at one place in dfixlat_program()
function. Move its definition and update there to avoid compilation
issues about variables being set but not used.
Signed-off-by: Yann Gautier <yann.gautier@st.com>
Change-Id: I6b5273365f1da26773fc46f898362cb9af170506
Refactor to be more robust. Removed duplicated and not used functions.
Add in ADMA read.
Change-Id: I1a5a00397ece6f9ccc5916225ab5317010b01b52
Signed-off-by: Jit Loon Lim <jit.loon.lim@intel.com>
Signed-off-by: Sieu Mun Tang <sieu.mun.tang@intel.com>
In order to avoid separate platform definitions when not using the
default timer functions, it is better to move these functions out of the
header file and into the source files, so that they can be built if
needed.
Move timer functions from delay_timer.h into generic_delay_timer.c. Add
them as callback functions which are then called in delay_timer.c.
Change-Id: I96a1eac8948b1a7b1e481899b67a083db4c9b97d
Signed-off-by: Abhi Singh <abhi.singh@arm.com>
DCC driver to support boot and runtime console scope
switch for dedicated boot and runtime consoles.
Change-Id: I7769dc44860a5fda99ca42ce17a3a6009288d7e7
Signed-off-by: Maheedhar Bollapalli <MaheedharSai.Bollapalli@amd.com>
When initialising a GICv3 compatible interrupt controller, we currently
assume that the GIC is still in its reset state, which means
the GICR_WAKER.ProcessorSleep bit is set. There is an "assert" in the
GIC setup function to check this.
However when using RESET_TO_BL31, there might be prior firmware running,
and it might have used the GIC already. This is for instance the case on
the Allwinner A523 SoC, where the BootROM initialises the GIC to use it
when handling the built-in USB debug protocol.
Drop the assert, which is not the right thing to do here anyway: it's
not checking an internal state. Instead return early when the
redistributor is already marked as active.
Also keep waiting if ChildrenAsleep is unexpectedly set, but warn about
this.
This fixes booting TF-A on an Allwinner A523 SoC when using the USB
debug mode.
Change-Id: I5be9e1b0489d33b8371fff484e526483d5f3d937
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Address issue CID 445362 and CID 445361 found during coverity scan.
Change-Id: I1ab460d2e1353b81517788e32de662f203b0352f
Signed-off-by: Maxime Méré <maxime.mere@foss.st.com>
Add driver to support DDR on STM32MP2 platform. It drives the DDR PHY
and its firmware, as well as the DDR controller.
Signed-off-by: Nicolas Le Bayon <nicolas.le.bayon@st.com>
Signed-off-by: Maxime Méré <maxime.mere@foss.st.com>
Change-Id: I93de2db1b9378d5654e76b3bf6f3407d80bc4ca5
Disabling AXI port, enabling host interface and both enabling/disabling
software self-refresh services are already present inside the driver
source code.
Factorize by gathering them as services inside the generic part,
and adapt driver to call these new functions.
Add services to manage quasi-dynamic registers. DDRCTRL contains
quasi-dynamic registers, which are dynamic only under some conditions
defined by the user guide (with 4 groups).
In our driver, out of reset state, only groups 3 and 4 are updated.
Group 4 needs only sw_done/sw_done_ack sequence, already available.
Group 3 sequence include more conditions, that are gathered in
specific services. stm32mp_ddr_disable_host_interface() has been added
to do this.
Add dedicated generic service to toggle rfshctl3.refresh_update_level
and wait for completion.
Manage AXI ports and HIF when updating QD3 registers. Quasi-dynamic
group 3 (QD3) registers are updated when DDR is not completely
initialized, i.e. when AXI ports are not enabled and possibly when
host interface (HIF) is not enabled too.
In that case, a specific mechanism is necessary to restore the same
conditions as before accessing QD3 registers.
Static functions have been added to get AXI ports and HIF states and
are used to determine the needed conditions to set/unset.
Signed-off-by: Nicolas Le Bayon <nicolas.le.bayon@st.com>
Change-Id: I880f88b1cb6fc76199ad9ea33e9d63a5c469aed4
It was only used for an error trace, with little added value. This
allows some gains in DDR driver data size.
Signed-off-by: Yann Gautier <yann.gautier@st.com>
Change-Id: I02e2fc75efd90ff188e62f39850b9bd4c3af1649
Instead of using hard-coded number in DDR driver, use macros.
Modify TIMEOUT_US_1S to DDR_TIMEOUT_US_1S to align with other defines.
Signed-off-by: Yann Gautier <yann.gautier@st.com>
Change-Id: I489084132821774b0049a4a5d7fc30db24a7bb11
Modify the DDR driver to handle the DDR power supplies when a PMIC
is present in the function stm32mp_board_ddr_power_init(), define
in the platform file.
This patch allows to easily modify the used DDR power supplies
for customer boards, when they don't use STPMIC1 PMU or when
the regulators are not connected as on the STMicroelectronics
boards.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Change-Id: I93ee6295ef7032ac20f03608d22cd460f7d87ef5
The digest size in bytes for sha1/256/384/512 were defined in multiple
places. Refactor the macros into a common header file.
Change-Id: I84ef3561486ff70345ae8c871d5d6e1564574ec2
Signed-off-by: Raghu Krishnamurthy <raghupathyk@nvidia.com>
The function 'update_stack_depth' modifies the value of the 'depth'
parameter passed by reference. Typically, the caller recevies this
parameter by value, and it is then passed to 'update_stack_depth' by
reference. This violates MISRA 17.8 rule. To address this issue, a new
local variable is introduced to store the value of 'depth'.
Change-Id: Ia37f4ede9e6558f778bdda17b7b195f1f50d0c30
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
Enable the DDR clock by setting up its reset block, the associated
partition and configuring the clock tree above the MC_CGM mux.
Change-Id: Idfed24b3e74a189df87f9782886a91b906cd2022
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
The DDR clock can be powered by either a PLL or an oscillator. An MC_CGM
mux selects between these two clock sources. A reset block, part of
partition 0, is also connected to this IP block. Therefore, all the
dependencies mentioned above must be configured to have a working clock.
Change-Id: Ia841428db9acb95c59ea851b6afeb0b7ff9230a2
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
Add the DDR PLL instance and configure it to operate at its maximum
allowed frequency.
Change-Id: I96efd68687de78f70759f631d10a0f611c234c8d
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
MC_ME is one of the leading hardware blocks responsible for partitions'
transition to and from a reset state. Not being the only one involved in
this role, it must cooperate with some other modules (MC_RGM, RDC) to
successfully bring a peripheral out of the reset state. As a result, the
partition management is isolated into a dedicated file, as parts of it
will later contribute to peripheral reset control.
Change-Id: I6a9dbf28008b1677bc847bbafa474b489c999d05
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
MC_RGM is a hardware block involved in resetting peripherals and
partitions. Here, the accessories for partition reset are added.
Change-Id: If00755fe0e93ba2e4841f95ed5ae3c87db20bebf
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
ERR051700 erratum is present on all S32CC-based SoCs and relates to
reset. Releasing multiple Software Resettable Domains (SRDs) from
reset simultaneously, may cause a false error in the fault control
unit.
The workaround is to clear the SRD resets sequentially instead of
simultaneously.
Change-Id: I883bc223bf6834907259e6964a5702d7186e4c7f
Signed-off-by: Alexandru-Catalin Ionita <alexandru-catalin.ionita@nxp.com>
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
The STPMIC2 embeds 15 regulators with various
properties, and is designed to supply the STM32MP2
SOC. This driver handles a minimal set of feature
to handle the boot of a board.
Signed-off-by: Pascal Paillet <p.paillet@st.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Signed-off-by: Maxime Méré <maxime.mere@foss.st.com>
Change-Id: Ibe0cacf8aec2871eb9a86ec16cbbd18d3745fe9e
Simplify the clock enablement mechanism from a usage perspective. With
this new approach, enabling a clock cascades the turn-on sequence of all
its parent clocks in the clock tree. Therefore, enabling the A53 clock
will also turn on the A53 PLL and the oscillator that feeds it.
Change-Id: Ifc2bee3e9edbb4baced34f9e809a961562f7d0a6
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
Bring in the implementation for the struct clk_ops->get_parent callback
for the S32G clock driver. The parent is established depending on the
clock object type. Usually, this is determined based on the parent
field, but not always.
Change-Id: I76a3d2636dc23ba2d547d058b8650dd0e99fe1fa
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
The UART clock initialization failed because the clock mux enablement
mechanism did not correctly recognize the PERIPH PLL mux. Therefore, it
was reported as an unknown mux ID.
Change-Id: I6cc72c87a8462a2ed2e7c360f59a74961bb2f3a1
Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
This new version uses the multipart PSA AEAD API;
the authentication tag is verified via
a call to psa_aead_verify.
Change-Id: If4b7e6258223ae6fead1794d3e8d0004f0f387b3
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
MbedTLS 3.6.1 fixed the issue which previously
produced this warning, so this hack is no longer
necessary.
Change-Id: I934adefbf2fed16e16b9d98bc8674125b70b08fc
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
Always support the regulator_set_voltage operation for the same voltage.
This patch prepares the DDR power configuration when the power supplies
are fixed.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Signed-off-by: Maxime Méré <maxime.mere@foss.st.com>
Change-Id: Idf032b03c6f0c95f997dec3ed8a0d38c54a3de15
Permit to override enable ramp-delay value from
the device tree in BL2.
Change-Id: Id8e803b368055a50fbd14d4527917c449b958ad9
Signed-off-by: Pascal Paillet <p.paillet@st.com>
Signed-off-by: Maxime Méré <maxime.mere@foss.st.com>
