Add support for Arm Ethos-N NPU multi-device.
The device tree parsing currently only supports one NPU device with
multiple cores. To be able to support multi-device NPU configurations
this patch adds support for having multiple NPU devices in the device
tree.
To be able to support multiple NPU devices in the SMC API, it has been
changed in an incompatible way so the API version has been bumped.
Signed-off-by: Laurent Carlier <laurent.carlier@arm.com>
Change-Id: Ide279ce949bd06e8939268b9601c267e45f3edc3
Disabling filter 0 causes inability to access DRAM.
An attempt leads to an abort.
ARM manual disallows to disable filter 0, but if we do
that from SRAM, nothing bad happens.
This patch prevents disabling of a filter 0, allowing to
reconfigure other filters from DRAM.
Note: this patch doesn't change the logic after reset.
It is only needed in case someone wants to reconfigure the
previously configured TZASC.
Change-Id: I196a0cb110a89afbde97f64a94df3101f28708a4
Signed-off-by: stsp@users.sourceforge.net
Add a new function that allows to enable or disabled filters on
configured regions dynamically. This will avoid the need to
reconfigure the entire attribute and just manage to
enable/disable filters.
Change-Id: If0937ca755bec6c45d3649718147108459682fff
Signed-off-by: Lionel Debieve <lionel.debieve@st.com>
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
The FMU is part of the GIC Distributor (GICD) component. It implements
the following functionality in GIC-600AE:
* Provides software the means to enable or disable a Safety Mechanism
within a GIC block.
* Receives error signaling from all Safety Mechanisms within other GIC
blocks.
* Maintains error records for each GIC block, for software inspection
and provides information on the source of the error.
* Retains error records across functional reset.
* Enables software error recovery testing by providing error injection
capabilities in a Safety Mechanism.
This patch introduces support to enable error detection for all safety
mechanisms provided by the FMU. Platforms are expected to invoke the
initialization function during cold boot.
The support for the FMU is guarded by the GICV3_SUPPORT_GIC600AE_FMU
makefile variable. The default value of this variable is '0'.
Change-Id: I421c3d059624ddefd174cb1140a2d2a2296be0c6
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
At the moment we have a GIC_ENABLE_V4_EXTN build time variable to
determine whether the GIC interrupt controller is compliant to version
4.0 of the spec or not. This just changes the number of 64K MMIO pages
we expect per redistributor.
To support firmware builds which run on variable systems (emulators,
fast model or FPGAs), let's make this decision at runtime.
The GIC specification provides several architected flags to learn the
size of the MMIO frame per redistributor, we use GICR_TYPER[VLPI] here.
Provide a (static inline) function to return the size of each
redistributor.
We keep the GIC_ENABLE_V4_EXTN build time variable around, but change
its meaning to enable this autodetection code. Systems not defining this
rely on a "pure" GICv3 (as before), but platforms setting it to "1" can
now deal with both configurations.
Change-Id: I9ede4acf058846157a0a9e2ef6103bf07c7655d9
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
At the moment we have a GIC_ENABLE_V4_EXTN build time variable to
determine whether the GIC interrupt controller is compliant to version
4.0 of the GIC spec or not.
In case of the GIC-600 multichip support we were somewhat abusing that
flag to differentiate between a GIC-700 and GIC-600 implementation
being used in the system.
To avoid a build time dependency on this flag, look at the GICD_IIDR
register and check if the hardware is a GIC-600 or not, to make this
decision at runtime. We then use the values for either GIC-700 or the
GIC-600, respectively.
Change-Id: I8c09ec1cd6fd60d28da879ed55ffef5506f9869d
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
For the GIC power management we need to identify certain GIC
implementations, so we have the IIDR values for some Arm Ltd. GIC models
defined.
We will need those number elsewhere very soon, so export them to a
shared header file, to avoid defining them again.
Change-Id: I1b8e2d93d6cea0d066866143c89eef736231134f
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
The GIC IP formerly known as "GIC Clayton" has been released under the
name of "GIC-700".
Rename occurences of Clayton in comments and macro names to reflect the
official name.
Change-Id: Ie8c55f7da7753127d58c8382b0033c1b486f7909
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
A RAS error may be triggered while offline core in OS. Error:
Uncorrected software error in the Distributor, with IERR=9,SERR=f.
Core put to sleep before its Group enables were cleared.
gicv3_cpuif_disable() will be called in offline core flow.
According to GIC architecture version 3 and version 4:
Architectural execution of a DSB instruction guarantees that
the last value written to ICC_IGRPEN0_EL1, ICC_IGRPEN1_EL1,
ICC_IGRPEN1_EL3 or GICC_CTLR.{EnableGrp0, EnableGrp1}is observed
by the associated Redistributor.
An ISB or other context synchronization operation must precede
the DSB to ensure visibility of System register writes.
Signed-off-by: Ming Huang <huangming@linux.alibaba.com>
Change-Id: Iff1475657f401374c761b5e8f2f5b3a4b2040e9d
Use helper functions to get SPI and ESPI INTID limit, to remove
several pieces of similar code in gicv3 driver.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: Iaf441fe5e333c4260e7f6d98df6fdd931591976d
Add helper function gicv3_get_espi_limit() to get the value of
(maximum extended SPI INTID + 1), so that some duplicated code can be
removed later.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I0355ca2647f872e8189add259f6c47d415494cce
Add newline(\n) to make the message clearer when the warnings are
really triggered.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I5e2574a52c9065db32ecb4d453b9b02445f69a82
Add helper function gicv3_get_spi_limit() to get the value of (maximum
SPI INTID + 1), so that some duplicated code can be removed later.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I160c8a88fbb71d22790b8999a84afbfba766f5e7
By default the Arm Ethos-N NPU will boot up in secure mode. In this mode
the non-secure world cannot access the registers needed to use the NPU.
To still allow the non-secure world to use the NPU, a SiP service has
been added that can delegate non-secure access to the registers needed
to use it.
Only the HW_CONFIG for the Arm Juno platform has been updated to include
the device tree for the NPU and the platform currently only loads the
HW_CONFIG in AArch64 builds.
Signed-off-by: Mikael Olsson <mikael.olsson@arm.com>
Change-Id: I65dfd864042ed43faae0a259dcf319cbadb5f3d2
The legacy console is gone. Re-add DCC console support based
on the multi-console framework.
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@xilinx.com>
Acked-by: Michal Simek <michal.simek@xilinx.com>
Change-Id: Ia8388721093bc1be3af40974530d7c9a9ae5f43e
A new function tzc400_it_handler() is created to manage TZC400
interrupts. The required helpers to read and clear interrupts are added
as well.
In case DEBUG is enabled, more information about the faulty access
(address, NSAID, type of access) is displayed.
Change-Id: Ie9ab1c199a8f12b2c9472d7120efbdf35711284a
Signed-off-by: Yann Gautier <yann.gautier@st.com>
* changes:
drivers/gicv3: also shift eSPI register offset in GICD_OFFSET_64()
drivers/gicv3: add debug log for maximum INTID of SPI and eSPI
drivers/gicv3: limit SPI ID to avoid misjudgement in GICD_OFFSET()
drivers/gicv3: fix logical issue for num_eints
drivers/gicv3: fix potential GICD context override with ESPI enabled
drivers/gicv3: use mpidr to probe GICR for current CPU
TZC_400_REGION_ATTR_FILTER_BIT_ALL is a simple constant definition, so
it can't get the real filter number to construct the bit flag for all
existing filters. If the platform doesn't have 4 filters, passing
FILTER_BIT_ALL to tzc400_configure_region() will cause assertion or
misconfiguration. So adjust the bit flag against the real filter
number.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: Ie5c48303485f3b5015772961ee7c34746121ee84
ESPI register offset should also be shifted right by REG##R_SHIFT to
keep consistent.
It is not a functional issue, for GICD_OFFSET_64() is only used for
GICD_IROUTER<E>, and IROUTER_SHIFT is 0.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I76eee5c50e4300890e78e80bddde135ce88daa2d
Add debug log for the maximum supported INTID of SPI and eSPI on the
current GIC implementation.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: Ie45ab1d85b39658c4ca4bc54ee433ac44e41d03f
The GICv3 architecture allows GICD_TYPER.ITLinesNumber to be 31, so
the maximum possible value for num_ints is 1024. The value must be
limited to (MAX_SPI_ID + 1), or GICD_OFFSET() will consider it as ESPI
INTID and return wrong register address.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: Iddcb83d3e5d241b39f4176c19c2bceaa2c3dd653
In function gicv3_spis_config_defaults(), the variable num_ints is set
to (maximum SPI INTID + 1), while num_eints is set to (maximum ESPI
INTID). It introduces not only inconsistency to the code, but also
logical bug in the "for" loops, for the INTID of num_eints is also
valid and the check should be inclusive.
Fix this by setting num_eints to (maximum ESPI INTID + 1) as well.
Fix similar issues in gicv3_distif_save() and
gicv3_distif_init_restore().
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I4425777d17e84e85f38853603340bd348640154f
RESTORE/SAVE_GICD_EREGS uses (int_id - (MIN_ESPI_ID - MIN_SPI_ID)) to
get the context array index for ESPI, which will override the space of
standard SPI starting from (MIN_SPI_ID + MIN_SPI_ID).
However, using TOTAL_SPI_INTR_NUM to replace the above MIN_SPI_ID
cannot totally fix the issue, for TOTAL_SPI_INTR_NUM is not well
aligned and the array index will be rounded down by the shifting
operation if being shifted more than 2 bits. It will cause buffer
override again when the existing maximum SPI reaches 1019.
So round up TOTAL_SPI_INTR_NUM with (1 << REG##R_SHIFT) for GICD
context arrays.
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I5be2837c42f381a62f8d46a4ecd778009b1fe059
In function gicv3_rdistif_probe(), line #1322 implies
gicv3_driver_data->mpidr_to_core_pos() may be null, but the original
code uses this interface to get current CPU index unconditionally.
It is better to use MPIDR to probe GICR which does not depend on
gicv3_driver_data->mpidr_to_core_pos().
Signed-off-by: Heyi Guo <guoheyi@linux.alibaba.com>
Change-Id: I64add055385040fe0a56b977e2299608e2309a6e
Usually, C has no problem up-converting types to larger bit sizes. MISRA
rule 10.7 requires that you not do this, or be very explicit about this.
This resolves the following required rule:
bl1/aarch64/bl1_context_mgmt.c:81:[MISRA C-2012 Rule 10.7 (required)]<None>
The width of the composite expression "0U | ((mode & 3U) << 2U) | 1U |
0x3c0U" (32 bits) is less that the right hand operand
"18446744073709547519ULL" (64 bits).
This also resolves MISRA defects such as:
bl2/aarch64/bl2arch_setup.c:18:[MISRA C-2012 Rule 12.2 (required)]
In the expression "3U << 20", shifting more than 7 bits, the number
of bits in the essential type of the left expression, "3U", is
not allowed.
Further, MISRA requires that all shifts don't overflow. The definition of
PAGE_SIZE was (1U << 12), and 1U is 8 bits. This caused about 50 issues.
This fixes the violation by changing the definition to 1UL << 12. Since
this uses 32bits, it should not create any issues for aarch32.
This patch also contains a fix for a build failure in the sun50i_a64
platform. Specifically, these misra fixes removed a single and
instruction,
92407e73 and x19, x19, #0xffffffff
from the cm_setup_context function caused a relocation in
psci_cpus_on_start to require a linker-generated stub. This increased the
size of the .text section and caused an alignment later on to go over a
page boundary and round up to the end of RAM before placing the .data
section. This sectionn is of non-zero size and therefore causes a link
error.
The fix included in this reorders the functions during link time
without changing their ording with respect to alignment.
Change-Id: I76b4b662c3d262296728a8b9aab7a33b02087f16
Signed-off-by: Jimmy Brisson <jimmy.brisson@arm.com>
In the function gicv2_set_spi_routing, the signed value proc_num is cast
to unsigned int before being compared to other unsigned values in two
assert calls. The value proc_num can be a negative value, and once the
negative value is cast to unsigned it becomes a very large number which
will trigger the assert. This patch changes the assert cast so that the
unsigned values are cast to signed instead, keeping the same functionality
but allowing proc_num to be negative.
This bug can be seen when running the SDEI RM_ANY routing mode test in
TFTF on the Juno platform.
This patch also makes the usage of the proc_num variable in other gicv2
functions more clear.
Signed-off-by: John Powell <john.powell@arm.com>
Change-Id: If1b98eebb00bd9b73862e5e995e5e68c168170a6
And from crash_console_flush.
We ignore the error information return by console_flush in _every_
place where we call it, and casting the return type to void does not
work around the MISRA violation that this causes. Instead, we collect
the error information from the driver (to avoid changing that API), and
don't return it to the caller.
Change-Id: I1e35afe01764d5c8f0efd04f8949d333ffb688c1
Signed-off-by: Jimmy Brisson <jimmy.brisson@arm.com>
A GICv3 interrupt controller will be instantiated for a certain number
of cores. This will result in the respective number of GICR frames. The
last frame will have the "Last" bit set in its GICR_TYPER register.
For platforms with a topology unknown at build time (the Arm FPGAs, for
instance), we need to learn the number of used cores at runtime, to size
the GICR region in the devicetree accordingly.
Add a generic function that iterates over all GICR frames until it
encounters one with the "Last" bit set. It returns the number of cores
the GICv3 has been configured for.
Change-Id: I79f033c50dfc1c275aba7122725868811abcc4f8
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This patch fixes defects 10.3, 10.4, 10.7, 20.7
reported by MISRA-2012 scan and adds braces for
conditional statements according to the TF-A
coding style.
Change-Id: If84ed31cdd55bc8e7cdd2a5f48c0dacc25792112
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
The Arm GIC-Clayton IP has the same power management requirements as
the GIC-600, when it comes to powering up the redistributors before
using them.
Add the IIDR value to the existing list of implementations requiring
the power sequence.
Change-Id: Ib965dfe278c40a4fff94f65a8d445c27a2ae6fd2
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This patch moves all GICv2 driver files into new added
'gicv2.mk' makefile for the benefit of the generic driver
which can evolve in the future without affecting platforms.
NOTE: Usage of 'drivers/arm/gic/common/gic_common.c' file
is now deprecated and platforms with GICv2 driver need to
be modified to include 'drivers/arm/gic/v2/gicv2.mk' in
their makefiles.
Change-Id: Ib10e71bdda0e5c7e80a049ddce2de1dd839602d1
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
This patch adds the IIDR value for GIC600-AE to the gicv3_is_gic600()
helper function. This helps platforms supporting this version of the
GIC600 interrupt controller to function with the generic GIC driver.
Verified with tftf-validation test suite
******************************* Summary *******************************
> Test suite 'Framework Validation'
Passed
> Test suite 'Timer framework Validation'
Passed
=================================
Tests Skipped : 0
Tests Passed : 6
Tests Failed : 0
Tests Crashed : 0
Total tests : 6
=================================
NOTICE: Exiting tests.
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
Change-Id: I518ae7b56f7f372e374e453287d76ca370fc3574
Cleaned up confused naming between TB_FW and FW_CONFIG.
Signed-off-by: Louis Mayencourt <louis.mayencourt@arm.com>
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Change-Id: I9e9f6e6ca076d38fee0388f97d370431ae067f08
It is desired to have the peripheral writes completed to clear the
interrupt condition and de-assert the interrupt request to GIC before
EOI write. Failing which spurious interrupt will occurred.
A barrier is needed to ensure peripheral register write transfers are
complete before EOI is done.
GICv2 memory mapped DEVICE nGnR(n)E writes are ordered from core point
of view. However these writes may pass over different interconnects,
bridges, buffers leaving some rare chances for the actual write to
complete out of order.
GICv3 ICC EOI system register writes have no ordering against nGnR(n)E
memory writes as they are over different interfaces.
Hence a dsb can ensure from core no writes are issued before the previous
writes are *complete*.
Signed-off-by: Sandeep Tripathy <sandeep.tripathy@broadcom.com>
Change-Id: Ie6362009e2f91955be99dca8ece14ade7b4811d6
The only difference between GIC-500 and GIC-600 relevant to TF-A is the
differing power management sequence.
A certain GIC implementation is detectable at runtime, for instance by
checking the IIDR register. Let's add that test before initiating the
GIC-600 specific sequence, so the code can be used on both GIC-600 and
GIC-500 chips alike, without deciding on a GIC chip at compile time.
This means that the GIC-500 "driver" is now redundant. To allow minimal
platform support, add a switch to disable GIC-600 support.
Change-Id: I17ea97d9fb05874772ebaa13e6678b4ba3415557
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
To support compatibility with previous GICv3 driver version
this patch:
- restores original API for gicr_read_ipriority() and
gicr_wrtite_ipriority() functions;
- adds accessor functions for GICR_XXX0,1 registers, e.g.
GICR_IGROUPR0, GICR_ICFGR0, GICR_ICFGR1, etc.
Change-Id: I796a312a61665ff384e3d9de2f4b3c60f700b43b
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
GIC-Clayton supports multichip operation mode which allows it to connect
upto 16 other GIC-Clayton instances. GIC-Clayton's multichip programming
and operation remains same as GIC-600 with a minor change in the
SPI_BLOCKS and SPI_BLOCK_MIN shifts to accommodate additional SPI
ranges. So identify if the GIC v4 extension is enabled by the platform
makefile and appropriately select the SPI_BLOCKS and SPI_BLOCK_MIN
shifts.
Change-Id: I95fd80ef16af6c7ca09e2335539187b133052d41
Signed-off-by: Vijayenthiran Subramaniam <vijayenthiran.subramaniam@arm.com>
This patch adds support for GICv4 extension.
New `GIC_ENABLE_V4_EXTN` option passed to gicv3.mk makefile
was added, and enables GICv4 related changes when set to 1.
This option defaults to 0.
Change-Id: I30ebe1b7a98d3a54863900f37eda4589c707a288
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
This patch provides support for GICv3.1 extended PPI and SPI
range. The option is enabled by setting to 1 and passing
`GIC_EXT_INTID` build flag to gicv3.mk makefile.
This option defaults to 0 with no extended range support.
Change-Id: I7d09086fe22ea531c5df51a8a1efd8928458d394
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
This patch moves all GICv3 driver files into new added
'gicv3.mk' makefile for the benefit of the generic driver
which can evolve in the future without affecting platforms.
The patch adds GICv3 driver configuration flags
'GICV3_IMPL', 'GICV3_IMPL_GIC600_MULTICHIP' and
'GICV3_OVERRIDE_DISTIF_PWR_OPS' described in
'GICv3 driver options' section of 'build-option.rst'
document.
NOTE: Platforms with GICv3 driver need to be modified to
include 'drivers/arm/gic/v3/gicv3.mk' in their makefiles.
Change-Id: If055f6770ff20f5dee5a3c99ae7ced7cdcac5c44
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
This patch provides separation of GICD, GICR accessor
functions and adds new macros for GICv3 registers access
as a preparation for GICv3.1 and GICv4 support.
NOTE: Platforms need to modify to include both
'gicdv3_helpers.c' and 'gicrv3_helpers.c' instead of the
single helper file previously.
Change-Id: I1641bd6d217d6eb7d1228be3c4177b2d556da60a
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
The cpu index was wrongly checked causing it to assert always.
Since this code path is exercised only during TF test "NODE_HW_STAT",
which queries Power state from SCP, this bug was not detected earlier.
Change-Id: Ia25cef4c0aa23ed08092df39134937a2601c21ac
Signed-off-by: Manish Pandey <manish.pandey2@arm.com>
Laurent Carlier
Stas Sergeev
André Przywara
Mark Dykes
Lionel Debieve
Varun Wadekar
Ming Huang
Heyi Guo
Mikael Olsson
Venkatesh Yadav Abbarapu
Yann Gautier
Manish Pandey
Joanna Farley
Mark Dykes
Jimmy Brisson
johpow01
Alexei Fedorov
Manish V Badarkhe
Sandeep Tripathy
Vijayenthiran Subramaniam