At the moment, the memory translation library allows to create memory
mappings of 2 types:
- Device nGnRE memory (named MT_DEVICE in the library);
- Normal, Inner Write-back non-transient, Outer Write-back
non-transient memory (named MT_MEMORY in the library).
As a consequence, the library code treats the memory type field as a
boolean: everything that is not device memory is normal memory and
vice-versa.
In reality, the ARMv8 architecture allows up to 8 types of memory to
be used at a single time for a given exception level. This patch
reworks the memory attributes such that the memory type is now defined
as an integer ranging from 0 to 7 instead of a boolean. This makes it
possible to extend the list of memory types supported by the memory
translation library.
The priority system dictating memory attributes for overlapping
memory regions has been extended to cope with these changes but the
algorithm at its core has been preserved. When a memory region is
re-mapped with different memory attributes, the memory translation
library examines the former attributes and updates them only if
the new attributes create a more restrictive mapping. This behaviour
is unchanged, only the manipulation of the value has been modified
to cope with the new format.
This patch also introduces a new type of memory mapping in the memory
translation library: MT_NON_CACHEABLE, meaning Normal, Inner
Non-cacheable, Outer Non-cacheable memory. This can be useful to map
a non-cacheable memory region, such as a DMA buffer for example.
The rules around the Execute-Never (XN) bit in a translation table
for an MT_NON_CACHEABLE memory mapping have been aligned on the rules
used for MT_MEMORY mappings:
- If the memory is read-only then it is also executable (XN = 0);
- If the memory is read-write then it is not executable (XN = 1).
The shareability field for MT_NON_CACHEABLE mappings is always set as
'Outer-Shareable'. Note that this is not strictly needed since
shareability is only relevant if the memory is a Normal Cacheable
memory type, but this is to align with the existing device memory
mappings setup. All Device and Normal Non-cacheable memory regions
are always treated as Outer Shareable, regardless of the translation
table shareability attributes.
This patch also removes the 'ATTR_SO' and 'ATTR_SO_INDEX' #defines.
They were introduced to map memory as Device nGnRnE (formerly called
"Strongly-Ordered" memory in the ARMv7 architecture) but were not
used anywhere in the code base. Removing them avoids any confusion
about the memory types supported by the library.
Upstream platforms do not currently use the MT_NON_CACHEABLE memory
type.
NOTE: THIS CHANGE IS SOURCE COMPATIBLE BUT PLATFORMS THAT RELY ON THE
BINARY VALUES OF `mmap_attr_t` or the `attr` argument of
`mmap_add_region()` MAY BE BROKEN.
Change-Id: I717d6ed79b4c845a04e34132432f98b93d661d79
`board_arm_def.h` contains multiple definitions of
`PLAT_ARM_MMAP_ENTRIES` and `MAX_XLAT_TABLES` that are optimised for
memory usage depending upon the chosen build configuration. To ease
maintenance of these constants, this patch replaces their multiple
definitions with a single set of definitions that will work on all ARM
platforms.
Platforms can override the defaults with optimal values by enabling the
`ARM_BOARD_OPTIMISE_MMAP` build option. An example has been provided in
the Juno ADP port.
Additionally, `PLAT_ARM_MMAP_ENTRIES` is increased by one to accomodate
future ARM platforms.
Change-Id: I5ba6490fdd1e118cc9cc2d988ad7e9c38492b6f0
The common topology description helper funtions and macros for
ARM Standard platforms assumed a dual cluster system. This is not
flexible enough to scale to multi cluster platforms. This patch does
the following changes for more flexibility in defining topology:
1. The `plat_get_power_domain_tree_desc()` definition is moved from
`arm_topology.c` to platform specific files, that is `fvp_topology.c`
and `juno_topology.c`. Similarly the common definition of the porting
macro `PLATFORM_CORE_COUNT` in `arm_def.h` is moved to platform
specific `platform_def.h` header.
2. The ARM common layer porting macros which were dual cluster specific
are now removed and a new macro PLAT_ARM_CLUSTER_COUNT is introduced
which must be defined by each ARM standard platform.
3. A new mandatory ARM common layer porting API
`plat_arm_get_cluster_core_count()` is introduced to enable the common
implementation of `arm_check_mpidr()` to validate MPIDR.
4. For the FVP platforms, a new build option `FVP_NUM_CLUSTERS` has been
introduced which allows the user to specify the cluster count to be
used to build the topology tree within Trusted Firmare. This enables
Trusted Firmware to be built for multi cluster FVP models.
Change-Id: Ie7a2e38e5661fe2fdb2c8fdf5641d2b2614c2b6b
This patch adds a link to the Cortex-A57 Software Optimization Guide
in the ARM CPU Specific Build Macros document to justify the default
value of the A57_DISABLE_NON_TEMPORAL_HINT build flag.
Change-Id: I9779e42a4bb118442b2b64717ce143314ec9dd16
This patch adds support for the `%p` format specifier in tf_printf()
following the example of the printf implementation of the stdlib used
in the trusted firmware.
FixesARM-software/tf-issues#292
Change-Id: I0b3230c783f735d3e039be25a9405f00023420da
The shared memory region on ARM platforms contains the mailboxes and,
on Juno, the payload area for communication with the SCP. This shared
memory may be configured as normal memory or device memory at build
time by setting the platform flag 'PLAT_ARM_SHARED_RAM_CACHED' (on
Juno, the value of this flag is defined by 'MHU_PAYLOAD_CACHED').
When set as normal memory, the platform port performs the corresponding
cache maintenance operations. From a functional point of view, this is
the equivalent of setting the shared memory as device memory, so there
is no need to maintain both options.
This patch removes the option to specify the shared memory as normal
memory on ARM platforms. Shared memory is always treated as device
memory. Cache maintenance operations are no longer needed and have
been replaced by data memory barriers to guarantee that payload and
MHU are accessed in the right order.
Change-Id: I7f958621d6a536dd4f0fa8768385eedc4295e79f
All files including plat/mediatek/mt8173/drivers/gpio/gpio.h were
using system includes instead of user includes, which may cause the
wrong version of the header to be included. Said includes have been
changed to user includes to make sure that the included file is the
wanted one.
Change-Id: I29bdfe96fbd9a7900875e2357bbb43f3ea431fa5
The folowing build options were missing from the User Guide and have been
documented:
- CTX_INCLUDE_FPREGS
- DISABLE_PEDANTIC
- BUILD_STRING
- VERSION_STRING
- BUILD_MESSAGE_TIMESTAMP
Change-Id: I6a9c39ff52cad8ff04deff3ac197af84d437b8b7
ARM Trusted Firmware supports 2 different interconnect peripheral
drivers: CCI and CCN. ARM platforms are implemented using either of the
interconnect peripherals.
This patch adds a layer of abstraction to help ARM platform ports to
choose the right interconnect driver and corresponding platform support.
This is as described below:
1. A set of ARM common functions have been implemented to initialise an
interconnect and for entering/exiting a cluster from coherency. These
functions are prefixed as "plat_arm_interconnect_". Weak definitions of
these functions have been provided for each type of driver.
2.`plat_print_interconnect_regs` macro used for printing CCI registers is
moved from a common arm_macros.S to cci_macros.S.
3. The `ARM_CONFIG_HAS_CCI` flag used in `arm_config_flags` structure
is renamed to `ARM_CONFIG_HAS_INTERCONNECT`.
Change-Id: I02f31184fbf79b784175892d5ce1161b65a0066c
Current code mandates loading of SCP_BL2/SCP_BL2U images for all
CSS platforms. On future ARM CSS platforms, the Application
Processor (AP) might not need to load these images. So, these
items can be removed from the FIP on those platforms.
BL2 tries to load SCP_BL2/SCP_BL2U images if their base
addresses are defined causing boot error if the images are not
found in FIP.
This change adds a make flag `CSS_LOAD_SCP_IMAGES` which if set
to `1` does:
1. Adds SCP_BL2, SCP_BL2U images to FIP.
2. Defines the base addresses of these images so that AP loads
them.
And vice-versa if it is set to `0`. The default value is set to
`1`.
Change-Id: I5abfe22d5dc1e9d80d7809acefc87b42a462204a
Prior to this patch, it was assumed that on all ARM platforms the bare
minimal security setup required is to program TrustZone protection. This
would always be done by programming the TZC-400 which was assumed to be
present in all ARM platforms. The weak definition of
platform_arm_security_setup() in plat/arm/common/arm_security.c
reflected these assumptions.
In reality, each ARM platform either decides at runtime whether
TrustZone protection needs to be programmed (e.g. FVPs) or performs
some security setup in addition to programming TrustZone protection
(e.g. NIC setup on Juno). As a result, the weak definition of
plat_arm_security_setup() is always overridden.
When a platform needs to program TrustZone protection and implements the
TZC-400 peripheral, it uses the arm_tzc_setup() function to do so. It is
also possible to program TrustZone protection through other peripherals
that include a TrustZone controller e.g. DMC-500. The programmer's
interface is slightly different across these various peripherals.
In order to satisfy the above requirements, this patch makes the
following changes to the way security setup is done on ARM platforms.
1. arm_security.c retains the definition of arm_tzc_setup() and has been
renamed to arm_tzc400.c. This is to reflect the reliance on the
TZC-400 peripheral to perform TrustZone programming. The new file is
not automatically included in all platform ports through
arm_common.mk. Each platform must include it explicitly in a platform
specific makefile if needed.
This approach enables introduction of similar library code to program
TrustZone protection using a different peripheral. This code would be
used by the subset of ARM platforms that implement this peripheral.
2. Due to #1 above, existing platforms which implements the TZC-400 have been
updated to include the necessary files for both BL2, BL2U and BL31
images.
Change-Id: I513c58f7a19fff2e9e9c3b95721592095bcb2735
Current code assumes `SCP_COM_SHARED_MEM_BASE` as the base address
for BOM/SCPI protocol between AP<->SCP on all CSS platforms. To
cater for future ARM platforms this is made platform specific.
Similarly, the bit shifts of `SCP_BOOT_CONFIG_ADDR` are also made
platform specific.
Change-Id: Ie8866c167abf0229a37b3c72576917f085c142e8
Functions to configure the MMU in S-EL1 and EL3 on ARM platforms
expect each platform to export its memory map in the `plat_arm_mmap`
data structure. This approach does not scale well in case the memory
map cannot be determined until runtime. To cater for this possibility,
this patch introduces the plat_arm_get_mmap() API. It returns a
reference to the `plat_arm_mmap` by default but can be overridden
by a platform if required.
Change-Id: Idae6ad8fdf40cdddcd8b992abc188455fa047c74
ARM PL061 GPIO driver requires the "PLAT_PL061_MAX_GPIOS" definition.
By default, it's defined to 32 in PL061 GPIO driver. If user wants
more PL061 controllers in platform, user should define the build
flag in platform.mk instead.
Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>
Each ARM Compute Subsystem based platform implements a System Security
Control (SSC) Registers Unit. The SSC_VERSION register inside it carries
information to identify the platform. This enables ARM Trusted Firmware
to compile in support for multiple ARM platforms and choose one at
runtime. This patch adds macros to enable access to this register.
Each platform is expected to export its PART_NUMBER separately.
Additionally, it also adds juno part number.
Change-Id: I2b1d5f5b65a9c7b76c6f64480cc7cf0aef019422
This patch moves the definition of some macros used only on
ARM platforms from common headers to platform specific headers.
It also forces all ARM standard platforms to have distinct
definitions (even if they are usually the same).
1. `PLAT_ARM_TZC_BASE` and `PLAT_ARM_NSTIMER_FRAME_ID` have been
moved from `css_def.h` to `platform_def.h`.
2. `MHU_BASE` used in CSS platforms is moved from common css_def.h
to platform specific header `platform_def.h` on Juno and
renamed as `PLAT_ARM_MHU_BASE`.
3. To cater for different sizes of BL images, new macros like
`PLAT_ARM_MAX_BL31_SIZE` have been created for each BL image. All
ARM platforms need to define them for each image.
Change-Id: I9255448bddfad734b387922aa9e68d2117338c3f
This patch moves the private GIC common accessors from `gic_common.h` to
a new private header file `gic_common_private.h`. This patch also adds
additional comments to GIC register accessors to highlight the fact
that some of them access register values that correspond to multiple
interrupt IDs. The convention used is that the `set`, `get` and `clr`
accessors access and modify the values corresponding to a single interrupt
ID whereas the `read` and `write` GIC register accessors access the raw
GIC registers and it could correspond to multiple interrupt IDs depending
on the register accessed.
Change-Id: I2643ecb2533f01e3d3219fcedfb5f80c120622f9
The code to set the interrupt priority for secure interrupts in the
new GICv2 and GICv3 drivers is incorrect. The setup code to configure
interrupt priorities of secure interrupts, one interrupt at a time, used
gicd_write_ipriorityr()/gicr_write_ipriority() function affecting
4 interrupts at a time. This bug did not manifest itself because all the
secure interrupts were configured to the highest secure priority(0) during
cold boot and the adjacent non secure interrupt priority would be configured
later by the normal world. This patch introduces new accessors,
gicd_set_ipriorityr() and gicr_set_ipriorityr(), for configuring priority
one interrupt at a time and fixes the the setup code to use the new
accessors.
FixesARM-software/tf-issues#344
Change-Id: I470fd74d2b7fce7058b55d83f604be05a27e1341
GICD_IPRIORITYR and GICD_ITARGETSR specifically support byte addressing
so that individual interrupt priorities can be atomically updated by
issuing a single byte write. The previous implementation of
gicd_set_ipriority() and gicd_set_itargetsr() used 32-bit register
accesses, modifying values for 4 interrupts at a time, using a
read-modify-write approach. This potentially may cause concurrent changes
by other CPUs to the adjacent interrupts to be corrupted. This patch fixes
the issue by modifying these accessors to use byte addressing.
FixesARM-software/tf-issues#343
Change-Id: Iec28b5f5074045b00dfb8d5f5339b685f9425915
Currently, `ccn_snoop_dvm_domain_common()` is responsible for providing
a bitmap of HN-F and HN-I nodes in the interconnect. There is a request
node (RN) corresponding to the master interface (e.g. cluster) that needs
to be added or removed from the snoop/DVM domain. This request node is
removed from or added to each HN-F or HN-I node present in the bitmap
depending upon the type of domain.
The above logic is incorrect when participation of a master interface in
the DVM domain has to be managed. The request node should be removed
from or added to the single Miscellaneous Node (MN) in the system
instead of each HN-I node.
This patch fixes this by removing the intermediate
`ccn_snoop_dvm_domain_common()` and instead reads the MN registers to
get the needed node Id bitmap for snoop(HN-F bitmap) and DVM(MN bitmap)
domains.
Additionally, it renames `MN_DDC_SET_OFF` to `MN_DDC_SET_OFFSET` to
be inline with other macros.
Change-Id: Id896046dd0ccc5092419e74f8ac85e31b104f7a4
When BL31 is compiled at `-O3` optimization level using Linaro GCC 4.9
AArch64 toolchain, it reports the following error:
```
services/std_svc/psci/psci_common.c: In function 'psci_do_state_coordination':
services/std_svc/psci/psci_common.c:220:27: error: array subscript is above
array bounds [-Werror=array-bounds]
psci_req_local_pwr_states[pwrlvl - 1][cpu_idx] = req_pwr_state;
^
```
This error is a false positive and this patch resolves the error by asserting
the array bounds in `psci_do_state_coordination()`.
FixesARM-software/tf-issues#347
Change-Id: I3584ed7b2e28faf455b082cb3281d6e1d11d6495
In the Cortex-A35/A53/A57 CPUs library code, some of the CPU specific
reset operations are skipped if they have already been applied in a
previous invocation of the reset handler. This precaution is not
required, as all these operations can be reapplied safely.
This patch removes the unneeded test-before-set instructions in
the reset handler for these CPUs.
Change-Id: Ib175952c814dc51f1b5125f76ed6c06a22b95167
The LDNP/STNP instructions as implemented on Cortex-A53 and
Cortex-A57 do not behave in a way most programmers expect, and will
most probably result in a significant speed degradation to any code
that employs them. The ARMv8-A architecture (see Document ARM DDI
0487A.h, section D3.4.3) allows cores to ignore the non-temporal hint
and treat LDNP/STNP as LDP/STP instead.
This patch introduces 2 new build flags:
A53_DISABLE_NON_TEMPORAL_HINT and A57_DISABLE_NON_TEMPORAL_HINT
to enforce this behaviour on Cortex-A53 and Cortex-A57. They are
enabled by default.
The string printed in debug builds when a specific CPU errata
workaround is compiled in but skipped at runtime has been
generalised, so that it can be reused for the non-temporal hint use
case as well.
Change-Id: I3e354f4797fd5d3959872a678e160322b13867a1
This allows reproducible builds (same source and same compiler produce
bit-identical results) and also allows coordinating the timestamp across
multiple projects, eg. with another firmware.
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Move up to Base FVP version 7.2 (build 0.8/7202) and Foundation FVP version
9.5 (build 9.5.41) in the user guide.
Change-Id: Ie9900596216808cadf45f042eec639d906e497b2
Replaced a long dash in a comment by the ASCII character '-'. Support
for multibyte character in the source character set is not enforced by
the C99 standard. To maximize compatibility with C processing tools
(e.g. compilers or static code analysis tools), they should be removed.
Change-Id: Ie318e380d3b44755109f042a76ebfd2229f42ae3
