The PE target mask is used to translate linear PE index (returned by
platform core position) to a bit mask used when targeting interrupts to
a PE, viz. when raising SGIs and routing SPIs.
The platform shall:
- Populate the driver data with a pointer to array that's to contain
per-PE target masks.
- Invoke the new driver API 'gicv2_set_pe_target_mask()' during
per-CPU initialization so that the driver populates the target mask
for that CPU.
Platforms that don't intend to target interrupts or raise SGIs need not
populate this.
Change-Id: Ic0db54da86915e9dccd82fff51479bc3c1fdc968
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>
Document the API in separate platform interrupt controller API document.
Change-Id: If18f208e10a8a243f5c59d226fcf48e985941949
Co-authored-by: Yousuf A <yousuf.sait@arm.com>
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>
The MEM_PROTECT support adds a MMAP region for DRAM2, which when
building with TBBR support and OP-TEE tsp requires an additional
entry in the MMAP region array in BL2 - PLAT_ARM_MMAP_ENTRIES is
increased.
The MEM_PROTECT support also adds a new region in BL31, and when
BL31 is placed in DRAM, the memory mappings require an additional
translation table - MAX_XLAT_TABLES is increased.
Change-Id: I0b76260da817dcfd0b8f73a7193c36efda977625
Signed-off-by: David Cunado <david.cunado@arm.com>
This patch adds functions to save and restore GICv3 ITS registers during
system suspend. Please note that the power management of GIC ITS is
implementation defined. These functions only implements the
architectural part of the ITS power management and they do not restore
memory structures or register content required to support ITS. Even if
the ITS implementation stores structures in memory, an implementation
defined power down sequence is likely to be required to flush some
internal ITS caches to memory. If such implementation defined sequence
is not followed, the platform must ensure that the ITS is not power
gated during system suspend.
Change-Id: I5f31e5541975aa7dcaab69b0b7f67583c0e27678
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
During system suspend, the GICv3 Distributor and Redistributor context
can be lost due to power gating of the system power domain. This means
that the GICv3 context needs to be saved prior to system suspend and
restored on wakeup. Currently the consensus is that the Firmware should
be in charge of this. See tf-issues#464 for more details.
This patch introduces helper APIs in the GICv3 driver to save and
restore the Distributor and Redistributor contexts. The GICv3 ITS
context is not considered in this patch because the specification says
that the details of ITS power management is implementation-defined.
These APIs are expected to be appropriately invoked by the platform
layer during system suspend.
FixesARM-software/tf-issues#464
Change-Id: Iebb9c6770ab8c4d522546f161fa402d2fe02ec00
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
This patch introduces the ability of the xlat tables library to manage
EL0 and EL1 mappings from a higher exception level.
Attributes MT_USER and MT_PRIVILEGED have been added to allow the user
specify the target EL in the translation regime EL1&0.
REGISTER_XLAT_CONTEXT2 macro is introduced to allow creating a
xlat_ctx_t that targets a given translation regime (EL1&0 or EL3).
A new member is added to xlat_ctx_t to represent the translation regime
the xlat_ctx_t manages. The execute_never mask member is removed as it
is computed from existing information.
Change-Id: I95e14abc3371d7a6d6a358cc54c688aa9975c110
Co-authored-by: Douglas Raillard <douglas.raillard@arm.com>
Co-authored-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
Co-authored-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Introduce a variant of the TLB invalidation helper function that
allows the targeted translation regime to be specified, rather than
defaulting to the current one.
This new function is useful in the context of EL3 software managing
translation tables for the S-EL1&0 translation regime, as then it
might need to invalidate S-EL1&0 TLB entries rather than EL3 ones.
Define a new enumeration to be able to represent translation regimes in
the xlat tables library.
Change-Id: Ibe4438dbea2d7a6e7470bfb68ff805d8bf6b07e5
Co-authored-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
Co-authored-by: Douglas Raillard <douglas.raillard@arm.com>
Co-authored-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
The current implementation of the memory mapping API favours mapping
memory regions using the biggest possible block size in order to
reduce the number of translation tables needed.
In some cases, this behaviour might not be desirable. When translation
tables are edited at run-time, coarse-grain mappings like that might
need splitting into finer-grain tables. This operation has a
performance cost.
The MAP_REGION2() macro allows to specify the granularity of
translation tables used for the initial mapping of a memory region.
This might increase performance for memory regions that are likely to
be edited in the future, at the expense of a potentially increased
memory footprint.
The Translation Tables Library Design Guide has been updated to
explain the use case for this macro. Also added a few intermediate
titles to make the guide easier to digest.
Change-Id: I04de9302e0ee3d326b8877043a9f638766b81b7b
Co-authored-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
Co-authored-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Since commit 769d65da77 ("xlat: Use MAP_REGION macro as compatibility
layer"), building with GCC 4.9 fails.
CC plat/arm/board/fvp/fvp_common.c
plat/arm/board/fvp/fvp_common.c:60:2: error: initializer element is not constant
ARM_MAP_SHARED_RAM,
^
plat/arm/board/fvp/fvp_common.c:60:2: error: (near initialization for 'plat_arm_mmap[0]')
make: *** [Makefile:535: build/fvp/release/bl1/fvp_common.o] Error 1
Taking into account that MAP_REGION(_FLAT) is widely used in array
initializers, do not use cast.
Fixes: 769d65da77 ("xlat: Use MAP_REGION macro as compatibility layer")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
On ARM platforms, the maximum size of the address space is limited
to 32-bits as defined in arm_def.h. In order to access DRAM2, which
is defined beyond the 32-bit address space, the maximum address space
is increased to 36-bits in AArch64. It is possible to increase the
virtual space for AArch32, but it is more difficult and not supported
for now.
NOTE - the actual maximum memory address space is platform dependent
and is checked at run-time by querying the PARange field in the
ID_AA64MMFR0_EL1 register.
Change-Id: I6cb05c78a63b1fed96db9a9773faca04a5b93d67
Signed-off-by: Roberto Vargas <roberto.vargas@arm.com>
mem_protect needs some kind of non-volatile memory because it has
to remember its state across reset and power down events.
The most suitable electronic part for this feature is a NVRAM
which should be only accesible from the secure world. Juno and
FVP lack such hardware and for this reason the MEM_PROTECT
functionality is implemented with Flash EEPROM memory on both
boards, even though this memory is accesible from the non-secure
world. This is done only to show a full implementation of
these PSCI features, but an actual system shouldn't use a
non-secure NVRAM to implement it.
The EL3 runtime software will write the mem_protect flag and BL2
will read and clear the memory ranges if enabled. It is done in
BL2 because it reduces the time that TF needs access to the full
non-secure memory.
The memory layout of both boards is defined using macros which
take different values in Juno and FVP platforms. Generic platform
helpers are added that use the platform specific macros to generate
a mem_region_t that is valid for the platform.
Change-Id: I2c6818ac091a2966fa07a52c5ddf8f6fde4941e9
Signed-off-by: Roberto Vargas <roberto.vargas@arm.com>
This commit introduces a new type (mem_region_t) used to describe
memory regions and it adds two utility functions:
- clear_mem_regions: This function clears (write 0) to a set
of regions described with an array of mem_region_t.
- mem_region_in_array_chk This function checks if a
region is covered by some of the regions described
with an array of mem_region_t.
Change-Id: I12ce549f5e81dd15ac0981645f6e08ee7c120811
Signed-off-by: Roberto Vargas <roberto.vargas@arm.com>
This patch adds the generic code that links the psci smc handler
with the platform function that implements the mem_protect and
mem_check_range functionalities. These functions are optional
APIs added in PSCI v1.1 (ARM DEN022D).
Change-Id: I3bac1307a5ce2c7a196ace76db8317e8d8c8bb3f
Signed-off-by: Roberto Vargas <roberto.vargas@arm.com>
Add new option rsa+ecdsa for TF_MBEDTLS_KEY_ALG, which selects
rsa or ecdsa depending on the certificate used.
Change-Id: I08d9e99bdbba361ed2ec5624248dc382c750ad47
Signed-off-by: Qixiang Xu <qixiang.xu@arm.com>
For Trusted Board Boot, BL1 RW section and BL2 need more space to
support the ECDSA algorithm. Specifically, PLAT_ARM_MAX_BL1_RW_SIZE
is increased on ARM platforms.
And on the Juno platform:
- BL2 size, PLAT_ARM_MAX_BL2_SIZE is increased.
- SCP_BL2 is loaded into the space defined by BL31_BASE ->
BL31_RW_BASE. In order to maintain the same size space for
SCP_BL2,PLAT_ARM_MAX_BL31_SIZE is increased.
Change-Id: I379083f918b40ab1c765da4e71780d89f0058040
Co-Authored-By: David Cunado <david.cunado@arm.com>
Signed-off-by: Qixiang Xu <qixiang.xu@arm.com>
The type `unsigned long` is 32 bit wide in AArch32, but 64 bit wide in
AArch64. This is inconsistent and that's why we avoid using it as per
the Coding Guidelines. This patch changes all `UL` occurrences to `U`
or `ULL` depending on the context so that the size of the constant is
clear.
This problem affected the macro `BIT(nr)`. As long as this macro is used
to fill fields of registers, that's not a problem, since all registers
are 32 bit wide in AArch32 and 64 bit wide in AArch64. However, if the
macro is used to fill the fields of a 64-bit integer, it won't be able
to set the upper 32 bits in AArch32.
By changing the type of this macro to `unsigned long long` the behaviour
is always the same regardless of the architecture, as this type is
64-bit wide in both cases.
Some Tegra platform files have been modified by this patch.
Change-Id: I918264c03e7d691a931f0d1018df25a2796cc221
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
In the S-EL1&0 translation regime we aren't using the higher VA range,
whose translation table base address is held in TTBR1_EL1. The bit
TCR_EL1.EPD1 can be used to disable translations using TTBR1_EL1, but
the code wasn't setting it to 1. Additionally, other fields in TCR1_EL1
associated with the higher VA range (TBI1, TG1, SH1, ORGN1, IRGN1 and
A1) weren't set correctly as they were left as 0. In particular, 0 is a
reserved value for TG1. Also, TBBR1_EL1 was not explicitly set and its
reset value is UNKNOWN.
Therefore memory accesses to the higher VA range would result in
unpredictable behaviour as a translation table walk would be attempted
using an UNKNOWN value in TTBR1_EL1.
On the FVP and Juno platforms accessing the higher VA range resulted in
a translation fault, but this may not always be the case on all
platforms.
This patch sets the bit TCR_EL1.EPD1 to 1 so that any kind of
unpredictable behaviour is prevented.
This bug only affects the AArch64 version of the code, the AArch32
version sets this bit to 1 as expected.
Change-Id: I481c000deda5bc33a475631301767b9e0474a303
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
The value of BL31_BASE currently depends on the size of BL31. This
causes problems in the RESET_TO_BL31 case because the value of
BL31_BASE is used in the model launch parameters, which often changes.
Therefore, this patch fixes BL31_BASE to the middle of Trusted SRAM,
to avoid further model parameter changes in future.
Change-Id: I6d7fa4fe293717d84768974679539c0e0cb6d935
Signed-off-by: David Cunado <david.cunado@arm.com>
This patch provides helper macros for both Device and Normal memory MAIR
encodings as defined by the ARM Architecture Reference Manual for ARMv8-A
(ARM DDI0487B.A).
Change-Id: I5faae7f2cf366390ad4ba1d9253c6f3b60fd5e20
Signed-off-by: David Cunado <david.cunado@arm.com>
Use the MAP_REGION to build the mmap_region_t argument in wrappers like
mmap_add_region(). Evolution of the mmap_region_t might require adding
new members with a non-zero default value. Users of MAP_REGION are
protected against such evolution. This commit also protects users of
mmap_add_region() and mmap_add_dynamic_region() functions against these
evolutions.
Also make the MAP_REGION macro implementation more explicit and make it
a mmap_region_t compound literal to make it useable as a function
parameter on its own and to prevent using it in initialization of
variables of different type.
Change-Id: I7bfc4689f6dd4dd23c895b65f628d8ee991fc161
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
This patch gives users control over logging messages printed from the C
code using the LOG macros defined in debug.h Users now have the ability
to reduce the log_level at run time using the tf_log_set_max_level()
function. The default prefix string can be defined by platform by
overriding the `plat_log_get_prefix()` platform API which is also
introduced in this patch.
The new log framework results in saving of some RO data. For example,
when BL1 is built for FVP with LOG_LEVEL=LOG_LEVEL_VERBOSE, resulted
in saving 384 bytes of RO data and increase of 8 bytes of RW data. The
framework also adds about 108 bytes of code to the release build of FVP.
FixesARM-software/tf-issues#462
Change-Id: I476013d9c3deedfdd4c8b0b0f125665ba6250554
Co-authored-by: Eleanor Bonnici <Eleanor.bonnici@arm.com>
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
This patch introduces tf_vprintf() and tf_string_print() APIs
which is needed by the logging framework introduced in a later
patch.
Change-Id: Ie4240443d0e04e070502b51e371e546dd469fd33
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
Erratum 855971 applies to revision r0p3 or earlier Cortex-A72 CPUs. The
recommended workaround is to disable instruction prefetch.
Change-Id: I7fde74ee2a8a23b2a8a1891b260f0eb909fad4bf
Signed-off-by: Eleanor Bonnici <Eleanor.bonnici@arm.com>
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>
Erratum 855972 applies to revision r1p3 or earlier Cortex-A57 CPUs. The
recommended workaround is to disable instruction prefetch.
Change-Id: I56eeac0b753eb1432bd940083372ad6f7e93b16a
Signed-off-by: Eleanor Bonnici <Eleanor.bonnici@arm.com>
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>
This patch does the required changes to enable CSS platforms
to build and use the SDS framework. Since SDS is always coupled with
SCMI protocol, the preexisting SCMI build flag is now renamed to
`CSS_USE_SCMI_SDS_DRIVER` which will enable both SCMI and SDS on
CSS platforms. Also some of the workarounds applied for SCMI are
now removed with SDS in place.
Change-Id: I94e8b93f05e3fe95e475c5501c25bec052588a9c
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
This patch fixes the inconsistency with regards as to how
BL2_BASE/BL2U_BASE and BL2_LIMIT/BL2U_LIMIT macros are defined
when building Juno to run in AArch32 mode at EL3. In this case,
BL32 is compiled for AArch32 whereas BL1 and BL2 are compiled
for AArch64. This resulted in BL32 conditionally compiling a
different definition of the above mentioned macros from BL1/BL2.
This is fixed by taking into consideration the
JUNO_AARCH32_EL3_RUNTIME build flag as well in the conditional
compilation check.
Change-Id: I27ac68aa4df0502089c1739c05366a9c509eb5be
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
The commit 3eb2d67 optimizes the memory map for BL2 when TSP
is not present. But this also broke OP-TEE as it was reusing
the TSP mapping. This patch fixes this problem by adding a
separate mapping for OP-TEE in the BL2 memory map table.
Change-Id: I130a2ea552b7b62d8478081feb1f4ddf5292a118
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
Platform may use specific cache line sizes. Since CACHE_WRITEBACK_GRANULE
defines the platform specific cache line size, it is used to define the
size of the cpu data structure CPU_DATA_SIZE aligned on cache line size.
Introduce assembly macro 'mov_imm' for AArch32 to simplify implementation
of function '_cpu_data_by_index'.
Change-Id: Ic2d49ffe0c3e51649425fd9c8c99559c582ac5a1
Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
This patch ensures that the ARM_MAP_TSP_SEC_MEM memory region is mapped
in BL2 only if the TSPD has been included in the build. This saves one
entry in the plat_arm_mmap[] array and avoids to map extra memory when
it's not needed.
Change-Id: I6ae60822ff8f0de198145925b0b0d45355179a94
Signed-off-by: Achin Gupta <achin.gupta@arm.com>
Signed-off-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
Add Call Frame Information assembler directives to vector entries so
that debuggers display the backtrace of functions that triggered a
synchronous exception. For example, a function triggering a data abort
will be easier to debug if the backtrace can be displayed from a
breakpoint at the beginning of the synchronous exception vector.
DS-5 needs CFI otherwise it will not attempt to display the backtrace.
Other debuggers might have other needs. These debug information are
stored in the ELF file but not in the final binary.
Change-Id: I32dc4e4b7af02546c93c1a45c71a1f6d710d36b1
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
Assembly routines are usually defined by using "func" and "endfunc":
func foo
...
endfunc foo
Currently, the "func" macro does not specify ".align" directive
by default. It causes unaligned instruction under some circumstances.
As far as I tested, this problem happens for GCC 5 or older. It did
not happen for GCC 6 or newer. Taking into account that GCC 4.x / 5.x
is still used, make sure that assembly code is at least 4 byte aligned.
[ How to reproduce the problem ]
For example, use GCC 5.3 downloaded from Linaro:
http://releases.linaro.org/components/toolchain/binaries/5.3-2016.05/
aarch64-linux-gnu/gcc-linaro-5.3.1-2016.05-x86_64_aarch64-linux-gnu.tar.xz
Expand mbedtls-2.4.2 to the current directory.
Try the following:
$ git log --oneline -1
77544ef Merge pull request #1071 from jeenu-arm/syntax-fix
$ aarch64-linux-gnu-gcc --version | head -1
aarch64-linux-gnu-gcc (Linaro GCC 5.3-2016.05) 5.3.1 20160412
$ make CROSS_COMPILE=aarch64-linux-gnu- PLAT=uniphier \
TRUSTED_BOARD_BOOT=1 MBEDTLS_DIR=mbedtls-2.4.2
( snip build log )
$ aarch64-linux-gnu-nm build/uniphier/release/bl1/bl1.elf | grep handler
00000000800088f4 T bl1_fwu_smc_handler
00000000800084c8 T bl1_smc_handler
000000008000a6e0 t _panic_handler
000000008000a8e0 W plat_error_handler
000000008000a8e8 W plat_panic_handler
000000008000a8d8 W plat_reset_handler
000000008000a39f T reset_handler
000000008000a367 t smc_handler
000000008000a2ef t smc_handler64
You will notice "smc_handler64", "reset_handler", etc. are not properly
aligned.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
At present, various CPU register macros that refer to CPUACTLR are named
ACTLR. This patch fixes that.
The previous register names are retained, but guarded by the
ERROR_DEPRECATED macro, so as not to break platforms that continue using
the old names.
Change-Id: Ia872196d81803f8f390b887d149e0fd054df519b
Signed-off-by: Eleanor Bonnici <Eleanor.bonnici@arm.com>
In order to avoid Undefined behavior, left operand in left-shift
expressions needs to be unsigned, and of sufficient size. The safest and
most consistent approach is to use unsigned long long type.
Change-Id: I9612f16a6e6ea4c7df62a02497d862abf19b8e1b
Signed-off-by: Eleanor Bonnici <Eleanor.bonnici@arm.com>