Starting with QEMU v3.1.0 (Dec 2018), QEMU's TCG emulation engine supports
the SVE architecture extension. In QEMU v7.1.0 (Aug 2022) it also gained
SME support.
As it stands today, running TF-A under QEMU with "-cpu max" makes Linux
hang, because SME and SVE accesses trap to EL3, but are never handled
there. This is because the Linux kernel sees the SVE or SME feature bits,
and assumes firmware has enabled the feature for lower exception levels.
This requirement is described in the Linux kernel booting protocol.
Enable those features in the TF-A build, so that BL31 does the proper
EL3 setup to make the feature usable in non-secure world.
We check the actual feature bits before accessing SVE or SME registers,
so this is safe even for older QEMU version or when not running with
-cpu max. As SVE and SME are AArch64 features only, do not enable them
when building for AArch32.
Change-Id: I5b718eb298a0bbcf36244479e8d42e54a2faca61
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Hardware video decoding is not working after enabling EMI MPU protection
for SCP.
According to coreboot DEVAPC setting, SCP belongs to domain 4 instead of
domain 3. So correct the permission setting.
BUG=b:249954378
TEST=play video and see codec irq count is incrementing.
Signed-off-by: Yidi Lin <yidilin@chromium.org>
Change-Id: If71de3eabf8682909f96924c159aa92f25deb96c
In order to sync drivers with MediaTek internal code base, we move lpm
drivers back to common folder.
Signed-off-by: Bo-Chen Chen <rex-bc.chen@mediatek.com>
Change-Id: I1066e092febe0abb9782a46f668613e137737c88
- Add cpu_pm driver for CPU idle and SMP flow.
- Add SMP driver for CPU power on/off control.
- Add CPC driver to handle CPU powered on/off in CPU suspend.
- Add mbox driver for tinysys support.
Signed-off-by: Edward-JW Yang <edward-jw.yang@mediatek.com>
Change-Id: I20141474e1c43cdfacb9f2c6a2285721e50a617c
The new helpers are created in STM32MP1 platform for prefetch and data
aborts.
While at it, put plat_report_exception() under DEBUG flag. If DEBUG is
not set, the weak function which does the same will be used.
This plat_report_exception() function can also be simplified, as it will
no more be used to report aborts.
Change-Id: Ibe989b28e236693f317cffb0545ea0611b7bdde4
Signed-off-by: Yann Gautier <yann.gautier@st.com>
New helper functions are created to handle data & prefetch aborts
in AARCH32. They call platform functions, just like what
report_exception is doing.
As extended MSR/MRS instructions (to access lr_abt in monitor mode)
are only available if CPU (Armv7) has virtualization extension,
the functions branch to original report_exception handlers if this is
not the case.
Those new helpers are created mainly to distinguish data and prefetch
aborts, as they both share the same mode.
This adds 40 bytes of code.
Change-Id: I5dd31930344ad4e3a658f8a9d366a87a300aeb67
Signed-off-by: Yann Gautier <yann.gautier@st.com>
The purpose of this code is to extract api_id from smc_fid but this masking
is done already in the code with using generic mask from smccc.h
(FUNCID_NUM_MASK). That's why remove FUNCID_MASK is which not needed and
actually also equal to already used FUNCID_NUM_MASK.
Signed-off-by: Michal Simek <michal.simek@amd.com>
Change-Id: I1113825baa5d9d58d9d7c5d9d5855fecf62e8d45
On one hand, there is currently no upstream platform supporting the
RSS. On the other hand, we are gradually introducing driver code for
RSS. Even though we cannot test this code in the TF-A CI right now, we
can at least build it to make sure no build regressions are introduced
as we continue development.
This patch adds support for overriding PLAT_RSS_NOT_SUPPORTED build
flag (which defaults to 1 on the Base AEM FVP) from the command
line. This allows introducing an ad-hoc CI build config with
PLAT_RSS_NOT_SUPPORTED=0, which will correctly pull in the RSS and MHU
source files. Of course, the resulting firmware will not be
functional.
Change-Id: I2b0e8dd03bf301e7063dd4734ea5266b73265be1
Signed-off-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
Neoverse Reference Design platform RD-Edmunds has been renamed to RD-V2
and so all corresponding references have been changed.
Signed-off-by: Joel Goddard <joel.goddard@arm.com>
Change-Id: I134f125f8ce9ec2f42988ecd742de307da936f2b
Neoverse Demeter CPU has been renamed to Neoverse V2 CPU.
Correspondingly, update the CPU library, file names and other
references to use the updated IP name.
Signed-off-by: Joel Goddard <joel.goddard@arm.com>
Change-Id: Ia4bf45bf47807c06f4c966861230faea420d088f
* changes:
feat(stm32mp1): add early console in SP_min
feat(st): properly manage early console
feat(st-uart): manage STM32MP_RECONFIGURE_CONSOLE
docs(st): introduce STM32MP_RECONFIGURE_CONSOLE
feat(st): add trace for early console
fix(stm32mp1): enable crash console in FIQ handler
feat(st-uart): add initialization with the device tree
refactor(stm32mp1): move DT_UART_COMPAT in include file
feat(stm32mp1): configure the serial boot load address
fix(stm32mp1): update the FIP load address for serial boot
refactor(st): configure baudrate for UART programmer
refactor(st-uart): compute the over sampling dynamically
According to the Generic Names Recommendation in the Devicetree
Specification Release v0.3, and the DT Bindings for the Renesas Reduced
Pin Count Interface, the node name for a Renesas RPC-IF device should be
"spi". The node name matters, as the node is enabled by passing a DT
fragment from TF-A to subsequent software.
Fix this by renaming the device node in the passed DT fragment from
"rpc" to "spi".
Fixes: 12c75c8886 ("feat(plat/rcar3): emit RPC status to DT fragment if RPC unlocked")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Change-Id: Idb43353947607611331abc344f8c8ae932a20408
Since patch on libc refactoring, there is a compilation error with
STM32MP_USB_PROGRAMMER=1:
plat/st/common/stm32cubeprogrammer_usb.c:81:35: error:
implicit declaration of function 'strnlen'
[-Werror=implicit-function-declaration]
length += strnlen((char *)&dfu->buffer[GET_PHASE_LEN],
The string.h header file should be included.
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Change-Id: I1fbb2d9714cbc0d0640cb5e3c5ae8201dbfbe14e
MISRA Violation: MISRA-C: 2012 R.10.1
- The operand to the operator does not have an essentially
unsigned type.
Signed-off-by: HariBabu Gattem <haribabu.gattem@amd.com>
Change-Id: I0f974e9d6f63dddfab55d55c952a57645d931e40
Update test CCA Platform token in fvp_plat_attest_token.c to be
up-to-date with RMM spec Beta0.
Change-Id: I0f5e2ac1149eb6f7a93a997682f41d90e109a049
Signed-off-by: Mate Toth-Pal <mate.toth-pal@arm.com>
DDR4 Chip is EOL during redesign of ls1043ardb pd version. The replacement from MT is MT40A1G8SA-062E:R.
New ddr configure is compatible with both pd and old version of ls1043ardb.
Signed-off-by: Chunlei Xu <chunlei.xu@nxp.com>
Change-Id: I714c091a2cf15046438d0723fb55a4410c386ef4
MISRA Violation: MISRA-C:2012 R.4.6
- Using basic numerical type int rather than a typedef
that includes size and signedness information.
Signed-off-by: HariBabu Gattem <haribabu.gattem@amd.com>
Change-Id: I3779f7b6e074e33cb66ace3bef2117029badce1e
Allow early console to be used at the beginning of SP_min, before
the clocks and UART have been reconfigured.
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Change-Id: I53d66938d42fcec830d9b81e5ef62b3790d0c3b3
The new flag STM32MP_RECONFIGURE_CONSOLE is managed in platform.mk.
It is used in stm32mp_setup_early_console() when calling
plat_crash_console_init(). This call is also under:
"#if defined(IMAGE_BL2)"
as this crash console init shouldn't be done by default in BL32.
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Change-Id: Ib6b89db83d80095b662a2016e18ceb3fa8668435
When the early console is configured with STM32MP_EARLY_CONSOLE,
display a message indicating it is enabled.
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Change-Id: Iafdfa5afef27eba823d707841853a8a46de0b42d
When a FIQ occurs and is trapped by SP_min, it is an unrecoverable
error. As kernel may have switched the UART console off, we should
re-enable it with plat_crash_console_init() for those failing states.
Signed-off-by: Yann Gautier <yann.gautier@st.com>
Change-Id: Ib02e1271b6213f8e383a062b74494abf8826188f
Move the definition of DT_UART_COMPAT in stm32mp1_def.h to be used
in several files.
Change-Id: I74d0350bcd971df9b15697f2b9ec04061d6a7656
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
For product with 128MB DDR size, the OP-TEE is located at the end
of the DDR and the FIP can't be loaded at the default location
because it overlap the OP-TEE final location. So the default value
for DWL_BUFFER_BASE is invalid.
To avoid this conflict the serial boot load address = DWL_BUFFER_BASE
can be modified with a configuration flags.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Change-Id: Ie27b87c10c57fea5d4c6200ce4f624e775b9a080
Update the FIP load address and size for serial boot to support
product with a DDR size = 128MB
1/ Move the FIP location at the end of the first 128MB
2/ Reduce the DWL_BUFFER_SIZE to 16MB, to be coherent with the value
indicated in USB enumeration
- for STM32MP13x: "@SSBL /0x03/1*16Me"
- for STM32MP15x: "@Partition3 /0x03/1*16Me"
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Change-Id: Id93bf00c64832c17426bfd78e060861275677ecc
Add the possibility to configure the UART baudrate; reused the
console configuration, defined in STM32MP_UART_BAUDRATE.
The default value remains 115200.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Change-Id: Ifcf2b36e8ac929265405bc88e824ee78be3b5bbb
The parameter over_sampling of stm32_uart_init_s is not required
as it can be computed dynamically from clock rate of the serial
device and the requested baudrate.
Oversampling by 8 is allowed only for higher speed
(up to clock_rate / 8) to reduce the maximum receiver tolerance
to clock deviation.
This patch update the driver, the serial init struct and the
only user, the stm32cubeprogrammer over uart support.
Change-Id: I422731089730a288defeb7fa49886db65d0902b2
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
RD-N2* variants of Neoverse reference design platforms could be
configured to boot from SRAM or DRAM. Having ARM_BL31_IN_DRAM set to 1
within the common makefile would deter these platforms from having this
flexibility. Remove the default override configuration for
`ARM_BL31_IN_DRAM`.
Signed-off-by: Rohit Mathew <rohit.mathew@arm.com>
Change-Id: I8d79969c003a984675cbe705de890b51a1f7f4ea
Update SRAM size for Neoverse reference design platforms from 256KB to
512KB. This is required to place and execute BL31 image from the
on-chip SRAM. Additionally, revise BL31 image size to accommodate
larger BL31 images of multi-chip platforms.
Signed-off-by: Rohit Mathew <rohit.mathew@arm.com>
Change-Id: I11c2672a1089f24a9fafcf6555b8e1d52032cfde
It is more suitable to place platform_def.h in mt8188/include.
Signed-off-by: Bo-Chen Chen <rex-bc.chen@mediatek.com>
Change-Id: I35720690ff4f2ca99c9430edb8bbe17edf9aefb9
Add MCUSYS drivers support for MT8188.
Signed-off-by: Edward-JW Yang <edward-jw.yang@mediatek.com>
Change-Id: I84107702a9fd021c37d2997ad25b321a483a1a66
DFD (Design for Debug) is a debugging tool, which scans flip-flops and
dumps to internal RAM on the WDT reset. After system reboots, those
values could be showed for debugging.
TEST=build pass.
BUG=b:244216434
Signed-off-by: Fengquan Chen <fengquan.chen@mediatek.corp-partner.google.com>
Change-Id: I468036131e941a46bc1ec12d33105146000730d8
DCM means dynamic clock management, and it can dynamically
slow down or gate clocks during CPU or bus idle.
1. Add MCUSYS related DCM drivers.
2. Enable MCUSYS related DCM by default.
Signed-off-by: Garmin Chang <garmin.chang@mediatek.com>
Change-Id: I131354d72bbc190af504e9639bcc85a720e2bb17
To use power manager function more easier, we add some drivers to let
the implementation easier.
Signed-off-by: Edward-JW Yang <edward-jw.yang@mediatek.com>
Change-Id: Ibc6e1680c4534592ed37de49da39b6667f468ea1
To support more LPM feature, we add more options for build helper.
Signed-off-by: Edward-JW Yang <edward-jw.yang@mediatek.com>
Change-Id: I17eeedbe0674e321f1891074ba0c72d858841dae
