If FEAT_PMUv3 is implemented and PMEVTYPER<n>(_EL0).MT bit is implemented
as well, it is possible to control whether PMU counters take into account
events happening on other threads.
If FEAT_MTPMU is implemented, EL3 (or EL2) can override the MT bit
leaving it to effective state of 0 regardless of any write to it.
This patch introduces the DISABLE_MTPMU flag, which allows to diable
multithread event count from EL3 (or EL2). The flag is disabled
by default so the behavior is consistent with those architectures
that do not implement FEAT_MTPMU.
Signed-off-by: Javier Almansa Sobrino <javier.almansasobrino@arm.com>
Change-Id: Iee3a8470ae8ba13316af1bd40c8d4aa86e0cb85e
Manish Pandey [Wed, 9 Dec 2020 22:44:44 +0000 (22:44 +0000)]
Merge changes from topic "versal-bug-fixes-and-new-apis" into integration
* changes:
plat: xilinx: versal: Add support of register notifier
plat: xilinx: versal: Add support to get clock rate value
plat: xilinx: versal: Add support of set max latency for the device
plat: versal: Add InitFinalize API call
xilinx: versal: Updated Response of QueryData API call
plat:xilinx:versal: Use defaults when PDI is without sw partitions
plat: xilinx: Mask unnecessary bytes of return error code
xilinx: versal: Skip store/restore of GIC during CPU idle
plat: versal: Update API list in feature check
xilinx: versal: Do not pass ACPU0 always in set_wakeup_source()
Olivier Deprez [Wed, 9 Dec 2020 15:08:27 +0000 (15:08 +0000)]
Merge changes from topic "secure_no_primary" into integration
* changes:
spm: provide number of vCPUs and VM size for first SP
spm: remove chosen node from SPMC manifests
spm: move OP-TEE SP manifest DTS to FVP platform
spm: update OP-TEE SP manifest with device-regions node
spm: remove device-memory node from SPMC manifests
Marek Vasut [Sun, 8 Nov 2020 18:13:32 +0000 (19:13 +0100)]
rcar_gen3: drivers: console: Treat log as device memory
The BL31 log driver is registered before the xlat tables are initialized,
at that point the log memory is configured as device memory and can only
be accessed with up-to-32bit aligned accesses. Adjust the driver to do
just that.
The memset() call has to be replaced by a loop of 32bit writes to the log,
the memcpy() is trivial to replace with a single 32bit write of the entire
TLOG word. In the end, this even simplifies the code.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Change-Id: Ie9152e782e67d93e7236069a294df812e2b873bf
Ravi Patel [Tue, 18 Sep 2018 09:14:12 +0000 (02:14 -0700)]
zynqmp: pm: Update flags in common clk divisor node
Current implementation doesn't support change of div1 value if clk
has 2 divisor because div1 clk is the parent of the div2 clk and div2
clk does not have SET_RATE_PARENT flag.
This causes div1 value to be fixed and only value of div2 will be
adjusted according to required clock rate.
Example:
Consider a case of nand_ref clock which has 2 divisor and 1 mux.
The frequency of mux clock is 1500MHz and default value of div1 and
div2 is 15 and 1 respectively. So the final clock will be of 100MHz.
When driver requests 80MHz for nand_ref clock, clock framework will
adjust the div2 value to 1 (setting div2 value 2 results final clock
to 50MHz which is more inaccurate compare to 100Mhz) which results
final clock to 100MHz.
Ideally the value of div1 and div2 should be updated to 19 and 1
respectively so that final clock goes to around 78MHz.
This patch fixes above problem by allowing change in div1 value.
Signed-off-by: Ravi Patel <ravi.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: Ibb98f6748d28653fdd1e59bf433b6a37ce9c1a58
Olivier Deprez [Wed, 25 Nov 2020 09:29:41 +0000 (10:29 +0100)]
spm: provide number of vCPUs and VM size for first SP
The primary VM concept is removed from the SPMC.
Update the SPMC manifests with number of Execution Contexts
and SP workspace size for the first Secure Partition (as it
is done for NWd secondary VMs and other SPs).
Yuchen Huang [Sat, 1 Aug 2020 08:23:12 +0000 (16:23 +0800)]
mediatek: mt8192: add uart save and restore api
When system resume, we want to print log as soon as possible.
So we add uart save and restore api, and they will be called
when systtem suspend and resume.
Change-Id: I83b477fd2b39567c9c6b70534ef186993f7053ae Signed-off-by: Yuchen Huang <yuchen.huang@mediatek.com> Signed-off-by: Roger Lu <roger.lu@mediatek.com>
G.Pangao [Fri, 6 Nov 2020 01:20:25 +0000 (09:20 +0800)]
mediatek: mt8192: modify sys_cirq driver
1.Modify this driver to make it more complete and more standard.
2.And makes this driver available for more IC services.
3.Solve some bugs in the software.
Manish Pandey [Mon, 7 Dec 2020 11:29:46 +0000 (11:29 +0000)]
Merge changes from topic "marvell-a3k-makefile" into integration
* changes:
plat: marvell: armada: a3k: Simplify check if WTP variable is defined
plat: marvell: armada: a3k: Split building $(WTMI_MULTI_IMG) and $(TIMDDRTOOL)
plat: marvell: armada: Maximal size of bl1 image in mrvl_bootimage is 128kB
plat: marvell: armada: Add missing FORCE, .PHONY and clean targets
plat: marvell: armada: a3k: Use make ifeq/endif syntax for $(MARVELL_SECURE_BOOT) code
plat: marvell: armada: a3k: Build $(WTMI_ENC_IMG) in $(BUILD_PLAT) directory
plat: marvell: armada: a3k: Do not remove external WTMI image files outside of TF-A repository
plat: marvell: armada: a3k: Do not modify $(WTMI_MULTI_IMG)
plat: marvell: armada: a3k: Do not modify $(WTMI_IMG)
Tejas Patel [Wed, 25 Nov 2020 09:56:57 +0000 (01:56 -0800)]
plat: xilinx: versal: Add support of register notifier
Add support of register notifier.
Signed-off-by: Tejas Patel <tejas.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I41ef4c63abcc9aee552790b843adb25a5fd0c23e
plat: xilinx: versal: Add support to get clock rate value
Add support to get clock's rate value.
Signed-off-by: Tejas Patel <tejas.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I3ed881053ef323b2ca73e13edd0affda860d381d
Tejas Patel [Wed, 25 Nov 2020 09:53:12 +0000 (01:53 -0800)]
plat: xilinx: versal: Add support of set max latency for the device
Add support of set max latency, to change in the maximum powerup latency
requirements for a specific device currently used by Subsystem.
Signed-off-by: Tejas Patel <tejas.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I8749886abb1a7884a42c4d156d89c9cd562a5b1a
Ravi Patel [Mon, 12 Aug 2019 10:10:10 +0000 (03:10 -0700)]
plat: versal: Add InitFinalize API call
Add support to call InitFinalize API in Versal which calls
corresponding LibPM API.
Signed-off-by: Ravi Patel <ravi.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I3428b7245b4db1ef6db8a90b7ad20b6e484ed3b2
Rajan Vaja [Mon, 23 Nov 2020 12:13:54 +0000 (04:13 -0800)]
xilinx: versal: Updated Response of QueryData API call
For the current XilPM calls, The handler of IPI returns information
with 16 Bytes data.
So during QueryData API call for the ClockName and PinFunctionName,
response data(name of clock or function) response[0..3] are used to
return name. And status is not being returned for such API.
Updated XilPM calls reply in a consistent way and The handler of IPI
return information with 32Bytes data. Where response[0] always set
to status.
For the version-2 of QueryData API, during call for the ClockName
and PinFunctionName, response data(name of clock or function) get as
response[1...4].
To support both the version of QueryData API, added version based
compatibility by the use of feature check.
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@xilinx.com> Signed-off-by: Amit Sunil Dhamne <amit.sunil.dhamne@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I336128bff7bbe659903b0f8ce20ae6da7e3b51b4
Ravi Patel [Mon, 23 Nov 2020 12:19:08 +0000 (04:19 -0800)]
plat: xilinx: Mask unnecessary bytes of return error code
Versal firmware adds extra error codes along with PM error codes
while sending response to driver. This makes incorrect error
identification at driver side.
To fix this, mask the unnecessary error bytes before sending the
error code to the driver.
Signed-off-by: Ravi Patel <ravi.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I18c2f3bd2d067e91344852c2f0c1bafea0e6eb23
Ravi Patel [Fri, 21 Jun 2019 12:00:49 +0000 (05:00 -0700)]
xilinx: versal: Skip store/restore of GIC during CPU idle
GIC registers needs to be stored/restored during system
suspend/resume only and not during CPU idle.
During CPU idle, minimum 1 CPU is in ON state.
Signed-off-by: Ravi Patel <ravi.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I5de2ce3a61bf4260f9385349202b0f592a47aaba
Add below API in feature check list which is actually present in
firmware:
- PM_GET_CHIPID
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@xilinx.com> Signed-off-by: Ravi Patel <ravi.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I98b82da74164f065c8835861f74b0f2855e9bcbf
Ravi Patel [Tue, 30 Jul 2019 11:10:07 +0000 (04:10 -0700)]
xilinx: versal: Do not pass ACPU0 always in set_wakeup_source()
Existing code passes ACPU0 to LibPM as node_id in set_wakeup_source()
call because last suspending core will be ACPU0 in most of the case.
Now it may be possible that user may disable the ACPU0 using hot-plug
and after that it suspends Linux. So in that case ACPU0 will not be
last suspending core.
To overcome above scenario, pass the current running processor ID
while calling set_wakeup_source().
Signed-off-by: Ravi Patel <ravi.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: If15354c2150b5bb1305b5f93ca4e8c7a81d59f0a
Pali Rohár [Mon, 23 Nov 2020 18:49:23 +0000 (19:49 +0100)]
plat: marvell: armada: a3k: Split building $(WTMI_MULTI_IMG) and $(TIMDDRTOOL)
These two targets are build by make subprocesses and are independent.
So splitting them into own targets allow make to build them in parallel.
$(TIMBUILD) script depends on $(TIMDDRTOOL) so specify it in Makefile.
Pali Rohár [Mon, 23 Nov 2020 18:45:28 +0000 (19:45 +0100)]
plat: marvell: armada: Add missing FORCE, .PHONY and clean targets
FORCE target is used as a dependency for other file targets which needs to
be always rebuilt. .PHONY target is standard Makefile target which specify
non-file targets and therefore needs to be always rebuilt.
Targets clean, realclean and distclean are .PHONY targets used to remove
built files. Correctly set that mrvl_clean target is prerequisite for these
clean targets to ensure that built files are removed.
Finally this change with usage of FORCE target allows to remove mrvl_clean
hack from the prerequisites of a8k ${DOIMAGETOOL} target which was used
just to ensure that ${DOIMAGETOOL} is always rebuilt via make subprocess.
Pali Rohár [Mon, 23 Nov 2020 18:14:40 +0000 (19:14 +0100)]
plat: marvell: armada: a3k: Do not modify $(WTMI_IMG)
$(WTMI_IMG) is used only by $(MAKE) subprocess in $(DOIMAGEPATH) directory.
So calling truncate on $(WTMI_IMG) after $(MAKE) in $(DOIMAGEPATH) has no
effect and can just damage input file for future usage. Therefore remove
this truncate call.
plat: xilinx: zynqmp: Enable log messages for debug
Save some space by enabling the log messages like bl33 address
only for debug builds. Also check the bl33 and bl32 address and
print only if this is not NULL.
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I58d846bf69a75e839eb49abcbb9920af13296886
Tejas Patel [Mon, 23 Nov 2020 07:37:55 +0000 (23:37 -0800)]
plat: zynqmp: Change macro name of PM_BOOT_HEALTH_STATUS_REG
For boot health status PMU Global General Storage Register 4 is
used. GGS4 can be used for other purpose along with boot health
status. So, change its name from PM_BOOT_HEALTH_STATUS_REG
to PMU_GLOBAL_GEN_STORAGE4.
Signed-off-by: Tejas Patel <tejas.patel@xilinx.com> Signed-off-by: Rajan Vaja <rajan.vaja@xilinx.com>
Change-Id: I2f5c4c6a161121e7cdb4b9f0f8711d0dad16c372
Update the xilinx platform makefile to include GICv2 makefile
instead of adding the individual files. Updating this change
as per the latest changes done in the commit #1322dc94f7.
Alexei Fedorov [Tue, 1 Dec 2020 13:22:25 +0000 (13:22 +0000)]
Aarch64: Add support for FEAT_MTE3
This patch provides the following changes:
- Adds definition for FEAT_MTE3 value in ID_AA64PFR1_EL1 register
- Enables Memory Tagging Extension for FEAT_MTE3.
Alexei Fedorov [Wed, 25 Nov 2020 14:07:05 +0000 (14:07 +0000)]
Aarch64: Add support for FEAT_PANx extensions
This patch provides the changes listed below:
- Adds new bit fields definitions for SCTLR_EL1/2 registers
- Corrects the name of SCTLR_EL1/2.[20] bit field from
SCTLR_UWXN_BIT to SCTLR_TSCXT_BIT
- Adds FEAT_PANx bit field definitions and their possible
values for ID_AA64MMFR1_EL1 register.
- Adds setting of SCTLR_EL1.SPAN bit to preserve PSTATE.PAN
on taking an exception to EL1 in spm_sp_setup() function
(services\std_svc\spm_mm\spm_mm_setup.c)
Jessica Clarke [Sun, 25 Oct 2020 18:10:24 +0000 (18:10 +0000)]
fdts: Remove "virtio-rng" from Morello FVP
This is not a standard string that any kernel recognises, nor do any of
the FDTs embedded in kernels specify this, nor does QEMU's virt machine.
Whilst its presence does no harm, it's not a thing code should consult
as a result, and so drop it in order to not cause confusion and risk
incorrect code being written to search for it.
Manish Pandey [Thu, 19 Nov 2020 11:18:26 +0000 (11:18 +0000)]
Merge changes from topic "mrvl_bootimage" into integration
* changes:
docs: marvell: Update build documentation to reflect mrvl_bootimage and mrvl_flash changes
plat: marvell: armada: Add new target mrvl_bootimage
plat: marvell: armada: a3k: Add support for building $(DOIMAGETOOL)
Pali Rohár [Tue, 10 Nov 2020 15:34:48 +0000 (16:34 +0100)]
plat: marvell: Update SUBVERSION to match Marvell's forked version
Marvell's TF-A fork has SUBVERSION set to devel-18.12.2.
The only differences between Marvell's devel-18.12.0 and devel-18.12.2
versions are documentation updates and cherry-picked patches from TF-A
upstream repository.
So upstream TF-A has already all changes from Marvell's TF-A devel-18.12.2
fork and therefore update SUBVERSION to reflect this state.
Pali Rohár [Thu, 29 Oct 2020 15:50:19 +0000 (16:50 +0100)]
plat: marvell: armada: Add new target mrvl_bootimage
This new target builds boot-image.bin binary as described in documentation.
This image does not contain WTMI image and therefore WTP repository is not
required for building.
Having ability to build just this boot-image.bin binary without full
flash-image.bin is useful for A3720 Turris MOX board which does not use
Marvell's WTP and a3700_utils.
To reduce duplicity between a8k and a3k code, define this new target and
also definitions for $(BUILD_PLAT)/$(BOOT_IMAGE) in common include file
marvell_common.mk.
For this purpose it is needed to include plat/marvell/marvell.mk file from
a3700_common.mk unconditionally (and not only when WTP is defined). Now
when common file plat/marvell/marvell.mk does not contain definition for
building $(DOIMAGETOOL), it is possible to move its inclusion at the top of
the a3700_common.mk file.
Pali Rohár [Thu, 29 Oct 2020 15:44:45 +0000 (16:44 +0100)]
plat: marvell: armada: a3k: Add support for building $(DOIMAGETOOL)
Current binary wtptp/linux/tbb_linux which is specified in $(DOIMAGETOOL)
variable points to external pre-compiled Marvell x86_64 ELF linux binary
from A3700-utils-marvell WTP repository.
It means that currently it is not possible to compile TF-A for A3720 on
other host platform then linux x86_64.
Part of the A3700-utils-marvell WTP repository is also source code of
$(DOIMAGETOOL) TBB_Linux tool.
This change adds support for building $(DOIMAGETOOL) also for a3k platform.
After running $(MAKE) at appropriate subdirectory of A3700-utils-marvell
WTP repository, compiled TBB_linux tool will appear in WTP subdirectory
wtptp/src/TBB_Linux/release/. So update also $(DOIMAGETOOL) variable to
point to the correct location where TBB_linux was built.
To build TBB_linux it is required to compile external Crypto++ library
which is available at: https://github.com/weidai11/cryptopp.git
User needs to set CRYPTOPP_PATH option to specify path to that library.
After this change it is now possible to build whole firmware for A3720
platform without requirement to use pre-compiled/proprietary x86_64
executable binaries from Marvell.
Alexei Fedorov [Fri, 13 Nov 2020 12:36:49 +0000 (12:36 +0000)]
TSP: Fix GCC 11.0.0 compilation error.
This patch fixes the following compilation error
reported by aarch64-none-elf-gcc 11.0.0:
bl32/tsp/tsp_main.c: In function 'tsp_smc_handler':
bl32/tsp/tsp_main.c:393:9: error: 'tsp_get_magic'
accessing 32 bytes in a region of size 16
[-Werror=stringop-overflow=]
393 | tsp_get_magic(service_args);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~
bl32/tsp/tsp_main.c:393:9: note: referencing argument 1
of type 'uint64_t *' {aka 'long long unsigned int *'}
In file included from bl32/tsp/tsp_main.c:19:
bl32/tsp/tsp_private.h:64:6: note: in a call to function 'tsp_get_magic'
64 | void tsp_get_magic(uint64_t args[4]);
| ^~~~~~~~~~~~~
by changing declaration of tsp_get_magic function from
void tsp_get_magic(uint64_t args[4]);
to
uint128_t tsp_get_magic(void);
which returns arguments directly in x0 and x1 registers.
In bl32\tsp\tsp_main.c the current tsp_smc_handler()
implementation calls tsp_get_magic(service_args);
, where service_args array is declared as
uint64_t service_args[2];
and tsp_get_magic() in bl32\tsp\aarch64\tsp_request.S
copies only 2 registers in output buffer:
/* Store returned arguments to the array */
stp x0, x1, [x4, #0]
plat: xilinx: Use fno-jump-tables flag in CPPFLAGS
From GCC-9 implementation of switch case was generated through jump tables,
because of which we are seeing 1MB increase in rodata section. To reduce
the size we are recommending to use fno-jump-tables.
Davorin Mista [Fri, 24 Aug 2018 15:09:06 +0000 (17:09 +0200)]
zynqmp: pm: update error codes to match Linux and PMU Firmware
All EEMI error codes start with value 2000.
Note: Legacy error codes ARGS (=1) and NOTSUPPORTED (=4) returned by
current ATF code have been left in place.
Signed-off-by: Davorin Mista <davorin.mista@aggios.com> Acked-by: Will Wong <WILLW@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Change-Id: I939afa85957cac88025d82a80f9f6dd49be993b6
Mirela Simonovic [Fri, 24 Aug 2018 15:09:07 +0000 (17:09 +0200)]
zynqmp: pm: Filter errors related to clock gate permissions
Linux clock framework cannot properly deal with these errors. When the
error is related to the lack of permissions to control the clock we
filter the error and report the success to linux. Before recent changes
in clock framework across the stack, this was done in the PMU-FW as a
workaround. Since the PMU-FW now handles clocks and the permissions to
control them using general principles rather than workarounds, it can
no longer distinguish such exceptions and it has to return no-access
error.
Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com> Acked-by: Will Wong <WILLW@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Change-Id: I1491a80e472f44e322a542b29a20eb1cb3319802
projects / arm-tf.git / log