Add support to export a /cpus node to the device tree.
This patch creates and populates the /cpus node in a device tree
based on the existing topology. It uses the minimum required nodes
and properties to satisfy the binding as specified in
https://www.kernel.org/doc/Documentation/devicetree/bindings/arm/cpus.txt
Signed-off-by: Javier Almansa Sobrino <javier.almansasobrino@arm.com>
Change-Id: I03bf4e9a6427da0a3b8ed013f93d7bc43b5c4df0
anzhou [Wed, 5 Aug 2020 14:34:13 +0000 (22:34 +0800)]
Tegra: common: disable GICC after domain off
The the GIC CPU interface should be disabled after cpu off. The
Tegra power management code should mark the connected core as asleep
as part of the CPU off sequence.
This patch disables the GICC after CPU off as a result.
anzhou [Fri, 26 Jun 2020 07:21:10 +0000 (15:21 +0800)]
Tegra: fixup CNTPS_TVAL_EL1 delay timer reads
The delay_timer driver for Tegra uses the CNTPS_TVAL_EL1 secure, physical,
decrementing timer as the source. The current logic incorrectly marks this
as an incrementing timer, by negating the timer value.
This patch fixes the anomaly and updates the driver to remove this logic.
Tegra: add platform specific 'runtime_setup' handler
Tegra SoCs would like the flexibility to perform chip specific actions
before we complete cold boot. This patch introduces a platform specific
'runtime_setup' handler to provide that flexibility.
The SVE CPU extension library reads the id_aa64pfr0_el1 register to
check if SVE is enabled. Tegra platforms disabled ENABLE_SVE_FOR_NS for
pre-8.2 platforms, but this flag can safely be enabled now that the
library can enable the feature at runtime.
This patch updates the makefile to remove "ENABLE_SVE_FOR_NS = 0"
as a result.
In the case of Juno AArch32, platform security configuration
gets done from both BL2 and SP_MIN(BL32) components when
JUNO_AARCH32_EL3_RUNTIME and RESET_TO_SP_MIN build options
are set.
Fix is provided to avoid Platform security configuration from
SP_MIN when it is already done in BL2.
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Change-Id: I702e91dacb4cdd2d10e339ddeaea91289bef3229
Varun Wadekar [Mon, 26 Aug 2019 17:20:53 +0000 (10:20 -0700)]
Tegra: memctrl: platform setup handler functions
The driver initially contained the setup steps to help Tegra186
and Tegra194 SoCs. In order to support future SoCs and make sure
that the driver remains generic enough, some code should be moved
to SoC.
This patch creates a setup handler for a platform to implement its
initialization sequence.
The stream ID security configuration settings shall be done by the
previous level bootloader. This change removes the same settings
from the Tegra194 platform code as a result.
The stream ID override configuration is saved during System Suspend
as part MB1 bct. This change removes the same support from the Tegra194
platform code as a result.
Varun Wadekar [Wed, 11 Dec 2019 21:22:21 +0000 (13:22 -0800)]
Tegra: debug prints indicating SC7 entry sequence completion
This patch adds prints to display the completion of System Suspend
programming sequence for Tegra platforms. The console needs to
be kept alive until the very end of the System Suspend sequence as
a result.
Tegra194: memctrl: update TZDRAM base at 1MB granularity
The Memory controller expects the TZDRAM base value at 1MB granularity
and the current driver does not respect that limitation. This patch
fixes that anomaly.
David Pu [Fri, 7 Jun 2019 22:30:17 +0000 (15:30 -0700)]
Tegra194: ras: split up RAS error clear SMC call.
In order to make sure SMC call is within 25us, this patch reduces number of RAS
errors accessed to 8 at most for each SMC call and takes a input/output
parameter to specify in progress RAS error record index.
The measured SMC call latency is about 20us under Linux test kernel driver.
Change-Id: Ia1b57c8673e0193dc341a36af0b5c09fb48f965f Signed-off-by: David Pu <dpu@nvidia.com>
Varun Wadekar [Fri, 15 Nov 2019 23:46:14 +0000 (15:46 -0800)]
Tegra194: add memory barriers during DRAM to SysRAM copy
This patch adds memory barriers to the trampoline code copying TZDRAM
contents to SysRAM during exit from System Suspend. These barriers
make sure that all the copies go through before we start executing in
SysRAM.
Anthony Zhou [Wed, 4 Dec 2019 06:58:23 +0000 (14:58 +0800)]
Tegra: sip: add VPR resize enabled check
The Memory Controller provides a control register to check
if the video memory can be resized. The previous bootloader
might have locked this feature, which will be reflected by
this register.
This patch reads the control register before processing
a video memory resize request. An error code, -ENOTSUP,
is returned if the feature is locked.
Change-Id: Ia1d67f7a94aa15c6b18ff5c9b9b952e179596ae3 Signed-off-by: Anthony Zhou <anzhou@nvidia.com>
Anthony Zhou [Wed, 13 Nov 2019 10:36:07 +0000 (18:36 +0800)]
Tegra194: add redundancy checks for MMIO writes
MMIO writes should verify that the writes actually went through.
Read the value back after the write operation, perform assert
if the read back value is not same as the write value.
Change-Id: Id2ceb014116f3aa6a9e86505ca1ae9911470a679 Signed-off-by: Anthony Zhou <anzhou@nvidia.com>
During boot the platform enables dual execution for Xavier CPUs.
This patch reads back the ACTLR_ELx register to verify that the bit
is actually set. It asserts if the bit is not set.
Tegra194: verify firewall settings before resource use
The firewall settings for the hardware resources are present in the
Security Configuration Registers. The firewall settings are programmed
by other software components and so must be verified for correctness
before touching the hardware resources they protect.
This patch reads the firewall settings during early boot and asserts
if the settings mismatch.
Coverity warns about the risk of unintended sign-exension in some of the
calculations in spmi_arb.c. While the actual numbers used are small
enough that this cannot happen in practice, it's still a good idea to
clean them up by explicitly making the constants used unsigned.
Signed-off-by: Julius Werner <jwerner@chromium.org>
Change-Id: Ia169e0f7c6b01b8041e8029e8c8d30ee596ba30d
Add support for hexadecimal and pointer format specifiers to snprintf()
The current implementation of snprintf() does not support pointer and
hexadecimal format specifiers, which can be needed, for instance, for
DTB manipulations.
This patch adds that functionality by borrowing some code from the
printf() implementation.
Signed-off-by: Javier Almansa Sobrino <javier.almansasobrino@arm.com>
Change-Id: I2076ea46693a73a04890982bf20e3c633c2767fb
Merge changes from topic "tegra194-spmd" into integration
* changes:
Tegra194: introduce support for `SPD=spmd`
Tegra: introduce backend support to compile libfdt
Tegra: disable signed comparison
plat: common: include "bl_common.h" from plat_spmd_manifest.c
Julius Werner [Wed, 5 Jun 2019 19:40:35 +0000 (12:40 -0700)]
qti: Add SPMI PMIC arbitrator driver
This patch adds a very rudimentary driver for the SPMI arbitrator used
to access the PMIC. It doesn't support all the controller's actual
arbitration features, so it should probably not be used concurrently
with a running kernel (and it's also not optimized for performance). But
it can be used to set a few registers during boot or on shutdown to
control reset handling, which is all we need it for.
Change-Id: I8631c34a2a89ac71aa1ec9b8266e818c922fe34a Signed-off-by: Julius Werner <jwerner@chromium.org>
Julius Werner [Tue, 25 Aug 2020 01:34:38 +0000 (18:34 -0700)]
qti/sc7180: Fix GIC-600 support setting
The patch adding platform support for sc7180 landed around roughly the
same time as the patch that changed GICV3_IMPL to GICV3_SUPPORT_GIC600.
Thus the sc7180 Makefile is still using the old variable name which now
no longer does anything, and it hangs on boot due to the lacking GIC-600
support. This patch fixes the issue.
Signed-off-by: Julius Werner <jwerner@chromium.org>
Change-Id: Id76ada1445c3c5ac9a5a3697b4e749088b89d796
plat: common: include "bl_common.h" from plat_spmd_manifest.c
This patch includes the bl_common.h from plat_spmd_manifest.c to
fix the following compilation errors
<snip>
plat/common/plat_spmd_manifest.c: In function 'plat_spm_core_manifest_load':
plat/common/plat_spmd_manifest.c:130:18: error: implicit declaration of function 'page_align' [-Werror=implicit-function-declaration]
130 | pm_base_align = page_align(pm_base, UP);
| ^~~~~~~~~~
plat/common/plat_spmd_manifest.c:130:38: error: 'UP' undeclared (first use in this function); did you mean 'UL'?
130 | pm_base_align = page_align(pm_base, UP);
| ^~
| UL
plat/common/plat_spmd_manifest.c:130:38: note: each undeclared identifier is reported only once for each function it appears in
plat/common/plat_spmd_manifest.c:146:38: error: 'DOWN' undeclared (first use in this function)
146 | pm_base_align = page_align(pm_base, DOWN);
| ^~~~
cc1: all warnings being treated as errors
<snip>
The SMMU configuration can get corrupted or updated by
external clients during boot without our knowledge.
This patch introduces a "verify" function for the SMMU
driver, to check that the boot configuration settings are
intact. Usually, this function should be called at the
end of the boot cycle.
This function only calls panic() on silicon platforms.
Change-Id: I2ab45a7f228781e71c73ba1f4ffc49353effe146 Signed-off-by: George Bauernschmidt <georgeb@nvidia.com>
Varun Wadekar [Thu, 22 Aug 2019 18:52:36 +0000 (11:52 -0700)]
Tegra: TZDRAM setup from soc specific early_boot handlers
TZDRAM setup is not required for all Tegra SoCs. The previous bootloader
can enable the TZDRAM fence due to architectural improvements in the
newer chips.
This patch moves the TZDRAM setup to early_boot handlers for SoCs to
handle this scenario.
Varun Wadekar [Fri, 4 Oct 2019 18:40:56 +0000 (11:40 -0700)]
Tegra: print GICC registers conditionally
The GICC interface exists only on the interrupt controllers following
the GICv2 specification.
This patch prints the GICC register contents from the platform's macro,
plat_crash_print_regs' only when TEGRA_GICC_BASE is defined. This
allows platforms using future versions of the GIC specification to
still use this macro.
SPE no longer requires the flush bit to be set to start transmitting
characters over the physical uart. Therefore, the flush bit is no
longer required when calling console_core_putc. However, flushing the
console still requires the flush bit.
This patch removes the flush bit from the mailbox messages in
console_core_putc to improve ACK latency.
Original change by: Mustafa Bilgen <mbilgen@nvidia.com>
tools: Get the tool's binary name from the main makefile
Currently, the tool's makefile override the tool's binary name
which is already been defined in the main makefile.
Hence fix is provided so that the tool's makefile get the tool's
binary name from the main makefile instead of overriding it.
Change-Id: I8af2bd391a96bba2dbcddef711338a94ebf5f038 Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Mark Dykes [Wed, 19 Aug 2020 19:11:33 +0000 (19:11 +0000)]
Revert "libc/memset: Implement function in assembler"
This reverts commit e7d344de01ad11b856233634717aafe9312697e4.
This reverts the patch https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/5313 due to a timing issue with the merge. The merge occurred at the same time as the additional comments and thusly were were not seen until the merge was done. This reverts the change and additional patches from Alexei will follow to address the concerns expressed in the orignal patch.
Olivier Deprez [Fri, 21 Aug 2020 14:18:57 +0000 (14:18 +0000)]
Merge changes from topic "spm-secondary-cores" into integration
* changes:
SPMC: embed secondary core ep info into to SPMC context
SPMC: manifest changes to support multicore boot
SPMD: secondary cores PM on and off SPD hooks relayed to SPMC
SPMD: handle SPMC message to register secondary core entry point
SPMD: introduce SPMC to SPMD messages
SPMD: register the SPD PM hooks
SPMD: add generic SPD PM handlers
SPMD: enhance SPMC internal boot states
SPMD: entry point info get helper
Olivier Deprez [Mon, 23 Mar 2020 08:53:06 +0000 (09:53 +0100)]
SPMD: secondary cores PM on and off SPD hooks relayed to SPMC
Define SPMD PM hooks for warm boot and off events. svc_on_finish handler
enters the SPMC at the entry point defined by the secondary EP register
service. The svc_off handler notifies the SPMC that a physical core is
being turned off through a notification message.
SPMD: handle SPMC message to register secondary core entry point
Upon booting, the SPMC running on the primary core shall register the
secondary core entry points to which a given secondary core being woken
up shall jump to into the SPMC . The current implementation assumes the
SPMC calls a registering service implemented in the SPMD for each core
identified by its MPIDR. This can typically happen in a simple loop
implemented in the early SPMC initialization routines by passing each
core identifier associated with an entry point address and context
information.
This service is implemented on top of a more generic SPMC<=>SPMD
interface using direct request/response message passing as defined by
the FF-A specification.
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com> Signed-off-by: Max Shvetsov <maksims.svecovs@arm.com>
Change-Id: I1f70163b6b5cee0880bd2004e1fec41e3780ba35
Olivier Deprez [Mon, 28 Oct 2019 08:52:45 +0000 (08:52 +0000)]
SPMD: entry point info get helper
This patch provides a helper to get the entry_point_info
structure used by the boot CPU as it is used to initialise
the SPMC context on secondary CPUs.
Jacky Bai [Tue, 7 Jan 2020 03:05:22 +0000 (11:05 +0800)]
plat: imx8m: Fix the race condition during cpu hotplug
CPU hotplug & cpuidle have some race condition when doing CPU hotplug
stress test. different CPU cores have the chance to access the same
GPC register(A53_AD), so lock is necessary to do exlusive access.
Signed-off-by: Jacky Bai <ping.bai@nxp.com>
Change-Id: I1296592e05fa78429c3f0fac066951521db755e3
Alexei Fedorov [Sun, 16 Aug 2020 15:01:13 +0000 (16:01 +0100)]
libc/memset: Implement function in assembler
Trace analysis of FVP_Base_AEMv8A model running in
Aarch32 mode with the build options listed below:
TRUSTED_BOARD_BOOT=1 GENERATE_COT=1
ARM_ROTPK_LOCATION=devel_ecdsa KEY_ALG=ecdsa
ROT_KEY=plat/arm/board/common/rotpk/arm_rotprivk_ecdsa.pem
shows that when auth_signature() gets called
71.84% of CPU execution time is spent in memset() function
written in C using single byte write operations,
see lib\libc\memset.c.
This patch replaces C memset() implementation with assembler
version giving the following results:
- for Aarch32 in auth_signature() call memset() CPU time
reduced to 24.84%.
- Number of CPU instructions executed during TF-A
boot stage before start of BL33 in RELEASE builds:
----------------------------------------------
| Arch | C | assembler | % |
----------------------------------------------
| Aarch32 | 2073275460 | 1487400003 | -28.25 |
| Aarch64 | 2056807158 | 1244898303 | -39.47 |
----------------------------------------------
The patch also replaces memset.c with aarch64/memset.S
in plat\nvidia\tegra\platform.mk.
David Pu [Thu, 8 Aug 2019 21:20:03 +0000 (14:20 -0700)]
Tegra: common: make plat_psci_ops routines static
This patch makes Tegra platform psci ops routines to static. These
routines are called by PSCI framework and no external linkage is
necessary. This patch also fixes MISRA C-2012 Rule 8.6 violations.
Change-Id: Idd2381809f76dc0fd578c1c92c0f8eea124f2e88 Signed-off-by: David Pu <dpu@nvidia.com>
Masahisa Kojima [Thu, 11 Jun 2020 12:46:44 +0000 (21:46 +0900)]
qemu/qemu_sbsa: enable SPM support
Enable the spm_mm framework for the qemu_sbsa platform.
Memory layout required for spm_mm is created in secure SRAM.
Co-developed-by: Fu Wei <fu.wei@linaro.org> Signed-off-by: Fu Wei <fu.wei@linaro.org> Signed-off-by: Masahisa Kojima <masahisa.kojima@linaro.org>
Change-Id: I104a623e8bc1e44d035b95f014a13b3f8b33a62a
projects / arm-tf.git / log