GitLab

Compiling BL31 for the Allwinner platform now produces a message about
the deprecation of gic_common.c.
Follow the advice and use include gicv2.mk instead.

Collect all includes at the beginning of the file on the way.

Change-Id: Iee46e21a630bfa831d28059f09aa7b049eb554bb
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

The non-secure world has no business accessing the CPU power switches in
the PRCM; those are handled by TF-A or the SCP. Only allow access to the
clock control part of the PRCM.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I657b97f4ea8a0073448ad3343fbc66ba168ed89e

The base address for both the GPIO and the clock unit of the H6 memory map
have been typo-ed. Fix them to match the Linux DT and the manual.

The H6 code use neither of them, so this doesn't change or fix anything
in the real world, but should be corrected anyway.

The issue was found and reported by Github user "armlabs".

Change-Id: Ic6fdfb732ce1cfc54cbb927718035624a06a9e08
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

Since now the generic console_t structure holds the UART base address as
well, let's use that generic location and drop the UART driver specific
data structure at all.

Change-Id: I5c2fe3b6a667acf80c808cfec4a64059a2c9c25f
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

If an SCP firmware is present and able to communicate via SCPI, then use
that to implement CPU and system power state transitions, including CPU
hotplug and system suspend. Otherwise, fall back to the existing CPU
power control implementation.

The last 16 KiB of SRAM A2 are reserved for the SCP firmware, and the
SCPI shared memory is at the very end of this region (and therefore the
end of SRAM A2). BL31 continues to start at the beginning of SRAM A2
(not counting the ARISC exception vector area) and fills up to the
beginning of the SCP firmware.

Because the SCP firmware is not loaded adjacent to the ARISC exception
vector area, the jump instructions used for exception handling cannot be
included in the SCP firmware image, and must be initialized here before
turning on the SCP.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I37b9b9636f94d4125230423726f3ac5e9cdb551c

The function names follow the naming convention used by the existing
ARM SCPI client.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I543bae7d46e206eb405dbedfcf7aeba88a12ca48

The SCP firmware is allocated the last 16KiB of SRAM A2. This includes
the SCPI shared memory area, which must be mapped as MT_DEVICE to
prevent problems with cache coherency between the AP CPUs and the SCP.
For simplicity, map the whole SCP region as MT_DEVICE.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Ie39eb5ff281b8898a3c1d9748dc08755f528e2f8

The ARISC vector area consists of 0x4000 bytes before the beginning of
usable SRAM. Still, it is technically a part of SRAM A2, so include it
in the memory definition. This avoids the confusing practice of
subtracting from the beginning of the SRAM region when referencing the
ARISC vectors.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Iae89e01aeab93560159562692e03e88306e2a1bf

PLATFORM_CORE_COUNT - Unsigned int
PLATFORM_CLUSTER_COUNT - Unsigned int
PLATFORM_MAX_CPUS_PER_CLUSTER - Unsigned int
PLATFORM_CORE_COUNT_PER_CLUSTER - Unsigned int
Signed-off-by: Deepika Bhavnani <deepika.bhavnani@arm.com>
Change-Id: I7aea86891e54522c88af5ff16795a575f9a9322d

Remove the general BL31 mmap region: it duplicates the existing static
mapping for the entire SRAM region. Use the helper definitions when
applicable to simplify the code and add the MT_EXECUTE_NEVER flag.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I7a6b79e50e4b5c698774229530dd3d2a89e94a6d

Now that there is plenty of space (32 KiB) available for NOBITS
sections, we can afford using an entire page for coherent memory. In
fact, because it simplifies the code, this is a beneficial change for
loaded image (.text) size, where we are still close to the size limit.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I0b899dabcb162015c63b0e4aed0869569c889ed9

Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Ia2f69e26e34462e113bc2cad4dcb923e20b8fb95

This frees up space in SRAM A2 that will be used by the SCP firmware and
SCPI shared memory.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I8ce035257451e2d142666fe0cd045e59d4d57b35

Chip ID checking and poweroff work just like they did before.
Regulators are now enabled just like on A64/H5.

This changes the signatures of the low-level register read/write
functions to match the interface expected by the common driver.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I14d63d171a094fa1375904928270fa3e21761646

Previously, the A64/H5 and H6 platforms' PMIC setup code was entirely
independent. However, some H6 boards also need early regulator setup.

Most of the register interface and all of the device tree traversal code
can be reused between the AXP803 and AXP805. The main difference is the
hardware bus interface, so that part is left to the platforms. The
remainder is moved into a driver.

I factored out the bits that were obviously specific to the AXP803;
additional changes for compatibility with other PMICs can be made as
needed.

The only functional change is that rsb_init() now checks the PMIC's chip
ID register against the expected value. This was already being done in
the H6 version of the code.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Icdcf9edd6565f78cccc503922405129ac27e08a2

This simplifies the code a bit. Verified to produce the same binary.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Ie1ec1ce2ea39c46525840906826c90a8a7eff287

As of a561e41b

 ("allwinner: power: add enable switches for DCDC1/5")
there are no longer regulators without an enable register provided.
Since it seems reasonable that this will continue to be the case, drop
the check.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Icd7ec26fc6450d053e6e6d855fc16229b1d65a39

should_enable_regulator() is already checked in the regulators subnode
loop before setup_regulator() is called, so there's no need to check it
again here.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Idb8b8a6e435246f4fb226bc84813449d80a0a977

These are used by the PMIC setup code, which runs during BL31
initialization, and the PSCI shutdown code, also a part of BL31.
They can't be needed before BL31, or it wouldn't be possible to boot.
Allwinner platforms don't generally build anything but BL31 anyway, but
this change improves clarity and consistency with allwinner-common.mk.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I24f1d9ca8b4256e44badf5218d04d8690082babf

The function is only used in this file, and it doesn't make sense for it
to be used anywhere else.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Iab18f082911edcdbc37ceeaff8c512be68e0cb0f

The action of last resort isn't going to change between SoCs. This moves
that code back to the PSCI implementation, where it more obviously
matches the code in sunxi_system_reset().

The two error messages say essentially the same thing anyway.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I62ac35fdb5ed78a016e9b18281416f1dcea38a4a

- Check the return value from sunxi_init_platform_r_twi().
- Print the PMIC banner before doing anything that might fail.
- Remove double prefixes in error messages.
- Consistently omit the trailing period.
- No need to print the unknown SoC's ID, since we already did that
  earlier in bl31_platform_setup().
- On the other hand, do print the ID of the unknown PMIC.
- Try to keep the messages concise, as the large string size in these
  files was causing the firmware to spill into the next page.
- Downgrade the banner from NOTICE to INFO. It's purely informational,
  and people should be using debug builds on untested hardware anyway.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Ib909408a5fdaebe05470fbce48d245dd0bf040eb

Ensure that the default (zero) value represents the case where we take
no action. Previously, if a PLAT=sun50i_a64 build was booted on an
unknown SoC ID, it would be treated as an H5 at shutdown.

This removes some duplicate code and fixes error propagation on H6.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: I4e51d8a43a56eccb0d8088593cb9908e52e782bc

The clock must be running for the module to be reset.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Ic8fafc946f3a1a697174b91288e357ffa033ab9a

The current range check for the offset is wrong: it is counting bytes,
while indexing an array of uint32_t. Since the offset is always zero,
the parameter is unnecessary. Instead of adding more code to fix the
check, remove the parameter to avoid the problem entirely.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Iadfc7d027155adc754e017b3462233ce9a1d64f6

A64-OLinuXino family boards (maybe others too) uses PG for USB vbus
enable/disable. However PG is supplied by DLDO4, which is not present
in the list of known regulators. This patch adds DLD04 to it.
Signed-off-by: Stefan Mavrodiev <stefan@olimex.com>
Change-Id: I31d3bb3e0004ccf5b282d08b530ee44979da0466

The new API becomes the default one.

Change-Id: Ic1d602da3dff4f4ebbcc158b885295c902a24fec
Signed-off-by: Ambroise Vincent <ambroise.vincent@arm.com>

As all Allwinner platforms are single-cluster A53 chips, we can disable
support for newer, unsupported architecture extensions. We can also
avoid some cache maintenance code, since no platform-specific setup is
required to enable coherency.

These changes reduce the size of .text on a default build with GCC 9.1
enough that .vectors again fits in the second half of a page, instead
of requiring its own page.

This commit was boot-tested on the Pinebook.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Ib90454ef0c798d5e714b7780c585be0b1ed49c6d

Some Allwinner H6 has a broken watchdog that doesn't
make the soc reboot.

Use the R_WATCHDOG instead.
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Change-Id: Ie95cc30a80ed517b60b30d6bc2e655a1b53f18ba

So far the DT node describing the AXP803 PMIC used in many Allwinner A64
boards had only one subnode, so our code just entering the first subnode
to find all regulators worked fine.

However recent DT updates in the Linux kernel add more subnodes *before*
that, so we need to make sure to explicitly enter the "regulators"
subnode to find the information we are after.

Improve some DT node parsing error handling on the way.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

Convert them to take an mpidr instead of a (cluster, core) pair. This
simplifies all of the call sites, and actually makes the functions a bit
smaller.
Signed-off-by: Samuel Holland <samuel@sholland.org>

This maximizes the amount of data protected by the MMU.
Signed-off-by: Samuel Holland <samuel@sholland.org>

Enforce full include path for includes. Deprecate old paths.

The following folders inside include/lib have been left unchanged:

- include/lib/cpus/${ARCH}
- include/lib/el3_runtime/${ARCH}

The reason for this change is that having a global namespace for
includes isn't a good idea. It defeats one of the advantages of having
folders and it introduces problems that are sometimes subtle (because
you may not know the header you are actually including if there are two
of them).

For example, this patch had to be created because two headers were
called the same way: e0ea0928 ("Fix gpio includes of mt8173 platform
to avoid collision."). More recently, this patch has had similar
problems: 46f9b2c3 ("drivers: add tzc380 support").

This problem was introduced in commit 4ecca339

 ("Move include and
source files to logical locations"). At that time, there weren't too
many headers so it wasn't a real issue. However, time has shown that
this creates problems.

Platforms that want to preserve the way they include headers may add the
removed paths to PLAT_INCLUDES, but this is discouraged.

Change-Id: I39dc53ed98f9e297a5966e723d1936d6ccf2fc8f
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

This patch makes the build system link the console framework code by
default, like it already does with other common libraries (e.g. cache
helpers). This should not make a difference in practice since TF is
linked with --gc-sections, so the linker will garbage collect all
functions and data that are not referenced by any other code. Thus, if a
platform doesn't want to include console code for size reasons and
doesn't make any references to console functions, the code will not be
included in the final binary.

To avoid compatibility issues with older platform ports, only make this
change for the MULTI_CONSOLE_API.

Change-Id: I153a9dbe680d57aadb860d1c829759ba701130d3
Signed-off-by: Julius Werner <jwerner@chromium.org>

The DCDC6 power rail is typically driving VDD_SYS in the SoC, so it is
on by default and uses the default voltage.

As there seems to be at least on board using a different voltage, add
the rail to the list of known voltage lines, so we can setup the right
voltage as early as possible.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

The DCDC1 and DCDC5 power rails didn't specify the enable bits. This
isn't critical, since those rails are on by default (and are needed for
every board), but it is inconsistent.

Add the respective enable bits for those two rails.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

The DRIVEVBUS pin setup was broken in two ways:
- To configure this pin as an output pin, one has to *clear* the bit in
  register 0x8f. It is 0 by default, but rebooting from Linux might have
  left this bit set.
- Doing this just configures the pin as an output pin, but doesn't
  actually drive power to it. This is done via bit 2 in register 0x30.

Fix the routine to both properly configure the pin and drive power to
it. Add an axp_clrsetbits() helper on the way.

Now this isn't really perfect, still:
We only need to setup the PMIC power rails that are needed for U-Boot.
DRIVEVBUS typically controls the VBUS voltage for the host function of
an USB-OTG port, something we typically don't want in U-Boot (fastboot,
using the USB *device* functionality, is much more common). The
BananaPi-M64 uses the regulator in this way, but the Remix Mini PC
actually controls the power of both its USB ports via this line.

Technically we should differentiate here: if DRIVEVBUS controls a
microUSB-B socket, the power should stay off, any host-type A sockets
should be supplied, though.
For now just always enable the power, that shouldn't really hurt the
USB-OTG functionality anyway.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

The legacy Allwinner ATF port used to setup some clocks, and U-Boot is
still relying on this. We don't need to setup the full set, as the SPL
is doing most of it, but it misses one clock (AHB2) and programs another
(AHB1) to quite conservative values.

Fix this up during the platform setup to improve USB and Ethernet
performance, iperf values go up by 31% in my setup with that patch.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

All identifiers, regardless of use, that start with two underscores are
reserved. This means they can't be used in header guards.

The style that this project is now to use the full name of the file in
capital letters followed by 'H'. For example, for a file called
"uart_example.h", the header guard is UART_EXAMPLE_H.

The exceptions are files that are imported from other projects:

- CryptoCell driver
- dt-bindings folders
- zlib headers

Change-Id: I50561bf6c88b491ec440d0c8385c74650f3c106e
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

PSCI requires a core to turn itself off, which we can't do properly by
just executing an algorithm on that very core. As a consequence we just
put a core into WFI on CPU_OFF right now.
To fix this let's task the "arisc" management processor (an OpenRISC
core) with that task of asserting reset and turning off the core's power
domain. We use a handcrafted sequence of OpenRISC instructions to
achieve this, and hand this data over to the new sunxi_execute_arisc_code()
routine.
The commented source code for this routine is provided in a separate file,
but the ATF code contains the already encoded instructions as data.
The H6 uses the same algorithm, but differs in the MMIO addresses, so
provide a SoC (family) specific copy of that code.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>