Rockchip SoCs bootrom design is like this:
- First 2KB or 4KB internal memory is for bootrom stack and heap;
- Then the first 4-byte suppose to be a TAG like 'RK33';
- The the following memory address end with '0004' is the first
instruction load and running by bootrom;
Let's use the boot0 hook to reserve the first 4-byte tag for all
the Rockchip SoCs.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
[Commit message taken from an older patch by:]
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
The API for get priv pointer is wrong, fix it.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
U-Boot widely uses error() as a bit noisier variant of printf().
This macro causes name conflict with the following line in
include/linux/compiler-gcc.h:
# define __compiletime_error(message) __attribute__((error(message)))
This prevents us from using __compiletime_error(), and makes it
difficult to fully sync BUILD_BUG macros with Linux. (Notice
Linux's BUILD_BUG_ON_MSG is implemented by using compiletime_assert().)
Let's convert error() into now treewide-available pr_err().
Done with the help of Coccinelle, excluing tools/ directory.
The semantic patch I used is as follows:
// <smpl>
@@@@
-error
+pr_err
(...)
// </smpl>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
[trini: Re-run Coccinelle]
Signed-off-by: Tom Rini <trini@konsulko.com>
Falcon mode, is updating DDR dt node configuration through
spl_fixup_fdt() so add appropriate DDR base and size through
dram_init_banksize.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
To support the new "same-as-spl" specifier in the boot-order on the
RK3399, this implements the chip-specific mapping from the information
obtainable from the BROM to a OF path name.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
It is often desirable to configure the spl-boot-order (i.e. the order
that SPL probes devices to find the FIT image containing a full U-Boot)
such that it contains 'the same device the SPL stage was booted from'
early on. To support this, we introduce the 'same-as-spl' specifier
for the spl-boot-order property.
This commit adds:
- documentation for the new board_spl_was_booted_from() function that
individual SoCs/boards should provide, if they can determine where
the SPL was booted from
- implements the new board_spl_was_booted_from() stub function
- adds support for handling the 'same-as-spl' specifier and calling
into the per-SoC/per-board support code.
This also updates the documentation for the 'u-boot,spl-boot-order'
property.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
In the expectation that the spl-boot-order code will eventually
gain use outside of mach-rockchip: let's add documentation on the
spl_node_to_boot_device() function, which is likely to become a
publicly exported function.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The later-stage spl_board_init (as opposed to board_init_f) should set
up board-specific details: these differ between the EVB-RK3399 and the
RK3399-Q7 (Puma).
This moves spl_board_init back into the individual boards and removes
the unneeded functionality from Puma.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
The boot mode for rk322x is stored in sysreg 0, update it.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Since the size of SPL can't be exceeded 0x8000 bytes in RK3288,
it is not possible add new SPL features like Falcon mode or etc.
So add TPL stage so-that adding new features to SPL is possible.
- TPL: DRAM init, clocks
- SPL: MMC, falcon, etc
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
configure_l2ctlr will be shared between SPL and TPL so
move them into asm/arch/sys_proto.h
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
L2CTLR read/write functions are common to armv7 so, move
them in to include/asm/armv7.h and use them where ever it need.
Cc: Tom Warren <twarren@nvidia.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
[Backed out the change to arch/arm/mach-tegra/cache.c:]
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Instead of directly calling into the back-to-bootrom code, the RK3399
common SPL implementation now uses BOOT_DEVICE_BOOTROM to trigger a
transfer back into the bootrom.
With this factored out, the spl_board_init function can not be
customised for each RK3399 board.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Since we have CONFIG_RAM framwork and its driver folder, move the driver
into it.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Since we have CONFIG_RAM framwork and its driver folder, move the driver
into it.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Since we have CONFIG_RAM framwork and its driver folder, move the driver
into it.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Since we have CONFIG_RAM framwork and its driver folder, move the driver
into it.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
After Simon's patch, the dtoc can work with 64bit address,
so we need to fix reg number for it.
Depend on Simon's patch set:
https://patchwork.ozlabs.org/cover/807266/
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
This patch adds support for Vyasa RK3288 initial board
from Amarula Solutions.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
On the RK3399, we will have either OF_PLATDATA or full OF_CONTROL
enabled: this allows the use of syscon to retrieve the addresses of
GRF and SGRF (except for the early debug UART setup, which runs so
early that the device-model is not initialised).
This removes the hard-coded addresses and goes through syscon to
retrieve the base-addresses of GRF and SGRF. After that, we use
the structure definitions to locate the respective registers.
In addition to this, the inclusion of header files is also cleaned up:
- all headers are included at the beginning (there was a spurious
inclusion of the grf header from within a function)
- all #include statements for unused headers are removed
- the remaining #include statements are sorted (while keeping common.h
included in front)
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Unfortunately, the integrated macphy default is enabled, which will
increase power consuming, if we do not use this PHY. So let's disable
it at first, which will save power consuming. If we really use it, then
enable it in driver level.
Signed-off-by: David Wu <david.wu@rock-chips.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
To support fastboot "fastboot reboot-bootloader" cmd.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
syscon id table need a dummy member as NULL ending, or else system
will panic while try to match a compatible in this table as a list.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Sometimes it's helpful to know the reset reason caused in the SoC.
Add reset reason detection for the RK3288 SoC.
This will set an environment variable which represents the reset reason.
Signed-off-by: Wadim Egorov <w.egorov@phytec.de>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
The DRAM start address is not 0, so need to update the last bank size
as:
DRAM start addr + DRAM_SIZE - last bank start addr
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
fdtdec.h is included, but not used in rk3399-board-spl.c: remove the
'#include'-statement.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
We are now using an env_ prefix for environment functions. Rename setenv()
for consistency. Also add function comments in common.h.
Suggested-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Simon Glass <sjg@chromium.org>
Disable the ddr secure region setting in SPL and the ddr memory
becomes non-secure, every one can access it. the trust firmware
like OPTEE should have the correct setting for it after SPL if
there is one.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
With the new way of doing things (i.e. the hierarchical selection of
SPL_LDSCRIPT via Kconfig) in place, this moves the SPL_LDSCRIPT setting
for the RK3368 from defconfig back into Kconfig.
With this done, there should be no lingering cases of SPL_LDSCRIPT
outside of Kconfig files.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
With the changes to split SPL/TPL for the RK3368, I apparently missed
some needed adjustments to the RK3188 Kconfig and rock_defconfig.
This fixes build-issues for the rock board after applying the RK3368
enablement (and SPL/TPL) set that resulted from TPL_SERIAL_SUPPORT,
TPL_ROCKCHIP_BACK_TO_BROM and TPL_TINY_MEMSET being separate symbols
now.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When using DM timers w/ the timer0 block within the RK3368, we no
longer depend on the ARMv8 generic timer counting. This allows us to
drop the secure timer initialisation from the TPL and SPL stages.
The secure timer will later be set up by ATF, which starts the ARMv8
generic timer. Thus, there will be a dependency from Linux to the ATF
through the ARMv8 generic timer... this seems reasonable, as Linux
will require the ATF (and PSCI) to start up the secondary cores anyway
(in other words: we don't add any new dependencies).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
We can finally drop TPL_STACK, TPL_TEXT_BASE and TPL_MAX_SIZE off the
whitelist (this time it's really happening!) and migrate the setting
(only used on the RK3368-uQ7 so far) into Kconfig.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The RK3368 needs to have a different base-address and stack-pointer
for its TPL stage. Now that we want to do this via Kconfig, we need
to tick the appropriate 'TPL_NEEDS_...' boxes.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The RK3368-uQ7 (codenamed 'Lion') is a micro-Qseven (40mm x 70mm,
MXM-230 edge connector compatible with the Qseven specification)
form-factor system-on-module based on the octo-core Rockchip RK3368.
It is designed, supported and manufactured by Theobroma Systems.
It provides the following features:
- 8x Cortex-A53 (in 2 clusters of 4 cores each)
- (on-module) up to 4GB of DDR3 memory
- (on-module) SPI-NOR flash
- (on-module) eMMC
- Gigabit Ethernet (with an on-module KSZ9031 PHY)
- USB
- HDMI
- MIPI-DSI/single-channel LVDS (muxed on the 'LVDS-A' pin-group)
- various 'slow' interfaces (e.g. UART, SPI, I2C, I2S, ...)
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
With SPL and TPL support for the RK3368 in place, mark SPL and TPL as
supported from Kconfig for the RK3368. As this is primarily tested on
the RK3368-uQ7, we'll leave it to board's individual defconfig to
enable.
Also enable DEBUG_UART_BOARD_INIT for the RK3368, so we get output
during the early boot-up, as we turn on TPL and SPL.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Adds SPL support for the RK3368 (assuming that our TPL stage has
initialised DRAM and set up the memory firewall).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
In order to reuse the support for the u-boot,spl-boot-order property
from the rk3399, we split it into a reusable module that can be
included by the SPL code for any of our boards.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This adds the TPL support for the RK3368, including the u-boot-tpl.lds.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This adds a DRAM controller driver for the RK3368 and places it in
drivers/ram/rockchip (where the other DM-enabled DRAM controller
drivers for rockchip devices should also be moved eventually).
At this stage, only the following feature-set is supported:
- DDR3
- 32-bit configuration (i.e. fully populated)
- dual-rank (i.e. no auto-detection of ranks)
- DDR3-1600K speed-bin
This driver expects to run from a TPL stage that will later return to
the RK3368 BROM. It communicates with later stages through the
os_reg2 in the pmugrf (i.e. using the same mechanism as Rockchip's DDR
init code).
Unlike other DMC drivers for RK32xx and RK33xx parts, the required
timings are calculated within the driver based on a target frequency
and a DDR3 speed-bin (only the DDR3-1600K speed-bin is support at this
time).
The RK3368 also has the DDRC0_CON0 (DDR ch. 0, control-register 0)
register for controlling the operation of its (single-channel) DRAM
controller in the GRF block. This provides for selecting DDR3, mobile
DDR modes, and control low-power operation.
As part of this change, DDRC0_CON0 is also added to the GRF structure
definition (at offset 0x600).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Handling TPL and SPL in the Makefile for mach-rockchip was based on
nested if checks and/or if-else-if paths. This can be simplified and
made more readable by using $(SPL_TPL_) and by introducing
intermediate variables for the aggregation of SPL and TPL features.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
In TPL we will need to configure security in the SGRF of the RK3368.
This change adds support for the SGRF as a syscon device, so we can
retrieve its address range through the syscon API in TPL (and can
avoid having to hard-code the address).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The RK3368 has both a limited TPL size (just 0x7000 bytes) and the
added challenge of booting in AArch64, which increases the code size
for TPL (particularily when using the LP64 programming model). For
this reason we expect the RK3368 to always use OF_PLATDATA for its
TPL stage.
This change adds support for the MSCH, PMUGRF and GRF register regions
in syscon, which are necessary for initialising the RK3368's DRAM
controller.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The RK3368 option in Kconfig referred to the RK3328 (copy-and-paste)
and had a few typos and unnecessarily used UTF-8 characters. While
fixing this, I also reformatted and further clarified the text
(e.g. made the grouping into a a big and little cluster of 4 cores
each explicit).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The utility functions in sdram_common.c will be useful both for some
SPL implementations (and if unused, the linked will discard these
anyway) and for the full U-Boot stage.
This changes selects sdram_common.o through the $(SPL_TPL_) macro to
allow better control of its inclusion through the CONFIG_ROM,
CONFIG_SPL_RAM or CONFIG_TPL_RAM options.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
With the finer-grained control over LIBCOMMON_SUPPORT for TPL/SPL (i.e.
with the newly introduced distinction between TPL_LIBCOMMON_SUPPORT and
SPL_LIBCOMMON_SUPPORT), we can simplify the #ifdef-check to simply use
CONFIG_IS_ENABELD.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
[fixed up to use 'puts' and LIBCOMMON:]
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
The back-to-bootrom option is rather unfortunately named
CONFIG_ROCKCHIP_SPL_BACK_TO_BOOTROM
instead of
CONFIG_SPL_ROCKCHIP_BACK_TO_BOOTROM
To make is selectable through CONFIG_IS_ENABLED(ROCKCHIP_BACK_TO_BOOTROM),
we need to rename it. At the same time, we introduce a TPL_ variant of
the option to give us finer-grained control over when it should be used.
This change is motivated by our RK3368 boot process, which returns to
the boot ROM only from the TPL stage, but not from the SPL stage.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
[added fix-up for evb-rk3229_defconfig and phycore-rk3288_defconfig:]
[fixed inverted CONFIG_IS_ENABLED test for rk3288:]
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
include/configs/rock.h: undef
The back-to-bootrom support for Rockchip is equivalent to an
(assembly) implementation of setjmp/longjmp (i.e. it saves the
stack-pointer, link-register and callee-saved registers). Up until
now, this had only been implemented for AArch32 (i.e. ARMv7 or older),
which puts the new ARMv8 devices (which boot in AArch64 mode) at a
slight disadvantage.
To allow use of the 'back-to-bootrom' feature on new devices (e.g. the
RK3368), this commit adds an implementation for AArch64.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
printf will increase the code size more than 1kb, but platform
like rk3036 has no enough space for it.
Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
function board_init_f_init_reserve will call memset, which
is implemented in lib, and enabled by CONFIG_SPL_LIBGENERIC_SUPPORT
in spl stage.
To reduce the code size, also enable SPL_TINY_MEMSET.
As rk3036 will return to bootrom immediately after dram
initialization, there is no need to run DM, so disable
SPL_DM_SERIAL.
Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
