When booting from ARM Trusted Firmware, U-Boot runs in EL1-NS.
The boot flow is as follows:
BL1 -> BL2 -> BL31 -> BL33 (i.e. U-Boot)
This boot sequence works fine for LD11 SoC (Cortex-A53), but LD20
SoC (Cortex-A72) hangs in U-Boot. The solution I found is to
read sctlr_el1 and write back the value as-is. This should be
no effect, but surprisingly fixes the problem for LD20 to boot.
I do not know why.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
In order to be able to migrate the various SoC EHCI CONFIG options we
first need to finish the switch from CONFIG_USB_EHCI to
CONFIG_USB_EHCI_HCD.
Cc: Marek Vasut <marex@denx.de>
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Marek Vasut <marex@denx.de>
The SCP (System Control Processor) or what we call STM (Stand-by
MPU) is integrated in LD4, Pro4, sLD8, LD6b, LD11, and LD20.
For these SoCs, show the information if STM is enabled.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This configuration is supposed to be used with ARM Trusted Firmware,
so the SYSTEM_RESET is implemented in BL31. Invoke PSCI instead of
U-Boot's own reset code because we need to coordinate with SCP
(System Control Processor) for the system-level power management.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
By making dram_init_banksize() return an error code we can drop the
wrapper. Adjust this and clean up all implementations.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Commit 04cd4e7215 ("ARM: uniphier: remove DRAM base address from
board parameters") accidentally unset the DRAM_SPARSE flag, and
changed the physical map of the DRAM channels. Revive the original
behavior.
Fixes: 04cd4e7215 ("ARM: uniphier: remove DRAM base address from board parameters")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reported-by: Shunji Sato <sato.shunji@socionext.com>
Re-use of routines embedded in the Boot ROM requires a function
pointer table for each SoC. This is not nice in terms of the
maintainability in a long run.
Implement simple eMMC load APIs that are commonly used for LD11,
LD20, and hopefully future SoCs.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
For LD11 and LD20 SoCs, the RST_n pin is asserted by default. If
the EXT_CSD[162], bit[1:0] (RST_n_ENABLE) is fused, the eMMC device
would stay in the reset state until its RST_n pin is deasserted by
software.
Currently, this is cared by an ad-hoc way because the eMMC hardware
reset provider is not supported in U-Boot for now. This code should
be re-written once the "mmc-pwrseq-emmc" binding is supported.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
If the DRAM clock duty does not meet the allowable tolerance,
it is marked in an efuse register. If the register is fused,
the boot code should compensate for the DRAM clock duty error.
Signed-off-by: Kotaro Hayashi <hayashi.kotaro@socionext.com>
[masahiro: simplify code, add git-log]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This parameter is redundant because we can know the number of
channels by checking if dram_ch[2].size is zero.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The current implementation has ugly switch statements here and there,
and duplicates similar code. Rework it using table lookups for SoC
data and reduce code duplication.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The spl_boot_mode() is unrelated to the other code in this file.
Besides, this function is only called from common/spl/spl_mmc.c,
so it is reasonable to guard with CONFIG_SPL_MMC_SUPPORT.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, arch/arm/mach-uniphier/boot-mode/boot-mode.c is messed up
with unrelated code; there is no reason why the "mmcsetn" command
must be placed in this file.
Split out the MMC code into arch/arm/mach-uniphier/mmc-first-dev.c.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The base address of each DRAM channel can be calculated from other
parameters, so does not need hard-coding. What we need is the size
of each DRAM channel and DRAM_SPARSE flag to decide the start address
of DRAM channel 1.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Since commit 26b09c022a ("ARM: uniphier: move SBC and Support Card
init code to U-Boot proper"), the System Bus is initialized by
board_init(). The show_board_info() is called from board_init_f()
by default, so the revision register of the Micro Support Card may
not be accessed at this point. Show its revision after the System
Bus is initialized.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Now the "for" loop here iterates on the detected memory banks.
It must skip unused DRAM banks.
Fixes: c995f3a3c5 ("ARM: uniphier: use gd->bd->bi_dram for memory reserve on LD20 SoC")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
If SG_MEMCONF_CH2_DISABLE bit is set, the DRAM channel 2 is unused.
The register settings for the ch2 should be ignored.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
We repeated partial moves for CONFIG_SYS_NO_FLASH, but this is
not completed. Finish this work by the tool.
During this move, let's rename it to CONFIG_MTD_NOR_FLASH.
Actually, we have more instances of "#ifndef CONFIG_SYS_NO_FLASH"
than those of "#ifdef CONFIG_SYS_NO_FLASH". Flipping the logic will
make the code more readable. Besides, negative meaning symbols do
not fit in obj-$(CONFIG_...) style Makefiles.
This commit was created as follows:
[1] Edit "default n" to "default y" in the config entry in
common/Kconfig.
[2] Run "tools/moveconfig.py -y -r HEAD SYS_NO_FLASH"
[3] Rename the instances in defconfigs by the following:
find . -path './configs/*_defconfig' | xargs sed -i \
-e '/CONFIG_SYS_NO_FLASH=y/d' \
-e 's/# CONFIG_SYS_NO_FLASH is not set/CONFIG_MTD_NOR_FLASH=y/'
[4] Change the conditionals by the following:
find . -name '*.[ch]' | xargs sed -i \
-e 's/ifndef CONFIG_SYS_NO_FLASH/ifdef CONFIG_MTD_NOR_FLASH/' \
-e 's/ifdef CONFIG_SYS_NO_FLASH/ifndef CONFIG_MTD_NOR_FLASH/' \
-e 's/!defined(CONFIG_SYS_NO_FLASH)/defined(CONFIG_MTD_NOR_FLASH)/' \
-e 's/defined(CONFIG_SYS_NO_FLASH)/!defined(CONFIG_MTD_NOR_FLASH)/'
[5] Modify the following manually
- Rename the rest of instances
- Remove the description from README
- Create the new Kconfig entry in drivers/mtd/Kconfig
- Remove the old Kconfig entry from common/Kconfig
- Remove the garbage comments from include/configs/*.h
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
CONFIG_CMD_ZIP is not defined by any board. I am moving
CONFIG_CMD_UNZIP to defconfig files except UniPhier SoC family.
I am the maintainer of UniPhier platform, so I know "select CMD_UNZIP"
is better for this platform.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Stefan Roese <sr@denx.de>
Acked-by: Ryan Harkin <ryan.harkin@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
It seems more readable to use arrays to get SoC specific parameters
instead of the crappy switch statement.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
For LD20 SoC, the last 64 byte of each DRAM bank is used for the
dynamic training of DRAM PHY. The regions must be reserved in DT to
prevent the kernel from using them. Now gd->bd->bi_dram reflects
the actual memory banks. Just use it instead of getting access to
the board parameters.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
U-Boot needs to set up available memory area(s) in dram_init() and
dram_init_banksize(). It is platform-dependent how to detect the
memory banks. Currently, UniPhier adopts the memory banks _alleged_
by DT. This is based on the assumption that users bind a correct DT
in their build process.
Come to think of it, the DRAM controller has already been set up
before U-Boot is entered (because U-Boot runs on DRAM). So, the
DRAM controller setup register seems a more reliable source of any
information about DRAM stuff. The DRAM banks are initialized by
preliminary firmware (SPL, ARM Trusted Firmware BL2, or whatever),
so this means the source of the reliability is shifted from Device
Tree to such early-stage firmware. However, if the DRAM controller
is wrongly configured, the system will crash. If your system is
running, the DRAM setup register is very likely to provide the
correct DRAM mapping.
Decode the SG_MEMCONF register to get the available DRAM banks.
The dram_init() and dram_init_banksize() need similar decoding.
It would be nice if dram_init_banksize() could reuse the outcome
of dram_init(), but global variables are unavailable at this stage
because the .bss section is available only after the relocation.
As a result, SG_MEMCONF must be checked twice, but a new helper
uniphier_memconf_decode() will help to avoid code duplication.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The two arrays ddrphy_{op,ip}_dq_shift_val, occupy more than 3.8 KB
memory footprint, which is significant in SPL.
There are PHY parameters for 5 boards, but they are actually not
board specific, but SoC specific. After all, we just need to have
2 patterns, for LD20 and LD21. Also, the shift values are small
enough to become "short" type instead of "int". This change will
save about 3 KB memory footprint.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
There are similar functions that look up SoC data by the SoC ID.
The new macro UNIPHIER_DEFINE_SOCDATA_FUNC will be helpful to
avoid the code duplication.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, uniphier_get_soc_type() converts the SoC ID (this is
read from the revision register) to an enum symbol to use it for SoC
identification. Come to think of it, there is no need for the
conversion in the first place. Using the SoC ID from the register
as-is a straightforward way.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The <common.h> includes too many headers. Actually, these files
needed to include it for udelay() declaration. Now we can replace
it with <linux/delay.h> thanks to commit 5bc516ed66 ("delay:
collect {m, n, u}delay declarations to include/linux/delay.h").
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
These files only need error number macros. Actually, IS_ERR(),
PTR_ERR(), ERR_PTR(), etc. are not useful for U-Boot. Avoid
unnecessary header includes.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
We may want to run different firmware before running U-Boot. For
example, ARM Trusted Firmware runs before U-Boot, making U-Boot
a non-secure world boot loader. In this case, the SoC might be
initialized there, which enables us to skip SPL entirely.
This commit removes "select SPL" to make it configurable. This
also enables the Multi SoC support for the UniPhier ARMv8 SoCs.
(CONFIG_ARCH_UNIPHIER_V8_MULTI) Thanks to the driver model and
Device Tree, the U-Boot proper part is now written in a generic way.
The board/SoC parameters reside in DT. The Multi SoC support
increases the memory footprint a bit, but the U-Boot proper does
not have strict memory constraint. This will mitigate the per-SoC
(sometimes per-board) defconfig burden.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The ddrphy_shift_rof_hws() never writes back the shifted delay value
to the register, which makes this function non-effective.
Signed-off-by: Kotaro Hayashi <hayashi.kotaro@socionext.com>
[masahiro: add git log]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Initialize SBC and Support Card in U-Boot proper instead of SPL.
We may run different firmware (ex. ARM Trusted Firmware) before
U-Boot, and basic SoC initialization may be done there. In that
case, SPL may not be used.
The motivation for preparing SBC and Support Card in SPL was to use
LED for early debugging, but this is not mandatory to boot SoCs.
With this commit, LED will be unavailable in SPL, but we can use a
debug serial instead. So, this change will not be a big deal.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Merge init-*.c into a single file using a table of callbacks because
the initialization flow is almost common among SoCs.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The code here is cluttered due to the switch statement. Introduce a
table of callbacks to clean up the initialization code across SoCs.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Splitting reset assertion (support_card_reset) and deassertion
(support_card_init) is not adding much value any more. Handle
all the initialization of Support Card in support_card_init(),
then remove support_card_reset().
Also, detect_num_flash_banks() can have a static qualifier.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, memconf-sld3.c and memconf-pxs2.c duplicate the code.
There are 3 patterns in terms of MEMCONF init:
- DRAM 2 channels: LD4, sLD8, Pro4, Pro5, LD11
- DRAM 3 channels: sLD3
- DRAM 3 channels (Ch2 is disable by MEMCONF[21]): Pxs2, LD20
All of them can be moved into a single file by a little more
refactoring.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The clock enable bits for UMC are more SoC-specific than for
the other hardware blocks. Separate the UMC clocks and the other
clocks for better code reuse across SoCs.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
At first, we thought the LD20 PLL setting would be board dependent,
but this argument turned out unneeded after all.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
