Import the device-tree from linux-amlogic/for-next (Linux 6.4-rc1)
and omit the NPU node from the A311D board variant dts as this is
not supported under U-Boot.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20230323143142.780306-7-christianshewitt@gmail.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Add configuration for the BananaPi M2-Pro board.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20230323143142.780306-5-christianshewitt@gmail.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Import the board dts from the linux-amlogic/for-next (6.4-rc1)
branch. This involves spliting the BPI-M5 dts into a dtsi and
then reusing this for the M2-Pro.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20230323143142.780306-4-christianshewitt@gmail.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
The bananapi-m5 docs are missing from the file, so add them.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20230323143142.780306-2-christianshewitt@gmail.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Add separate documentation for the ODROID-HC4 board to ensure
users build U-Boot using the HC4 defconfig that enables PCIe
SATA boot. This avoids user frustration trying to boot after
using the C4 recipe which only works from SD card.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-34-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
GS-King-X is also supported with the beelink-s922x FIP sources and can use
the GT-King defconfig. Add a board document with instructions.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-33-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Improve documentation. Notably we can now support U-Boot install to
the internal eMMC storage in addition to SD cards.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-22-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Improve documentation. Notably we can now support U-Boot install to
the internal eMMC storage in addition to SD cards.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-21-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Improve documentation. Notably mention the ODROID-N2+ and the option
to use FDTDIR not FDT for automatic device-tree selection.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-19-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Improve documentation. Notably drop references to the ODROID-HC4 board
as its support has evolved and we will add an HC4 specific document in
a later patch.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-17-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Improve documentation. Notably we can now support U-Boot install to
the removable eMMC storage module in addition to SD cards.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-15-christianshewitt@gmail.com
[narmstrong: fixed doc build]
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Improve documentation. Notably we remove all the board names from the
feature matrix to reduce table width. SoC types have been added to the
titles in individual board documents so readers can still correlate a
board against the features. This makes it easier to add new boards to
to the document in the future.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Link: https://lore.kernel.org/r/20230320114609.930145-2-christianshewitt@gmail.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Add support for both the BananaPi BPI-CM4 module and the BananaPi
baseboard which is compatible with the RaspberryPi CM4IO baseboard.
The BananaPi BPI-CM4 module follows the CM4 specifications at [1],
but with a single HDMI port and a single DSI output.
The current CM4IO baseboard DT should work fine on the Raspberry CM4
baseboard and other derivatives baseboards, but proper DT should
be written for other baseboards.
[1] https://datasheets.raspberrypi.com/cm4io/cm4io-datasheet.pdf
Link: https://lore.kernel.org/r/20230307-u-boot-cm4-v1-2-43f5a393cd37@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
The biggest change is some refactoring of the H616 DRAM driver, which
allows better fine-tuning for each board, and is the base for pending
LPDDR3 and LPDDR4 support, needed by new boards. The sun8i-emac
Ethernet driver sees some refactoring that enables it for the Allwinner
D1 EMAC IP. The sunxi HDMI driver is now using more DT properties. Also
the early SPL code now supports some odd H616 SoC variant.
There are some more patches pending, that require the final review
touches and some testing, I will send a separate PR for them later.
The gitlab CI completed successfully, and I boot tested a few boards
with different SoCs, via FEL and SD card, into Linux.
- DDR Training sequence happens very fast. The speedup in boot time is
negligible by skipping the training sequence during 2nd boot or after.
So remove the check and skip.
- This change improves the robustness of DDR training. If u-boot crashed
during DDR training, the training could be left in a limbo state, where
the BootROM has recorded that it is already in a 2nd boot. The training
must be repeated in this scenario to get out of this limbo state, but due
to the check it cannot be performed.
Signed-off-by: Tony Dinh <mibodhi@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Add a new config option CONFIG_MVEBU_SPL_SATA_BLKSZ for specifying block
size of SATA disk. This information is used during building of SATA
kwbimage and must be correctly set, otherwise BootROM does not load SPL.
For 4K Native disks CONFIG_MVEBU_SPL_SATA_BLKSZ must be set to 4096.
Signed-off-by: Pali Rohár <pali@kernel.org>
Tested-by: Martin Rowe <martin.p.rowe@gmail.com>
Currently kwboot expected that sector size for SATA image is always 512
bytes. If SATA image cannot be parsed with sector size of 512 bytes, try
larger sector sizes which are power of two and up to the 32 kB. Maximal
theoretical value is 32 kB because ATA IDENTIFY command returns sector size
as 16-bit number.
Signed-off-by: Pali Rohár <pali@kernel.org>
SATA kwbimage contains offsets in block size unit, not in bytes.
Until now kwbimage expected that SATA disk always have block size of 512
bytes. But there are 4K Native SATA disks with block size of 4096 bytes.
New SATA_BLKSZ command allows to specify different block size than 512
bytes and therefore allows to generate kwbimage for disks with different
block sizes.
This change add support for generating SATA images with different block
size. Also it add support for verifying and dumping such images.
Because block size itself is not stored in SATA kwbimage, image
verification is done by checking every possible block size (it is any
power of two value between 512 and 32 kB).
Signed-off-by: Pali Rohár <pali@kernel.org>
This allows image type print_header() callback to access struct
image_tool_params *params.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Find SATA block device by blk_get_devnum_by_uclass_id() function and read
from it the real block size of the SATA disk. In case of error, fallback
back to 512 bytes.
Signed-off-by: Pali Rohár <pali@kernel.org>
