The PCIe Controllers on the J784S4 and AM69 SoCs support Root-Complex
mode of operation. PCIe0 instance of PCIe on both of the SoCs is brought
out on the Starter-Kit (AM69) and EVM (J784S4) boards. Hence, enable
the configs required for Root-Complex mode of operation.
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Fixes: 79f3e77133 ("Subtree merge tag 'v6.16-dts' of dts repo [1] into dts/upstream")
The PCIe Controllers on the J784S4 SoC support Root-Complex mode of
operation. Hence, enable it.
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Fixes: 79f3e77133 ("Subtree merge tag 'v6.16-dts' of dts repo [1] into dts/upstream")
The PCIe Controllers in the K3 SoCs have 4 GB Address Windows in the
64-bit address space to map System (CPU) Addresses to PCIe Bus Addresses.
The physical addresses for these Address Windows across PCIe instances
across SoCs is as follows:
+--------+----------------+----------------+----------------+----------------+
| SoC | PCIe0 | PCIe1 | PCIe2 | PCIe3 |
+--------+----------------+----------------+----------------+----------------+
| AM64 | 0x6_0000_0000 | NA | NA | NA |
| J722S | 0x6_0000_0000 | NA | NA | NA |
| AM68 | NA | 0x41_0000_0000 | NA | NA |
| J7200 | NA | 0x41_0000_0000 | NA | NA |
| J721S2 | NA | 0x41_0000_0000 | NA | NA |
| J742S2 | 0x40_0000_0000 | 0x41_0000_0000 | NA | NA |
| AM69 | 0x40_0000_0000 | 0x41_0000_0000 | 0x42_0000_0000 | 0x43_0000_0000 |
| J721E | 0x40_0000_0000 | 0x41_0000_0000 | 0x42_0000_0000 | 0x43_0000_0000 |
| J784S4 | 0x40_0000_0000 | 0x41_0000_0000 | 0x42_0000_0000 | 0x43_0000_0000 |
+--------+----------------+----------------+----------------+----------------+
Two regions for a 1:1 mapping from virtual addresses to physical addresses
catering to all of the above will be required, which are:
1. For AM64 and J722S SoCs
=> Start: 0x6_0000_0000 Size: 0x1_0000_0000
2. For AM68, AM69, J7200, J721E, J721S2, J742S2 and J784S4 SoCs
=> Start: 0x40_0000_0000 Size: 0x4_0000_0000
Since the 'Flash Peripherals' region from 0x5_0000_0000 to 0x8_7FFF_FFFF
includes the mapping for AM64 and J722S SoCs, only the second region
mentioned above needs to be added.
Hence, add the region to support 64-bit address space for PCIe.
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Fixes: 79f3e77133 ("Subtree merge tag 'v6.16-dts' of dts repo [1] into dts/upstream")
Drivers should extract device-tree data before probing via the
.of_to_plat hook.
Implement it for stm32_dsi driver. By doing so, it also solve a
variable shadowing in stm32_dsi_probe() where &clk was used as
peripheral clock and ref clock.
For readability some struct have been renamed such as:
* struct stm32_dsi_priv *dsi -> struct stm32_dsi_priv *priv
* struct clk clk -> struct clk pclk
Signed-off-by: Raphael Gallais-Pou <rgallaispou@gmail.com>
Reviewed-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Add Kconfig option OF_SYSTEM_SETUP=y to support booting boards with a
RK3582 SoC. CPU and GPU cores are failed based on ip-state and policy.
Tested on a ROCK 5C Lite v1.1:
cpu-code: 35 82
ip-state: 10 00 00 (otp)
ip-state: 30 9e 04 (policy)
remove cpu-map cluster1
rename cpu-map cluster2
fail gpu
fail rkvdec1
fail rkvenc1
fail cpu cpu@400
fail cpu cpu@500
and on a Radxa E52C:
cpu-code: 35 82
ip-state: 00 04 00 (otp)
ip-state: c0 9e 04 (policy)
remove cpu-map cluster2
fail gpu
fail rkvdec1
fail rkvenc1
fail cpu cpu@600
fail cpu cpu@700
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Quentin Schulz <quentin.schulz@cherry.de>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
The RK3582 SoC is a variant of the RK3588S with some IP blocks disabled.
What blocks are disabled/non-working is indicated by ip-state in OTP.
This add initial support for RK3582 by using ft_system_setup() to mark
any cpu, gpu and/or vdec/venc node with status=fail as indicated by
ip-state.
This apply same policy as vendor U-Boot for RK3582, i.e. two big cpu
cores, the gpu and one vdec/venc core is always failed/disabled.
Enable Kconfig option OF_SYSTEM_SETUP in board defconfig to make use of
the required DT fixups for RK3582 board variants.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
The USB OTG U3 port may have been disabled early, add support to the
COMBPHY driver to re-enable the U3 port.
This matches changes made in the Linux commit 7bb14b61b7d0 ("phy:
rockchip: naneng-combphy: Enable U3 OTG port for RK3568").
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Acked-by: Quentin Schulz <quentin.schulz@cherry.de>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
RK3588 has two known memory gaps when using 16+ GiB DRAM,
[0x3fc000000, 0x3fc500000) and [0x3fff00000, 0x400000000).
The vendor TPL blob encodes this information in the DDR_MEM tag data
passed to later boot phases. U-Boot proper will normally use this
information to configure the DRAM banks.
When a DDR_MEM tag cannot be found the fallback is to use the entire
range above 4 GiB. Something that will cause issues when OS try to use
memory from the two known memory gaps.
Add a weak dram init banksize fixup function and implement it for RK3588
to fix issues observed when trying to RAM boot RK3588 boards with 16+
GiB DRAM into Linux.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Add a simple unit test for testing the RAUC bootmethod. Provide only the
very basic tests for now, running a scan and list, to verify correct
detection of the RAUC bootmethod. More advanced boot tests of this
bootmethod can be added in a separate patch.
This requires another mmc image (mmc10) to contain the following
partitions:
1. boot A: contains a dummy boot.scr
2. root A: contains an empty root filesystem
3. boot B: contains a dummy boot.scr
4. root B: contains an empty root filesystem
The bootmeth_rauc scans all four partitions for existence and expects a
boot script in each boot partition.
Also add BOOTMETH_RAUC as a dependency on sandbox so that we can test
this with:
$ ./test/py/test.py -B sandbox --build -k test_ut # build the mmc10.img
$ ./test/py/test.py -B sandbox --build -k bootflow_rauc
Signed-off-by: Martin Schwan <m.schwan@phytec.de>
Reviewed-by: Simon Glass <simon.glass@canonical.com>
[trini: mmc9 is now in use, switch to mmc10]
Signed-off-by: Tom Rini <trini@konsulko.com>
FIT images don't work without having to explicitly specify physical
load addresses. Digging into that it looks like a flaw in
bootm_load_os().
It duplicates images->os for convenience. However, the code handling
"kernel_noload" images then updates the load address in the copy with
the value lmb_alloc_mem() returned. Later there's another call to
lmb_alloc_mem() that uses the old value. This leads to havoc due
to subsequent calls of lmb_alloc_mem() picking too low addresses.
The "fix" is to mark the local variable const to avoid accidental
assignments. This works but IMO the logic is still flawed somehow as
this leads to overlapping lmb reservations. I guess the fixed
reservation should only be done when the noload path wasn't hit.
Without the change:
+ bootm 0x40200000#qemu-arm 0x40200000#qemu-arm 0x40000000
Using 'qemu-arm' configuration
Verifying Hash Integrity ... OK
Trying 'kernel' kernel subimage
Description: Linux kernel
Created: 2026-02-24 14:10:09 UTC
Type: Kernel Image (no loading done)
Compression: gzip compressed
Data Start: 0x402000b8
Data Size: 12227440 Bytes = 11.7 MiB
Hash algo: sha256
Hash value: 7ea661fdecdd1127edd419cfbf8bff52e2d5ac55c...
Verifying Hash Integrity ... sha256+ OK
Using 'qemu-arm' configuration
Verifying Hash Integrity ... OK
Trying 'ramdisk' ramdisk subimage
Description: Initial ramdisk
Created: 2026-02-24 14:10:09 UTC
Type: RAMDisk Image
Compression: uncompressed
Data Start: 0x40da9528
Data Size: 1067114 Bytes = 1 MiB
Architecture: AArch64
OS: Linux
Load Address: unavailable
Entry Point: unavailable
Hash algo: sha256
Hash value: 2a711dcb5f58615187645ccec615c67eddcfbb3138...
Verifying Hash Integrity ... sha256+ OK
Booting using the fdt blob at 0x40000000
Working FDT set to 40000000
Uncompressing Kernel Image (no loading done) to 13a400000
Loading Ramdisk to 400fb000, end 401ff86a ... OK
device tree - allocation error
FDT creation failed!
resetting ...
Bloblist at 0 not found (err=-2)
alloc space exhausted ptr 400 limit 0
Bloblist at 0 not found (err=-2)
[reset]
After:
+ bootm 0x40200000#qemu-arm 0x40200000#qemu-arm 0x40000000
Using 'qemu-arm' configuration
Verifying Hash Integrity ... OK
Trying 'kernel' kernel subimage
Description: Linux kernel
Created: 2026-02-24 14:10:09 UTC
Type: Kernel Image (no loading done)
Compression: gzip compressed
Data Start: 0x402000b8
Data Size: 12227440 Bytes = 11.7 MiB
Hash algo: sha256
Hash value: 7ea661fdecdd1127edd419cfbf8bff52e2d5ac55ce...
Verifying Hash Integrity ... sha256+ OK
Using 'qemu-arm' configuration
Verifying Hash Integrity ... OK
Trying 'ramdisk' ramdisk subimage
Description: Initial ramdisk
Created: 2026-02-24 14:10:09 UTC
Type: RAMDisk Image
Compression: uncompressed
Data Start: 0x40da9528
Data Size: 1067114 Bytes = 1 MiB
Architecture: AArch64
OS: Linux
Load Address: unavailable
Entry Point: unavailable
Hash algo: sha256
Hash value: 2a711dcb5f58615187645ccec615c67eddcfbb3138...
Verifying Hash Integrity ... sha256+ OK
Booting using the fdt blob at 0x40000000
Working FDT set to 40000000
Uncompressing Kernel Image (no loading done) to 13a400000
Loading Ramdisk to 13a2fb000, end 13a3ff86a ... OK
Loading Device Tree to 000000013a1f8000, end 000000013a2fafff ... OK
Working FDT set to 13a1f8000
Starting kernel ...
Signed-off-by: Ludwig Nussel <ludwig.nussel@siemens.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Rafał Miłecki <rafal@milecki.pl> says:
I found flash_write_buf() somehow more difficult to follow than needed.
Some code turned out to be inconsistent and some not really needed.
There is a set of 6 patches that should be small enough to make them easy
to review. I was wondering if I should squash patches 5 and 6 but ended up
with sending two smaller diffs.
Those changes shouldn't really change any behaviour (except for smaller mem
alloc). Let me know if there are any issues with this!
Link: https://lore.kernel.org/r/20260221100011.29049-1-zajec5@gmail.com
Recent cleanups made some variables redundant. Both: "blocklen" and
"erase_len" ended up being used as temporary variables used locally in
some short code paths.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Reviewed-by: Tom Rini <trini@konsulko.com>
Calculate "erasesize" in the same place where other offsets and lengths
/ sizes are calculated. It makes code more consistent and will allow
further cleanups.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
In both code paths "blockstart" is calculated the same way. Unify it.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Reviewed-by: Tom Rini <trini@konsulko.com>
We already have DIV_ROUND_UP() to simplify code so add ROUND_UP() as
well. This makes code in flash_write_buf() easier to follow.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
When dealing with env data not aligned to flash blocks flash_write_buf()
has to use an extra buffer. It reads existing flash content to it,
modifies required part and writes it back.
While reading and writing a size stored in "write_total" is used. It's
what should be used when allocating the buffer too.
In some cases allocating memory of "erase_len" size could result in
allocating too big buffer. That wouldn't break anything but it was
making code less intuitive.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Use "erasesize" instead of "blocklen" in flash_write_buf()'s loop. This
change touches code executed for bad NAND blocks so it doesn't affect
any behaviour (for NAND flashes "blocklen" and "erasesize" are equal).
This just makes code a bit more consistent as "erasesize" is what is
used all around inside the writing loop.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
The upstream devicetree am33xx.dtsi does not have a "ti,am33xx-usb"
compatible, it uses "ti,sysc-omap4" for the same node. The
implementation of ti-musb uses a wrapper driver that binds to
ti,am33xx-usb and creates new devices ti-musb-host and
ti-musb-peripheral depending on the dr_mode property.
To avoid this wrapper driver with the upstream devicetree, add
compatibles for "ti,musb-am33xx" to both ti-musb-host and
ti-musb-peripheral. Add a bind function that checks for the correct
dr_mode value and rejects binding if it is not the correct driver.
Reviewed-by: Mattijs Korpershoek <mkorpershoek@kernel.org>
Signed-off-by: Markus Schneider-Pargmann (TI.com) <msp@baylibre.com>
Reviewed-by: Marek Vasut <marek.vasut@mailbox.org>
For the upstream DT the ctrl_mod node is using a relative register
address which is not translated by the current code.
Make address parsing understand relative addresses.
Reviewed-by: Mattijs Korpershoek <mkorpershoek@kernel.org>
Signed-off-by: Markus Schneider-Pargmann (TI.com) <msp@baylibre.com>
Reviewed-by: Marek Vasut <marek.vasut@mailbox.org>
Philip Molloy <philip@philipmolloy.com> says:
This series updates the maintainers for the ADI ADSP platform. It
follows Greg's series adding support for ADI ADSP SoCs.
Timesys spent years developing and maintaining Linux support for ADI
ADSP SoCs. The maintenance contract has ended and ADI has brought that
effort in-house. Additionally, Timesys was acquired by another company.
Thanks to everyone at Timesys for all of their hard work over the years!
Link: https://lore.kernel.org/r/20260226111136.354009-1-philip@philipmolloy.com
After years of developing the ADI ADSP platform, Timesys was purchased
by another company and is no longer contracted to maintain the platform.
Signed-off-by: Philip Molloy <philip.molloy@analog.com>
Reviewed-by: Greg Malysa <malysagreg@gmail.com>
After years of developing the ADI ADSP platform, Timesys was purchased
by another company and is no longer contracted to maintain the platform.
Utsav is leaving ADI after contributing to ADSP SoCs for the last 3
years. He is a founding member of the in-house team supporting the
chips.
Linux support at ADI has been consolidated. Use the company-wide mailing
list and git repository.
Signed-off-by: Philip Molloy <philip.molloy@analog.com>
Reviewed-by: Greg Malysa <malysagreg@gmail.com>
More strict checks in GCC 15 expose a new warning:
tools/atmelimage.c: In function ‘atmel_find_pmecc_parameter_in_token’:
tools/atmelimage.c:64:31: error: assignment discards ‘const’ qualifier from pointer target type [-Werror=discarded-qualifiers]
64 | param = strstr(token, "=");
| ^
cc1: all warnings being treated as errors
Add 'const' qualifier to variable 'param' to prevent build failing
due to -Werror.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
A few fixes/missing changes for UFS:
- remove unused ufs_post_bind() declaration
- Disable UTP command timeout in slow mode
- Missing MediaTek UFS PHY Driver to be used with the UFS driver
Remove the "phandle = <0x..>;" properties from the DT diff between
unpatched base DT and U-Boot augmented DT if DEVICE_TREE_DEBUG=1.
The phandle numbers are only generated by DTC, but not referenced
anywhere in the DT, because the original references are specifically
not replaced by phandle numbers when recent DTC is invoked with the
-I dts -O dts flags . The phandle number are therefore only a noise
in the diff, filter them out.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Tested-by: Quentin Schulz <quentin.schulz@cherry.de>
Reviewed-by: Quentin Schulz <quentin.schulz@cherry.de>
The exported sub_make_done variable leaks into the environment of all
child processes. When make targets like tcheck spawn independent make
invocations with O=, those child makes inherit sub_make_done=1, skip
the KBUILD_OUTPUT setup and try to build in the source tree.
There is a workaround that resets sub_make_done to 0 for specific test
targets, but this isn't great since it has tolist every target that
spawns independent make invocations.
Instead, unexport sub_make_done once we are in the final make
invocation. The direct sub-make already has the value in its
environment from the export, and no further propagation is needed.
This also allows the per-target workaround to be removed.
Fixes: 27529f1cb0 ("kbuild: skip parsing pre sub-make code for recursion")
Signed-off-by: Simon Glass <simon.glass@canonical.com>
The bootmenu_conv_key() function is shared with expo subsystem for key
input. Adding alphanumeric-to-BKEY_SHORTCUT conversion there causes expo
to swallow typed characters instead of inserting them as text, since
BKEY_SHORTCUT falls in the range that expo treats as a command key
rather than passing through.
Move the shortcut-key detection into bootmenu_loop() where it is
only used in the bootmenu context.
Fixes: 8c986521c3 ("cmd: bootmenu: permit to select bootmenu entry with a shortcut")
Signed-off-by: Simon Glass <simon.glass@canonical.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
The ofnode_to_fdt() function may return a NULL pointer in multiple cases.
Or, this function's return value is often passed directly to functions such
as fdt_getprop() which end up dereferencing it, thus causing a NULL pointer
exception.
Don't allow ofnode_to_fdt() to return NULL, to avoid a NULL pointer
dereference.
Reviewed-by: Raphaël Gallais-Pou <raphael.gallais-pou@foss.st.com>
Signed-off-by: Romain Gantois <romain.gantois@bootlin.com>
Reviewed-by: Simon Glass <simon.glass@canonical.com>
CI: https://source.denx.de/u-boot/custodians/u-boot-imx/-/pipelines/29404
- Miscelanous improvements for Siemens Capricorn board.
- Convert i.MX6 IPUv3 driver to use clock framework.
- Skip voltage switching for fixed 1.8V regulator on fsl_esdhc_imx.
- Support printing imx8m pinmux.
- Enter fastboot on USB boot by default on phycore-imx93.
- Use arch override for env_get_location() on imx95.
The get_ram_size() function fails to restore the original RAM data when
the data cache is enabled. This issue was observed on an AM625 R5 SPL
with 512MB of RAM and is a regression that became visible with
commit bc07851897 ("board: ti: Pull redundant DDR functions to a common
location and Fixup DDR size when ECC is enabled").
Observed boot failure messages:
Warning: Did not detect image signing certificate. Skipping authentication to prevent boot failure. This will fail on Security Enforcing(HS-SE) devices
Authentication passed
Starting ATF on ARM64 core...
The system then hangs. This indicates that without a data cache flush,
data in the cache is not coherent with RAM, preventing the system from
booting. This was verified by printing the content of this address when
the issue occurs.
Add a data cache flush after each restore operation to resolve this
issue.
Fixes: bc07851897 ("board: ti: Pull redundant DDR functions to a common location and Fixup DDR size when ECC is enabled")
Fixes: 1c64b98c1e ("common/memsize.c: Fix get_ram_size() when cache is enabled")
Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
Reviewed-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
Tested-by: Francesco Dolcini <francesco.dolcini@toradex.com> # Toradex Verdin AM62
Commit 067c1b0332 ("ufs: Call ufs_scsi_bind() from uclass .post_bind")
inlined ufs_scsi_bind() into ufs_post_bind() as trivial
device_bind_driver() call.
ufs_scsi_bind() is no longer referenced anywhere in the codebase, so
drop its declaration from include/ufs.h.
Drivers used to include <ufs.h> to include prototype of ufs_scsi_bind()
function, so we can now safely remove such includes.
Fixes: 067c1b0332 ("ufs: Call ufs_scsi_bind() from uclass .post_bind")
Signed-off-by: Julien Stephan <jstephan@baylibre.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patch.msgid.link/20260203-cleanup-ufs-header-v1-1-4c10424485f0@baylibre.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Like commit b9e48705e0 ("arm: imx: imx9: Use arch override for
env_get_location()"), use arch-level implementation here so that
env_get_location() can be used on board-level.
Signed-off-by: Sébastien Szymanski <sebastien.szymanski@armadeus.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Marek Vasut <marek.vasut@mailbox.org>
The i.MX9 pinctrl match table currently lists all SoC compatibles
unconditionally, which may lead to unused entries being included when
building for specific SoC variants. Guard each compatible entry with
the corresponding CONFIG_IMX9[X] option so only the required SoC
entries are compiled in, which reduces unnecessary data.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Add support for printing pin names and current mux configuration on i.MX91
when CMD_PINMUX is enabled by adding full pin descriptor table for i.MX91
pads.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Add support for printing pin names and current mux configuration on
i.MX93 when CMD_PINMUX is enabled.
- A full pin descriptor table for i.MX93 pads.
- Implementation of get_pins_count(), get_pin_name(), and
get_pin_muxing() in the i.MX9 pinctrl driver.
There is no good way to add real mux names, so just dump the function ID
from the mux register.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
We are going to add pinctrl data support for both i.MX93 and i.MX91.
Since these SoCs share the same pinctrl driver structure, rename
pinctrl-imx93.c to a more generic pinctrl-imx9.c and update all related
variable and function names accordingly.
This prepares the driver for supporting additional i.MX9 family SoCs.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
The i.MX8M pinctrl match table currently lists all SoC compatibles
unconditionally, which may lead to unused entries being included when
building for specific SoC variants. Guard each compatible entry with
the corresponding CONFIG_IMX8M[X] option so only the required SoC
entries are compiled in, which reduces unnecessary data.
Signed-off-by: Peng Fan <peng.fan@nxp.com>
