commits
I think the regulators already defined are supposed to be under a
regulators section inside the pmic.
- needs SCLK_MAC/SCLK_MAC_SRC support in clock driver
Signed-off-by: Alex Bee <knaerzche@gmail.com>
- based on sdram_rk322x.c
- phy is different
- replaced magic numbers
- removed DM/DT to fit in SRAM
- supports LPDDR2/DDR3 (all types as per DS)
- addtionally set trfc/trefi based on detected cap
- use sdram_common
- set axi priorities as vendor blob does
- provide serveral dram timings to make it "future proof"
(hopefully nobody will have fiddle to make it still fit in SRAM
when adding a new board)
Signed-off-by: Alex Bee <knaerzche@gmail.com>
The currently existing cs_cap calculation converts to MB, but the values
compared to are in Byte.
Example: 8Gb DDR3, 32-bit (15 cols, 10 rows) == 0x400 (1024 MB)
Fix it.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
test_addr is initialized as void __iomem *, but unlike test_addr1,
CFG_SYS_SDRAM_BASE is not properly casted when it is assigend.
Fix it.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
In order to reduce the size of defconfig when more boards get added for
this SoC provide Kconfig defaults.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
In order to prepare the switch to TPL/SPL for RK3128 provide basic Soc
init, which currently supports getting the boot device, initialization of
all 3 uart channels pins or usb uart, and disable forcing to JTAG mode of
the sdmmc/uart2 pins.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Support USB2 PHY in RK3128 which is used by it's HOST and OTG USB controllers.
Also enable it for evb-rk3128 which enables these in it's device tree.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
RK3128's efuse is currently only supported by u-boot, so add it to
rk3128-u-boot.dtsi and imply the options required to get the serial
and the ethernet mac address(es) in SoC Kconfig defaults.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Move RK3128 to OF_UPSTREAM. Properties that were used by evb-rk3128 and have
only been present in the former u-boot rk3128.dtsi have been moved to
rk3128-evb-u-boot.dtsi.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Similar to other ARMv7 Rockchip SoCs the linux kernel is unable to access an
fdt blob or initrd if it is put in DRAM above 512 MB. Add fdt_high and
initrd_high in order to prevent that.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
RK3128's uart clocks are not supported by u-boot's clock driver, so add
clock-frequency property to SoC's u-boot.dtsi
Signed-off-by: Alex Bee <knaerzche@gmail.com>
MMC controllers in RK3128 do not have the IDMAC feature and instead use PL330
for DMA which is unsupported by u-boot, thus we can use them in fifo-mode only.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Those enums are not used anywhere for RK3128 and they will cause problems
if both RK3128's and RK3288's grf header is included in a driver, because
RK3288's grf header defines enums with members which have exactly the same
name as these.
If they are required, they can be defined inline as it is done for most
other Rockchip SoCs.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
GRF registers from soc_con0 onwards are currently unusable because of a missing
offset before these, which is currently compensated by a to big reserved space
after those registers (to make check_member happy).
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Support the reset controller part of RK3128's CRU.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Support setting and getting all required clocks rates for sdmmc and emmc.
Also replace function prefix from rockchip_ to more appropriate rk3128_.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
A PLL is locked if the respective bit is set, thus we have to wait until
is set.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Adjust legoev3_defconfig to reduce the binary output size.
As u-boot has bloated a bit over the years, the legoev3_defconfig can no
longer build something that fits in the 256kB size limit of the EV3.
This drops a few unused features, but the real difference-makers are
enabling thumb instructions and using link time optimization to reduce
the size.
This reduced u-boot.bin from 279,920 to 198,416 bytes on my local
machine with arm-none-eabi-gcc (15:13.2.rel1-2) 13.2.1 20231009.
HAS_BOARD_SIZE_LIMIT is also added to catch any future regressions.
Signed-off-by: David Lechner <david@lechnology.com>
The mx6sabresd U-Boot proper binary size has grown beyond the
CONFIG_BOARD_SIZE_LIMIT.
Reduce its size by removing the CONFIG_MULTI_DTB_FIT, BOOTM_PLAN9 and
BOOTM_RTEMS options.
According to doc/README.multi-dtb-fit:
"Usually the DTB is selected by the SPL and passed down to U-Boot. But some
platforms don't use the SPL. In this case MULTI_DTB_FIT can used to provide
U-Boot with a choice of DTBs"
mx6sabresd uses SPL, so MULTI_DTB_FIT can be safely dropped as the DTB
selection in SPL is done by board_fit_config_name_match().
Tested boot on the imx6dl and imx6q variants.
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Pull request efi-2025-07-rc6
Documentation:
* Describe usage of U-Boot environment variables in extlinux.conf
UEFI:
* Add missing variable initialization in
efi_bootmgr_update_media_device_boot_option
This variable might end up being uninitialized if we exit early.
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
This option is very useful for A/B boot setups with read-only
filesystems: Letting U-Boot fill in the rootfs (and possibly related
parameters) allows keeping all boot parameters except the actual slot
selection in the extlinux.conf file, where they can be updated easily.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Cc: Tom Rini <trini@konsulko.com>
I'm moving to a new address, so map my old one to it.
Also update the MAINTAINERS entries.
Signed-off-by: Duje Mihanović <duje@dujemihanovic.xyz>
Since commit 51ec8db23280 ("pylibfdt: correct license information"), the
License classifiers are gone so I assume setuptools now extract the
license from the license argument to setuptools.setup() function.
It's always been incorrect as far as I could tell, so let's fix this
with the appropriate info from the SPDX License identifier at the top of
the file. It was missing GPL-2.0-or-later and we disambiguate by using
BSD-2-Clause instead of simply BSD.
Fixes: 6b08fb5cc44f ("fdt: Move to setuptools")
Signed-off-by: Quentin Schulz <quentin.schulz@cherry.de>
Heinrich Schuchardt <heinrich.schuchardt@canonical.com> says:
spl_fit_get_image_name() used to lack a detection of malformed image name
properties in FIT images. The change in commit 3704b888a4ca ("common/spl:
fix potential out of buffer access in spl_fit_get_image_name function")
tried to fix this but led to function spl_fit_get_image_name() no longer
detecting if a property at index > 1 does not exist.
This patch is reverted.
An explicit check for malformed image name properties is introduced.
Link: https://lore.kernel.org/u-boot/38f5d078-3328-4bdb-9c95-4fb5fe89ddc2@gmx.de/T/#u
Link: https://lore.kernel.org/r/20250624153431.46986-1-heinrich.schuchardt@canonical.com
A malformed FIT image could have an image name property that is not NUL
terminated. Reject such images.
Reported-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Tested-by: E Shattow <e@freeshell.de>
The change in commit 3704b888a4ca ("common/spl: fix potential out of buffer
access in spl_fit_get_image_name function") led to function
spl_fit_get_image_name() no longer detecting if a property does not exist
at a non-zero buffer.
Link: https://lore.kernel.org/u-boot/38f5d078-3328-4bdb-9c95-4fb5fe89ddc2@gmx.de/T/#m59f3a23e675daa992c28d12236de71cae2ca2bb9
Fixes: 3704b888a4ca ("common/spl: fix potential out of buffer access in spl_fit_get_image_name function")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Tested-by: E Shattow <e@freeshell.de>
The log level "emergency" cannot be used as that results in compilation
failure. Correct the macro name used to print at this level.
Fixes: 249679658750 ("log: provide missing macros")
Signed-off-by: Akashdeep Kaur <a-kaur@ti.com>
Reviewed-by: Quentin Schulz <quentin.schulz@cherry.de>
gunzip is used in cmd/config.c
Signed-off-by: Marius Dinu <m95d+git@psihoexpert.ro>
Reviewed-by: Quentin Schulz <quentin.schulz@cherry.de>
Observing the crash when we add the i2c-arbitrator node in the device
tree as per the DT bindings. The issue is with the child node of
i2c-arbitrator@72 i.e., i2c@f1950000->i2c-arbitrator@72->i2c-arb, as the
arbitrator uses the uclass of mux(UCLASS_I2C_MUX) and the mux uclass driver
checks for the "reg" property using the i2c_mux_child_post_bind() function,
if it won't find the "reg" property it will return -EINVAL which is leading
to the crash.
So, add the logic to check whether the child node has the "reg" property,
if the "reg" property exists then read the "reg" and update the channel.
https://www.kernel.org/doc/Documentation/devicetree/bindings/i2c/i2c-arb.txt
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
The value passed to dcd_skip of build_container() should be obtained by
parsing .cfg file, and should not be fixed to false. For i.MX8QXP, dcd
data needs to be skipped, in which case dcd_skip should be true.
Fixes: 5f28a6599f01("tools: imx8image: add i.MX95 support")
Signed-off-by: Alice Guo <alice.guo@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
Reported-by: Enric Balletbo i Serra <eballetb@redhat.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Commit 6fe50e395080 ("arm: asm/system.h: mrc and mcr need .arm if
__thumb2__ is not set") is not a proper fix for the LTO link error
mentioned in its description. It causes 32-bit arm instructions to be
mixed with thumb instructions, which the Kirkwood SoCs do not support.
For example, board_init_r() is mostly generated in Thumb-1 mode as
expected since the build flags contain -mthumb -mthumb-interwork. The
MCR instruction corresponding to writefr_extra_feature_reg() is also
correcly emitted as a 32-bit ARM instruction (it cannot be encoded in
Thumb-1 anyways). The problem is, the compiler inlines the MCR without
generating the BX or BLX instruction which are needed to transition
between the ARM and the Thumb-1 states. From the objdump output:
006186a0 <board_init_r>:
board_init_r():
/home/jerome/work/u-boot/common/board_r.c:799
6186a0: b5f0 push {r4, r5, r6, r7, lr}
6186a2: b0ab sub sp, #172 @ 0xac
get_gd():
/home/jerome/work/u-boot/./arch/arm/include/asm/global_data.h:127
6186a4: 464a mov r2, r9
...
/home/jerome/work/u-boot/arch/arm/mach-kirkwood/cpu.c:242
619aae: 9b15 ldr r3, [sp, #84] @ 0x54
writefr_extra_feature_reg():
/home/jerome/work/u-boot/./arch/arm/include/asm/arch/cpu.h:100
619ab0: ee2f3f11 mcr 15, 1, r3, cr15, cr1, {0}
^^^^^^^^
32-bit ARM instruction
Further investigation is needed to understand how to fix the issue so
that the code size is minimal for all boards. In the mean time, this
fix disables LTO for the two problematic files (common/board_f.c and
common/board_r.c). This makes the Kirkwood-based boards bootable again.
The binary size is increased by 1048 bytes which is perfectly
acceptable.
Fixes: 6fe50e395080 ("arm: asm/system.h: mrc and mcr need .arm if __thumb2__ is not set")
Reported-by: Tony Dinh <mibodhi@gmail.com>
Tested-by: Tony Dinh <mibodhi@gmail.com>
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Tom Rini <trini@konsulko.com>
Shiji Yang <yangshiji66@outlook.com> says:
This patchset fixes some compilation errors that I caught in version
v2025.07-rc4 and branch next. If they are acceptable, please apply
them to the master branch. If anyone has a better way to fix these
issues, it's fine to ignore this patchset.
Link: https://lore.kernel.org/r/OSBPR01MB16702ED24460D23A7ED63440BC7DA@OSBPR01MB1670.jpnprd01.prod.outlook.com
The function definition of fit_check_sign() is guarded by
"#ifdef CONFIG_FIT_SIGNATURE" in "tools/image-host.c". If we try
to build it without CONFIG_FIT_SIGNATURE, we will get an error:
/usr/bin/ld: tools/fit_check_sign.o: in function `main':
fit_check_sign.c:(.text.startup+0x165): undefined reference to `fit_check_sign'
collect2: error: ld returned 1 exit status
Fixes: 9c79c8fe70da ("tools/fit_check_sign: make key optional")
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
The symbol TOOLS_IMAGE_PRE_LOAD doesn't depend on TOOLS_LIBCRYPTO.
If we choose to build tools without openssl, rsa_verify_openssl()
will attempt to call the unavailable openssl library functions.
Fixes: 942c8c8e6697 ("rsa: Add rsa_verify_openssl() to use openssl for host builds")
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Device bus address might not be valid for direct access when the bus
address and CPU address are not the same. Use dev_bus_to_phys() to
translate bus address back to CPU address.
Fixes: 3d98b8c504e15 ("net: designware: Invalidate RX buffer cache before freeing the DMA descriptor")
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
As stated in 5e847f7729b3 ("efi_loader: call efi_init_early() earlier"):
efi_init_early() creates an event hook for block device probing.
It has to be called before any block device is probed.
Indeed, efi_bl_init() registers EVT_DM_POST_PROBE event, which calls
efi_disk_probe() whenever any block device is probed. And to make that
hook work, the initialization of all block devices was put after
efi_init_early() in initcall_run_r():
INITCALL(efi_init_early);
INITCALL(initr_nand);
INITCALL(initr_onenand);
INITCALL(initr_mmc);
Because LDFW firmware is being read from MMC, attempt to load LDFW in
board_init() causes MMC driver to be probed. And because board_init() is
executed before efi_init_early(), the hook mentioned above won't work
for MMC devices anymore. So EFI disk objects won't be created, which in
turn makes the EFI subsystem non-functional, showing next symptoms:
- 'efidebug dh' output is empty
- attempt to add boot devices in 'eficonfig' shows this message:
"No block device found!"
- 'bootefi selftest $fdtcontroladdr' shows this warning:
"Cannot persist EFI variables without system partition"
- booting GRUB with 'bootefi' runs minimal GRUB shell which doesn't
see any block devices as well, probably because EFI vars weren't
passed
Load LDFW in board_late_init() instead, as it's called after
efi_init_early(). This fixes the described problem and makes it possible
to run EFI apps like GRUB correctly, add entries in 'eficonfig', and
makes 'efivar --list' command in Linux rootfs actually show EFI
variables.
The only user of LDFW at the moment is the TRNG driver, and it's probed
later, only when it's going to be used (e.g. on "rng" command). So it's
fine to load LDFW in board_late_init(). Now the corresponding call order
will look like this:
efi_init_early()
initr_mmc()
mmc_probe()
EVT_DM_POST_PROBE -> efi_disk_probe()
board_late_init()
load_ldfw() -> fs_read(), blk_dread()
exynos_trng_probe()
Fixes: ccfd8de541a8 ("board: samsung: e850-96: Report LDFW loading failures")
Fixes: f04e58cc9788 ("board: samsung: e850-96: Load LDFW firmware on board init")
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Pull request doc-2025-07-rc5
* in wget documentation remove erroneous note about CAs
The note about U-Boot not being able to verify server certificates is
false now that WGET_CACERT and WGET_BUILTIN_CACERT have been added.
Remove it.
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Add me as maintainer for the RaidSonic ICY BOX ib62x0.
Signed-off-by: Tony Dinh <mibodhi@gmail.com>
This series from Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
fixes some regressions related to handling of FIT images with broken
contents that was introduced in this merge window.
Link: https://lore.kernel.org/r/20250610095632.1085431-1-mikhail.kshevetskiy@iopsys.eu
This fix a possible NULL pointer dereference.
There is also a risk of memory leaking within the same portion of code.
The leak will happen if loaded image is bad or damaged. In this case
u-boot-spl will try booting from the other available media. Unfortunately
resources allocated for previous boot media will NOT be freed.
We can't fix that issue as the memory allocation mechanism used here
is unknown. It can be different kinds of malloc() or something else.
To somewhat reduce memory consumption, one can try to reuse previously
allocated memory as it's done in board_spl_fit_buffer_addr() from
test/image/spl_load.c.
The corresponding comment was put to the code as well.
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Reviewed-by: Anshul Dalal <anshuld@ti.com>
load_simple_fit() returns -EPERM for the images with broken signatures.
Unfortunately this may conflict with image loaging selection on the base
of boot phase. See commit 873112db9ce68c38984ff25808dde726f8dd5573
("spl: Support selecting images based on phase in simple FIT").
Thus loading of
configurations {
uboot {
description = "u-boot";
firmware = "atf";
loadables = "atf", "tee", "uboot";
};
};
with damaged "tee" image may finish without errors. This may results in
board bricking.
This patch fixes commit 873112db9ce68c38984ff25808dde726f8dd5573
("spl: Support selecting images based on phase in simple FIT")
by replacing EPERM with EBADSLT places where it should be done.
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
The current code have two issues:
1) ineffective NULL pointer check
str = strchr(str, '\0') + 1
if (!str || ...
The str here will never be NULL (because we add 1 to result of strchr())
2) strchr() may go out of the buffer for the special forms of name variable.
It's better use memchr() function here.
According to the code the property is a sequence of C-string like
shown below:
'h', 'e', 'l', 'l', 'o', '\0', 'w', 'o', 'r', 'l', 'd', '\0', '!', '\0'
index is the string number we are interested, so
index = 0 => "hello",
index = 1 => "world",
index = 2 => "!"
The issue will arrise if last string for some reason have no terminating
'\0' character. This can happen for damaged or specially crafted dtb.
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Reviewed-by: Tom Rini <trini@konsulko.com>
According to include/sysreset.h the reset_cpu() function does not take any args
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Reviewed-by: Tom Rini <trini@konsulko.com>
- Move early SPL stack on R-Car V4H boards
The CONFIG_CUSTOM_SYS_INIT_SP_ADDR=0xeb300000 does not make it into
board .config on either R-Car V4H White Hawk or Sparrow Hawk, remove
this configuration option. The early SPL stack is however pointing
into the RT-VRAM and may corrupt payload loaded into the RT-VRAM by
the BootROM. Set the early SPL stack at fixed location at the end of
System RAM instead, where it cannot interfere with the payload loaded
into RT-VRAM.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Patrick Delaunay <patrick.delaunay@foss.st.com> says:
Fix and add documentation/tests for selection by string for known
partition type GUID introduced by bcb41dcaefac ("uuid: add
selection by string for known partition type GUID"):
- split list_guid for short name (used also for partiton
description with type parameter) and full name to display
information
- as the function are uuid_str_to_bin() / uuid_guid_get_str()
are no more under CONFIG_PARTITION_TYPE_GUID, since commit
31ce367cd100 ("lib/uuid.c: change prototype of uuid_guid_get_str()")
and commit c1528f324c60 ("lib: compile uuid_guid_get_str if
CONFIG_LIB_UUID=y") move the content of array under EFI_PARTITION
and linker will remove it is not used it (in SPL)
- Add and fix documentation for gpt command
- Add test test_gpt_write_part_type to test "type=" parameters
This first patch solves an issue for the "system" shortcut for ESP,
removed by commit d54V3 version solve issue for "ESP" support when
CONFIG_CMD_EFIDEBUG and CONFIG_EFI is not activated
for example for test with qemu-arm-sbsa defconfige1004b8b1 ("lib/uuid.c: use unique name
for PARTITION_SYSTEM_GUID") but used in 2 location (at least):
1- board/samsung/e850-96/e850-96.env:10:
partitions=name=esp,start=512K,size=128M,bootable,type=system;
partitions+=name=rootfs,size=-,bootable,type=linux
2- arch/arm/mach-stm32mp/cmd_stm32prog/stm32prog.c:1151
case PART_ESP:
/* EFI System Partition */
459219 type_str = "system"
....
offset += snprintf(buf + offset,
buflen - offset,
",type=%s", type_str);
CI: https://source.denx.de/u-boot/custodians/u-boot-stm/-/pipelines/26699
Link: https://lore.kernel.org/r/20250616112749.17311-1-patrick.delaunay@foss.st.com
Add sandbox test on gpt command with partition type for known type.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Add information on type partition, copied from README.gpt.
I also correct issue for gpt_parts variable and add example of
"gpt read" usage.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Add support of optional shortname for parameter 'type' of gpt
command (limited by UUID_STR_LEN) and a separate 'description'
for UID format "%pUs" used in 'part list' output.
When 'description' is absent in list_guid[], the optional
shortname is used as fallback.
Many partition types for EFI have no shortcut yet, but only
description as they are only used to display information.
This patch also restores the "system" as short name for EFI
System Partition (ESP).
Fixes: d54e1004b8b1 ("lib/uuid.c: use unique name for PARTITION_SYSTEM_GUID")
Tested-by: Patrice Chotard <patrice.chotard@foss.st.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
- based on sdram_rk322x.c
- phy is different
- replaced magic numbers
- removed DM/DT to fit in SRAM
- supports LPDDR2/DDR3 (all types as per DS)
- addtionally set trfc/trefi based on detected cap
- use sdram_common
- set axi priorities as vendor blob does
- provide serveral dram timings to make it "future proof"
(hopefully nobody will have fiddle to make it still fit in SRAM
when adding a new board)
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Those enums are not used anywhere for RK3128 and they will cause problems
if both RK3128's and RK3288's grf header is included in a driver, because
RK3288's grf header defines enums with members which have exactly the same
name as these.
If they are required, they can be defined inline as it is done for most
other Rockchip SoCs.
Signed-off-by: Alex Bee <knaerzche@gmail.com>
Adjust legoev3_defconfig to reduce the binary output size.
As u-boot has bloated a bit over the years, the legoev3_defconfig can no
longer build something that fits in the 256kB size limit of the EV3.
This drops a few unused features, but the real difference-makers are
enabling thumb instructions and using link time optimization to reduce
the size.
This reduced u-boot.bin from 279,920 to 198,416 bytes on my local
machine with arm-none-eabi-gcc (15:13.2.rel1-2) 13.2.1 20231009.
HAS_BOARD_SIZE_LIMIT is also added to catch any future regressions.
Signed-off-by: David Lechner <david@lechnology.com>
The mx6sabresd U-Boot proper binary size has grown beyond the
CONFIG_BOARD_SIZE_LIMIT.
Reduce its size by removing the CONFIG_MULTI_DTB_FIT, BOOTM_PLAN9 and
BOOTM_RTEMS options.
According to doc/README.multi-dtb-fit:
"Usually the DTB is selected by the SPL and passed down to U-Boot. But some
platforms don't use the SPL. In this case MULTI_DTB_FIT can used to provide
U-Boot with a choice of DTBs"
mx6sabresd uses SPL, so MULTI_DTB_FIT can be safely dropped as the DTB
selection in SPL is done by board_fit_config_name_match().
Tested boot on the imx6dl and imx6q variants.
Signed-off-by: Fabio Estevam <festevam@gmail.com>
This option is very useful for A/B boot setups with read-only
filesystems: Letting U-Boot fill in the rootfs (and possibly related
parameters) allows keeping all boot parameters except the actual slot
selection in the extlinux.conf file, where they can be updated easily.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Cc: Tom Rini <trini@konsulko.com>
Since commit 51ec8db23280 ("pylibfdt: correct license information"), the
License classifiers are gone so I assume setuptools now extract the
license from the license argument to setuptools.setup() function.
It's always been incorrect as far as I could tell, so let's fix this
with the appropriate info from the SPDX License identifier at the top of
the file. It was missing GPL-2.0-or-later and we disambiguate by using
BSD-2-Clause instead of simply BSD.
Fixes: 6b08fb5cc44f ("fdt: Move to setuptools")
Signed-off-by: Quentin Schulz <quentin.schulz@cherry.de>
Heinrich Schuchardt <heinrich.schuchardt@canonical.com> says:
spl_fit_get_image_name() used to lack a detection of malformed image name
properties in FIT images. The change in commit 3704b888a4ca ("common/spl:
fix potential out of buffer access in spl_fit_get_image_name function")
tried to fix this but led to function spl_fit_get_image_name() no longer
detecting if a property at index > 1 does not exist.
This patch is reverted.
An explicit check for malformed image name properties is introduced.
Link: https://lore.kernel.org/u-boot/38f5d078-3328-4bdb-9c95-4fb5fe89ddc2@gmx.de/T/#u
Link: https://lore.kernel.org/r/20250624153431.46986-1-heinrich.schuchardt@canonical.com
The change in commit 3704b888a4ca ("common/spl: fix potential out of buffer
access in spl_fit_get_image_name function") led to function
spl_fit_get_image_name() no longer detecting if a property does not exist
at a non-zero buffer.
Link: https://lore.kernel.org/u-boot/38f5d078-3328-4bdb-9c95-4fb5fe89ddc2@gmx.de/T/#m59f3a23e675daa992c28d12236de71cae2ca2bb9
Fixes: 3704b888a4ca ("common/spl: fix potential out of buffer access in spl_fit_get_image_name function")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Tested-by: E Shattow <e@freeshell.de>
Observing the crash when we add the i2c-arbitrator node in the device
tree as per the DT bindings. The issue is with the child node of
i2c-arbitrator@72 i.e., i2c@f1950000->i2c-arbitrator@72->i2c-arb, as the
arbitrator uses the uclass of mux(UCLASS_I2C_MUX) and the mux uclass driver
checks for the "reg" property using the i2c_mux_child_post_bind() function,
if it won't find the "reg" property it will return -EINVAL which is leading
to the crash.
So, add the logic to check whether the child node has the "reg" property,
if the "reg" property exists then read the "reg" and update the channel.
https://www.kernel.org/doc/Documentation/devicetree/bindings/i2c/i2c-arb.txt
Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
The value passed to dcd_skip of build_container() should be obtained by
parsing .cfg file, and should not be fixed to false. For i.MX8QXP, dcd
data needs to be skipped, in which case dcd_skip should be true.
Fixes: 5f28a6599f01("tools: imx8image: add i.MX95 support")
Signed-off-by: Alice Guo <alice.guo@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
Reported-by: Enric Balletbo i Serra <eballetb@redhat.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Commit 6fe50e395080 ("arm: asm/system.h: mrc and mcr need .arm if
__thumb2__ is not set") is not a proper fix for the LTO link error
mentioned in its description. It causes 32-bit arm instructions to be
mixed with thumb instructions, which the Kirkwood SoCs do not support.
For example, board_init_r() is mostly generated in Thumb-1 mode as
expected since the build flags contain -mthumb -mthumb-interwork. The
MCR instruction corresponding to writefr_extra_feature_reg() is also
correcly emitted as a 32-bit ARM instruction (it cannot be encoded in
Thumb-1 anyways). The problem is, the compiler inlines the MCR without
generating the BX or BLX instruction which are needed to transition
between the ARM and the Thumb-1 states. From the objdump output:
006186a0 <board_init_r>:
board_init_r():
/home/jerome/work/u-boot/common/board_r.c:799
6186a0: b5f0 push {r4, r5, r6, r7, lr}
6186a2: b0ab sub sp, #172 @ 0xac
get_gd():
/home/jerome/work/u-boot/./arch/arm/include/asm/global_data.h:127
6186a4: 464a mov r2, r9
...
/home/jerome/work/u-boot/arch/arm/mach-kirkwood/cpu.c:242
619aae: 9b15 ldr r3, [sp, #84] @ 0x54
writefr_extra_feature_reg():
/home/jerome/work/u-boot/./arch/arm/include/asm/arch/cpu.h:100
619ab0: ee2f3f11 mcr 15, 1, r3, cr15, cr1, {0}
^^^^^^^^
32-bit ARM instruction
Further investigation is needed to understand how to fix the issue so
that the code size is minimal for all boards. In the mean time, this
fix disables LTO for the two problematic files (common/board_f.c and
common/board_r.c). This makes the Kirkwood-based boards bootable again.
The binary size is increased by 1048 bytes which is perfectly
acceptable.
Fixes: 6fe50e395080 ("arm: asm/system.h: mrc and mcr need .arm if __thumb2__ is not set")
Reported-by: Tony Dinh <mibodhi@gmail.com>
Tested-by: Tony Dinh <mibodhi@gmail.com>
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Tom Rini <trini@konsulko.com>
Shiji Yang <yangshiji66@outlook.com> says:
This patchset fixes some compilation errors that I caught in version
v2025.07-rc4 and branch next. If they are acceptable, please apply
them to the master branch. If anyone has a better way to fix these
issues, it's fine to ignore this patchset.
Link: https://lore.kernel.org/r/OSBPR01MB16702ED24460D23A7ED63440BC7DA@OSBPR01MB1670.jpnprd01.prod.outlook.com
The function definition of fit_check_sign() is guarded by
"#ifdef CONFIG_FIT_SIGNATURE" in "tools/image-host.c". If we try
to build it without CONFIG_FIT_SIGNATURE, we will get an error:
/usr/bin/ld: tools/fit_check_sign.o: in function `main':
fit_check_sign.c:(.text.startup+0x165): undefined reference to `fit_check_sign'
collect2: error: ld returned 1 exit status
Fixes: 9c79c8fe70da ("tools/fit_check_sign: make key optional")
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
The symbol TOOLS_IMAGE_PRE_LOAD doesn't depend on TOOLS_LIBCRYPTO.
If we choose to build tools without openssl, rsa_verify_openssl()
will attempt to call the unavailable openssl library functions.
Fixes: 942c8c8e6697 ("rsa: Add rsa_verify_openssl() to use openssl for host builds")
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Device bus address might not be valid for direct access when the bus
address and CPU address are not the same. Use dev_bus_to_phys() to
translate bus address back to CPU address.
Fixes: 3d98b8c504e15 ("net: designware: Invalidate RX buffer cache before freeing the DMA descriptor")
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
As stated in 5e847f7729b3 ("efi_loader: call efi_init_early() earlier"):
efi_init_early() creates an event hook for block device probing.
It has to be called before any block device is probed.
Indeed, efi_bl_init() registers EVT_DM_POST_PROBE event, which calls
efi_disk_probe() whenever any block device is probed. And to make that
hook work, the initialization of all block devices was put after
efi_init_early() in initcall_run_r():
INITCALL(efi_init_early);
INITCALL(initr_nand);
INITCALL(initr_onenand);
INITCALL(initr_mmc);
Because LDFW firmware is being read from MMC, attempt to load LDFW in
board_init() causes MMC driver to be probed. And because board_init() is
executed before efi_init_early(), the hook mentioned above won't work
for MMC devices anymore. So EFI disk objects won't be created, which in
turn makes the EFI subsystem non-functional, showing next symptoms:
- 'efidebug dh' output is empty
- attempt to add boot devices in 'eficonfig' shows this message:
"No block device found!"
- 'bootefi selftest $fdtcontroladdr' shows this warning:
"Cannot persist EFI variables without system partition"
- booting GRUB with 'bootefi' runs minimal GRUB shell which doesn't
see any block devices as well, probably because EFI vars weren't
passed
Load LDFW in board_late_init() instead, as it's called after
efi_init_early(). This fixes the described problem and makes it possible
to run EFI apps like GRUB correctly, add entries in 'eficonfig', and
makes 'efivar --list' command in Linux rootfs actually show EFI
variables.
The only user of LDFW at the moment is the TRNG driver, and it's probed
later, only when it's going to be used (e.g. on "rng" command). So it's
fine to load LDFW in board_late_init(). Now the corresponding call order
will look like this:
efi_init_early()
initr_mmc()
mmc_probe()
EVT_DM_POST_PROBE -> efi_disk_probe()
board_late_init()
load_ldfw() -> fs_read(), blk_dread()
exynos_trng_probe()
Fixes: ccfd8de541a8 ("board: samsung: e850-96: Report LDFW loading failures")
Fixes: f04e58cc9788 ("board: samsung: e850-96: Load LDFW firmware on board init")
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
This fix a possible NULL pointer dereference.
There is also a risk of memory leaking within the same portion of code.
The leak will happen if loaded image is bad or damaged. In this case
u-boot-spl will try booting from the other available media. Unfortunately
resources allocated for previous boot media will NOT be freed.
We can't fix that issue as the memory allocation mechanism used here
is unknown. It can be different kinds of malloc() or something else.
To somewhat reduce memory consumption, one can try to reuse previously
allocated memory as it's done in board_spl_fit_buffer_addr() from
test/image/spl_load.c.
The corresponding comment was put to the code as well.
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Reviewed-by: Anshul Dalal <anshuld@ti.com>
load_simple_fit() returns -EPERM for the images with broken signatures.
Unfortunately this may conflict with image loaging selection on the base
of boot phase. See commit 873112db9ce68c38984ff25808dde726f8dd5573
("spl: Support selecting images based on phase in simple FIT").
Thus loading of
configurations {
uboot {
description = "u-boot";
firmware = "atf";
loadables = "atf", "tee", "uboot";
};
};
with damaged "tee" image may finish without errors. This may results in
board bricking.
This patch fixes commit 873112db9ce68c38984ff25808dde726f8dd5573
("spl: Support selecting images based on phase in simple FIT")
by replacing EPERM with EBADSLT places where it should be done.
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
The current code have two issues:
1) ineffective NULL pointer check
str = strchr(str, '\0') + 1
if (!str || ...
The str here will never be NULL (because we add 1 to result of strchr())
2) strchr() may go out of the buffer for the special forms of name variable.
It's better use memchr() function here.
According to the code the property is a sequence of C-string like
shown below:
'h', 'e', 'l', 'l', 'o', '\0', 'w', 'o', 'r', 'l', 'd', '\0', '!', '\0'
index is the string number we are interested, so
index = 0 => "hello",
index = 1 => "world",
index = 2 => "!"
The issue will arrise if last string for some reason have no terminating
'\0' character. This can happen for damaged or specially crafted dtb.
Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Reviewed-by: Tom Rini <trini@konsulko.com>
The CONFIG_CUSTOM_SYS_INIT_SP_ADDR=0xeb300000 does not make it into
board .config on either R-Car V4H White Hawk or Sparrow Hawk, remove
this configuration option. The early SPL stack is however pointing
into the RT-VRAM and may corrupt payload loaded into the RT-VRAM by
the BootROM. Set the early SPL stack at fixed location at the end of
System RAM instead, where it cannot interfere with the payload loaded
into RT-VRAM.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Patrick Delaunay <patrick.delaunay@foss.st.com> says:
Fix and add documentation/tests for selection by string for known
partition type GUID introduced by bcb41dcaefac ("uuid: add
selection by string for known partition type GUID"):
- split list_guid for short name (used also for partiton
description with type parameter) and full name to display
information
- as the function are uuid_str_to_bin() / uuid_guid_get_str()
are no more under CONFIG_PARTITION_TYPE_GUID, since commit
31ce367cd100 ("lib/uuid.c: change prototype of uuid_guid_get_str()")
and commit c1528f324c60 ("lib: compile uuid_guid_get_str if
CONFIG_LIB_UUID=y") move the content of array under EFI_PARTITION
and linker will remove it is not used it (in SPL)
- Add and fix documentation for gpt command
- Add test test_gpt_write_part_type to test "type=" parameters
This first patch solves an issue for the "system" shortcut for ESP,
removed by commit d54V3 version solve issue for "ESP" support when
CONFIG_CMD_EFIDEBUG and CONFIG_EFI is not activated
for example for test with qemu-arm-sbsa defconfige1004b8b1 ("lib/uuid.c: use unique name
for PARTITION_SYSTEM_GUID") but used in 2 location (at least):
1- board/samsung/e850-96/e850-96.env:10:
partitions=name=esp,start=512K,size=128M,bootable,type=system;
partitions+=name=rootfs,size=-,bootable,type=linux
2- arch/arm/mach-stm32mp/cmd_stm32prog/stm32prog.c:1151
case PART_ESP:
/* EFI System Partition */
459219 type_str = "system"
....
offset += snprintf(buf + offset,
buflen - offset,
",type=%s", type_str);
CI: https://source.denx.de/u-boot/custodians/u-boot-stm/-/pipelines/26699
Link: https://lore.kernel.org/r/20250616112749.17311-1-patrick.delaunay@foss.st.com
Add support of optional shortname for parameter 'type' of gpt
command (limited by UUID_STR_LEN) and a separate 'description'
for UID format "%pUs" used in 'part list' output.
When 'description' is absent in list_guid[], the optional
shortname is used as fallback.
Many partition types for EFI have no shortcut yet, but only
description as they are only used to display information.
This patch also restores the "system" as short name for EFI
System Partition (ESP).
Fixes: d54e1004b8b1 ("lib/uuid.c: use unique name for PARTITION_SYSTEM_GUID")
Tested-by: Patrice Chotard <patrice.chotard@foss.st.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>