Fix-ups for the BCM root complex when it is located behind an AXI
bridge and clocked with 54MHz. Some are from kernel commit
377bced88c326, some where picked by Oleksii off a now-stale older
branch. All reworked for the simpler setup code in U-Boot.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Co-authored-by: Oleksii Moisieiev <oleksii_moisieiev@epam.com>
Tested-by: Pedro Falcato <pfalcato@suse.de>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Rework the setup of inbound PCIe windows: use the convenience functions
from Linux kernel commit ae6476c6de187 to calculate the BAR offsets and
factor out the setup code into a separate function.
The Linux kernel first allocates and populates an array of inbound_win[]
and sets the BARs from it later, while U-Boot does it all on the fly,
in one go, so the code is not 1:1 comparable.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Tested-by: Pedro Falcato <pfalcato@suse.de>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
A driver for Broadcom rescal reset controllers ported from
linux/drivers/reset/reset-brcmstb-rescal.c to U-Boot.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Co-authored-by: Oleksii Moisieiev <oleksii_moisieiev@epam.com>
Tested-by: Pedro Falcato <pfalcato@suse.de>
A driver for Broadcom reset controllers ported from
linux/drivers/reset/reset-brcmstb.c to U-Boot.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Co-authored-by: Oleksii Moisieiev <oleksii_moisieiev@epam.com>
Tested-by: Pedro Falcato <pfalcato@suse.de>
The Linux kernel driver already had support for multiple hardware
variants when the bcm2712 was added (see e.g. linux commit
10dbedad3c818 which is the last in a longer set of changes). This
patch brings in this required infrastructure and adds a
differentiation between 2711 and 2712 register layouts on top.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Co-authored-by: Oleksii Moisieiev <oleksii_moisieiev@epam.com>
Tested-by: Pedro Falcato <pfalcato@suse.de>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
The linux kernel assigns a new domain for every Root Complex where bus
numbering starts from 0 for each domain. U-Boot does not have domains
and uses a flattened bus numbering scheme instead. This means that any
device or bridge on the second enumerated RC will receive a bus number
equal to the last assigned one +1. This bus number contributes to the
address written into the index register, which will select the
configuration space to be read. Compensate for this contribution by
subtracting the base bus number.
Signed-off-by: Andrea della Porta <andrea.porta@suse.com>
Signed-off-by: Torsten Duwe <duwe@suse.de>
Tested-by: Pedro Falcato <pfalcato@suse.de>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
Add a mapping region for the PCIe bus address spaces to the BCM2712
memory controller setup. Generously merging the PCIe address spaces
works sufficiently well for a boot loader.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Co-authored-by: Oleksii Moisieiev <oleksii_moisieiev@epam.com>
Tested-by: Pedro Falcato <pfalcato@suse.de>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
ulong is 64 bits on 64-bit platforms. Hence, simple_xtoa can
produce up to 16 hex characters + NULL byte. The initrd_filesize
buffer is only 10 bytes which can cause a buffer overflow on
every PXE boot that loads an initrd on an address greater than
4GB.
Increase buffer size to 17 bytes to hold the maximum hex
representation of a 64-bit address.
Signed-off-by: Francois Berder <fberder@outlook.fr>
Reviewed-by: Jerome Forissier <jerome.forissier@arm.com>
dhcp_message_type() scans DHCP options looking for a 0xff
end-of-options marker with no check that the scan pointer stays
within the received packet. A server can send a crafted OFFER with
no 0xff terminator and large option length fields, advancing the
pointer past bp_vend[312] into adjacent heap memory.
This is the same class of bug as CVE-2024-42040, which fixed the
related bootp_process_vendor() call site. Fix it by adding an end
parameter to dhcp_message_type() and checking that popt is lower
than end.
Signed-off-by: Francois Berder <fberder@outlook.fr>
Reviewed-by: Jerome Forissier <jerome.forissier@arm.com>
dhcp6_parse_options() verifies that an option's declared data fits
within the packet, but does not check that option_len is large
enough for the fixed-size read each case performs. A malicious
DHCP server can send an ADVERTISE with a zero-length IA_NA,
STATUS_CODE, SOL_MAX_RT, or BOOTFILE_PARAM option, causing the
parser to read 2-4 bytes past the option's declared data.
Check option_len value before each dereference of option_ptr.
Signed-off-by: Francois Berder <fberder@outlook.fr>
The net_boot_file_name is a 1024 byte buffer.
However, based on DHCPv6 RFC, bootfile-url length is
specified by option_len, a 16-bit unsigned integer
(valid range: 0-65535).
Hence, one needs to make sure that option_len is less
than the size of net_boot_file_name array before copying
bootfile-url to net_boot_file_name.
Signed-off-by: Francois Berder <fberder@outlook.fr>
Reviewed-by: Jerome Forissier <jerome.forissier@arm.com>
Currently, the sntp_handler uses data in the UDP packet
regardless of the actual packet size. A OOB read can occur
if the packet is too small.
Fix it by checking the packet length before extracting
seconds from a SNTP packet.
Signed-off-by: Francois Berder <fberder@outlook.fr>
Reviewed-by: Jerome Forissier <jerome.forissier@arm.com>
portname consists of dev->parent->name ("icssg0-eth",
"icssg1-eth", or "ethernet") and dev->name is the port node
name ("port@0" or "port@1"). Every board DTS in the repository
produces a string that overflows the buffer:
"icssg1-eth-port@0" 17 chars + NUL = 18 bytes (AM642 EVM, IoT2050)
"ethernet-port@0" 15 chars + NUL = 16 bytes (SR-SOM, phyboard)
This commits increases portname to 64 bytes and replaces sprintf
by snprintf so that any future DT node name cannot overflow it
regardless of length.
Signed-off-by: Francois Berder <fberder@outlook.fr>
Reviewed-by: Jerome Forissier <jerome.forissier@arm.com>
Correct the rule in .gitignore to skip u_boot_logo.bmp.S artifact which
is generated for any board with CONFIG_VIDEO_LOGO enabled.
Also, correct the stale u_boot_logo name in CLEAN_FILES in top-level
Makefile.
Signed-off-by: Denis Mukhin <dmukhin@ford.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Remove an extra leading dash from the KBUILD_CFLAGS assignment
under the CONFIG_CC_OPTIMIZE_FOR_DEBUG conditional block. The
extra dash breaks the build when CONFIG_CC_OPTIMIZE_FOR_DEBUG
is enabled.
Fixes: 56ae3c2a44 ("Makefile: repair CONFIG_CC_OPTIMIZE_FOR_DEBUG support")
Signed-off-by: Giovanni Santini <santigio2003@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
If the filesystem creation attempted by the FsHelper class fails, the
invocation of the cleanup function will cause a TypeError exception,
because the path of the filesystem itself, fed to os.remove(), will be
None. This will lead to a test failure even in case a skip is instead
wanted.
Such an exception will lead to a backtrace like this:
test/py/tests/test_fs/conftest.py:269: in fs_obj_basic
fsh.mk_fs()
test/py/tests/fs_helper.py:70: in mk_fs
self.fs_img = mk_fs(self.config, self.fs_type, self.size_mb << 20,
test/py/tests/fs_helper.py:246: in mk_fs
check_call(f'mkfs.{fs_lnxtype} {mkfs_opt} {fs_img}', shell=True,
/usr/lib64/python3.14/subprocess.py:420: in check_call
raise CalledProcessError(retcode, cmd)
E subprocess.CalledProcessError: Command '<...>' returned non-zero exit status 1.
During handling of the above exception, another exception occurred:
test/py/tests/test_fs/conftest.py:272: in fs_obj_basic
pytest.skip('Setup failed for filesystem: ' + fs_type + '. {}'.format(err))
E Skipped: Setup failed for filesystem: ext4. Command '<...>' returned non-zero exit status 1.
During handling of the above exception, another exception occurred:
test/py/tests/test_fs/conftest.py:277: in fs_obj_basic
fsh.cleanup()
test/py/tests/fs_helper.py:91: in cleanup
os.remove(self.fs_img)
E TypeError: remove: path should be string, bytes or os.PathLike, not NoneType
Fix this by checking if the variable containing the filesystem path is
valid before attempting to call os.remove() on it.
Fixes: 3691b1e4ce ("test: Convert fs_helper to use a class")
Signed-off-by: Francesco Valla <francesco@valla.it>
Add armada-375-xhci to the compatible list in XHCI MVEBU driver.
Tested with WD MyCloud Gen2 NAS.
Signed-off-by: Tony Dinh <mibodhi@gmail.com>
Reviewed-by: Marek Vasut <marek.vasut+usb@mailbox.org>
Add support for the Geist board based on the Renesas R8A779MD (M3Le) SoC, a
register-compatible variant of the R8A77965 (M3N) with reduced peripherals.
The Geist board design references the Renesas Salvator-X/XS boards, adapting
their configuration for the R8A779MD SoC.
The board will be switched to OF_UPSTREAM once the DTs land in upstream.
Signed-off-by: Huy Bui <huy.bui.pz@bp.renesas.com>
Signed-off-by: Nguyen Tran <nguyen.tran.pz@bp.renesas.com>
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
With commit 17ceb774a1 ("rpi_arm64: Enable MBEDTLS/LWIP/WGET and
WGET_HTTPS"), we can drop the tests for switching from the legacy stack
to lwIP.
Signed-off-by: Tom Rini <trini@konsulko.com>
sdhci_brcmstb_init_2712() reads host->mmc->host_caps to decide whether
to force card-detect for a non-removable eMMC, or to route the CD signal
for a removable SD card. At the time this function runs from
sdhci_bcmstb_probe(), however, host->mmc->host_caps is still zero, that
field is only populated later by the MMC uclass, after the driver's
probe returns. mmc_of_parse() has already filled plat->cfg.host_caps
from the device tree by this point, so check that field instead.
Without the fix, every BCM2712 SDHCI instance takes the else branch and
writes SDIO_CFG_SD_PIN_SEL = SDIO_CFG_SD_PIN_SEL_CARD (0x02), including
the non-removable eMMC on boards such as CM5 on Home Assistant Yellow.
The SDIO_CFG block lies outside the SDHCI core's reset scope, so this
value persists across SDHCI_RESET_ALL into the next stage. On the
BCM2712, having SD_PIN_SEL set to "SD" when the Linux kernel performs
its first set_power(MMC_POWER_UP) write racily prevents the SDHCI
POWER_ON bit from latching (see [1] for the whole backstory) - the
voltage bits stick but POWER_ON drops - which wedges the first CMD0 the
full 10 s software timeout. On Home Assistant Yellow this manifested as
a ~20 s eMMC probe delay on roughly one in two Linux boots when U-Boot
was the previous stage. Booting directly from the Pi firmware (no U-Boot
in between) left SD_PIN_SEL at its default and did not exhibit the race.
Reading plat->cfg.host_caps lets init_2712 see the "non-removable"
property and take the correct branch, leaving SD_PIN_SEL untouched for
the eMMC.
[1] https://github.com/home-assistant/operating-system/pull/3700#issuecomment-4430229511
Fixes: 10127cdbab ("mmc: bcmstb: Add support for bcm2712 SD controller")
Signed-off-by: Jan Čermák <sairon@sairon.cz>
Reviewed-by: Ivan T. Ivanov <iivanov@suse.de>
CI: https://source.denx.de/u-boot/custodians/u-boot-stm/-/pipelines/30256
- Add support dynamic A/B bank bootup for STM32MP15
- Increase SYS_MALLOC_F_LEN for stm32mp15_defconfig to fix boot with optee-4.10.0
- Enable Arm SMC watchdog for STM32MP1
- Update part number for TM32MP251/3 SoC's family
For this configuration, the watchdog (iwdg1) is secured and managed by
OP-TEE. Add an watchdog node with arm,smc-wdt compatible, and disable
iwdg2 node which is then no more used.
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
No watchdog was enabled for STM32MP13 platform. Add the required flags to
support it. As done for STM32MP15 (in SCMI config) and STM32MP2x, we use
the Arm SMC watchdog. The required nodes were already present in Linux
imported DT files (stm32mp13.dtsi & stm32mp135f-dk.dts).
To enable this SMC watchdog on other platforms based on STM32MP13, check
that both the following flags are enabled in the dedicated config file:
CONFIG_WDT=y
CONFIG_WDT_ARM_SMC=y
And that there is a node in Linux board DT that enables the feature, as
it is done in stm32mp135f-dk.dts:
&arm_wdt {
timeout-sec = <32>;
status = "okay";
};
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Enable the arm watchdog over SMC driver. This allows using a secure
watchdog, based on IWDG1 peripheral and managed by OP-TEE.
The driver will be probed if a watchdog node with "arm,smc-wdt"
compatible is enabled.
Signed-off-by: Lionel Debieve <lionel.debieve@foss.st.com>
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Factorize FWU multi-bank support code common to STM32MP1 and
STM32MP2 platforms into a dedicated shared source file.
No functional change intended.
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Following commit 4300f9f4c5 ("board: st: stm32mp25: support dynamic
A/B bank bootup"), this patch enables automatic detection of the active
A/B bank on STM32MP15 platforms by retrieving partition GUIDs from FWU
metadata.
This ensures the system correctly identifies the bootable partitions
even in multi-bank scenarios, falling back to a standard bootable flag
scan if the UUIDs are missing.
To enable A/B bank bootup on STM32MP15 boards, add the following Kconfig
options to the stm32mp15[_basic]_defconfig:
CONFIG_FWU_MULTI_BANK_UPDATE=y
CONFIG_FWU_MDATA=y
CONFIG_FWU_NUM_BANKS=2
CONFIG_FWU_NUM_IMAGES_PER_BANK=3
CONFIG_CMD_FWU_METADATA=y
CONFIG_FWU_MDATA_V2=y
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
The bcm2835 SDHCI driver sets up the MMC host configuration via
sdhci_setup_cfg(), but does not parse generic MMC device tree
properties.
As a result, properties such as bus-width are ignored. On Raspberry Pi
Compute Module 4, the eMMC node describes an 8-bit bus, but U-Boot
initialized the device as 4-bit.
Call mmc_of_parse() before sdhci_setup_cfg() so that generic MMC
properties are folded into the host configuration before the MMC core
selects the bus width.
Before this change, mmc info reported:
Bus Speed: 52000000
Bus Width: 4-bit
After this change, mmc info reports:
Bus Speed: 52000000
Bus Width: 8-bit
Tested on Raspberry Pi Compute Module 4 with onboard eMMC.
Signed-off-by: Liel Harel <liel.harel@gmail.com>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
Tested-by: Peter Robinson <pbrobinson@gmail.com> # on the CM4 as well
Enable LWIP and HTTPS on the Raspberry Pi arm64 platform to be able to
use it in the boot process.
Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Matthias Brugger <mbrugger@suse.com>
The 'brcm,bcm2712-hdmi0' compatible string is used on RPi5.
There appears to be no change that impacts early boot output
on the display controller so add the RPi5 compatible string.
Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Matthias Brugger <mbrugger@suse.com>
A file like rm-cfg.yaml accidentally left in the source tree root
shadows the board-specific copy. binman builds the wrong YAML, the
resulting rm-cfg.bin may match a different SoC, and we end up with
the following error:
k3_system_controller sysctrler: k3_sysctrler_start:
Boot Notification response failed. ret = -110
Move the board directory ahead of the srctree root so that the
most-specific match wins.
Signed-off-by: Wadim Egorov <w.egorov@phytec.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Our official domain is now u-boot-project.org, so update all in-tree
references to use the correct domain.
Reviewed-by: Tony Dinh <mibodhi@gmail.com>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Currently the sequence to enable caches for the A53/A72 core on K3
devices looks as follows:
1. Map entire DDR banks
2. Setup page tables (done by mmu_setup)
3. Enable MMU
4. Unmap reserved-memory regions
5. Enable caches
However there is a brief period of execution between #3 and #4 where the
core can issue speculative accesses to the entire DDR space (including
the reserved-memory regions) despite the caches being disabled.
A firewall exception is triggered whenever such speculative access is
made to secure DDR region of TFA or OP-TEE. This patch fixes the issue
by re-ordering the sequence as follows:
1. Map entire DDR banks
2. Setup page tables
3. Unmap reserved-memory regions
4. Enable MMU
5. Enable caches
Fixes: f1c694b8fd ("mach-k3: map all banks using mem_map_from_dram_banks")
Reported-by: Suhaas Joshi <s-joshi@ti.com>
Signed-off-by: Anshul Dalal <anshuld@ti.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Currently mmu_setup for ARMv8 performs two functions, first it sets up
the page tables based the memory map provided by the board and then it
enables the MMU.
However for some platforms runtime fixes to the generated page tables
are required before the MMU can be enabled, such as K3 family of SoCs.
Therefore this patch moves the enablement of the MMU out of mmu_setup
and to a standalone mmu_enable function to give more granular control to
the platforms.
Note that no functional changes are intended from this patch.
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: Anshul Dalal <anshuld@ti.com>
babae80169 removed bootm_size from ti_common.env to allow K3 boards
to process images larger than 256MB, but preserved it in
ti_armv7_keystone2.env for ARMv7 Keystone2 boards. AM57xx (also ARMv7)
was not covered by that preservation.
Without bootm_size, env_get_bootm_size() falls back to gd->ram_size,
causing initrd_high to be computed as the top of all RAM. On ARM32
boards with more RAM than the DMA zone (e.g. AM572x IDK with 2GiB),
this places the ramdisk above 0xafe00000 (HighMem), which is not
directly accessible by the kernel after MMU setup, causing a silent
crash.
With bootm_size=0x10000000, initrd_high is constrained to
0x80000000 + 0x10000000 = 0x90000000, keeping the ramdisk in the
DMA zone and allowing the kernel to access it correctly.
Fixes: babae80169 ("include: env: ti_common: remove bootm_size")
Reviewed-by: Neha Malcom Francis <n-francis@ti.com>
Signed-off-by: Moteen Shah <m-shah@ti.com>
When fpga load is called with a misaligned buffer address, the
versal_align_dma_buffer() function shifts the pointer forward to the
next aligned boundary and uses memcpy() to copy the data. Since the
destination is ahead of the source and the regions overlap, memcpy()
produces undefined behavior; in practice U-Boot's generic memcpy()
copies forward, repeating the first ARCH_DMA_MINALIGN-aligned chunk
throughout the buffer.
Replace memcpy() with memmove() which correctly handles overlapping
regions by copying backwards when the destination is ahead of the
source.
Fixes: 26e054c943 ("arm64: versal: fpga: Add PL bit stream load support")
Signed-off-by: Pranav Tilak <pranav.vinaytilak@amd.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20260507113359.3665220-1-pranav.vinaytilak@amd.com
