| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
| |
After upstreaming our changes, the file in PTXdist 2024.02.0 is now the
same as ours, so we can remove it here.
Suggested-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
|
|
|
|
|
|
|
| |
On v7a with NEON, we want to have the NEON optimized SHA1
implementation.
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This saves us 124K, which we dearly need, because otherwise barebox
would overlap the environment partition. The environment partition
is reserved in the genimage config, but it's used due to the description
in the barebox device tree. We may want to change both, but for now, buy
us some time by switching from LZO to XZ.
Reported-by: Michael Olbrich <mol@pengutronix.de>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240318140633.1997386-1-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The latest barebox release incorporates the single patch we had on top
of v2024.01.0, so let's drop it by updating.
Config changes are those of olddefconfig with the exception of
CONFIG_CMD_PARTED.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-8-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
| |
Like done for v8a, also update the v7a kernel to the latest release.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-7-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
v2023.12.0 first broke i.MX8MQ support by mistake. This was fixed in
v2024.03.0, so update to this version.
The config changes are those of olddefconfig with the exception of
CONFIG_CMD_PARTED: This command allows us to partition block devices
from within the barebox shell.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-6-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
config changes are those of olddefconfig with the notable exception of
CONFIG_BLK_DEV_WRITE_MOUNTED: This is enabled by default as not to break
some existing userspace, but we shouldn't have any of that, so disable it.
This will cause the kernel to refuse writes to mounted block devices.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-5-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When Linux probed the PCI on the i.MX8M-EVK, the system hangs
completely. We've seen the same effect with other i.MX8M EVKs too,
where between EVK revisions, NXP changed how the PCI PHY is clocked
and because the DT only supported one variant, probing PCI on the
other variant led to system hang while probing the PCI address space.
If this turns out to be the issue on the i.MX8M-EVK as well, we should
probably try to identify what variant we run on and then fix up the DT
in the bootloader. For now though, just take the easy route and disable
PCI, which we don't use anyway.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-4-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This lets us git rid of a fair number of patches we have, but in return
requires us to import three patches:
- Fix link error for imx8mq-evk: The image got too big with addition
of DRAM retention support, so this is now disabled by default:
https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/27277
- Support for automatic console base detection on i.MX8MQ. This was
already in our patch stack, but wasn't submitted upstream, unlike
the support for i.MX8MM, i.MX8MN and i.MXP. This has been rectified:
https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/27278
- STM32MP1 support with SP_min as BL31 instead of OP-TEE is broken
since v2.10 was first released. This is due to PMU changes that
were apparently only tested on ARMv8. Fixed here:
https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/27162
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-3-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This lets us make use of the updated OP-TEE packages for the incoming
STM32MP13 support. Support in OP-TEE 3.20 is incomplete and will lead to
a crash on boot up:
I/TC: Platform stm32mp1: flavor 135F_DK - DT stm32mp135f-dk.dts
I/TC: Non-secure SYSRAM [0xddfff000 0xddffffff]
I/TC: DTB enables console (non-secure)
I/TC: Primary CPU switching to normal world boot
E/TC:0 tzc_it_handler:26 TZC permission failure
E/TC:0 dump_fail_filter:420 Permission violation on filter 0
E/TC:0 dump_fail_filter:425 Violation @0xfcff7fe0, non-secure privileged write, AXI ID 420
E/TC:0 Panic at core/arch/arm/plat-stm32mp1/plat_tzc400.c:30 <tzc_it_handler>
E/TC:0 TEE load address @ 0xde000000
E/TC:0 Call stack:
E/TC:0 0xde00316d
E/TC:0 0xde011b37
E/TC:0 0xde005245
E/TC:0 0xde011201
E/TC:0 0xde037155
E/TC:0 0xde000350
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240315211240.3016716-2-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The idea behind enabling the USB gadget by default was to allow
quick experimentation with only a USB connection by exposing both
fastboot and a ACM serial gadget.
However, Fastboot over UDP, isn't as useful:
- the user first has to find out the IP address assigned to the board
- getting a DHCP lease on every boot takes a bit of time
- The shell is unusable until the DHCP request succeeds or times out.
Ctrl+C can't be used as it's not called interactively
Therefore, disable Fastboot over UDP and leave only USB enabled by
default. User wishing to use Fastboot over UDP can just set
nv.fastboot.net.autostart=1
manually to restore the old behavior.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240306184640.628394-1-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
| |
Already enabled for v8a. Tested on Beaglebone Black and RIoTboard.
Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240306092530.1448784-4-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Backport PTXdist commit 0d360df876bc26f709cf (2024-02-02, Alexander
Dahl: "blspec-entry: Fix kernel entry for FIT image"), otherwise
barebox fails when trying to boot a FIT image like this:
ERROR: could not open /mnt/mmc0.1//boot//boot/linux.fit: No such file or directory
ERROR: blspec: Booting failed
Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240306092530.1448784-3-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
| |
The latest and greatest. Has the nice support for 'noload' sub-images
in FIT images conforming for spec.
Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240306092530.1448784-2-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Update to the new PTXdist version with default options. Note that
HOST_DTC, HOST_FLEX and HOST_M4 only vanish from the platformconfigs and
are still enabled in ptxconfig.
Remove our systemd patch, which is now included in systemd 255.2.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20240219121717.3507053-1-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
| |
With this entry the /dev/disk/by-usage/rootfs0 and /dev/disk/by-usage/rootfs1
device links are created, that are required to install RAUC updates on this
board.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240129093259.172908-2-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
| |
CONFIG_CRYPTO_CCM is no longer needed by any driver, so we can disable
CONFIG_CRYPTO_AES_ARM64_CE_CCM.
Inspired-by: Michael Olbrich <m.olbrich@pengutronix.de>
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
| |
In eb939a84ef844769bc26eb0eed95245d1f7160f2, memory compaction was
switched off. However, the help text claims it should be on, and the
Pengutronix internal 'reason' checker does also want to have it on.
Re-enable it.
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is the result of a review of the whole kernel config.
The options have been selected by the following rules:
In general, the kernel config is optimized to keep the kernel small. So
any features that are not used at all are disabled. This includes things
like extra scheduling governors that require explicit configuration to
be used.
All drivers that are not used by the boards configured in DistroKit are
now disabled.
If there was no other strong preference then the value from the
defconfig was used.
Signed-off-by: Michael Olbrich <m.olbrich@pengutronix.de>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://lore.pengutronix.de/20240126170432.740899-1-m.olbrich@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
| |
This allows updating the root filesystem via RAUC.
The magic value was selected by fair dice roll.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-8-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
| |
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-7-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The generated image can be used to boot the board via SD card.
The dev.mmc1.broken_cd barebox environment entry is required because
the board's card detect does not reliably detect an inserted USB-SD-Mux,
resulting in the board failing to boot.
This should not (negatively) affect the other v8a machines.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-6-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
Having a place to store state allows us to use RAUC on i.MX8M based
boards. The 1Mb size is quite large for a state partition, which
uses a compact representation, but using this size means we do not
have to change the partition alignment.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-5-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The barebox environment is configured in the board devicetrees to
be at 896k and 128k in size (e.g. the last 128k before the 1M mark).
This is for compatibility with the fdisk behaviour of placing the
start of the first partition at the 1M mark.
Not everyone knows this though, so it can easily happen that someone
increases the size of the barebox partition, making it overlap with
the environment "partition".
Make this less likely to happen by listing the barebox state in
the genimage config.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-4-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The 896K assigned to the barebox partition are getting ever closer to
overflowing (barebox-nxp-imx8mn-evk.img is as of now 883K).
Increasing the barebox partition also requires changes to the respective
devicetrees because the location of the environment is hardcoded to
be at 896K there.
Move the root-a partition further back to free up some space so we
can add a state partition now and eventually move the environment
partition as well and then increase the barebox partition size.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-3-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This helps us on the TQ MBa8MPxL i.MX8MP development board,
that uses a different UART for debug output than the EVKs we
currently support.
Other i.MX8M boards should not be affected by the
IMX_BOOT_UART_BASE=auto configuration option because the automatic
selection will choose the correct UART.
Signed-off-by: Leonard Göhrs <l.goehrs@pengutronix.de>
Link: https://lore.pengutronix.de/20240117133440.944800-2-l.goehrs@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There's ongoing work to have Linux generate a FIT image with all enabled
DTs[1]. Let's prepare for that by enabling FIT image support for the
ARMv8 boards.
[1]: https://lore.kernel.org/all/20231202035511.487946-1-sjg@chromium.org/
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
LGTMed-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20240112170529.1970537-6-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This lets us drop two patches that fixed issues with v2023.09.0.
This update also enables some options that are useful for the hardware
we support:
- on STM32MP1, support for the NAND controller is enabled
- on STM32MP1, we enable HWRNG and stack protector support
The latter could be enabled universally for all platforms that are no
size constrained, but it prints an ugly error message if there's no
HWRNG as using the fixed fallback canary is insecure. We'll just
postpone enabling stack protector on other platforms until we enable
HWRNG support there.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240112170529.1970537-5-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
There's not much new features relevant to our v8a platform in this
version bump, but there are a number of bug fixes, that may be relevant,
like improved support for HAB on i.MX8M and reworked regulator
initialization.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240112170529.1970537-4-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
| |
The RAVE SP is a board management controller connected over serial link.
We have no boards that have this controller built-in, so there's no need
to include its driver.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240112170529.1970537-3-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- We don't boot uImages anymore
- We can reinstate the default environment using saveenv -z
- We have a TFTP file system and don't need an extra tftp command
- PSCI communication can be tested with reset -r psci and doesn't
need a dedicated command
Therefore disable all this commands, as replacements for their functionality
are already enabled.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240112170529.1970537-2-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
barebox supports Ethernet and USB OTG in host and device mode on the
Rockchip and i.MX boards, so there's little use if any for RATP.
Disable it and reclaim the space.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20240112170529.1970537-1-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
| |
Let's update to the latest and greatest.
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
The DistroKit image is capable of booting both from SD-Card and eMMC
user area. For boot from eMMC boot partition, we can keep reusing the
same image for the user area, but need to place the boot firmware into
the boot partition as well. Add an extra image that does just that.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20231213165014.1279779-2-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
By default, TF-A located on eMMC boot partition will only boot
FIP from eMMC user area. Toggle STM32MP_EMMC_BOOT, so FIP image is first
looked for at offset 256K on the eMMC boot partition, before booting
from GPT partition in user area.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.pengutronix.de/20231213165014.1279779-1-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
| |
We will use this to find out whether we can run systemd-repart.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231213144750.1899431-3-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
systemd v254 doesn't know about the partition type GUID for big-endian
MIPS root partitions yet, so systemd-repart fails to parse the
"Type=root" setting in its config, and fails to start. Backport a patch
from systemd 255-rc1, which will wire up the respective GUID.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231213144750.1899431-2-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
| |
'reason' suggests to enable an optimized SHA1 implementation on ARM.
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Rockchip RK35* and i.MX8M* share the same barebox environment, but RAUC
and bootchooser were only configured for the Rockchip boards, rendering
the i.MX boards unbootable. Fix this by adding bootchooser as boot
target only for the radxa,rock3a.
While at it, we switch to using the new of_compatible command. This is
likey more stable than the Model string and allows easy extension in
future by additional compatibles.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Reviewed-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231205095929.3469672-3-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
Boot up is broken on some of the Raspberry Pi and STM32MP1 board we
support, because boot method is changed to bootchooser, but state isn't
defined. Fix this by extending state definition to the missed boards.
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Reviewed-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231205095929.3469672-2-a.fatoum@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This partly reverts commit e0ca556db7983a99acc3330dc32cdae395e84003.
As Ahmad reports, the kernel device tree of STM32MP157C-EV1 has no
aliases at all, which is a precondition for enabling MMCBLKDEV_ROOTARG
in barebox. Restore the old state which detects the PARTUUID of the root
partition and passes it to the kernel.
Link: https://lore.distrokit.org/distrokit/9afb1d8a-f90d-7d74-781d-c33f6b7a2efb@pengutronix.de
Fixes: e0ca556db7983a99acc3 (2023-10-20, "v7a: barebox: pass root=mmcblkXpN instead of root=PARTUUID=… to kernel")
Reported-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
|
|
|
|
|
|
|
|
|
| |
Enable the update bundle so v8a platforms can be updated via RAUC, and
put something useful in the bundle description.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231104003525.1673736-5-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Add a state partition to the genimage configuration (offset is not
needed for the root partition since genimage can calculate that based
on the size of the previous partition and the alignment; and the
barebox-state partition is autodetected based on the partition GUID)
* Add a device tree fragment to barebox containing the state backend
* Add appropriate variables to the barebox defaultenv so the bootchooser
can decide into which partition to boot
* Map boot device (SD card) in rauc-udev compatibility layer
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231104003525.1673736-4-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
| |
Enable two useful commands for debugging.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231104003525.1673736-3-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
| |
Update to the new version with default options. This also includes
detection of GPT partitions with the barebox-state GUID for state
backends, which we'll make use of in the next commits.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231104003525.1673736-2-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now that every image is in GPT format, we can use the devnodes that
include the partition label instead of the partition number to boot our
rootfs. These names are more readable and also more robust against
changing the order of the partitions. Make the change on all bareboxes
that didn't have that yet.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231103225253.1349209-11-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now that all our images have a GPT header which contains partitions
with the labels "root-A", "root-B", and "data", we can use those to
generate the symlinks in /dev/disk/by-usage/ instead of using the
hardcoded, number-based kernel device names. This gives us the
flexibility to change the order of the partitions on disk as long as
their GPT partition labels stay the same.
Since the symlinks in /dev/disk/by-partlabel/ are not necessarily unique
(see [1]), we cannot use these here. Instead, rework the current match
clauses to determine the devpath of the boot device based on the board
compatible, and only create by-usage symlinks for partitions on the boot
device.
[1]: https://github.com/systemd/systemd/pull/29219
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231103225253.1349209-10-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
These partitions are now generated on the first boot by systemd-repart.
The second rootfs is not needed for booting, and overwritten by RAUC
when a bundle is installed, so it can as well be empty. Remove the two
partitions to save space in the generated images.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231103225253.1349209-9-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
| |
The license is for the source of the package and this
package has none. Use 'ignore' here.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231103225253.1349209-8-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
[rsc: mol asked for 'ignore', as the license is not for the recipe. Fixed up.]
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The data partition has not been mounted since /etc/fstab was removed in
commit 5eb165a3722013d2c80d (2016-08-18, "ptxdist: version bump
2016.07.0 -> 2016.08.0"). Create the mount point, and add a systemd
mount unit based on the /dev/disk/by-usage/data symlink set up by our
custom rauc-udev rule.
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Link: https://lore.pengutronix.de/20231103225253.1349209-7-rhi@pengutronix.de
Signed-off-by: Robert Schwebel <r.schwebel@pengutronix.de>
|