Age | Commit message (Collapse) | Author |
|
Do not depend on U-Boot environment for setting the Linux kernel command
line from the boot script.
Use standard APPEND variable to specify the kernel command line, stop
using ${defargs}, ${setup} and ${vidargs} from U-Boot environment.
APPEND is the standard OE variable to set the kernel command line, it is
for example used in OE-core rootfs-postcommands.bbclass to add "ro" to
the command line and part of the official documentation.
U-Boot ${tdxargs} is still supported and its content is added to the
command line.
Remove consoleblank=0 since this is already the kernel default.
Link: https://docs.yoctoproject.org/dev/kernel-dev/faq.html#how-do-i-change-the-linux-kernel-command-line
Related-to: ELB-5619
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
|
|
The removed patch is relevant to the downstream colibri-imx7 kernel
which is no longer supported.
Related-to: ELB-5611
Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
(cherry picked from commit 8f7abc1ea1a1dff9192894acb3b874013802f4eb)
Conflicts:
recipes-bsp/u-boot/u-boot-toradex_2022.07.bb
|
|
CONFIG_OF_UPSTREAM=y uses device sources directly out of the linux
source tree.
The location of the dtb artefacts change depending on OF_UPSTREAM
used or not.
Check for OF_UPSTREAM and pick the dtb file from the used location.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
The problematic change was reverted in mainline, commit 3c3a73424e6d
("Revert "cmd: Set modern hush as default shell""), remove the
workaround so we know in case it break again.
Related-to: ELB-5543
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
|
|
At least imx-boot requires u-boot. Bitbake would thus not consider
our upstream U-Boot recipe as a valid provider and the build will
fail for machines not using binman within U-Boot to build the
boot container.
So the change is required once we start building the upstream
U-Boot for Apalis iMX8 and Colibri iMX8X.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
To build tiboot3 binman uses additional python modules. Add them
to the dependencies.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
U-Boot mainline after v2024.01 has a new hush parser added with by
default is used, however we have the option to still use the old one.
Our boot script fails with the new parser, thus change the u-boot config
to keep our master builds working.
Related-to: ELB-5543
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
The following error was spotted during the boot of Colibri iMX7S on Tezi
6:
```
U-Boot 2022.07-6.4.0-devel+git.e092e3250270 (Jul 11 2022 - 13:42:58 +0000)
CPU: Freescale i.MX7S rev1.2 800 MHz (running at 792 MHz)
CPU: Extended Commercial temperature grade (-20C to 105C) at 29C
Reset cause: POR
DRAM: initcall sequence 8786b820 failed at call 87818e09 (err=-3)
\### ERROR ### Please RESET the board ###
```
This patch solves this issue by calling fdt_increase_size() on
colibri_imx7.c file.
Related-to: TEI-1138
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit 30484414981f3652caf43a0a71b3922e269a643b)
|
|
This commit introduces patches to the defconfig of Apalis/Colibri iMX6,
Colibri iMX7 and Colibri iMX6ULL to enable fastboot support and adds a
logic to the bootcmd to select SDP or Fastboot based on the configs
enabled.
Related-to: TEI-1136
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit 266bdf4d7b1ce31bcbc35dd92f52ca45c60eadba)
|
|
Add a patch to fix device tree build errors from other boards that are
not related to Colibri.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit b49784d54dbdc8532fc87c163f951ebff3073701)
|
|
The last commit [0cdb7f9e] removed the need of the 'nand_padding' function
within our recipes, so remove it.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit 97ac673d8e85079e6803246aace7165752ffacd2)
|
|
This commit backports two patches submitted to the U-Boot mainline, both of
which relocate the 'nand_padding' function to the Makefile of 'mach-imx.'
The first patch introduces the 'u-boot-nand.imx' target to the Makefile,
producing a binary file with 1024 leading zeros. The second patch adds a 'dd'
command to the Makefile, appending 10k zeros to the end of the binary.
These changes will eliminate the need of the 'nand_padding' function.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit 678579d8783e2d24666ca1b6af58544f67579aa0)
|
|
Change to the most "open" MIT license. While at it also update the
summary, description, drop the bogus inherit systemd and update the
SRC_URI to be fully in-line with btnxpuartconf.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
(cherry picked from commit a4ba6f572792adf2ac1703963ecc1f0edc53abfc)
|
|
As the power-down pin (PD#) is shared and handled by +mmc-pwrseq for
mwifiex_sdio, make sure mwifiex gets loaded first.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
(cherry picked from commit db93e09dcc520be7efbe38c8c6717f77c289825e)
|
|
Fix missing Upstream-Status tag on recently merged patch.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit af82ccce6b01f2f282cc2149f1bbbaf43e14de7a)
|
|
Add patch to enable variant detection durint boot time. Depending on the
board configuration, the variant will be set to "-emmc" or cleared in
case it is a NAND device.
This patch improves variant detection during system initialization
through USB recovery mode.
Related-to: TEI-1112
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit 16e5f73b2b294855fe579e8252b149154bdb7322)
|
|
Rename mwifiexap to mwifiexconf as it now handles more than just the AP
mode with disabling power saving on Verdin AM62 as well.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
(cherry picked from commit cd2bef82186ae5862d1d9b6948697b4a7e755cf0)
|
|
On the Verdin AM62 which uses the IW416 chipset for Wi-Fi add module
load options for the MWiFiEx driver to disable power save for now as
recommended by NXP.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
(cherry picked from commit 239f170b4daa160dd4bd2ae3333d6b05fc98d09a)
|
|
Follow meta-ti downstream dependency change on ti-sci-fw and
change to do the atf, dm, optee dependency through packageconfig.
See also commits:
22b8a0a3 ("recipes-bsp:ti-sci-fw: Add k3r5 as COMPATIBLE_MACHINE")
69056610 ("recipes-bsp: u-boot-ti.inc: Remove ti-sci-fw from DEPENDS for k3 machine")
ti-sci-fw is only needed for the k3r5 build, removing it from k3 fixes:
| ERROR: Nothing PROVIDES 'ti-sci-fw' (but .../meta-toradex-bsp-common/recipes-bsp/u-boot/u-boot-toradex_mainline.bb DEPENDS on or otherwise requires it)
| ti-sci-fw was skipped: incompatible with machine verdin-am62 (not in COMPATIBLE_MACHINE)
ti-dm-fw is only needed for the k3 build, removing it fixes errors
during recipe parse, e.g.:
| ERROR: Nothing RPROVIDES 'ldconfig' (but mc:k3r5:.../openembedded-core/meta/recipes-devtools/dpkg/dpkg_1.22.0.bb RDEPENDS on or otherwise requires it)
Related-to: ELB-5428
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
The correct override matching all i.MX based machines is now
imx-generic-bsp, not imx. Fix that.
Fixes 7a2d6ff ("u-boot-toradex: split into a release/pinned and latest head recipe")
Related-to: ELB-5357
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
This reduces confusion about the current solution on what version gets
built with use-head-next.
This drops using the `use-head-next` override in favour of having two
recipes. Which recipe is to be built will then be set in the various
<machine.conf> files.
To reduce code duplication split out common code into a inc file and
further split out i.MX and TI specific stuff into two specific inc
files.
Related-to: ELB-5357
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Stop deploying u-boot-with-spl.imx, this is not required to properly flash
the board nor for debugging. Separated SPL and U-Boot proper binaries available,
used by Tezi and can be easily flashed using update_uboot and update_spl helper
scripts.
This enables some simplification on Tezi.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
(cherry picked from commit 81a446adde4296c7e68f916bc536414a1caa1397)
|
|
We no longer provide a kernel based on the NXP downstream BSP.
Drop patching U-Boot to support booting such a kernel on a Colibri iMX7.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
This enables building upstream U-Boot for Verdin AM62 including building
and deploying tiboot3 and tispl with binman.
Related-to: ELB-5331
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
The class now handles to provide an imx-boot symlink to flash.bin.
Also drop redundant entries in DEPENDS which are now part of the
generic u-boot include file.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
U-Boot release 2023.07 branched away from the master branch which
is reflected in OE u-boot recipe.
Switch this back to the master branch, so that for use-head-next we
take latest on master rather than latest on branch u-boot-2023.07.y.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
With the kernel moving 32 bit device trees in subdirectories starting
with kernel 6.5 the toradex machine configuration files provide the
variable KERNEL_DTB_PREFIX.
The fit image configuration name is constructed from KERNEL_DTB_PREFIX
plus the dtb file name. Make use of the KERNEL_DTB_PREFIX variable to
construct the configuration used for booting.
As we now also set KERNEL_DTB_PREFIX for the 64 bit machines the inline
python works for all machines, thus drop the DTB_PREFIX overrides.
Related-to: ELB-5262
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Add a patch to fix a DDR RAM issue that can cause bit flips at
temperatures above 60°C.
Related-to: ELB-5211
Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
(cherry picked from commit 63e4e72c5c71d510106c6504074697c30d396e47)
|
|
Add TI-TFL ("TEXAS INSTRUMENTS TEXT FILE LICENSE") to TI tezi image artifact,
while removing the NXP one.
Current license file come from [1].
Disable the download of an updated license from a specific url, formerly used
for NXP EULA, due to missing proper link.
[1] https://git.ti.com/cgit/graphics/ti-img-rogue-umlibs/tree/?h=linuxws/dunfell/k5.10/1.15.6133109
Related-to: ELB-5161
Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
(cherry picked from commit a44240a3d58d353d86be7fed4d79c086dc2f2312)
|
|
Since commit 4f16c57dda5f49cef1648ba829d9bf10cb4010dc:
[ u-boot: distro-boot-script: support booti automatic decompression ]
For the boards that support "booti automatic decompression", the
kernel_addr_load had been set to ${loadaddr} rather than
${ramdisk_addr_r} for fitImage, this is breaking fitImage boot due to
overlapping DTB/DTBO addresses.
Change it back to boot from ${ramdisk_addr_r} for fitImage.
Signed-off-by: Ming Liu <ming.liu@toradex.com>
(cherry picked from commit 4d6e71f6b40415841d402afdbd54c9c89adf10f5)
|
|
Not like ARM machines, which the devicetrees locate at
arch/arm/boot/dts in kernel source and the devicetree names directly
built into fitImage by bitbake.
For i.MX8 machines instead, the devicetrees locate at a vendor specific
subdirectory, take NXP as a example, they are under
arch/arm64/boot/dts/freescale, so bitbake builds them into fitImage
also with a "freescale_" prefix to the real devicetree name.
We should distinguish these two situations in boot.scr, let's introduce
a variable KERNEL_DTB_PREFIX to achieve that, and set it to
"freescale_" for i.MX8 machines.
Signed-off-by: Ming Liu <ming.liu@toradex.com>
(cherry picked from commit f22922827754e40a0cbdbd929826b13e68e0a6c7)
|
|
When setupargs is not already set we set it using a fallback.
Remove escaping of the value set as we do want variables evaluated
here.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
(cherry picked from commit 66b323f88297c78eadf3e014d26aca2edf484204)
|
|
Add patch to use the in band bad block table on Colibri iMX7.
Related-to: ELB-4989
Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
(cherry picked from commit 86e64d70123b2ff55924ec6581c6154e628f661b)
|
|
The u-boot-with-spl.imx binary, now is deployed for each u-boot configuration
type and the condition that suppressed the error in case u-boot-with-spl.imx
was not generated has been removed.
Signed-off-by: Stefano Viola <stefano.viola@toradex.com>
(cherry picked from commit 15ec9fe118984cc977016269c069571c4530ef8f)
|
|
Update the distro config env memory layout for the Apalis iMX6,
Colibri iMX6, Colibri iMX6ULL and Colibri iMX7.
This fixes a potential issue caused by the compressed kernel being
relocated on top of the ramdisk causing its corruption.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
(cherry picked from commit 84c7638a7261561e30bfd574b69e1fddcac546cc)
|
|
Add patch to properly size the memory on Verdin iMX8M Plus.
Related-to: ELB-4711
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
(cherry picked from commit 6931dff87db18d38a14414fe35ef49805bcf5238)
|
|
Keep the file name with the inital U-Boot environment variables
constant, independent of the recipe name U-Boot is built with.
Our image creation and libubootenv expect a fixed name.
Related-to: ELB-4680
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Fix build issue with patches not found
Relates-to: ELB-4680
Fixes: 7ad862bbbaaf ("u-boot: convert bbappend into our own recipe")
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
|
|
The version mismatch between what we want to build and what the recipe
in openembedded-core builds can get big so that CVE backports and other
changes makes our build fail.
Rather create our own recipe reusing the include file from
openembedded-core to reduce the risk of build failures.
Relates-to: ELB-4680
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Support booti automatic decompression with the following memory layout:
- loadaddr=0x48200000 allows for 128MB area for uncompressing (ie FIT
images, kernel_comp_addr_r, kernel_comp_size)
- fdt_addr_r = loadaddr + 128MB - allows for 128MB kernel
- scriptaddr = fdt_addr_r + 512KB - allows for 512KB fdt
- ramdisk_addr_r = scriptaddr + 512KB - allows for 512KB script
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
We've decided not maintain these machines in BSP6, let's drop the
machine references.
There is a special varialbe UBOOT_KERNEL_TYPE, it's only being used by
apalis-tk1 machine, so also drop it.
Related-to: ELB-4471
Signed-off-by: Ming Liu <ming.liu@toradex.com>
|
|
Add patch to enable `boot` command on colibri-imx7, submitted in
https://lore.kernel.org/all/20220803112009.271563-1-francesco.dolcini@toradex.com/
Related-to: ELB-4642
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
|
|
Update U-Boot v2022.07 patches, backport `board/toradex`
patches from current denx/master git.
This updates the config block interactive creation procedure.
Related-to: ELB-4626
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
|
|
lzop-native is not needed during u-boot compilation since LZO
compression is not enabled in u-boot.
Related-to: ELB-4534
Signed-off-by: Ming Liu <ming.liu@toradex.com>
|
|
Remove patch 0005-toradex-tdx-cfg-block-add-0068-i.mx-8m-mini-sku.patch
from the patches applied when building master, since it is now merged with
commit a2da29a4e2a0 ("toradex: tdx-cfg-block: add 0068 i.mx 8m mini sku").
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
|
|
Currently we have downstream/upstream Toradex uboot recipes and we dont
want to add fw_env.config for both of them to avoid code redundant.
Let's add fw_env.config to libubootenv, which will be installed for
both donwstream/upstream distros.
Change libubootenv to a machine arch recipe since now it has machine
specific fw_env.config files.
Related-to: ELB-4536
Signed-off-by: Ming Liu <ming.liu@toradex.com>
|
|
The class makes sure that we see the git hash the U-Boot is built from
in the U-Boot banner.
Related-to: ELB-4090
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
oe-core now applies 0001-i2c-fix-stack-buffer-overflow-vulnerability-in-i2c-m.patch
fixing a CVE in 2022.01. As we bumped to 2022.07 the commit is already applied
and applying it a second time makes do_patch fail.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Apply patches needed for Toradex SoM which were applied after v2022.07
or are still in flight.
xxxxxxxxxxx toradex: tdx-cfg-block: add 0068 i.mx 8m mini sku
1cf4e79f577 toradex: tdx-cfg-block: add new toradex oui range
7e27ce16c5d toradex: tdx-cfg-block: extend assembly version
494ef10c3bf toradex: tdx-cfg-block: use defines for string length
39ff0624bc5 toradex: tdx-cfg-block: use only snprintf
Related-to: ELB-4601
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Note that once in openembedded-core u-boot is bumped to v2022.07 the
LIC_FILES_CHKSUM and PV change can be dropped.
Related-to: ELB-4600
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|