beaglev-ahead update - #685
Conversation
df231ef to
ece320c
Compare
|
I updated my branch with full mainline support (though it's not finalized). | wic.WicError: BeagleV boot plugin: U-Boot image exceeds the 1 MiB boot area Adn Ehab Elkawass (@adn-dodo): could you propose a change to your WIC plugin so that we can make the U-Boot area 4 MiB big? |
|
Hi Michael Opdenacker (@michaelopdenacker), I attached the patch here. Let me know if you want me to change anything. |
ece320c to
4c0a346
Compare
|
Pushed a new update. |
c26ca92 to
4c0a346
Compare
|
Khem Raj (@kraj) The checks are failing because of the need to add "blacksail" to LAYERSERIES_COMPAT Happy to rebase my branch when this is done. |
Hi Adn Ehab Elkawass (@adn-dodo) |
4c0a346 to
cffc4fe
Compare
|
Rebased the branch against the latest master. |
|
Once these are merged, I'll use them to add support for LicheePi 4a (same SoC, fully upstream too). |
cffc4fe to
e5615e3
Compare
This allows to remove the "opensbi-revyos" recipe, which has no more users in this layer. Signed-off-by: Michael Opdenacker <michael.opdenacker@rootcommit.com>
Keeping the same kernel patches that all apply and look still relevant. This allows to remove the "linux-beaglev-dev" recipe. Add configuration fragment to support USB. Signed-off-by: Michael Opdenacker <michael.opdenacker@rootcommit.com>
For DDR training on TH1520 Make the "beaglev-ahead" machine use it. Signed-off-by: Michael Opdenacker <michael.opdenacker@rootcommit.com>
Mainline U-Boot no longer fits in the existing 1 MiB boot area. Increase the available U-Boot area to 4 MiB and relocate the GPT entry array accordingly. Move the first WIC partition to an 8 MiB boundary so it does not overlap the relocated GPT metadata. Signed-off-by: Adn Elkawass <elkawasadn@gmail.com>
Using 2 patches from https://github.com/knaerzche/u-boot/tree/th1520-beaglev-ahead These patches are necessary to use training on single-rank DDR memory on TH1520. Enable support for U-Boot "bootflow" commands https://docs.u-boot-project.org/en/latest/develop/bootstd/overview.html#bootflow Signed-off-by: Michael Opdenacker <michael.opdenacker@rootcommit.com> Signed-off-by: Alper Ak <alperyasinak1@gmail.com>
The bootcmd scanned all bootdevs in order, so a board that has an image on
eMMC did not boot from an SD card that is inserted, and a board with no
bootable medium waited for the ethernet bootdev.
Scan the SD card bootdev first and the eMMC bootdev after it. The bootcmd sets
rootdev before each scan and extlinux uses it in the root argument, so one
extlinux.conf works on both media and the deploy step no longer rewrites a
copy of it with sed.
Using U-Boot's "bootflow" command requires to declare the boot partition
as "bootable" ("--active" in the .wks file).
Signed-off-by: Alper Ak <alperyasinak1@gmail.com>
Signed-off-by: Michael Opdenacker <michael.opdenacker@rootcommit.com>
Flashing with fastboot is more complicated and flashes partition by partition only. When U-Boot supports USB storage, we will be able to reflash an existing image with the "ums" command. Signed-off-by: Michael Opdenacker <michael.opdenacker@rootcommit.com>
e5615e3 to
b8df9e3
Compare
|
Removed a few obsolete comments |
Khem Raj (kraj)
left a comment
Khem Raj (kraj)
left a comment
There was a problem hiding this comment.
I am happy with the state of this patchset to merge. Lets address any other improvements via separate PRs
6 participants
Description
Updating support for the beaglev-ahead machine:
Checklist
docs/DEPRECATED.md,
then it is maintenance-only, i.e. it does not add support for a
previously-removed BSP (or significant features to one slated for removal)
bitbake-registry/have been updated, if necessary
kasfile has been provided inkas/Signed-off-byand any other tags to each commitHow has this been tested?
Tested on a newly built image.
The image is first booted for SD, and the SD image is used to flash eMMC with the same wic image.
Then checked that the board can boot without the SD card.
All CPU cores are detected now, and we have open-source DDR training.