Closed delhiryder closed 8 months ago
Can one of the admins verify this patch?
@resin-jenkins ok to test
@delhiryder please redo the git commit logs. <component>
is just a placeholder which should be replaced with the actual component name that the commit modifies. Have a look at examples of commit logs in this repository or any other balena boards repositories. There are plenty to take inspiration from. As a quick example https://github.com/balena-os/balena-beaglebone/pull/544/commits
@delhiryder balena-yocto-scripts
, contracts
, meta-balena
submodules do not need to be updated manually because they get automatically updated. Please remove them from this PR
@delhiryder
balena-yocto-scripts
,contracts
,meta-balena
submodules do not need to be updated manually because they get automatically updated. Please remove them from this PR
Understood. What about the meta-ti
, meta-arm
and meta-openembedded
submodules ? These still need to be updated to the versions in the master branch, right ?
@delhiryder
balena-yocto-scripts
,contracts
,meta-balena
submodules do not need to be updated manually because they get automatically updated. Please remove them from this PRUnderstood. What about the
meta-ti
,meta-arm
andmeta-openembedded
submodules ? These still need to be updated to the versions in the master branch, right ?
Also, the poky
submodule was also updated between my feature branch creation last year and the latest PR date. So these should be part of the PR, right ?
@delhiryder
balena-yocto-scripts
,contracts
,meta-balena
submodules do not need to be updated manually because they get automatically updated. Please remove them from this PRUnderstood. What about the
meta-ti
,meta-arm
andmeta-openembedded
submodules ? These still need to be updated to the versions in the master branch, right ?Also, the
poky
submodule was also updated between my feature branch creation last year and the latest PR date. So these should be part of the PR, right ?
Also, the commit hash that my branch's meta-balena
submodule used to point to appears to have been removed (or at least is no longer available) - see the section of the failure logs in the the Flowzone/Versioned source
test below:
Error: fatal: remote error: upload-pack: not our ref ba96a9ea69aa82f84eb792a503c66ec16cf6f142
Error: fatal: Fetched in submodule path 'layers/meta-balena', but it did not contain ba96a9ea69aa82f84eb792a503c66ec16cf6f142. Direct fetching of that commit failed.
Error: The process '/usr/bin/git' failed with exit code 128
@delhiryder
balena-yocto-scripts
,contracts
,meta-balena
submodules do not need to be updated manually because they get automatically updated. Please remove them from this PRUnderstood. What about the
meta-ti
,meta-arm
andmeta-openembedded
submodules ? These still need to be updated to the versions in the master branch, right ?Also, the
poky
submodule was also updated between my feature branch creation last year and the latest PR date. So these should be part of the PR, right ?
Unless you need a poky update, let's not update it. The poky submodule also get automatic updates
@delhiryder
balena-yocto-scripts
,contracts
,meta-balena
submodules do not need to be updated manually because they get automatically updated. Please remove them from this PRUnderstood. What about the
meta-ti
,meta-arm
andmeta-openembedded
submodules ? These still need to be updated to the versions in the master branch, right ?
The meta-openembedded also gets automatic updates. Do you need to have the BSP (meta-ti and meta-arm) updated for this board?
Also, the commit hash that my branch's
meta-balena
submodule used to point to appears to have been removed (or at least is no longer available) - see the section of the failure logs in the theFlowzone/Versioned source
test below:Error: fatal: remote error: upload-pack: not our ref ba96a9ea69aa82f84eb792a503c66ec16cf6f142 Error: fatal: Fetched in submodule path 'layers/meta-balena', but it did not contain ba96a9ea69aa82f84eb792a503c66ec16cf6f142. Direct fetching of that commit failed. Error: The process '/usr/bin/git' failed with exit code 128
@floion, I have reverted my local feature branch's submodules to the ones from the time I checked out my branch (July 2023). However, I am unable to update my layers/meta-balena
to the exact commit hash ba96a9ea69aa82f84eb792a503c66ec16cf6f142
because it appears to have been removed from the meta-balena
repository.
Here is what my local command prompt prints out:
ubuntu@ip-172-31-16-228:~/git-debugging/balena-beaglebone$ git submodule update --init --recursive -- layers/meta-balena
fatal: remote error: upload-pack: not our ref ba96a9ea69aa82f84eb792a503c66ec16cf6f142
fatal: Fetched in submodule path 'layers/meta-balena', but it did not contain ba96a9ea69aa82f84eb792a503c66ec16cf6f142. Direct fetching of that commit failed.
fatal:
My submodules are otherwise correctly pinned to the versions they had last July (see below):
ubuntu@ip-172-31-16-228:~/git-debugging/balena-beaglebone$ git submodule
0008306e565a95aa51d04a8a0fcc318df3df59a2 balena-yocto-scripts (v1.19.41)
8dfe06b57576e15e8579d4083bb316c9c7671488 contracts (v2.0.91)
c39bb4ce3b60b73d35c5fb06af012432e70d6b38 layers/meta-arm (4.0.2-1-gc39bb4ce)
+069243961adb123830eb4073a6245b2fa1e6f8b3 layers/meta-balena (v5.1.49)
4da92ed9be41734f6ced46b981958e2e868cbff2 layers/meta-openembedded (4da92ed9b)
1ccba22923b1c0ce3558075f3de3846ba983155c layers/meta-ti (09.00.00.008)
471318ae2f6b3c142822001f4a18e2fed8c78f1a layers/poky (yocto-4.0.11-65-g471318ae2f)
Should use the next latest release tag (from my branch point in July, 2023) this would appear to be v4.1.6
or something else entirely ?
Let me know.
Thanks,
Sidd
@floion - I found and used the following meta-balena submodule commit 3df341 (v 3.1.3). It is from August 2023. All other submodule updates commits have been removed.
Reproducing some questions from the zulip chat here:
@Sidd Gupta (guest) what need is for the second machine, the beaglebone-play-k3-r5? what is the difference to beaglebone-play?
@floion - This was the only way for me to get the so called Multiconfiguration (MC) build to succeed. My understanding is that think the beaglebone-play has two MCUs on it, and both targets need to be build simultaneously.
second question, why do we declare the machine conf files here? https://github.com/balena-os/balena-beaglebone/pull/659/commits/3786a3ec932083f21f535e5516d015220e27c456
@floion - I chose at the time to create a beaglebone-play
machine type, to be consistent with the rest of the naming conventions (e.g. beaglebone-black etc). This required setting up the beaglebone-play
machine type(s) as an alias for beagleplay
, which is indeed set up in the meta-ti
layer.
@floion - I have changed the MACHINE type from beaglebone-play
to beagleplay
. Please take a look.
Closing this pull request, in favor of the much newer and shinier https://github.com/balena-os/balena-beaglebone/pull/666.
Rolled up PR, with representative checkins showing which logical components were changed. Built and tested locally. Ready to be added to the test automation.