coreboot version bump from 4.22.01 to 24.02.01 (Q35/ivy/sandy/haswell/librems) #1715
Conversation
Signed-off-by: Thierry Laurion <[email protected]>
Signed-off-by: Thierry Laurion <[email protected]>
grep -R 24.02.01 boards/ | awk -F "/" {'print $2'} | while read board; do sudo make BOARD=$board coreboot.save_in_oldconfig_format_in_place; done
Signed-off-by: Thierry Laurion <[email protected]>
Signed-off-by: Thierry Laurion <[email protected]>
tlaurion
changed the title
…712" This reverts commit c43b6fc, reversing changes made to fb9c558. Signed-off-by: Thierry Laurion <[email protected]>
|
@aluciani roms have built successfully. T430 can be tested and results posted here. Artifacts for t430 with either hotp or not available for 30 days:
|
tlaurion
changed the title
tlaurion
changed the title
|
Tested https://output.circle-artifacts.com/output/job/a3a29799-197a-43a6-ad76-ec3b0e1b238d/artifacts/0/build/x86/x230-maximized/heads-x230-maximized-v0.2.0-2215-g21ab781.zip working (not causing brick). BOOTSPLASH IS GONE |
I was going to report |
|
Putting back as draft: cbmem -c | grep -ie error -ie warning shows errors with memalign and out of memory errros when trying to map bootsplash. Also shows PNP: missing read resources errors. |
|
Testing local builds pointing to v0.2.0 docker image which contains libnss3-dev new coreboot requirement from built docker image from 21ab781 which was #1713 This is done locally by calling (latest currently points to v0.1.9, not v0.2.0 to reflect master's currently used v0.1.9 docker image) |
|
Replicates the no bootsplash issue. So can be tested with qemu, which will be more easy to report to coreboot and involve coreboot to fix this without real hardware. |
@JonathonHall-Purism I guess you get the same on librems? |
|
Attempting to build for coreboot 2405 fils with coreboot buildstack dependencies pointing to unavailable tarballs, as can be seen with this CircleCI build attempt at https://app.circleci.com/pipelines/github/tlaurion/heads/2642/workflows/92c0c23d-340a-4bd5-8cd7-9b5b479b00bf for testing branch https://github.com/tlaurion/heads/tree/xxx0_coreboot2405 testing commit tlaurion@481c173 with commit log stating error that will show in build logs there.... coreboot regressions all around it seems. TLDR: coreboot 2405 depends on https://distfiles.macports.org/acpica/acpica-unix-20230628.tar.gz which doesn't exist. |
@JonathonHall-Purism can you cache https://downloadmirror.intel.com/783534/acpica-unix-20230628.tar.gz under purism mirror please? |
@JonathonHall-Purism I just came to realize that acpica-unix is not cached under packages/x86/... |
@JonathonHall-Purism that was our fault, we override the acpica-unix url defined under our mirroring using script which I hacked now as of tlaurion@71a6035 |
|
Would you like to send link to maximized T530, T420(dgpu - if you like to also test dgpu), and T440p(dgpu - if you like to also test dgpu) versions? I would take the risk on me to have to spend hours on each device externally reflashing it. @fhvyhjriur : We know bootsplash is not working, since coreboot changed its memory manager for jpg processing. This will require to change all Heads bootsplashes to be smaller. As for testing roms from this PR, the same applies as documented at https://osresearch.net/Downloading#downloading-heads-from-circleci If something missing there (like same applies to PR) please provide a PR to change heads-wiki there following community guidelines for contributions. |
|
I think i have find out the files for this PR. I expected same job-id in the link, but it seems like the job-id should differ from thinkpad model to other thinkpad model. When i am correct, this should be the image for T530: https://output.circle-artifacts.com/output/job/dc335a65-ce66-4455-ade6-298db6863f2a/artifacts/0/build/x86/UNMAINTAINED_t530-maximized/heads-UNMAINTAINED_t530-maximized-v0.2.0-2215-g21ab781.rom Should i wait for the new release with the updated bootsplash jpg file and report if its working fine with the different screen resolutions? They differ from thinkpad to thinkpad and this would probably be a good test. |
|
Asked coreboot devels at:
TLDR of current issue under commit message at tlaurion@8a4ac33 |
Mirror is updated with deps for 24.02.01 (this PR) and 24.05 (#1716). Thanks for the reminder. |
…ailure on bootsplash and defaulting to 4mb of heap when bootsplash Add upstreeam patches to downstream: git fetch https://review.coreboot.org/coreboot refs/changes/75/83475/1 && git format-patch -1 --stdout FETCH_HEAD > 0001-prevent_bootsplash_heap_allocation_failure_to_booting.patch git fetch https://review.coreboot.org/coreboot refs/changes/76/83476/2 && git format-patch -1 --stdout FETCH_HEAD > 0002-increase_heap_from1mb_to4mb_when_bootpslash_enabled.patch Rebuilding coreboot version with patches applied: docker run -e DISPLAY=$DISPLAY --network host --rm -ti -v $(pwd):$(pwd) -w $(pwd) tlaurion/heads-dev-env:v0.2.0 -- make BOARD=qemu-coreboot-whiptail-tpm1 real.remove_canary_files-extract_patch_rebuild_what_changed Remove file added by patch when patching fails: sudo rm build/x86/libgpg-error-1.46/src/syscfg/lock-obj-pub.powerpc64le-unknown-linux-musl.h Applying patches touching coreboot version 24.02.01 defiend in board config to all boards: grep -R 24.02.01 boards/ | awk -F "/" {'print $2'} | while read board; do sudo make BOARD=$board coreboot.save_in_oldconfig_format_in_place; done Adding added/modified files to git tracked files for commit: git status | grep -e modified -e added | awk -F ":" {'print $2'}| xargs git add Signed-off-by: Thierry Laurion <[email protected]>
…s from default vs master Signed-off-by: Thierry Laurion <[email protected]>
…iptail-tpm1-hotp.config, remove unneeded diff from defconfig TODO: - all xx20/xx30/xx4x have wifi support, top down allocation disabled etc: create issue to track what is needed/unneeded Signed-off-by: Thierry Laurion <[email protected]>
-CONFIG_BOOTSPLASH_CONVERT=y -CONFIG_BOOTSPLASH_CONVERT_QUALITY=70 Signed-off-by: Thierry Laurion <[email protected]>
grep -R 24.02.01 boards/ | awk -F "/" {'print $2'} | while read board; do sudo make BOARD=$board coreboot.save_in_oldconfig_format_in_place; done
Signed-off-by: Thierry Laurion <[email protected]>
|
@JonathonHall-Purism I added your fixes, tested on qemu and works. This invalidates #1720 which I will close. |
|
@JonathonHall-Purism posted PR to oreboot version bump to pr pointing to their coreboot fork with fixes included at #1721 @JonathonHall-Purism : another PR will need to change CircleCI board cache dependencies: all board will be able to share 24.02.01 caches once both are merged. |
@fhvyhjriur once 8d3c5cf CircleCI builds are successful, the roms produces for xx20/xx30/xx4x will need testing. @JonathonHall-Purism Thanks for your precious contribution upstream! |
|
@JonathonHall-Purism since PR title is also for Librems, I could merge your PR here so that another commit adds CircleCI boards dependencies to reduce time for ROM production if all boards sharing 24.02.01 could be built from same caches? |
Signed-off-by: Thierry Laurion <[email protected]>
|
@JonathonHall-Purism testing gains of shared coreboot 24.02.01 buildstack at https://github.com/tlaurion/heads/tree/test_240201_shared_coreboot_buildstack building at https://app.circleci.com/pipelines/github/tlaurion/heads/2667/workflows/5086fa75-30a6-44ac-aa72-69bf140de9c7 If works/useful: will add that commit here. |
Build failed for some boards. The build for the boards i have did not fail. For example this looks to be the build for T530: https://output.circle-artifacts.com/output/job/cf336210-0ac2-4458-82a0-a4bd8db904fb/artifacts/0/build/x86/UNMAINTAINED_t530-maximized/heads-UNMAINTAINED_t530-maximized-v0.2.0-2225-g9e9ab76.rom |
Fails since coreboot forks and coreboot releases tarballs cannot be mixed to cross use their buildstacks. This will be postponed. Will open issue and unrelated to this coreboot 24.02.01 being common. Will open issue to track and fix. |
@aluciani there is no gain outside of sharing coreboot buildstack (crossgcc) and linux built artifacts to speed up CircleCI as said in previous comment, and as you witnessed, ti doesn't work out of the box today under CircleCI building
When it comes to testing PRs, when you scroll down, you will see all tests and links to CirlceCI to download roms. Note that this PR final commit (unless bugs are found under specific platforms) is 7454d58 which is atop of master, and completed producing rom/zipfiles artifacts for all boards currently defined under CircleCI under master. Note: 8d3c5cf was commit prior of 7454d58. CircleCI interrupted building the prior because I commited the latter and CircleCI interrupts builds for the same branch when newer commti exists. In PR, we can dismiss commits with red X if a green checkmark is provided by CircleCI: this means that the latest commit under a PR successfully produced roms/zip files for all boards contained under .circleci/config.yaml and that those roms are ready to test for willing board testers for coreboot/linux version bumps, which are the riskiest thing to test and requires an external programmer in case the roms produce bricks for tester to revert to master and report results in testing PR |
|
I have put this PR as draft because i'm concerned in differences between defconfig changes between 4.22.01 and 24.02.01, and since doing a call for testing from all #692 was historically long, I would like this PR to contain proper coreboot changes to be tested by all community testers to fasten this merge as much as possible. |
|
Something is corrupting heads repo from coreboot forks branch. Traces of what I try to accomplish here. Try to generate all coreboot defconfig to see deviation from coreboot specific version: Then from failed boards, pick one board per shared coreboot version and build: Then regenerate coreboot configs in defconfig: Unfortunately, heads repo gets replaced by Dasharo when doing some randomly. Will revert to backup and snapshot qube in between to figure out what is happening with |
This was me calling docker in parallel for multiple boards. |
|
Deprecated in favor of #1723 |
Coreboot version bump to 24.02.01 for all platforms that depended on coreboot 4.20.01.
Testable roms available under green mark, select board, then go under artifacts and select zip file to internally upgrade. Do backup first from recovery shell and be ready to externally flash backup if bricked.
EDIT: 8d3c5cf inclues @JonathonHall-Purism patches (that can be seen under
patches/coreboot-24.02.01of this PR):Tested:
Old (prior of 8d3c5cf)
Tested
All boards untested will be prepended UNTESTED, then all boards UNTESTED will go UNMAINTAINED.