plan for maru-0.7 #154
Comments
|
@utzcoz Sounds good! I will get maru-0.7 branches set up soon and will update this issue as I make progress. 2021 Q1 is a good target for 0.7. |
|
Something about mflinger. I can provide libX11, libXfixes, libXext, libXdamage, libXi and libXrandr source to be built with Android tree so someone can try to merge mflinger and mclient into one program. |
|
@pdsouza if you wanna base yourself on my manifest: https://github.com/pintaf/manifest. It's already building some devices. It's not working 100% especially nash, but it's on the good path. PME workd quite well. Let me know if you have any questions. |
|
@pintaf Thanks, I was already referencing your manifest as soon as I started. I'm just getting the official maru-0.7 to a buildable state; then you can send me PRs for most of the changes from your repos and I'll review and merge them in. |
|
Alright guys, I've pushed up a manifest for Everything builds without error when I tested with a Nexus 5 (hammerhead). The core services like Happy to take PRs from @pintaf on any other modifications he thinks are worth including for 0.7. Just send over the PRs and I'll review and merge. |
|
Hi @pdsouza Excellent news! Once everything seems nice do you think about building the devices officialy with the new buster image ? @utzcoz , looking forward working with you on integrating your external settings. It would be nice that this external settings could download the container image and decompress it. It would reduce rom size, and enable us to provide more easilly diferrent images. Debian, ubuntu... |
|
@pintaf That sounds great! I can't wait to ship an official image for the HTC10. And yes, absolutely, as soon as 0.7 is stable and all is working well I will start up the automated builds for |
|
@pdsouza for me, out of the box, the only device you shared in your 0.7 manifest, the hammerhead is failing to build: Seems like a partition size issue. |
|
Ah yes, I will have to disable dex pre-optimization for hammerhead since
the system image is too small on newer versions of Android. I didn't build
with a full-size container image yet so I didn't hit this error. Another
reason why I am looking forward to having our container images downloaded
straight to the userdata partition with our proposed app to bypass these
size restrictions on legacy devices.
…On Fri, Nov 27, 2020, 5:05 AM loïc poisot ***@***.***> wrote:
@pdsouza <https://github.com/pdsouza> for me, out of the box, the only
device you shared in your 0.7 manifest, the hammerhead is failing to build:
Out of space? The tree size of /mnt/disk/maru07/out/target/product/hammerhead/obj/PACKAGING/target_files_intermediates/maru_hammerhead-target_files-eng.ubuntu/SYSTEM is 1133555712 bytes (1081 MB), with reserved space of 0 bytes (0 MB).
The max image size for filsystem files is 1073741824 bytes (1024 MB), out of a total partition size of 1073741824 bytes (1024 MB).
Seems like a partition size issue.
Appart from that it seems to build correctly so I will integrate the htc10
and Aquarius X2.
Maybe also the Pixel XL 2 from bootlessxfly.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#154 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACMM4XLJVDVLHUEYBZKSCLSR52W5ANCNFSM4TVMYHVQ>
.
|
|
@pdsouza I do have an issue while building HTC10 pme: In the past I had solved it in this commit: Do you have any suggestion about why I hit this error with PME and not with hammerhead ? This seems so weird because the issue is linked to mcprepare allow rules, so it shouldn't be device specific. At first I though that you had forgot to call maru code in the hamerhead device repo, but no, you called it. Maybe is it because you disabled selinux on your build ? |
|
Just posting comments as documentation for my future PRs: Solved by adding |
@pintaf , could you send a PR to |
|
Yes, will do |
I remember the SELinux is |
|
PR made. It includes HTC 10 device, the devicesettings repo solving the missing string issue. |
|
@pdsouza can you add me as maintainer for: As of today, I cannot push new maru-0.7 branches: |
|
@pintaf Just added you to two teams within our org (marlin and pme) that have access to those repos. |
@pintaf if you have finished the work of porting marlin and sailfish, you can push LOS 16 source code to maru-0.7 branch, and send your patches with a PR, I can test it locally. |
|
@utzcoz I probably won't have time to PR this until wednesday. So if you want to try it out as of today, you can head over this: https://github.com/pintaf/manifest/blob/maru-0.7-pintaf/default.xml |
Okay, I will test after pushed. The pr can help to show the real changes of device and kernel to boot Maruos, and can be searched directly. |
|
two PR made for pixel kernel and device sources. |
|
The sailfish and marlin boot correctly now with extracted vendor blobs. But the rootfs didn't been extracted correctly, so I will try to fix this problem. |
|
Hello @utzcoz So you ended up on the one thing I was not able to fix. |
|
Also, I did not spend too much time on that, because since we plan to transition to your external settings, the mcprepare could potentially be run after downloading the rootfs image. Instead of including it at build time. |
Looks like its a selinux problem. |
|
When I try to port perspective service in |
|
I though that maybe we could get in touch with @ppoffice from lxc-android. |
|
I'm trying to integrate |
Okay, I found the reason, the |
Maybe this was causing @pintaf's boot hang issue? |
I remember @pintaf has pointed out the permission problem. Maybe he missed something after |
@dianaxxyyzz welcome back. I'm sorry to miss it. I also can confirm it, and I have saw your work and your sharing in maruos' Google Groups. So sorry to miss it. |
|
Yes true. I was thankful that @dianaxxyyzz shared the full sources, because there were modifications that he forgot to share as patches, so full source helped me to put them in git repos and make it available for build to everyone. |
|
Hey guys. I think I have 2 possible solutions for low system space problem.
|
|
@dianaxxyyzz I hadn't. But somebody here had.
|
|
It is possible to download tarballs from external server and not to ship it with system. Or maybe it is possible to build Debian container from scratch on device using something like Linux Deploy. |
Downloading container images is the next step once we have our base The /system space issue is mostly on our legacy devices like hammerhead. Our old Nexus 7 had space issues even on maru 0.6, which is why we have discontinued it for now. We should be able to bring these old devices back online once we download the images from the Maru app. |
|
I will start to port dynamic input switch based on @twaik 's suggestion. |
Update: the new version for dynamic input switching based on @twaik 's suggestion with |
|
After tile service merged into |
|
This is great! I will get started on setting up automated builds for 0.7.
…On Sun, May 30, 2021, 2:31 AM utzcoz ***@***.***> wrote:
After tile service merged into MaruSettings, I think all features are all
merged or refactored into maru-0.7. There is a found issue that showing
internal problem dialog caused by matrix compatibility on Pixel devices. I
will try to fix it recent days. Hi @pdsouza <https://github.com/pdsouza>
maybe we can prepare a alpha release for maru-0.7.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#154 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACMM4XTFPDNCG5BBCC4KZ3TQHLUBANCNFSM4TVMYHVQ>
.
|
|
Just merged security updates from LOS16 through May 2021 into Remember to |
|
I will test with my sailfish and Marlin devices. |
|
Another thing, @pdsouza could you help to change default branch to |
Done! All |
|
@pintaf @pdsouza @utzcoz The solution requires us to add mcprepare to dac_override_allowed so that mcprepare is allowed to use dac_override. This is inside of domain.te. The one big issue I came across with this solution was that mcprepare type is not loaded early enough, so simply adding mcpreapre to dac_override_allowed causes the build to fail since mcprepare is an unknown type at that point(gets loaded later in the sepolicy check). As it is though, I think it may be beneficial to create a maru specific system/sepolicy project, as this would allow for the above problem to be solved. I would simply moved the check into mcprepare.te, add a custom dac_override_allowed which includes mcprepare and comment it out inside domain.te. This allows us to move the check to a point where sepolicyundertands what mcprepare type is instead of failing when adding to the allowed list. This keeps the original check in place, it just moves into into vendor/maruos. Here is the system/sepolicy change: Also, I've hit two issues where system_server is explicitly nerverallowed. This causes two issues: Im working on a vendor/sepolicy specific work around for this by adding a wrapper to default_android_hwservice and default_android_service I also submitted a PR to vendor/maruos that should fix the sepolicy issues with treble devices first noted by @pintaf: |
|
@bootlessxfly Thanks your work, I think this thread is the correct place for sharing. I have left some comments about your PRs. For tracking Pixel 2 and Pixel 2L, you can contact @pdsouza to grant permission to create related projects under MaruOS, and push your source code to those projects. We can discuss how to process vendor binaries at email.
Sorry, I know
I believe those rules work correctly at my Pixel 3a and Pixel 3a XL. Could you post your failed avc denied log at your related PR page? We can discuss it at that page. |
|
Potential new features we can do for 0.7(ideas from @twaik):
|
|
Update for 0.7: we can't release 0.7 alpha successfully last year, but we will try it again this year. |
and others
Hi @pdsouza , we know @pintaf has ported maruos to Android 9.0, can you push base LOS to repositories with maru-0.7 branch, and @pintaf and other people start to send pr with porting patches? The people of community can help to test and review, and you just need to review lastly and merge the pr. If we can start recent days, maybe we can release maru-0.7 the Q1 of the next year.
The text was updated successfully, but these errors were encountered: