Manjaro Broke Asahi Linux... AGAIN!!!

Key Takeaways

reset the timers boys it's time once again for a completely unnecessary mistake and do I even need to tell you the group that is responsible for making the change menjaro not of the x86 variety this time of the arm side instead once again broke something massive now it's one thing to break it on Manjaro to break it in the Manjaro packages which everybody outside of Manjaro can very easily ignore it's another thing to make a patch and then Upstream it over the head of the affected project breaking it on everything based on Arch Linux arm now to be fair the Arch Linux arm maintainer is absolutely at fault for letting this patch go through but this patch should have never existed in the first place so a couple of days ago this report was made on the Asahi Linux subreddit M1 MacBook Air Linux Asahi Edge won't boot now for anyone completely unaware RC Linux is the project to get Linux running on the Apple silicon now in the case of this report while it looks like it's not booting that's not the case what's actually happening is it's booting perfectly fine but the instant the xor loads it crashes out now no one really suspected that anything was off here Hector Martin otherwise known as Markin just thought xorg is terrible and don't use xlog and the problem will probably go away but it turns out the problem is a little bit deeper than that now to understand what the problem is and why this is happening we need to go back a couple of months to October of 2022. extra slash Mesa add us he driva to Arch 64. this is being made on the pack field repo for Arch Linux arm basically this is a repo that includes all of the build scripts for everything in the Arch Linux arm repo this is just the general way that Arch Linux manages stuff and it works basically the exact same way over on the x86 side as well we've looked at this reefer before and if you're ever curious about what's going on with the package this is always a good place to come to but hold on a second this is Arch Linux and Arch Linux ah Manjaro doesn't maintain this so how are they the ones that fault we'll get to that in just a moment before then let's have a look at the pull request itself this is by strict Mesa has been working on the ushi driver for about a year and is still in pretty rapid development the assay driver is used by Apple based devices with the M1 and M2 Apple silicon chips add this driver so it can be used on Apple hardware and the commit is very simple it's basically nothing really changing it's changing the package version upping it to 1.1 or the package row and also changing a simple build option the gallium option including Asahi this is going to make sure the RC driver is compiled into Mesa and is able to be used okay then so what's the big deal here when you're using Linux on Apple silicon right now today you probably shouldn't be using Mainline Mesa what you should be using instead is something called Mesa Asahi Edge this is a modified version of Mesa and set up to be the exact version needed to make everything work the way that it should I'm not going to say a good experience but it's going to be an optimal experience based on what the developers understand however by all accounts Mainland Mesa should be working and it shouldn't be crashing X or org at a bare minimum it should be giving you a usable system now whilst it says closed over here it's not closed without merger so it was actually merged in a separate PR being this one here which added a separate thing which isn't important today but as we can see down here this added the usahi driver and this was added all the way back in January and with that merger a new version of the Mesa package was released and this sat in the Arch Linux arm repos for a couple of months without anybody really noticing a problem most people using Asahi know they should be using Misa ASI Edge so no one even realized there was an actual problem here until that Reddit post was made and when Hector realized the problem he had this to say over on his Mastodon lovely another Arch Linux R moment the reason why EXO crashes on Asahi when you use the new kernel without our Downstream Mesa is that alarm arched Linux arm helpfully started enabling the Asahi driver in their Upstream packages some time ago which is useless and was only there for testing without asking us even though it's not part of the default driver set for arm 64. now unlike a lot of the other upstreaming dramas out there this isn't a matter of pushing a patch that adds extra functionality into the project this driver is already present in Mesa it's just not accessible when you normally go and run it and normally go and compile it but why would there be broken code in Upstream Mesa that just doesn't seem to make any sense well this is a pretty normal development practice the Upstream code is hidden behind a build option that Mesa specifically does not set and does not instruct distributions to set some code is merged as part of the upstreaming process I.E for build testing in CI but it's not intended to be used by end users and is not what we're shipping in Asahi Linux to my knowledge this has not been an issue with any other distribution since even a minute of testing would reveal that it doesn't work literally booting it with this patch enabled it will not load xorg the patch shouldn't be there Mesa will be back porting a fix to neutralize the option at any rate it never occurred to me that someone would go out of their way to ship a broken build of Mesa and Hector also had more to say on this dude it was broken by design it is your job as a distro not to do stupid things that breaks users that includes not enabling non-default buildup options without testing them it doesn't matter that it's in a stable release you can't just randomly change build options without testing or understanding anything then blame Upstream when things break for everyone setting build options properly is your job as a distro and if you can't do it you should leave it at the default having something like this in Upstream Mesa is completely normal always read the documentation before touching any of the build options if you don't understand the build options do not touch them because you're probably going to break something now this takes us into where this is a Manjaro problem you might have noticed that strit signed this off as danger Hansen strict mengiaro.org but who is shrit is this just some random no name and Gyro developer that you know just happened to submit a patch and happen to use the Manjaro email no no it's actually not this is someone who should absolutely know better and is not shipping something that is completely broken on arm let's have a look at his website and also the Manjaro team page so stritz.dk Dan Johansson Project Lead on the menjaro armor project and over here arm lead developer works on arm additions including tooling and infrastructure Dan is a hardware guy he enjoys fiddling with boards this is the project leader of Manjaro arm submitting patches to Arch Linux armor that he did zero testing on because if you did any testing on this patch you would know it was broken this isn't one of those patches where you know you run it and it doesn't break it straight away it's like you know under this really weird specific condition it causes a crash this causes a crash if you load the system with exorg that is all the only way you could think this patch was stable is if you didn't even try it with exorg you only used it on Weyland and if by some chance he doesn't even have an Apple device to test this on he should not be submitting any patches involving Asahi Linux if it cannot be tested on Apple silicon you are not the person who should be doing this now once again I want to be very clear Manjaro are not the only ones at fault here but they do play a major part Stritch should absolutely have tested his patch realize within 10 seconds the patch did not work and not submitted it however the same can be said for this Arch Linux arm maintainer that merged the patch without testing it maybe they compiled it and saw Mesa compiles just fine and didn't really do anything more than that because they clearly didn't test this on Apple silicon and if you're not going to test something on Apple silicon you shouldn't be accepting patches involving Apple silicon I get these Apple devices are expensive but if nobody on your team has the devices to test these you shouldn't even be in the position where you're accepting packages for them why are there packages that aren't being tested at all and Hector didn't just comment on Mastodon he also commented on this thread as well saying FYI this PR is the reason why alarm Mesa Arch Linux are Mesa has been broken with Edge kernels on xorg please do not send nonsense like this without understanding the consequences that Upstream driver is useless until the uapi stabilizes and only intended to be built for testing purposes it cannot and has never worked in this state and actively broke things when enabled to this archlinks arm maintainer please consult with this before merging random ushi stuff from Clueless menjaro Developers thanks and I can understand why Hector is really annoyed with Manjaro like this is not the first time that Manjaro has shipped things or tried to patch things that involved his project and were completely broken why is it so difficult if you're not in a position to test things involving Asahi shooting Hector a DM asking hey is it all right if I do this or this is it all right if I ship this version is it all right if I add this patch or that patch just send Hector a DM and he'll respond pretty quickly like Hector's DMS are open and he'll get back to you especially if you're involved in a project like Manjaro arm and just that little bit of effort is going to ensure that 99 of these issues just vanish because what he's gonna say is ship the thing that we know works on usahi and don't ask about anything else also as of recording this the patch has been reverted and it's no longer an issue anymore now I know in these videos it can come off as if I'm saying that everybody involved in Manjaro is incompetent nobody should use Manjaro it's just a waste of a project but there are people in the project who are good actors who are trying to make things better for example this person that was responding to Hector and Alyssa hello Markin I am part of the Manjaro arm team I am really sorry for what the team has been doing with apple silicon support I've been following us he chats and I'm also testing it locally I'll discuss this mistake with the team and see how we can avoid such major mistakes but honestly it seems like Hector is just done with this and has no interest in dealing with Manjaro heck I'd argue that bad actors in the open source collaboration space like Manjaro should be ignored when it comes to stuff like this because it's not everyone else's job to work around the explosions they cause but let me know your thoughts do you think it's that big of a deal do you think Hector is overreacting do you think this makes perfect sense and yeah stop messing with this project just do the thing that he's already said works and stop trying to change that I would love to know so let me know your thoughts and comment section down below if you liked the video go and like the video and if you really like the video and you want to become one of these amazing people over here check out the patreon scrubs alibera pay linked down below that's gonna be it for me and um uh I don't have an outro bye [Applause] [Music] foreign [Music]