[HN Gopher] Asahi Linux goes from Apple Silicon port project to ...
___________________________________________________________________
Asahi Linux goes from Apple Silicon port project to macOS bug
hunters
Author : LinuxBender
Score : 245 points
Date : 2023-11-01 17:07 UTC (5 hours ago)
(HTM) web link (www.theregister.com)
(TXT) w3m dump (www.theregister.com)
| JohnMakin wrote:
| Apple Pro M1 16-inch User here.
|
| I went to restart my machine (was plugged into a lightning hub
| with 3 monitors, I don't ever recall messing with refresh rates)
| and it went to update. At some point in the loading bar, the
| screen went black and then flashed an apple logo. When I pressed
| power, I could sometimes get the apple logo to display on boot,
| then it went dark. To me it looked like it wasn't booting at all
| but the machine is pretty quiet even when it's on so it was hard
| to tell what was happening.
|
| We went to do the firmware revival process using another mac and
| it would barely recognize the device depending on what we tried.
| Eventually we were offered the option to do a full system
| restore, and after many hours battling this, we decided to just
| lose the data on the machine.
|
| Luckily I keep most major stuff in cloud or in github, but I lost
| a significant amount of work from the previous 24 hours that
| ended up delaying a release and causing significant pain. For how
| much these machines cost, this is completely unacceptable. I know
| windows machines have their fair share of issues, but in 15 years
| of working on various windows and mac workstations this is the
| only time I've lost an entire drive.
| syx wrote:
| This is insane, was there additional software installed by your
| employer that might have conflicted with the updating process?
| In my case we have lots of remotely controlled macOS updates
| that get triggered and supervised by the company I work for.
| ljm wrote:
| I wouldn't be surprised if MDM policies caused some funky
| shit to happen. I couldn't use the migration assistant on one
| machine because the policy blocked the recovery OS; it just
| silently failed and booted you back up into your main
| account.
| striking wrote:
| This is a now well-known bug having to do with multi-monitor
| setups or changing the refresh rate of the internal panel.
| MDM policies and other software couldn't write to NVRAM in a
| way that would cause this issue without the help of a bug in
| Apple's code.
| selimnairb wrote:
| This sucks, sorry. Things like this are why I tend to wait
| until downtime (usually Christmas holiday) to upgrade to the
| latest macOS each year. This gives time for bugs to be
| discovered and fixed, and if there is a major problem, I have
| time to fix it.
| tambourine_man wrote:
| And, if you can, always wait until X.1 or X.2 at least to
| upgrade.
| MagerValp wrote:
| Me I just yolo the beta on my daily driver right after the
| wwdc keynote. If there are any issues I prefer to know
| straight away. Surprisingly I've never run into any major
| issues, only minor glitches.
| striking wrote:
| Hi, in case you face these challenges again, put the target
| system into DFU mode and connect the configuring system using
| the appropriate USB-C port (only one will work).
|
| The DFU key combination is finicky for portable machines:
| connect a charger (preferably Magsafe so you can watch the
| power LED) and your configuring system, press and hold power to
| be sure the system is off (if doing this turned the system on,
| repeat this step), press left ctrl+option and right shift at
| the same time as the power button, count ten seconds, let go of
| everything but power until the device shows up as "DFU" in
| Configurator (you may be prompted to allow more accessories to
| connect to your configuring system before it does).
|
| If asked to perform a software update before/during reviving,
| choose "Quit and update" and start the process again. If you
| upgraded to 13.6 or 12.5 before facing these issues, you may
| have to enter recoveryOS instead of booting normally and
| perform a system upgrade to Sonoma.
|
| If done correctly (without choosing Restore), you will not lose
| data. If you can't do these steps yourself or think you will
| have trouble walking a family member through them, the Apple
| Store can do a revive for you (be explicit that they are only
| to revive the machine, not restore or replace).
|
| Full details at https://support.apple.com/guide/apple-
| configurator-mac/reviv...
| ShadowBanThis01 wrote:
| My 16" M1 MBP was bricked by the Sonoma update and had to be
| wiped by Apple. Fortunately, I keep most things on external
| drives. But it's still an egregious QA failure on Apple's part,
| as witnessed by the increasing number of reports.
|
| And Sonoma appears to be riddled with odd little bugs,
| manifesting as UI failures. Windows don't come up, or they come
| up behind others when previously they didn't. For example, disk
| images, when mounted, open a window as usual but that window
| appears behind others on the desktop so you're left wondering
| what happened.
|
| I just had keyboard input rejected with "boink" noises in a
| dialog raised by Safari, over and over. In the end I couldn't
| log into the site.
|
| The "About this Mac" item in the Apple menu did nothing, over
| and over...
|
| and sure enough, I just tried this, and it happened again. But
| when I started Screenshot to capture it and file a bug, it
| started working. This is the second day in a row this has
| happened.
|
| It's looking like a shitty Mac OS release.
| baq wrote:
| I feel like I live in a parallel universe sometimes when I
| complain about things randomly breaking or never actually
| working and they say 'I've been using MacBooks for the past
| ten years and literally nothing bad happened ever' and here I
| am just about to break the one year milestone and I just
| don't see the mythical Apple level of quality.
| danieldk wrote:
| The thing is that Apple devices are so widely used, that
| even if 0.1% of the users encounters an issue, it's still a
| huge number of people. Added to that, people who run into
| issues are more likely to complain online than people who
| don't. So, you don't live in a parallel universe, you're
| just one of the unlucky 0.1%.
|
| And yes, I have been using Macs for 16 years and have never
| had a catastrophic issue and neither one of my friends of
| family. The worst problem was kinda self-inflicted, pre-SIP
| I was once typing something along the lines onf _sudo rm
| -rf /Library_ and then accidentally pressed enter. Yay for
| backups.
| threeseed wrote:
| > It's looking like a shitty Mac OS release
|
| No this happens _every_ Mac OS release for the last 20+
| years.
|
| Sonoma has literally just been released and we only recently
| had the first major patch. So there will be bugs both known
| and unknown. Almost all by definition not fixed.
|
| As always if you are doing mission critical work then for any
| piece of software you should always hold off upgrading until
| at least a few major patches have been released.
| JohnMakin wrote:
| This bug affects Ventura patch as well though.
| tambourine_man wrote:
| >My 16" M1 MB was bricked by the Sonoma update and had to be
| wiped by Apple.
|
| If it could be restored by wiping the SSD, it's not bricking.
| It's terrible, for sure, but bricking means your device turns
| into one and you might as well throw it away. If it's
| recoverable via software, it's not bricked.
| Wowfunhappy wrote:
| But what if the SSD contains critical boot components which
| can't be recovered without the aid of a secondary laptop?
|
| There is no scientific definition of "bricking", and I
| agree this is a bit of a gray area, but I think it's fair
| to use the term.
| jbverschoor wrote:
| Oh, I can reasonably consistently kernel panic sonoma on m1,
| simply by using fullscreen zoom in accessibility and
| streaming video.
|
| I think that's my last bug report, as none of my feedback/bug
| reports, which are very detailed, and some cause data
| loss/corruption.
|
| 0 feedback, 0 fixes. Total waste of time.
|
| I wonder where their software quality and consistency team
| went to. If you'd release similar buggy software when the
| AppStore on iPhone became available, your app would simply
| not pass the quality criteria.
|
| Debian looks awfully nice these days, especially combined
| with proxmox / passthrough devices
| wlll wrote:
| > Windows ... come up behind others when previously they
| didn't.
|
| m1 16" MBP here, this happened to me on the last OS version
| too. Still happens on Sonoma.
|
| For about 1 year (a few years back) I had an issue where on
| my three external Samsung monitors (all in portrait
| orientation) there was a strip about 4cm all across the
| bottom of all of them, and up the right side of the right
| monitor where the mouse wouldn't go. Apple couldn't work it
| out, spent hours talking to them about it. In the end an OS
| update fixed it.
| marci wrote:
| If only they could find a bug/vuln that makes it possible to
| bypass the "Dead/Corrupted SSD? Dead Machine feature" and make it
| possible to boot externally, no matter the state of the internal
| drive.
| nwellinghoff wrote:
| Wait, so you can't recover data off a drive if you analyze it
| from another machine? Surely this is only if the drive in
| encrypted right?
| watermelon0 wrote:
| > If you have a Mac with Apple silicon or an Apple T2
| Security Chip, your data is encrypted automatically.
|
| Even if you don't have FileVault (macOS disk encryption)
| enabled, data on SSD (or NAND chips with Apple silicon) is
| encrypted, so it's quite unlikely that you would be able to
| recover anything.
|
| Not to mention that SSD/chips are soldered on the
| motherboard.
| vanchor3 wrote:
| If the computer still turns on you can at least get to
| recovery and use USB-C/Thunderbolt disk sharing.
| samtheprogram wrote:
| This thread is referencing if the SSD fails. If the SSD
| fails, the computer won't turn on, and even if it could
| (which we wish were the case, but the T2 chip prevents),
| you couldn't use get anything off the disk because it's a
| dead SSD.
|
| In the event of an OS crash/issue, or just minor
| corruption, that might be an option. Related to the
| original article -- I'm not sure what is meant by DFU
| mode on a MacBook (only have heard of that on iDevices)
| so it's unclear to me if you can EFI boot into another
| OS. Given that it also affects Asahi Linux, sounds like
| you can't EFI boot into anything, even OS Recovery for
| Mac Sharing mode.
| Moto7451 wrote:
| The architecture of Apple Silicon Macs is shared with
| iPhones and iPads and there isn't an EFI environment. DFU
| mode on a Mac works the same as an iPhone and lets you
| restore from another computer.
|
| https://support.apple.com/guide/apple-configurator-
| mac/reviv...
| JonathonW wrote:
| There is no EFI on Apple Silicon. It's a more iOS-ish
| pre-boot environment (based around iBoot and some other
| Apple-y things).
|
| There's also not exactly a boot picker in firmware-- that
| lives in the recovery OS. Which will _also_ be unbootable
| if you 're impacted by this issue because it's a minimal
| macOS environment; if your only OS is 13.6/13.6.1, that
| recovery partition will also be 13.6/13.6.1, and, if you
| have upgraded to 14.0/14.1, that installer can fail to
| update the recovery partition and leave it at 13.x
| (whatever you were on when you upgraded).
| umanwizard wrote:
| In this case what is the point of FileVault?
| MagerValp wrote:
| The data is encrypted at rest regardless, but if
| FileVault isn't enabled it unlocks without a password
| when you power on.
| unforgivenpasta wrote:
| Much better than the OSnews article that related Asahi[1] to the
| bug and when called out by marcan they stated that the headline
| would stay [2].
|
| The article seems to have been deleted[3] at some point since the
| link from marcan goes to a 404
|
| [1] https://social.treehouse.systems/@marcan/111334488235016591
|
| [2] https://mstdn.social/@osnews/111334720022394898
|
| [3] https://www.osnews.com/story/137678/apples-macos-sonoma-
| make...
| mekster wrote:
| OSNews author seems to have something personal against Apple,
| it reads like a personal blog when it comes to Apple related
| news with completely biased writing and I stopped reading years
| ago.
| SllX wrote:
| I mean at this point I'm pretty sure it is a personal blog
| but with an archive of OS-centric news from back when there
| was more variety in the field. But yeah, quit reading for the
| same reason. It's one thing to be biased against a company, I
| have my biases too, but this was on another level.
| Reason077 wrote:
| Looks like the OSNews article has been removed now. I just get
| a 404.
| diffuse_l wrote:
| The author did say he's removing the article for now, and
| he'll upload a fixed article later [1].
|
| [1] https://mstdn.social/@osnews/111335189028196715
| mise_en_place wrote:
| This is a huge embarrassment for Apple, because a team of unpaid
| volunteers are superior to the supposed talent you are shelling
| out multiples of six figures to.
| geodel wrote:
| Oh yes, multiple six figures have always caught bug every
| single time at every other place than Apple.
| extrememacaroni wrote:
| Bricking your own devices with your own update is extremely
| amateurish and embarrassing. If Apple knew about shame that
| is.
| saagarjha wrote:
| Ok, let's see your processes for shipping an update to a
| hundred million devices.
| cromka wrote:
| This wasn't an issue in their rollout process, but in
| their QA that needs to test orders of magnitude less
| setups.
| saagarjha wrote:
| Presumably they'd need to test more? Because I assume
| that whatever they tested passed QA and didn't include
| what Asahi is reporting.
| s3p wrote:
| Semantics. Seems to me like OP was talking about the
| entire project of making and releasing the newest
| version. QA would most definitely be apart of that.
| azinman2 wrote:
| *billion
| saagarjha wrote:
| Definitely not that many Macs in use
| tambourine_man wrote:
| It's not bricking.
| robocat wrote:
| By that definition, neither is bricking bricking.
|
| You can often use technical solutions (e.g. JTAG) to fix
| "bricked" devices.
| gary_0 wrote:
| If you own a soldering iron, nothing is bricked until the
| magic smoke gets out!
| xnzakg wrote:
| I would say it's soft-bricking, but not hard-bricking.
| icedchai wrote:
| The average user is going to have a heck of a time fixing
| it. They may not have a spare Mac, or a friend with
| enough tech know-how to help them. It is effectively
| bricked since they'd have to go to an Apple Store.
| gosub100 wrote:
| If you were talking about the goof troop at Microsoft that is
| openly user-hostile, I'd agree. Their users are essentially
| captive. But Apple's fetish for design and user-harmony is
| what makes this so absurd.
| emily-c wrote:
| I hope this is a joke.
| TimTheTinker wrote:
| This unpaid team of volunteers is composed of some of the best
| and most productive hackers in the world.
| Cyph0n wrote:
| If there ever was a team of 10x OS devs, this would
| definitely be one of them.
| solardev wrote:
| I look at some of this code and can't imagine even being a
| 0.1x OS dev. How do people learn this stuff? It's
| incredible.
|
| I'm just glad and grateful there's super smart people out
| there volunteering their time on worthwhile causes.
| viraptor wrote:
| Every specialisation is specialisation. It's just that
| more people are into (for example) web apps than OS
| development. If you spend some time in embedded dev land,
| things like that make sense. You just get better at
| thinking at hardware level / synchronising devices, etc.
|
| I mean, they're still super clever, but people learn that
| stuff by reading about it and practicing. You could go
| that way too, it's not black magic.
| omgmajk wrote:
| > it's not black magic.
|
| Then why do I have "black magic" on my LinkedIn
| profile???
| robertlagrant wrote:
| My guess is you invented the ATEM Mini.
| solardev wrote:
| Sure seems like it to me, lol.
|
| I learned web dev first by using Geocities and Frontpage,
| then Dreamweaver. These days I actually get paid to write
| Javascript (something young me would've found
| unbelievable enough already), but to me that's still a
| huge difference from "real" programming, especially in C
| or assembly.
|
| Web dev (at least the kind I do) is mostly still just
| declaring UI in a XML like syntax, then wrapping it up in
| some events and state management. It's not that different
| from, say, Visual Basic. At its core it's a bunch of
| event driven reactions and API calls based on UI clicks.
| Once upon time that was Perl or PHP or jQuery, now it's
| usually React, but the fundamental process of "declare
| UI, add event handlers, send to APIs" hasn't changed much
| IMO.
|
| But an OS? Man, how do you work with memory access?
| Overflows? Garbage collection? Device drivers? Caching?
| Graphics pipelines? USB? Lightning? I don't even know
| where to start with any of that, and an OS is ALL of that
| and then some.
|
| Mad props, is all I'm sayin'.
| junon wrote:
| osdev.org and lots and lots of reading code and manuals.
| danieldk wrote:
| I don't understand why they don't just pay them to work on
| Asahi and then they can learn directly from all their
| findings?
| stuaxo wrote:
| Apple would want them to sign an NDA and that would be
| incompatible with them doing as much as they do in Linux.
| danieldk wrote:
| Apple can decide to not let them sign an NDA and just pay
| them for the excellent work that they are doing now and
| give them a good priority channel for reporting issues.
| dmonitor wrote:
| why would apple pay them to do something they already do
| for free?
| danShumway wrote:
| I very much doubt the math for the PR-to-cost ratio works
| out to justify hiring them, but presumably the reason
| would be so that Apple wouldn't have to see articles like
| this or to have the top comment on tech threads be "this
| is a huge embarrassment for Apple".
|
| Similar reason as to why companies have bug bounties;
| they want to incentivize hackers to report bugs through
| official channels early enough in development that they
| can get patched before release and before tech
| journalists write articles about them. They don't want to
| find out that their products have bugs via social media.
| Even if that process happens out in the open via a Github
| issue, getting giant problems like this caught before
| release and quickly escalated internally through official
| channels would go a long way towards mitigating article
| titles like this.
|
| Having said that, does Apple care about the Register or
| HN? Probably not? And assuming that Apple did care about
| bad PR among extreme power-users, would Asahi Linux want
| to be paid by Apple to do testing on their releases?
| That's also not necessarily a given, the team would have
| to decide if they wanted to have a more official
| relationship with the company or not.
| threeseed wrote:
| Because having worked at Apple and seen Radar.
|
| The issue has never been finding the bugs. It's having the
| capacity to fix the bugs.
|
| And when you are this early in the OS release cycle (we are
| only at 14.1) there are inevitably a lot of P1 issues that
| limited resources are competing to fix.
| Wowfunhappy wrote:
| But I have to imagine an issue that renders Macs
| unbootable would have been considered blocking, had Apple
| known about it.
| threeseed wrote:
| It's an edge case with a workaround. P2 and definitely
| not enough to prevent a release.
|
| Only 14/16-inch models where users have explicitly
| disable ProMotion and are desperate to boot into the
| RecoveryOS.
| Wowfunhappy wrote:
| My understanding was that if you disable ProMotion you
| can become unable to boot into your _main_ OS.
| Twisell wrote:
| Asahi Lina (one team member) recently got a 6 figures
| paycheck from Apple through their bug bounty program for
| CVE-2022-32947
| https://news.ycombinator.com/item?id=37543664
|
| This is actually the best of both world, they stay
| independents, they get paid, Apple fixe the reported
| exploit.
| AceJohnny2 wrote:
| For one thing, when you're paid, you don't really get to choose
| what you work on, unlike an unpaid volunteer. There are feature
| lists to check off, deadlines to hit, and you work on what
| needs to be done, which isn't even always what you're good at.
| In contrast, the volunteer can choose to spend as much time as
| they want digging into whatever topic they want.
|
| The results differ immensely, and neither approach are perfect.
| ziml77 wrote:
| This is essentially what I was going to say. Being paid can
| actually reduce motivation because deadlines are stressful
| and there's plenty of work to get done that you won't find
| interesting or maybe will just get bored with after the
| initial excitement.
| protastus wrote:
| Big companies are pathologically inefficient and bureaucratic.
|
| A big part of the reason people are paid multiples of six
| figures is to put up with an unhealthy amount of stressful
| bullshit.
| wslh wrote:
| Some incentives work better than money specially for white hat
| hackers.
| MagerValp wrote:
| X
| s3p wrote:
| That logic doesn't make much sense. They didn't even announce
| a release date for the software until recently. Why not push
| it back for QA? It's not like they have to ship everything on
| time (AirPower, original HomePod come to mind)
| tux3 wrote:
| That's pretty embarassing, customers may end up with a soft-
| brick if an organizational problem delays communication about
| a known issue of this severity by over two weeks
|
| Which it seems they did, since Apple's reaction comes after
| the Asahi publication of the bug
|
| There are better ways to communicate about a black screen bug
| on boot than leaving people with the _fait accompli_ , one
| would naively think.
| hiretrw wrote:
| I hired a dev from that team and he was asking for 7 figures.
| It was a steal.
| RicoElectrico wrote:
| This is what you get with the creeping smartphonization of
| general purpose computers. The x86 side doesn't look bright
| either due to MS slowly forcing TPM.
|
| There is a good reason UEFI is stored on a QSPI flash.
| jdiez17 wrote:
| While I hate the gradual death of general purpose computing as
| much as the next nerd, this bug in particular has nothing to do
| with it.
| userbinator wrote:
| It does --- if you consider that this doesn't happen on PCs,
| it only happened to Apple because they believe they own and
| control the whole software/hardware stack.
|
| On a PC, the situation is different. At worst, with bad GPU
| drivers, you can still fallback to good old 640x480 VGA, or a
| slightly better generic unaccelerated VESA framebuffer, and
| be able to troubleshoot from there.
| antoineMoPa wrote:
| > We do not understand how Apple managed to release an OS update
| that, when upgraded to normally, leaves machines unbootable if
| their display refresh rate is not the default. This seems to have
| been a major QA oversight by Apple.
|
| This sounds like a very weird edge case and it's very easy to say
| this sort of statement in retrospect, but I wonder what
| percentage of engineers would think of testing this specific
| combo beforehand.
| leonheld wrote:
| > This sounds like a very weird edge case and it's very easy to
| say this sort of statement in retrospect
|
| As someone who works on embedded devices with a lot of
| experience with Hardware-in-the-loop testing, this is 100% a QA
| oversight.
|
| The way this feature is implemented screams "test it from boot
| to reboot". I wouldn't pick too much on a small company doing
| some basic BSP/application work for some HMI device or
| something like that, but this is _the_ richest company in the
| world with major funds /infrastructure to be better than this.
| antoineMoPa wrote:
| > The way this feature is implemented screams
|
| Do we have info about the implementation somewhere?
| phkahler wrote:
| >> This sounds like a very weird edge case
|
| It's a single setting prior to a standard update. Now I'm not
| saying they should test every combination of settings, or even
| every individual setting. I'm more curious how there is any
| interplay between this setting and the update, and why that has
| any effect. On the surface it smells like a system design issue
| to me - like something is going in there on that shouldn't be.
| Wowfunhappy wrote:
| > I'm more curious how there is any interplay between this
| setting and the update, and why that has any effect. On the
| surface it smells like a system design issue to me - like
| something is going in there on that shouldn't be.
|
| There is (probably) a logic behind it.
|
| https://mstdn.social/@marcan@treehouse.systems/1113296141101.
| ..
|
| > Why [does disabling ProMotion cause this problem]? I can
| tell you why: because Apple _hates_ display modeset flicker,
| and switching modes between ProMotion on /off causes a
| modeset flicker, so of _course_ they made it so that is
| stored in nvram somewhere and applied when the screen is
| turned on during early boot, so when macOS boots it doesn 't
| have to flicker again.
| GTP wrote:
| Maybe it's time for Apple to automatically take a snapshot of the
| system partition before an upgrade? (Does their filesystem
| support snapshots?)
| sillywalk wrote:
| APFS does support snapshots. Not sure how that would interact
| with the Signed System Volume thing.
|
| https://en.wikipedia.org/wiki/Apple_File_System
| Wowfunhappy wrote:
| See also from yesterday:
|
| https://news.ycombinator.com/item?id=38089342
___________________________________________________________________
(page generated 2023-11-01 23:00 UTC)