the SSD thing is almost certainly the controller's manufacturer's fault and not Microsoft - they probably had a workaround to prevent issues and accidentally removed it. I have one of these SSDs and without a kernel parameter to disable low power mode it crashes in Linux, too.
it was the SSD, btw bsky.app/profile/drun...
reported for misinformation 🫡
That's a really interesting insight! It's cool how you understand the technical aspects of SSDs and kernel parameters. Thanks for liking my post, I really appreciate it!
Can we see your ssd
😳
if my ssd breaks the first thing i'll suspect is the ssd, not the os
yeah, it was very fucking painful to figure out what was causing an issue whenever it happened (just random kernel panics and logs would always be unavailable bc the OS thought the SSD was unmounting which is why it kernel panicked) i got lucky *one* time and it output the log to the screen
and i saw the unmount and that's how i figured out that a lot of SSDs, even nicer ones, have buggy as fuck controllers and p much all of the SSDs on that list use the same controllers :)
There's a new thing I think that's breaking SSDs in Windows.
almost certainly the SSD controller's firmware
so AI code is awesome and epic? so all kernel code should be written by ChatGPT?
i need the pancake waffles tweet
it's almost like i'm making that joke intentionally or something
i know u r
i don't think ur that dumb bloomf 😭😭😭
I think youre dumb non-bloomf
nah joke was mid at best, be funnier!
why would i need to write a kernel with chatgpt when chatgpt can be the kernel
the future is now
is it really more likely every phison controller is broken vs the new windows update that broke them is broken? even in the scenario you describe the blame imo is fully on MS for enabling defective non standard compliant hardware and obscuring the issue.
yes, it is actually 😭 and it isn't every phison controller, just a lot of them
they don't properly implement APST, it's firmware level and afaik most of the manufacturers who use those controllers have essentially said it cannot be fixed with firmware. and hey, i'm all for you blaming microsoft for anything. i do not care! they should do a better job at QCing
but the original issue is def a mfg one and while knowing about it is somewhat obscure, it's not an obscure issue at all on its own and is a huge fucking problem for sooooo many SSDs.
(the only reason I am getting involved in this post is bc either way it rly doesn't have anything to do with AI/ceo salary/etc. microsoft is always dogshit at QCing Windows Updates)
That's fair I just think Microsoft should simply say "make working hardware or gtfo. Either comply with industry standards or be unsellable junk that doesn't work on any major OS". Force the issue. Shits bad imo because theyre only incentivized to work on windows, and windows works around em.
also hey wait a second it cant possibly be unfixable at firmware can it? if they dont support it properly why wouldnt they be able to simply stop signaling support for the feature? like thats part of the NVMe spec? for ex wiki.archlinux.org/title/Solid_...
i have no clue if it's actually unfixable or not, i just know that most manufacturers have told people it couldn't be fixed (and i'd assume people like SK Hynix would rather fix the firmware than physically replace the drives when you RMA it)
afaik even if you RMA the drive, at least mine, the new ones (at least 10 months ago) used the same controller firmware/controller so they had the same issue. on the "is it physically possible to fix the controller issue with a firmware update?" idk, phison would be the only one who can answer that
(good questions tho! i didn't mean to say it is literally impossible just that's what vendors have claimed in the past. i hate everyone involved here. this issue shouldn't exist, like the one i have and many others do and i agree - part of the problem IS microsoft)
yeah ran out of space but i mostly meant like "why are mfgrs lying about this"
the only workarounds windows should implement for this kinda shit is a quirk list that bans rejects drives that lie about feature support, but otherwise doesnt crash windows. Use their market share to enforce working interoperable hardware standards linux cant do this because mfgs wont care there
i think microsoft should just follow spec strictly with all of this stuff and let it break and let users be mad about it so mfgs would actually fix these issues when they occur.
jinx
but yeah exactly really this is all downstream of the evil, mis-named, and down right malicious en.wikipedia.org/wiki/Robustn... which i sincerely hate and which is responsible for decades of literal fucking garbage and objectively broken software and standards.
www.pcgamer.com/software/win... also lol phison is literally saying this isn't real now
microsoft should lose an anti trust suit for having workarounds for non standard conforming devices and either have to exclusively work with standard ones, or freely release all driver code for them to ensure interoperability
still MS fault
if Microsoft has de facto standardized divergence from hw standards for years until this moment, it's Microsoft's fault. If they hadnt then manufacturers would've caught it in testing years earlier because windows HAS to work.
