shoutouts to msi, firmware developer knows best

[123689.084271] Buffer I/O error on device dm-0, logical block 97262419
[123689.084288] Aborting journal on device dm-0-8.
[123689.084296] EXT4-fs error (device dm-0): ext4_journal_check_start:87: comm syslogd: Detected aborted journal
[123689.084305] EXT4-fs (dm-0): Delayed block allocation failed for inode 20863169 at logical offset 0 with max blocks 3 with error 30
[123689.084307] EXT4-fs (dm-0): This should not happen!! Data will be lost
[123689.084307] 
[123689.084310] EXT4-fs error (device dm-0) in ext4_do_writepages:2944: Journal has aborted
[123689.084312] EXT4-fs (dm-0): Delayed block allocation failed for inode 22707117 at logical offset 0 with max blocks 1 with error 30
[123689.084316] EXT4-fs (dm-0): This should not happen!! Data will be lost
[123689.084316] 

you know the evening is going great when dmesg starts screaming at you

@domi I see an ext4 moment! It's been so long!!!

@toast @domi ext4 has always been rock solid for me

@fun @domi I've seen it do really funny things, though that *was* like a decade ago

@toast @domi hmm, makes me feel better now :) I'm quite careful about new filesystems until they have been proven to be rock solid. Ext4 *was* experimental back in its inception, but with a lot of time it became something that is extremely solid.

The same will happen to btrfs and such, just give it time.

@fun @domi ext4 wasn't considered experimental in the mid 2010s ^^;;
main btrfs issue I've seen has been the RAID56 write hole which is just how it be tbh
most recently I've surprisingly mostly had issues with xfs (??)

@fun @domi regardless in this case it's dmraid complaining about stuff so indeed most likely SATA controller or similar; I just had NEURON ACTIVATION

@toast sadly, this moment was sponsored by

[123528.741804] nvme nvme0: I/O tag 659 (c293) opcode 0x1 (I/O Cmd) QID 10 timeout, aborting req_op:WRITE(1) size:53248
[123559.002675] nvme nvme0: I/O tag 512 (0200) opcode 0x1 (I/O Cmd) QID 6 timeout, reset controller
[123621.988455] nvme nvme0: Abort status: 0x371
[123621.988460] nvme nvme0: Abort status: 0x371
[123621.988460] nvme nvme0: Abort status: 0x371
[123621.988461] nvme nvme0: Abort status: 0x371
[123621.988462] nvme nvme0: Abort status: 0x371
[123622.002954] nvme nvme0: 12/0/0 default/read/poll queues
[123622.065394] nvme nvme0: Ignoring bogus Namespace Identifiers
[123689.044224] nvme nvme0: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0x10
[123689.044229] nvme nvme0: Does your device have a faulty power saving mode enabled?
[123689.044230] nvme nvme0: Try "nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off" and report a bug
[123689.076225] nvme 0000:01:00.0: enabling device (0000 -> 0002)
[123689.076302] nvme nvme0: Disabling device after reset failure: -19

resident Firmware Expert™ told me to just pcie_aspm=off and carry on, i trust her

@domi oh lol I love the controller not coming back up (pcie_aspm=off sounds about right to me)

@toast @domi > ext4 wasn't considered experimental in the mid 2010s

What I mean is that I tend to trust filesystems that have been around for a long time and battle tested for years

@fun @toast FAT12 /j

@domi @toast and with actual practical use today :^)

FAT32 is too limited and so is FAT16 (yes, I know FAT32 supports up to ~2 TB)

@fun @domi UFS

@fun @domi 1

@fun @toast i wonder if anyone has added iso9660 r/w support to the kernel. you know, boot from CD-R, commit your files in sessions…

@fun @toast ah, i guess that would fall into impractical. oh well, best I can do is XFS

@domi @fun I would have unironically done this around the timeframe I had the funny ext4 failures! sounds fun

@domi dmesg &> /dev/null

What problem?

@toast @domi

@elly @domi @toast it's over that's MacBook coded

@fluttersh @domi @toast MacBook-coded?

@elly @domi @toast I recently got a 2017 or 2016 MacBook and I had to disable suspend too even needed special speaker drivers and camera drivers 😩

@elly @domi @toast also the battery reporting was neva accurate

@fluttersh @domi @toast S3 != ASPM. I don't remember having to do either.

Of course Intel-based MacBooks also need special drivers, Apple did a fair bit of custom stuff with T1 and T2.
I tried to run Linux on 2019 16in MacBook Pro circa 2022, things mostly worked but you basically has to maintain your own kernel (which was a fairly big no-no in my employer's book so I gave up on that).
Worst part was that you could easily fry your speakers and Broadcom WiFi couldn't connect to 5GHz networks.

Not sure if anyone made it work in upstream since then, but I don't think it would be difficult to make Linux run on those machines at all

@elly @domi @toast works great for QCNFA222 (that spams dmesg with ASPM errors and also has broken 5GHz)

@elly @domi @toast it's cooked fr, I wish someone sold an adapter so I could put a real wifi card in ze MacBook also no hw accel fractional scaling is double cooked

@fun @domi @toast You can get FAT32 up to 16 TB if you use disks with 4k sectors :)

@jernej__s @domi @toast omw to source 16 TB qualcomm UFS modules to do that (these use 4k sectors)

@fun @domi @toast Many 3.5″ USB enclosures will present the drive as having 4k sectors for some reason; a lot of Seagate drives can be reformatted to 4k sectors by using SeaChest tools, and similarly a lot of NVMe drives can be formatted to 4k sectors (note: don't do this with DRAM-less WD NVMe drives, because the firmware is buggy and will start crashing your machine).

@jernej__s @domi @toast but sourcing 16 TB qualcomm UFS modules is more fun