Debian 12.3 Delayed Due To An EXT4 Data Corruption Bug Being Addressed

[phoronix]

Phoronix: Debian 12.3 Delayed Due To An EXT4 Data Corruption Bug Being Addressed

Due to a problematic patch back-ported from Linux 6.5 causing interference between EXT4 and iomap code, there's the possibility of a data corruption bug on older kernels -- most notably recent Linux 6.1 LTS point releases that can currently be found in the likes of Debian 12...

https://www.phoronix.com/news/Debian-123-Delayed-EXT4-Corrupt

[cynic]

where are all those saying that they'll never us ZFS or btrfs because ext4 is so much more reliable? 😬

[user1]

Turns out, when it comes to filesystems, the most reliable setup is NTFS on Windows

But seriously, I know that Windows NT is closed source, but despite the fact that its userland may suck, I don't think the NT kernel suffers from this amount of serious regressions like Linux lately does, especially in filesystems.

[SkyWarrior]

NTFS and windows are both regression themselves.

[OneTimeShot]

I can't comment on these particular bugs, but the reason the NT kernel seems to not suffer from these regressions is that (1) They test the hell out of it, (2) they don't change things constantly - seriously, you can use a 25 year old book and code Windows Kernel stuff.

It's also not true that Windows is hugely compatible, as anyone who has to maintain software on the Windows platform knows - when Microsoft release a new patch, you test your stuff and sometimes you need to fix something.

The difference is that Microsoft actually take responsibility. If an application breaks, they fix it in Windows (detecting your software, emulating bugs, and even replacing libraries). Compare the Linux experience - even libc isn't always backwards compatible. At the moment, everything needs to get rewritten to support GTK4 or Wayland or whatever ffmpeg-wrapping codec plugin system is flavour of the month, etc...

[RealNC]

where are all those saying that they'll never us ZFS or btrfs because ext4 is so much more reliable? 😬

Ext4 is fine, Debian just happened to break it.

[Volta]

where are all those saying that they'll never us ZFS or btrfs because ext4 is so much more reliable? 😬

Here. If you didn't notice ext4 issue is in any way as serious as zfs. Furthermore, it's bad backport not ext4 bug.

[mrg666]

where are all those saying that they'll never us ZFS or btrfs because ext4 is so much more reliable? 😬

At least read the first sentence before posting. This is not an ext4 bug; it is a backport mishap. My ext4 on 6.6.5 kernel is fine.

[stargeizer]

It seems You people don't know how to read properly, or don't know the "backport" concept (too much Windows users here maybe??).

If the concept is alien, allow me to explain a little: when bugs are fixed in the lastest and greatest versión of the kernel, depending on how usefull is the fix, it gets backported to older, but supported LTS versions of the kernel. The problem HERE is not the backport itself, but the patch is not fixing "as is", probably since the difference between the Debían prepared kernel and the mainland kernel is just too much in this área, so the patch must be remade, reapplied and retested as long is required to be declared "stable". It has nothing to do with reliability of certain file systems. Bugs are bugs and are to be fixed. No matter the os, or the file systems involved. It just people cares to backport fixes to the Long Term Supported versions of the Linux kernel.

Microsoft never backported any fixes of their FS of their OS, since is expected that you are using their lastest offering (and nowadays FORCING You to do so). NTFS only had 2 or3 breaking changes at the time of Windows NT 4.0 and since then, they only added support for features at the OS level, rather than really fix the underlying FS. That's why you can read an older NTFS formatted drive with Windows 10, but You cannot reliably read a Windows 10 NTFS formatted drive in Windows 2000/XP pre sp 2, that have features and fixes done in newer systems.

[Volta]

I can't comment on these particular bugs, but the reason the NT kernel seems to not suffer from these regressions is that (1) They test the hell out of it, (2) they don't change things constantly - seriously, you can use a 25 year old book and code Windows Kernel stuff.

It's also not true that Windows is hugely compatible, as anyone who has to maintain software on the Windows platform knows - when Microsoft release a new patch, you test your stuff and sometimes you need to fix something.

The difference is that Microsoft actually take responsibility. If an application breaks, they fix it in Windows (detecting your software, emulating bugs, and even replacing libraries). Compare the Linux experience - even libc isn't always backwards compatible. At the moment, everything needs to get rewritten to support GTK4 or Wayland or whatever ffmpeg-wrapping codec plugin system is flavour of the month, etc...

Nothing is true in your claims. Ms takes no responsibility at all. It's the worst company when comes to dealing with serious bugs and issues. You're not aware of this, because Phoronix don't talk about ms problems.

https://www.bleepingcomputer.com/news/microsoft/windows-10-ntfs-corruption-bug-gets-unofficial-temporary-fix/

It is unknown when Microsoft plans to fix this bug, so if you are concerned threat actors could abuse it on your computer, this is a good alternative while you wait.

Yeah, very responsible company.