Warning, Qemu 4.1 has data corruption bug(s?)

Found out about this here


Basically if you use or were hoping to use qemu 4.1, consider not doing so yet. Also check your vm’s for corruption. I’m not really clear about the specifics of whats going on, but figured a lot of you may want a heads up.


Bump for visibility. I may bump it again this evening and then leave it at that.

Cursory reading seems to to be a QCOW2 bug, not Qemu as a whole. Doesn’t seem RAW formatted disks are affected.

Worth noting, 4.0.0 doesn’t seem to be affected either, so downgrading is a safe option.

Also seems to be two closely related, but different bugs in play here. One in the 4.1 Qemu release, and another in the XFS filesystem. The Qemu bug affects all qcow images, regardless of filesystem, but XFS users get a double-whammy.

Looks like the fix has been applied to both XFS and Qemu upstream, and will be released in the 4.1.1 release happening Soon™.

Now that I’ve had some time to look at what’s going on, you have a fantastic summary.

While I haven’t been bit (using raw), what’s a bit disappointing are that these dataloss bugs are still “Importance:Undecided”, a month after discovery for one. Maby I’m just spoiled by the urgency which the ZFS team responds to this sort of thing.

Huh, looks like the fix commit was proposed 04/11/19, according to the last post of the first issue tracker link?