current kernel from apt repo (6.16.8-edge-sunxi64) still vulnerable to local privilege escalation bug CVE-2026-31431 "copy.fail"

[akabulous]

Steps to repeat the bug:

1) use the cross platform PoC written in C, the Python one that everyone is sharing contains obfuscated code (bad ju-ju) and is x86_64 specific

`git clone https://github.com/tgies/copy-fail-c`

`cd copy-fail-c`

2) compile either on your target device natively, or do what I did and cross-compile it as a static binary using an aarch64-linux-musl toolchain (this made it easy to test on different SBCs)

`PREFIX="/opt/toolchains/aarch64-linux-musl-cross" CC=aarch64-linux-musl-gcc LD=aarch64-linux-musl-ld CFLAGS="-static -fPIC -I/opt/toolchains/aarch64-linux-musl-cross/include -L/opt/toolchains/aarch64-linux-musl-cross/lib" LDFLAGS="-static -fPIE -L/opt/toolchains/aarch64-linux-musl-cross/lib" make -j$(nproc --all)`

3) pass the resulting binaries "payload" and "exploit" to your target device (if you cross compiled)

4) from an unprivileged user account not in the sudo group, run the exploit

 

I'm not here to point fingers but I would like to see AT LEAST an advisory of this potentially devastating bug with a public exploit available on the Armbian homepage, radio silence for over a week seems completely inappropriate to me

[Igor]

59 minutes ago, akabulous said:

radio silence for over a week seems completely inappropriate

 

We understand the concern, and we appreciate the effort put into testing and documenting the issue. At the same time, it is important to understand the realities of the embedded Linux ecosystem. Armbian supports a very large combination of SoCs, vendor kernels, boot chains, and downstream modifications across several hundred boards. Security response and validation in this environment is significantly more complex than in standardized desktop/server distributions.

Explained here:
https://github.com/armbian/build/issues/6937#issuecomment-4366571379

 

This is not a matter of ignoring the issue, but of limited engineering resources, kernel fragmentation, and the high cost of validating fixes safely across multiple platforms. Project can only finance security from your contributions https://github.com/sponsors/armbian volonteers or sponsors. Until none is taking this seriously, there is little what existing team members can do.

 

We already attempted mitigation work on one of the most widely used kernel branches:

https://github.com/armbian/linux-rockchip/pull/475 but even targeted fixes require substantial testing effort and may (i am sure it will) introduce regressions on affected hardware families.

 

Current resources barely sustain even our regular release and maintenance process:
https://docs.armbian.com/Process_Release-Model/

 

For users who need receiving upstream fixes faster and are willing to accept a higher risk of regressions on hardware feature breakage, there is always an option to switch to rolling/daily builds, where fix may already be available:
https://docs.armbian.com/User-Guide_Armbian-Config/System/#rolling

Tradeoff between stability, validation cost, hardware compatibility, and update speed is unfortunately a sad reality of embedded Linux maintenance. 

 

[akabulous]

I understand that you're overworked and underpaid. This is free software, so I won't demand more labor from any of you. That said, I think it would build trust to at least have a little temporary pop-up (not the right word but I dont know what it's called) on the homepage that says something like "Current kernels may be vulnerable to this bug, we're working to resolve this, here are some relevant links" and then post the kernel dot org patch for this bug and point people to the Armbian build system. Surely the amount of effort it would take to do that is equal to or even less than it took to give me such a thorough response (which I appreciate btw). 

[bedna]

This, even though serious, it's "only" a privilege escalation, not remote code execution (RCE), so if someone were to use this on you, they would first need to have physical access OR access remotely (RCE) to be able to run the exploit and escalate to root. And if you have a malware with RCE, you are already f**ed.

 

In time, I'm pretty sure upstream patches will spread down to armbian builds too.

 

In the meantime if you think someone might be able to get access and escalate, you can disable algif_aead (or try to apply patches yourself) as mentioned in https://xint.io/blog/copy-fail-linux-distributions

 

Quote

Remediation

Patch the kernel. The fix reverts AF_ALG AEAD to out-of-place operation, eliminating page cache pages from the writable scatterlist.

Update your distribution's kernel package. Major distributions should ship the fix through normal kernel package updates.

 

For immediate mitigation, block AF_ALG socket creation via seccomp or blacklist the algif_aead module:

echo "install algif_aead /bin/false" > /etc/modprobe.d/disable-algif-aead.conf rmmod algif_aead 2>/dev/null

 

Another solution is for you to pay a developer to do changes/testing needed

If you don't "demand more labor", then why are you requesting more labor? (don't answer, it's a rhetorical question)

Edited 5 hours ago by bedna

[akabulous]

Really disappointing that both responses to the original post (although the first was at least polite) are basically "f*ck you, pay me" 🙄

 

I've already mitigated this on my own devices, my concern is for other users of this distribution.

 

And yeah, LPE isn't RCE, but it still deserves an advisory. Consider how easy it would be for an attacker to embed this exploit in a malicious file download + god knows what kind of payload and turn your computer into a zombie in a botnet, a cryptominer, hit you with ransomware, etc. 

 

When mentioning a real security issue gets a response this crappy, it doesn't bode well for the future of the project. Igor was at least professional. But go ahead, Bedna, tell me more about why the homepage being updated to include an advisory costs money that I should be sending instead of making bug reports 🙄 pathetic 

[eselarm]

4 hours ago, akabulous said:

And yeah, LPE isn't RCE, but it still deserves an advisory. Consider how easy it would be for an attacker to embed this exploit in a malicious file download + god knows what kind of payload and turn your computer into a zombie in a botnet, a cryptominer, hit you with ransomware, etc. 

You know there is a fruit company in Cambridge UK that is proud to tell they have sold 75 million pieces and 75% is to non-noobs. They provide passwordless sudo by default and a self-modifying rpi-update script by default in their OS. So why so complicated with this copy-fail exploit, every script kiddy had and has an easy task of keeping the noobs on a leash for more than a decade.