Correcting the above: in terms of LTS kernels it's 5.10+, but the 2020 commit came in Linux 5.6 (so would affect for instance some Ubuntu 20.04 kernels as well on 5.8)
Exploits are now appearing targeting pidfd, which is forced into all Linux kernels since 5.10 (2020), no module or initcall to blacklist this time, must patch ASAP!
We've just sent a detailed mail to all customers notifying them of this issue, with split-out fixes available for 5.15, 6.6, and 6.18. We'll share more information on our Knowledge Base as it becomes available.
We've just published a Knowledge Base article with more information about the vulnerability, current published/unpublished exploits, and current mitigations. We still recommend patching ASAP.
Exploits are now appearing targeting pidfd, which is forced into all Linux kernels since 5.10 (2020), no module or initcall to blacklist this time, must patch ASAP!
We don't recommend the '3' setting (which rejects all attach-like ptrace_may_access() requests, regardless of privilege level) as once set, it is locked to that value until reboot.
As mentioned at https://t.co/Te5MKCkBu8 , distro users can echo 2 > /proc/sys/kernel/yama/ptrace_scope to mitigate. Keep in mind this will break some normal usage, like preventing unprivileged use of strace, gdb, etc entirely.
We've just sent a detailed mail to all customers notifying them of this issue, with split-out fixes available for 5.15, 6.6, and 6.18. We'll share more information on our Knowledge Base as it becomes available.
We've uploaded new patches for 5.15, 6.6, and 6.18 to address an obfuscated upstream Linux logic vulnerability that should exist in all kernel versions: https://t.co/qdnNlRzRTf
The commit message makes no mention of it, but we believe the goal of exploiting the vulnerability would be to target a suid root binary that drops its privileges while retaining privileged access to a file on exit that via the vuln an unprivileged user could gain access to.
We've uploaded new patches for 5.15, 6.6, and 6.18 to address an obfuscated upstream Linux logic vulnerability that should exist in all kernel versions: https://t.co/qdnNlRzRTf
We've published a detailed KB article for customers on the two vulnerabilities involved in Dirty Frag and the associated public exploits, feel free to reach out with any questions.
The upstream rxrpc vulnerability affects >= 6.5. The second vuln reusing the same "Dirty Frag" naming requires unprivileged userns to exploit as an unprivileged user, disabled in grsecurity since it first existed.
Fixes for Dirty Frag (https://t.co/K9PbJrmenw) were already included in our current 6.6 and 6.18 patches from May 4th. No distro builds in CONFIG_AF_RXRPC, so MODHARDEN should be generally effective against unprivileged exploitation.
We'll have updated 5.15 and 6.6 patches available shortly (despite 5.15 being EOL) along with split-out patches available for both 5.15 and 6.6 for those on older kernels who need CONFIG_CRYPTO_USER_API_AEAD enabled (which shouldn't be anyone)
We'll have updated 5.15 and 6.6 patches available shortly (despite 5.15 being EOL) along with split-out patches available for both 5.15 and 6.6 for those on older kernels who need CONFIG_CRYPTO_USER_API_AEAD enabled (which shouldn't be anyone)
For RHEL/RHEL-derived configurations, this approach will work (the function name has been stable since 2015 and initcall_blacklist has been supported since 2014): https://t.co/DJh67yMqh0
Creating a separate post so more people see this: the mitigation recommended by https://t.co/e2Cwqzet1X for https://t.co/1DsFYvMk41 *WILL NOT WORK* for any RHEL or RHEL-derived distro, including CentOS, Fedora, Oracle, and Alma as the vulnerable code is built-in.
Don't rely on mitigation suggestions from others floating around: non-readable suid binaries only does something for this specific exploit, it's exploitable in other ways that don't involve suid binaries at all.