Mitigating Infinity Core (CVE-2026-98260, CVE-2026-98256, CVE-2026-80775, CVE-2026-89754) on Rocky Linux 9, 10, and LTS Variants

Arsalan Zaidi
Customer Support Engineer

Introduction

Infinity Core is a collection of six local privilege escalation (LPE) vulnerabilities in the Linux kernel, disclosed on 2026-10-16. The vulnerabilities target core kernel subsystems (exec, signal handling, futex, and memory management) and can be used for privilege escalation on Linux servers and container escape. Public exploit code is available.

Four of the six CVEs affect CIQ kernel variants:

  • CVE-2026-98260 (CVSS 3.1: 7.8, Important) is a use-after-free in the exec/POSIX timer subsystem. When exec() fails in a non-leader thread, a race with POSIX CPU timer cleanup produces a UAF on the timer structure.
  • CVE-2026-98256 (CVSS 3.1: 7.8, Important) is a sigqueue use-after-free caused by a race between exec() in a non-leader thread and cleanup of the private pending signal queue. The freed signal structure can be reused, leading to arbitrary write.
  • CVE-2026-80775 (CVSS 3.1: 7.0, Important) is a use-after-free in the futex subsystem. Two tasks sharing an mm_struct through nested vfork() calls race to allocate the initial reference counter for FUTEX_PRIVATE_HASH, producing a UAF.
  • CVE-2026-89754 (CVSS 3.1: 7.0, Important) is an out-of-bounds write in the memory management subsystem. Retry state from the page table walker leaks into an upper level, causing an OOB write to the temporary result buffer used by mincore().

Two additional CVEs from the same disclosure do not affect CIQ kernel variants. CVE-2026-97941 (CVSS 5.5, SLUB list corruption causing kernel panic) and CVE-2026-98385 (CVSS 7.0, scheduler cache UAF requiring CONFIG_SCHED_CACHE) affect only upstream stable kernels. The vulnerable code paths were never present in any CIQ or Rocky Linux kernel branch.

Problem

The four affecting CVEs target core kernel subsystems that are compiled into every kernel configuration. Unlike module-based vulnerabilities where the attack surface can be removed by blocking a kernel module, exec, signal handling, futex, and memory management are fundamental kernel facilities that cannot be disabled. Any system running an affected kernel version is vulnerable to local exploitation.

All four exploits require only local unprivileged access. They do not require any special capabilities, user namespaces, or kernel module loads. The attacks run entirely in userspace via standard syscalls. This makes the attack surface broader than module-dependent CVEs like Dirty Frag or CIFSwitch, where blocking the vulnerable module is a viable interim measure.

Not all four CVEs affect the same kernel versions. The scope per CVE is:

CVE Subsystem CVSS Affected CIQ Variants
CVE-2026-98260 exec/POSIX timers 7.8 LTS 9.2, LTS 9.6, RLC 9, RLC 10
CVE-2026-98256 exec/signal queue 7.8 RLC 10
CVE-2026-80775 futex/vfork 7.0 RLC 9, RLC 10
CVE-2026-89754 mm/mincore 7.0 RLC 10

Treat the following product variants as affected unless they are running a patched kernel (see Patched Kernels):

  • RLC Pro LTS 9.2 and 9.6
  • RLC Pro 9
  • RLC Pro 10
  • RLC Pro FIPS variants built on el9 kernels
  • RLC Pro Hardened variants built on el9 or el10 kernels
  • CIQ SIG/Cloud Next 9 and 10

Rocky Linux 8, all el8 LTS variants (8.6, 8.8, 8.10), RLC Pro 8, FIPS 8.x, and CIQ Bridge / CentOS 7 are not affected. The vulnerable code paths were either introduced after the el8 kernel fork point or were never backported into the 4.18 kernel series.

Status

  • Patched kernels are available for all supported variants as of 2026-10-22. See the Patched Kernels table below for NVRs and release dates.
  • No kernel-level mitigation available. The four CVEs target built-in kernel subsystems (exec, signal, futex, memory management) that cannot be disabled, blocked, or restricted. There is no sysctl, modprobe override, or configuration change that mitigates these CVEs at the kernel level. The only remediation is installing a patched kernel and rebooting. Operational controls (restricting local access, auditing accounts) can reduce exposure until the patch is applied.
  • Recommended action: Install the patched kernel for your variant and reboot. See Installing the Update below for variant-specific instructions. Systems with local shell users, CI/CD runners, shared application accounts, or workloads that execute untrusted code should be treated as highest priority.
  • Rocky Linux community releases: CIQ does not publish patched kernels for community Rocky Linux editions. For community release patching status, refer to Rocky Linux Errata.
  • Open a support case if you need help assessing exposure across your fleet or tracking patched kernel availability for a specific CIQ variant.

Patched Kernels

Variant Patched Kernel Version Repo Released
RLC Pro LTS 9.2 kernel-5.14.0-284.30.1+43.1.el9_2_ciq rlc-9.2-lts.x86_64 2026-10-15
RLC Pro LTS 9.6 kernel-5.14.0-570.60.1+25.1.el9_6_ciq rlc-9.6-lts.x86_64 2026-10-15
RLC Pro FIPS 9.2 kernel-5.14.0-284.30.1+43.1.el9_2_ciq fips-9.2-compliant-x86_64 2026-10-15
RLC Pro FIPS 9.6 kernel-5.14.0-570.60.1+25.1.el9_6_ciq fips-9.6-compliant-x86_64 2026-10-15
RLC Pro 9 kernel-5.14.0-687.57.1+2.1.el9_8_ciq rlc-9-core.x86_64 2026-10-19
CIQ SIG/Cloud Next 9 kernel-5.14.0-687.57.1+2.1.el9_8_ciq ciq-sigcloud-next-9.x86_64 2026-10-19
RLC Pro 10 kernel-6.12.0-211.65.1+2.1.el10_2_ciq rlc-10-core.x86_64 2026-10-19
CIQ SIG/Cloud Next 10 kernel-6.12.0-211.65.1+2.1.el10_2_ciq ciq-scn-10.x86_64 2026-10-19

CIQ updates this table as tested kernels become available for each variant.

Confirm what is running on a given system with:

uname -r

If your system reports the patched version above (or newer), the fix is in place.

Installing the Update

RLC Pro LTS and FIPS Variants

For RLC Pro LTS variants (including LTS 9.2 and LTS 9.6) and FIPS compliant variants (FIPS 9.2 and FIPS 9.6), the patched kernel is available directly from the configured CIQ repository. No additional repository configuration is required:

sudo dnf update kernel*
sudo reboot

RLC Pro and SIG/Cloud Next

For RLC Pro 9, RLC Pro 10, and CIQ SIG/Cloud Next variants, refresh your DNF metadata and update:

sudo dnf clean all && sudo dnf makecache
sudo dnf update kernel*
sudo reboot

If the update does not appear, confirm that the appropriate depot channel is enabled for your product:

sudo depot enable lts-9.2      # RLC Pro LTS 9.2
sudo depot enable lts-9.6      # RLC Pro LTS 9.6
sudo depot enable rlc-pro-9    # RLC Pro 9 / CIQ SIG Cloud Next 9
sudo depot enable rlc-pro-10   # RLC Pro 10 / CIQ SIG Cloud Next 10

For Rocky Linux community releases, refer to Rocky Linux Errata for patched kernel availability and update instructions.

Verification

After installing the patched kernel and rebooting, confirm the running kernel:

uname -r

The output should match the patched version listed in the Patched Kernels table for your variant.

Resolution

Install the patched kernel for your variant and reboot:

sudo dnf update kernel*
sudo reboot

After rebooting, verify the running kernel matches the patched version:

uname -r

No mitigation was applied for these CVEs, so there is nothing to revert.

Compensating Controls

There is no kernel-level mitigation for these CVEs. The only fix is a patched kernel. For systems that cannot be patched immediately (for example, systems on a variant where the patched kernel is still pending, or systems that require a maintenance window before rebooting), the following compensating controls can reduce exposure.

Reducing the Attack Surface

All four Infinity Core exploits require local unprivileged access. They cannot be exploited remotely on their own. Every account that can get a shell on an affected system is a potential attack vector.

Audit local accounts and SSH access on affected systems. Identify accounts that have shell access but do not need it, and restrict them:

sudo usermod -s /sbin/nologin <service-account>

Review /etc/passwd for accounts with interactive shells (/bin/bash, /bin/sh, /bin/zsh) and confirm each one is still needed. Service accounts, application accounts, and accounts used only for automated processes should not have interactive shells.

Systems with CI/CD runners, shared compute environments, container runtimes executing untrusted images, or any workload that runs code from external sources should be treated as highest priority for patching. These environments provide the local access the exploits require without needing a compromised user account.

Restricting SSH access to known administrators and disabling password-based SSH authentication (in favor of key-based authentication) further reduces the pool of accounts that could be used to run an exploit.

Exploit Attempt Detection

Kernel-level exploits that use standard syscalls are difficult to detect reliably. The following monitoring can surface failed exploitation attempts, but a successful exploit will not necessarily trigger any of these indicators. This monitoring is supplemental and is not a substitute for patching.

If the exploit crashes or partially fails, it may produce a kernel oops or BUG message in the kernel ring buffer. A periodic check of dmesg output for signatures related to the affected subsystems can flag these events:

dmesg --level=err,warn | grep -iE 'BUG:|WARNING:|unable to handle|posix.cpu.timer|sigqueue|futex|mincore'

On systems with kdump configured (default on CIQ LTS variants with crashkernel= in the boot command line), a failed exploitation attempt that crashes the kernel will produce a vmcore in /var/crash/. This provides forensic evidence that an exploit was attempted. Check for unexpected crash dumps with:

ls /var/crash/

If crash dumps appear outside of known maintenance or testing windows, they warrant investigation. The vmcore-dmesg.txt file inside each dump contains the kernel panic trace, which can confirm whether the crash matches the Infinity Core call signature (for example, posix_cpu_timers_exit in the call trace for CVE-2026-98260).

On systems with auditd running, rules that log execve calls from unusual contexts or rapid vfork and futex syscall sequences can provide additional telemetry. However, these syscalls are common in normal workloads. Without a SIEM or log analysis platform to correlate events and filter false positives, raw audit logs will be noisy and difficult to act on. Audit-based detection is only practical for environments that already have centralized log analysis in place.

A successful privilege escalation produces no kernel error. The most reliable post-compromise indicator is an unexpected process running as root, or an unexpected uid=0 session appearing in last, utmptool, or your session logging infrastructure.

Notes

  • No kernel-level interim mitigation exists. All four CVEs target built-in kernel subsystems that cannot be disabled or restricted. Unlike module-based vulnerabilities (Dirty Frag, CIFSwitch) where blocking a kernel module is effective, the only remediation for Infinity Core is a patched kernel. See Compensating Controls above for steps to reduce exposure on systems that cannot be patched immediately.
  • x86_64 exploits, architecture-independent bugs. The published exploit code targets x86_64. The underlying kernel bugs exist on all architectures, so aarch64 and other platforms should also be patched, even though no public aarch64 exploit exists at this time.
  • Public exploit availability. Working exploits are publicly available. Treat affected systems with local users or untrusted workloads as exploitable now.
  • Not all CVEs affect all variants. CVE-2026-98260 has the broadest scope (all listed variants). CVE-2026-98256 affects only RLC 10. CVE-2026-80775 affects RLC 9 and RLC 10. CVE-2026-89754 affects only RLC 10. A single kernel update for each variant covers all applicable CVEs.
  • el8 is safe. The vulnerable code was either introduced after the el8 kernel fork point or was not backported to the 4.18 kernel series.
  • Two additional CVEs from the same disclosure (CVE-2026-97941 and CVE-2026-98385) do not affect CIQ products. CVE-2026-97941 is a SLUB list corruption causing kernel panic, and CVE-2026-98385 is a scheduler cache UAF requiring CONFIG_SCHED_CACHE. Neither vulnerable code path is present in any CIQ or Rocky Linux kernel branch.