From mboxrd@z Thu Jan 1 00:00:00 1970 From: AL-KERNEL To: kernel-cve@kernelcve.org Subject: [CVE-2026-98366][MODERATE REGULAR] RDMA/rxe: validate access flags before swapping the MR's PD [ Upstream Date: Thu, 08 Oct 2026 07:32:56 -0400 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-AL-KERNEL-CVE: CVE-2026-98366 X-AL-KERNEL-Priority: MODERATE REGULAR X-AL-KERNEL-Severity: MODERATE REGULAR X-AL-KERNEL-Base-Severity: MODERATE X-AL-KERNEL-KPANIC: NO X-AL-KERNEL-ActionableScore: 4 X-AL-KERNEL-ActionableScore-Lower: unknown X-AL-KERNEL-Commit: 08f12745cb72981aa2cabe214af0c7a42a856f0a List-Id: CVE: CVE-2026-98366 Priority: MODERATE REGULAR AL-KERNEL base severity: MODERATE KPANIC flag: NO Patch: RDMA/rxe: validate access flags before swapping the MR's PD [ Upstream Commit: 08f12745cb72981aa2cabe214af0c7a42a856f0a Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=08f12745cb72981aa2cabe214af0c7a42a856f0a Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-98366 Analysis date: Thu, 08 Oct 2026 07:32:56 -0400 ActionableScore: 4 ActionableScore lower bound: unknown Actionable bucket: Borderline / manual review recommended Manual review required: YES Summary: No one-sentence report phrase was found. ====================================================================== ABOUT THIS REPORT ====================================================================== The original Linux kernel CVE announcement for CVE-2026-98366 is available here: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-98366 The original announcement does not normally provide a security severity estimate, CVSS assessment, or enough information to determine whether the reported kernel bug represents a practically relevant security issue. This report was generated by AL-KERNEL, an AI-assisted Linux kernel vulnerability analysis system developed by Alexander Larkin. It combines an autonomous classifier with LLM-assisted technical analysis and a separate ActionableScore mechanism. The purpose of this report is to prioritize Linux kernel CVEs before manual review, identify cases that require prompt investigation, and support automatic closure of issues that are unlikely to have meaningful security impact. Published priority for this report: MODERATE REGULAR Manual review required: YES A detailed explanation of the methodology and priority rules is included at the end of this message. ====================================================================== AL-KERNEL CLASSIFICATION RESULT ====================================================================== CVE-2026-98366 MODERATE https://www.kernelcve.org/list/?q=CVE-2026-98366 CHECK WITH IMPACT FROM ORIG NN IMPORTANT | AEGIS IMPACT MODERATE_KPANIC AEGIS_AS=4 LEGACY=MODERATE RESULT=DIFF (legacy and Aegis reconciliation differ; DECISION: legacy MODERATE vs Aegis MODERATE_KPANIC) Maybe valid. Check manually. Hints by AL-KERNEL: The best (paranoid) CVSS is 'AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H';*CWE-416;*CWE-664;*CWE-20;Other CVSS 'AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H';BEST CVSS score: '7';DESCR 'A use after free write exists in the RXE RDMA driver because rxe_rereg_user_mr can change an MR protection domain before validating requested access flags. A local process can combine protection domain reassignment with invalid access flags, leaving the MR associated with a new protection domain while RDMA core reference accounting remains associated with the original protection domain. Later destruction of the MR can incorrectly drop the new protection domain reference count to zero while a memory window still references it. Memory window cleanup can then write through a freed protection domain object and cause a kernel crash or potentially memory corruption. For the CVSS the PR:L rating is used because an attacker needs local access to the RDMA uverbs interface and permission to create and manipulate RXE RDMA objects. The issue is not directly network reachable because the vulnerable sequence is performed through local uverbs control operations. The base impact is local denial of service. The paranoid impact includes confidentiality and integrity risk because the patch and KASAN trace demonstrate a use after free write in kernel memory.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Borderline / manual review recommended (with actual score 4) MAYBE DANGER UAF NETWORK LINUS DECREASED_TO_MODERATE70_BASED_ON_ACTIONABLESCORELOWERTHAN5 DECREASED_TO_MODERATEREG_BASED_ON_FALSEPOSCHECKOFKP - - checked ====================================================================== ACTIONABLESCORE ANALYSIS ====================================================================== ActionableScore=4 PrimitiveClass=CLASS_B RealUAF=YES HostAdminRequired=UNKNOWN NamespaceScopedPrivilege=UNKNOWN ImportantCandidate=NO ManualReview=RECOMMENDED AutoDowngradeAllowed=YES 1. ActionableScore * Raw conservative score: 4 * Conservative score: 4 * Raw paranoid score: 4 * Paranoid score: 4 * Final recommended bucket: **Borderline / manual review recommended**:: No floor or cap changed the score. 2. Signal breakdown Positive signals: * +2 Real UAF and lifetime violation. The new PD can be freed while a memory window retains a pointer to it. `rxe_mw_cleanup()` subsequently calls `__rxe_put()` through that freed PD pointer. KASAN directly confirms a slab use after free write. * +1 Attacker-influenced active-object lifetime manipulation. A caller with usable RDMA uverbs access can create PD, MR, and memory-window objects, invoke MR re-registration with combined PD and invalid access flags, and drive object destruction through the normal uverbs lifecycle. * +1 Reliable kernel crash or strong DoS. The demonstrated stale refcount write can produce a kernel fault or corruption during RDMA object cleanup. The supplied KASAN report establishes the invalid post-free write. Important non-awarded signals: * +0 Controlled reclaim or replacement is not demonstrated. The evidence does not establish that attacker-controlled data can occupy the freed PD allocation before `rxe_mw_cleanup()` performs `__rxe_put()`. * +0 Arbitrary or directed kernel write is not demonstrated. The observed write is a constrained refcount-like write in `__rxe_put()`, not an attacker-selected write value or destination. * +0 Strong LPE plausibility is not awarded. A kernel UAF write exists, but no controlled reuse, callback control, type confusion, credential corruption, or practical escalation chain is established. * +0 Structured stale-object leverage is not separately awarded. The stale use occurs in memory-window cleanup and the supplied evidence only establishes the refcount mutation and UAF write. * +0 Concrete repeated release or refcount signal is not awarded. The patch establishes a stale `rxe_mw_cleanup()` reference after the PD was freed through inconsistent PD use-count accounting, but it does not clearly establish that the exact memory-window-held reference itself was previously released and then released a second time. * +0 Paranoid repeated-refcount leverage is withheld for the same reason. The evidence confirms a stale refcount-like write, but not the stricter exact-reference double-release sequence required by that rule. * +0 Additional availability impact is not awarded. It would double-count the same kernel crash or fault. * +0 Confidentiality and integrity impact are not awarded. The UAF write is real, but no read-back, protected-object modification, or attacker-controlled post-reuse state is demonstrated. * +0 Remote reachability is not established. The vulnerable sequence is performed through local RDMA uverbs object-management operations. * No race penalty. This is an ordered error-path state inconsistency, not a narrow timing race. * No host-admin penalty. The supplied material does not prove that host-root-equivalent privilege is required for the complete uverbs sequence. * No niche-subsystem penalty. RXE is specialized, but the input does not establish that a reachable RXE uverbs device is unusually difficult for the relevant attacker to access. 3. Reachability analysis The evaluated attacker is a local process using the RDMA uverbs interface. The trigger requires access to an RXE RDMA device and sufficient permission to create and manipulate PD, MR, and memory-window objects. The supplied evidence does not resolve whether the relevant `/dev/infiniband/uverbs*` device is accessible to an ordinary user, a user in an RDMA-related group, a delegated service account, a container workload, or only a host administrator. Therefore, `HostAdminRequired=UNKNOWN` is appropriate. RXE device provisioning may require prior network or RDMA configuration, but the patch does not establish whether an attacker can create the required RXE device and complete all operations from an attacker-controlled namespace. Therefore, `NamespaceScopedPrivilege=UNKNOWN`. No special physical RDMA hardware is required for the software RXE driver itself. However, an RXE device and uverbs access must exist. The path is not directly reachable through RDMA network packets alone because the vulnerable operation is a local uverbs re-registration and teardown sequence. This is not a timing race. The caller can deterministically request both `IB_MR_REREG_PD` and `IB_MR_REREG_ACCESS`, provide unsupported access flags, then drive the later MR, PD, and MW teardown sequence. 4. Primitive and severity interpretation The exact primitive is a real active-object lifetime violation involving an RXE protection-domain object. `rxe_rereg_user_mr()` changes `mr->ibmr.pd` before validating access flags. On validation failure, RDMA core error handling does not roll back the altered PD association or its corresponding use-count ownership. This causes a mismatch between the MR-visible PD pointer and the RDMA-core PD accounting. The new PD can be freed while a memory window still retains a pointer to it. A later `rxe_mw_cleanup()` executes `__rxe_put()` on this stale pointer, producing the confirmed KASAN slab use after free write. The stale object is an active structured object, namely an RXE protection domain, but the demonstrated stale use is in cleanup processing. The supplied evidence establishes a constrained post-free refcount-like write, not controlled object replacement, arbitrary write, callback invocation, information disclosure, or privilege escalation. This is therefore `CLASS_B`, not `CLASS_A`. The demonstrated practical consequence is a local kernel fault or DoS. A stronger consequence such as kernel memory corruption after suitable allocator reuse is technically possible but remains speculative because controlled reclaim and attacker-shaped replacement data are not established. `ImportantCandidate=NO` is selected because the currently demonstrated path is cleanup-associated and lacks concrete reuse or security-boundary leverage beyond the post-free refcount write. `ManualReview=RECOMMENDED` is still appropriate because this is a real kernel UAF write rather than a validation-only failure. `AutoDowngradeAllowed=YES` is appropriate under the supplied evidence because the stronger exploitation primitives remain unproven and the exact repeated-release exception is not established. Likely operational severity is actionable local DoS review priority, above an ordinary bookkeeping issue but below a confirmed Important-class kernel corruption exploit. 5. One-sentence report phrase A local process with access to an RXE RDMA uverbs device can trigger a use after free write during memory-window cleanup by combining MR protection-domain reassignment with invalid access flags, potentially causing a kernel crash. 6. Manual review recommendation MANUAL CHECK RECOMMENDED The patch and KASAN trace establish a real slab use after free write in kernel memory, but the available evidence does not establish controlled reclaim, arbitrary write, privilege escalation, or a complete below-host-admin attacker path. 7. Call-site confidence Call-site confidence: medium ====================================================================== UPSTREAM PATCH SUMMARY ====================================================================== Patch: RDMA/rxe: validate access flags before swapping the MR's PD [ Upstream Commit: 08f12745cb72981aa2cabe214af0c7a42a856f0a Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=08f12745cb72981aa2cabe214af0c7a42a856f0a Changed files: drivers/infiniband/sw/rxe/rxe_verbs.c Diff excerpt: Not included in this email. See the upstream URL for the full patch. Full patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=08f12745cb72981aa2cabe214af0c7a42a856f0a ====================================================================== DETAILED REPORT METHODOLOGY ====================================================================== The original Linux kernel CVE announcement for CVE-2026-98366 can be found here: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-98366 The original CVE announcement normally does not include a security-level estimate. In particular, it may not contain a CVSS assessment, an impact level, or enough information to determine whether the reported bug is a practically relevant security issue. One purpose of this parallel CVE list is to provide that missing technical and prioritization information. The original goal of the AL-KERNEL project was to prioritize Linux kernel CVE analysis automatically before manual review. The system can also help identify non-security issues that may be suitable for automatic closure. This report was generated by AL-KERNEL, an AI-assisted Linux kernel vulnerability analysis system developed by Alexander Larkin. The first analysis stage combines an autonomous classifier with additional LLM-based analysis. The autonomous classifier runs locally on a CPU and is based on a backpropagation neural network. Together, these mechanisms produce a technical vulnerability description, identify likely weakness types, estimate CVSS severity, and provide input for ActionableScore. Two CVSS estimates are retained because incomplete kernel vulnerability information often permits more than one defensible interpretation: Conservative CVSS vector: AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H The Best / paranoid CVSS vector: AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H The Best / paranoid CVSS score: 7 The conservative vector represents a lower-impact interpretation. The Best/paranoid vector intentionally represents a plausible upper-bound interpretation and should not automatically be treated as demonstrated real-world impact. CVSS may also need to be adjusted for a particular Linux deployment, because actual reachability, privileges, enabled kernel configuration, hardware, namespaces, exposed device nodes, and other environmental conditions can differ significantly between systems. A separate ActionableScore mechanism evaluates practical remediation urgency. Its analysis may include reachability, attack prerequisites, subsystem exposure, memory-corruption characteristics, denial-of-service reliability, and possible confidentiality, integrity, or privilege-escalation impact. Conservative ActionableScore: unknown Paranoid ActionableScore: 4 The final base severity is taken directly from the second tab-separated field of the AL-KERNEL classification result. ActionableScore does not replace or independently override that final AL-KERNEL decision, and if ActionableScore adjusted impact level of ALKERNEL, then you would see self-readable flags above like INCREASED_TO_HIGH_BASED_ON_ACTIONABLESCOREHIGHEREQTHAN7. For an AL-KERNEL result of MODERATE, this report uses the following additional presentation split: ActionableScore below 5 -> MODERATE REGULAR ActionableScore 5 or more -> MODERATE 7.0 The distinction between MODERATE REGULAR and MODERATE 7.0 makes it possible to identify Moderate issues that should receive manual analysis and fixes before lower-priority MODERATE REGULAR issues. In many cases, MODERATE REGULAR fixes may wait for a later rebase or routine update. There is one override in which MODERATE REGULAR becomes MODERATE 7.0 even when the ActionableScore is below 5. When the AL-KERNEL result contains the KPANIC flag, a MODERATE result is always presented as MODERATE 7.0. The KPANIC flag selected with few regexps without usage of AI at all, so it helps to detect cases when Kernel Crash happens and similar (to filter False-Negative results from the LLM usage). KPANIC indicates that a reliable kernel crash, kernel panic, or similarly serious kernel availability impact was identified by the classification workflow. AL-KERNEL base severity for this report: MODERATE KPANIC detected for this report: NO Published priority for this report (same as in Subject): MODERATE REGULAR These results are intended to support engineering triage. They are machine-generated estimates, and cases marked for manual review should be validated by a human security engineer before final disposition. For more info read docs linked from here: https://kernelcve.org/ (and you can submit you own patch there to generate such a report for non-existant CVE-id yet). Note that in many cases this AI tool selects higher severity, than real is (means you can expect Importants instead of Moderate 7.0 or Moderates 7.0 instead of regular Moderates). If you see such cases, please use reply email interface to add additional manual analyses info to this particular CVE. And please, please, let me know when you see Lows instead of Importants or Important instead of Low (because particular for such cases I need to tune this AI tool to make it better for this one and next similar). My contact email for such notifications is alexanjelausa@gmail.com (and both send reply to CVE record itself too and see "reply" button below for howto reply).