Title : Security is failure: Every Patch Is a Guess
Author : Tedsig42
|=-----------------------------------------------------------------------=| |=-----------=[ Security is failure: Every Patch Is a Guess ]=-----------=| |=-----------------------------------------------------------------------=| |=-------=[ Ted "Tedsig42" Kouhouenou <[email protected]> ]=--------=| |=-----------------=[ every-patch-is-a-guess-full.txt ]=-----------------=| |=-----------------------------------------------------------------------=| "Every patch encodes an assumption about what can go wrong. Your job is to find the assumption -- and watch what it mutates into." ~ Dedicated to my parents ~ Armel Kouhouenou, 17/08/1975-31/03/2022 Julienne Kouchokeho, 16/02/1977-18/09/2026 --[ Table of Contents 1 - What This Paper Is 2 - The Mutation Model 3 - The Family: Logical Engines 4 - The Differential Audit Loop 5 - The Toolkit: Semgrep, Weggli, ctags 6 - Case Study: nf_tables 6.1 - Architecture: Netfilter Transaction Model 6.2 - Architectural Scatter (Independent Collapse): CVE-2023-35001 and CVE-2024-1086 6.3 - Incomplete-Fix Variant: CVE-2023-4244 -> CVE-2024-1085 6.4 - From Candidate to Exploit 6.5 - Attacker's Takeaway: Where to Hunt Next 7 - Limits: What the Framework Does Not Find 8 - Epilogue: The Pattern is the Weapon 9 - Acknowledgments 10 - References 11 - About the Author 12 - Attachments 13 - Extended Warm-Ups: Node.js and Apache MINA 13.1 - Warm-Up: Node.js Path Normalization (Predictive Oracle) 13.2 - Warm-Up: Apache MINA Deserialization (Logical Engine) 14 - Extended Case Study: Predictive Oracles (eBPF/JIT) 14.1 - Architecture: eBPF Verifier and JIT Pipeline 14.2 - Trigger A: CVE-2020-27194 [7] (ALU32 Bounds) 14.3 - Trigger B: CVE-2021-3490 [8] (ALU32 XOR/AND) 14.4 - Structural Symmetry Bypass: CVE-2022-23222 [9] (OR_NULL) 14.5 - Structural Symmetry Bypass: CVE-2023-2163 [10] (State Pruning) 14.6 - The Shock: CVSS 10.0 and the Irrelevance of KASLR 14.7 - Attacker's Takeaway: Where to Hunt Next 15 - Extended Case Study: Concurrent Proxies (io_uring) 15.1 - Architecture: Submission Queue and Shared Memory 15.2 - Architectural Scatter: CVE-2021-20226 -> CVE-2021-3491 15.3 - Structural Symmetry Bypass: CVE-2023-2598 -> CVE-2024-0582 15.4 - The Shock: Freeing Pages the User Still Owns 15.5 - Attacker's Takeaway: Where to Hunt Next 16 - Cross-Domain Proof: Windows Authentication Reflection Chain 16.1 - Architecture: Windows Authentication FSM 16.2 - Five CVEs, One Root Cause 16.3 - The Framework Maps Cleanly 17 - Extended References 18 - Extended Appendix 18.1 - Why the Parser Chokes: AST Limits in Kernel Code 18.2 - Family-Specific Seams: Concurrent Proxies 18.3 - Differential Patch Auditing (Complete Reference) 18.4 - Quick-Reference: Tool Patterns by Family 18.5 - Family-Specific Seams: Logical Engines 18.6 - Family-Specific Seams: Predictive Oracles 18.7 - The Hunting Checklist 18.8 - Triage: Separating Candidates from Exploits 18.9 - The Historical Invariant 18.10 - The Hunting Questions 18.11 - What the Framework Does Not Find 18.12 - Fix Analysis 18.13 - Why the Fix Was Incomplete 18.14 - The Shock: From Bug to SLUB Corruption --[ 1 - What This Paper Is This is a framework for n-day / variant analysis: given a fixed CVE, how do you find the next bug in the same family before the attacker does? The framework uses two tools -- weggli and semgrep -- to scale a human hypothesis about "what did this patch not cover?" across an entire subsystem. It is demonstrated end to end on a single Linux kernel subsystem, nf_tables, through a chain of four real CVEs. By "bypass" we mean: a bug in the same structural family remains reachable after a fix lands -- whether because the fix missed a twin, or because it exposed a pre-existing weakness. Not: we defeated a mitigation. This is not an exploit development guide, and the framework finds candidates, not confirmed vulnerabilities -- Section 7 states the honest limits (false positives, false negatives, what a classifier validated on 62 sampled CVEs can and cannot claim). All rules, the CVE dataset, and validation scripts referenced throughout are in the uuencoded archive described in Section 12. --[ 2 - The Mutation Model When a developer fixes a CVE, they encode an implicit model of the bug: "it was in this function, on this path, under this condition." The fix closes that path. For many fixes it does not close every structurally equivalent path sharing the same underlying weakness -- that is the case this paper is built around, though it is not true of every patch: some fixes genuinely close a weakness outright, and a patch can also introduce a new bug of its own rather than merely fail to cover an old one. Where it applies, the pattern looks like mutation: the vulnerability does not respawn identically, it relocates. CVE-2023-4244's fix removed the per-element busy mark from nf_tables' GC path; five months after that fix landed (August 2023), a related weakness surfaced elsewhere in the same subsystem, as CVE-2024-1085 (fixed January 2024, Section 6.3). This is the operative case study for everything that follows. The audit loop in Section 4 surfaces two distinct failure modes, both shown in Section 6. Incomplete-Fix Variant: the patch closes one expression of a weakness, but a related path remains reachable after the fix lands. Two flavors exist: the patch missed a structurally equivalent twin of what it fixed (the intuitive case, not demonstrated here), or the patch removed a guard that had been hiding a pre-existing bug elsewhere (CVE-2023-4244 -> CVE-2024-1085, Section 6.3 demonstrates this second flavor). Architectural Scatter: two independent bugs of the same family, neither a bypass of the other (CVE-2023-35001 and CVE-2024-1086, Section 6.2) -- proof the same loop also works as proactive hunting, not only post-patch analysis. --[ 3 - The Family: Logical Engines This paper works with one architectural family out of three the full methodology classifies: Predictive Oracle and Concurrent Proxy are the other two. A component is a Logical Engine when it manages objects with explicit lifecycles: create, destroy, abort, rollback. The failure mode is an asymmetric state transition -- an abort or cleanup path that does not maintain the same invariants as the happy path. This label describes a lifecycle-management shape, not "a logical bug" in the general sense -- logic errors can and do occur in validators and concurrency code too; that is a separate axis from which family a component belongs to. The hunting question: does this component manage objects with explicit lifecycles? If you cannot classify a component one way or another, you likely do not yet understand its architecture well enough to audit it. --[ 4 - The Differential Audit Loop The loop is always the same three steps. Run it after reading any fix commit for a Logical Engine component. -------------------------------------------------------------------- STEP 1 -- IDENTIFY THE ADDED CODE in the fix commit. -------------------------------------------------------------------- Every added line encodes an assumption -- and the assumption is what the bug will mutate around: + if (!nft_is_active_next(net, x)) continue; -> Are there other functions still using the current-gen check instead of the next-gen check? git show <fix_commit> > fix.patch grep "^+" fix.patch | grep -v "^+++" | head -40 -------------------------------------------------------------------- STEP 2 -- FIND PARALLEL CODE NOT MODIFIED. -------------------------------------------------------------------- Search for functions operating on the same data structure that perform similar operations (_deactivate, _destroy, _free): grep "^+" fix.patch | grep "struct\|->use\|->refcount\|genmask" ctags -R net/netfilter/ && grep 'nft_setelem' tags -------------------------------------------------------------------- STEP 3 -- VERIFY the twins were not patched with the same fix. -------------------------------------------------------------------- weggli 'nft_is_active_next(_, _)' net/netfilter/nf_tables_api.c Question 4 of the checklist below is the single most productive question for CVE bypass discovery: "If the patch adds a check to function X, list every function in the file with the same structural role. Were they patched too?" Post-hoc it flags the exact call site behind CVE-2024-1085 (Section 6.3). The same question, asked of the concurrent-proxy family in io_uring, frames the bug class of CVE-2024-0582 (reported by Google Project Zero). Hunting checklist, in order: 1. What family is this component? (Section 3) 2. What is the nominal path, and what are the exceptional paths (abort, cancellation, concurrent modification)? Do the exceptional paths keep the same invariants? 3. What does each function assume about the state it receives? Read the first ten lines; write down the implicit preconditions; is there a caller that violates one? 4. Is the fix symmetric across every structurally equivalent function? (see above) 5. What changed recently for performance or new features in this subsystem? New operation types are where an old invariant is most likely to have been forgotten. --[ 5 - The Toolkit: Semgrep, Weggli, ctags Semgrep and weggli are not oracles. Given a kernel tree and no architectural model in your head, they return thousands of syntactically valid hits and zero insight. The tool amplifies a hypothesis you already hold -- it does not generate the hypothesis. Every pattern below exists because a researcher first answered the hunting questions by hand, then wrote a query to scale that one insight across a subsystem. Structural question for Logical Engines: is cleanup symmetric to allocation? The seams to audit, in priority order: -------------------------------------------------------------------- 1. Abort and rollback handlers -- least-tested paths in any transactional system, written under time pressure after the happy path is done. -------------------------------------------------------------------- semgrep: pattern = "goto $L; ... $L: ... free($OBJ)" (check state invariants hold before all goto targets) ctags: ctags -R net/netfilter/ && grep "nft_abort\|_rollback" tags weggli: weggli '{ _ $abort(...); free($obj); }' net/netfilter/ -------------------------------------------------------------------- 2. Deactivation functions that differ from free functions -- check whether _deactivate() is always followed by _free(). -------------------------------------------------------------------- semgrep: pattern-not "_free($OBJ)" following "_deactivate($OBJ)" in all paths -------------------------------------------------------------------- 3. Generation bitmask checks: current vs. next generation -- nft_is_active() vs. nft_is_active_next(). Weggli cannot exclude calls natively, so pair it with grep -v: -------------------------------------------------------------------- weggli 'nft_is_active(_, _)' net/netfilter/ \ | grep -v nft_is_active_next rule: rules/semgrep/logical-engines/is-active-without-next.yml -------------------------------------------------------------------- 4. Union pointer stride mismatch: union { T1; T2; } *ptr where sizeof(T1) != sizeof(T2) -- the stride silently becomes sizeof(largest member) for every access. -------------------------------------------------------------------- rule: rules/semgrep/logical-engines/union-pointer-stride.yml Triage scorecard once a candidate is found (extended scorecard in the archive): +2 if the candidate lives in a different file than the patched function, +1 if the subsystem had a kernel CVE in the last two years, +1 if the fix landed in the last three releases. Score >= 6: investigate. Score >= 8: prioritize. These coefficients are a proposed heuristic encoding which signals were useful during this work's triage, not a statistically validated model calibrated against a labelled exploit/non-exploit dataset. How much noise does each pattern produce? The benchmark runs in the archive (validation/benchmark_results.md, Section 4) report: the nft_is_active-without-_next pattern yields 9 low-false-positive candidates; the io_uring multiple-SQE-read pattern yields ~300 functions needing per-function triage (moderate FP); broad array access patterns yield ~2000 high-FP hits; and ~8 patterns do not match at all due to weggli/semgrep expressiveness limits -- a real false-negative source. These are triage signals, not verdicts. To evaluate the classifier itself: `python3 validation/run_validation.py --csv-text` scores 67.7% on unseen CVE text -- the honest figure. Treat it as a triage heuristic, not an oracle. --[ 6 - Case Study: nf_tables "State is not a bug. Asymmetric state management is." --[ 6.1 - Architecture: Netfilter Transaction Model nf_tables [3] manages a hierarchy: tables contain chains; chains contain rules; rules reference sets and objects. Every control-plane change (add/delete a rule, element, or object) goes through a two-phase commit: changes are staged, validated, then committed atomically across a generation counter. Two independent code paths can touch the same element: the netlink control plane (user-issued transactions) and the garbage collector (async reclamation of stale set elements). Both are expected to respect the generation counter. Section 6.3 is what happens when one of them does not. --[ 6.2 - Architectural Scatter (Independent Collapse): CVE-2023-35001 and CVE-2024-1086 These two CVEs are not a bypass chain -- they demonstrate Architectural Scatter: the same failure mode (an under-validated numeric field) expressed independently at two isolated call sites. Patching one did not touch the other. CVE-2023-35001 [2] (NFT_BYTEORDER Union Alignment OOB) Date: July 2023 | Type: OOB Read/Write | Fixed: kernel 6.4.4 [TRIGGER] nft_byteorder_eval() declares union { u32; u16 } *d at function scope. sizeof(union) is always 4 (the largest member), so pointer arithmetic on d[i] strides by 4 bytes regardless of which member the code intends to access. For priv->size == 2, the loop bound is priv->len/2, but d[i] still steps by 4 -- with priv->dreg at byte offset 32 and len capped at 48, the loop runs 24 times and the last write reaches byte 125 -- roughly 46 bytes past the 80-byte register file: an attacker-controlled OOB read/write in adjacent kernel stack memory. [ASSUMPTION] "priv->len bytes from the base register are sufficient to bound the access." [BLIND SPOT] The union's stride is 4 bytes, not 2. Fix: removed the union, used typed pointers directly. CVE-2024-1086 [1][4] (Verdict Domain Confusion Double-Free) Date: January 2024 | Type: Use-After-Free Fixed: stable 6.7.3 / mainline 6.8-rc2, commit f342de4e2f33 (backported to 6.6.15, 6.1.76, 5.15.149) [TRIGGER] nft_verdict_init() validates a verdict code by masking its lower 8 bits against NF_VERDICT_MASK, then falls through to treat anything matching as NFT_CONTINUE/BREAK/RETURN. The sequence that follows, confirmed against the revert commit's own rationale: 1. Userspace sets a verdict whose low byte is NF_DROP, with an arbitrary value in the upper bits -- a field nf_hook_slow() expects to hold a negative errno (-EPERM, -EHOSTUNREACH) or 0. 2. nft_verdict_init() never validates that upper-bits field; it only checks the low byte against NF_VERDICT_MASK. 3. The verdict reaches nf_hook_slow(), which reads the upper bits as the drop error via NF_DROP_GETERR(). 4. A positive value there -- one that numerically resembles NF_ACCEPT (1) -- is misread as a non-error outcome instead of a genuine drop. 5. The packet's skb is freed on the drop path, but the misread outcome causes a second free on the same skb downstream: a double-free, confirmed exploitable to a working root shell (Notselwyn, "Flipping Pages"). [ASSUMPTION] "Masking against NF_VERDICT_MASK rejects all invalid verdict values." [BLIND SPOT] The mask only checks the lower 8 bits; the fallthrough accepts any value whose low byte happens to match. Audit strategy: whenever any nf_tables CVE adds numeric-range validation to one function, enumerate every other function in nf_tables_api.c, nft_byteorder.c, nft_cmp.c, and nft_immediate.c that accepts a user-controlled numeric field. Each is a candidate. --[ 6.3 - Incomplete-Fix Variant: CVE-2023-4244 -> CVE-2024-1085 This pair IS an incomplete-fix variant -- but the causal chain is subtler than "the fix missed a twin". CVE-2024-1085's own fix commit states the bug was "uncovered after the recent removal of the element busy mark (a2dd0233cbc4)", and its Fixes: tag points to the 2021 commit that introduced catch-all set elements (aaa31047a6d2) -- not to CVE-2023-4244's fix. The asymmetric-generation check in nft_setelem_catchall_deactivate() had been dormant since 2021, masked by the busy mark. CVE-2023-4244's fix did not miss a twin of the function it patched; removing the busy mark took away the mechanism that had been hiding an unrelated, older bug. The mutation model still holds -- an asymmetric state transition survived one fix and surfaced after another -- but "exposed a pre-existing weakness" is the accurate description, not "missed a twin". CVE-2023-4244 [5] (UAF in GC Race) Date: September 2023 | Type: Use-After-Free | Fixed: kernel 6.5 [TRIGGER] The GC asynchronously reclaims stale set elements via nft_setelem_data_deactivate() + nft_set_gc_batch_add(). Concurrent netlink processing can call nft_setelem_deactivate() on the same element. When these race, the element's bound data is released twice. [ASSUMPTION] "Generation locking protects concurrent element access." [BLIND SPOT] Generation locking protects visibility, not the refcount integrity of bound objects across independent deactivation paths. Fix: replaced the busy-mark mechanism with a gc_seq sequence counter, and removed the per-element busy mark entirely. [BLIND SPOT, again] Removing the busy mark exposed a latent bug in nft_setelem_catchall_deactivate() -- one that predates this fix by two years and was never a structural twin of anything CVE-2023-4244 touched. CVE-2024-1085 [6] (Double-Free) Date: January 2024 | Type: Double-Free Fixed: kernel 6.7.2 (also backported to 6.6.14, 6.1.75, 5.15.148) [TRIGGER] nft_setelem_catchall_deactivate() checks nft_is_active() (current generation) instead of nft_is_active_next() (next generation): list_for_each_entry(catchall, &set->catchall_list, list) { ext = nft_set_elem_ext(set, catchall->elem); if (!nft_is_active(net, ext)) /* BUG */ continue; /* deactivate and free the element ... */ } Within a transaction this deactivates a still-live element and double-frees it. [ASSUMPTION] "A current-generation check is the correct semantic for this path -- the surrounding transaction logic will handle any edge case." [BLIND SPOT] It is not. Every sibling path uses nft_is_active_next(): the transaction path deactivates objects in the next generation, not the current one. This check was written before catch-all elements could be deactivated mid-transaction, and nobody revisited it when that became possible. Run post-hoc against net/netfilter/ -- on Linux 7.1-rc1 (a8e7ef3cec99), a post-fix tree, so the rules locate the twin function, not the vulnerable check -- the archive's semgrep rules surface the twin in two complementary ways. The is-active-without- next rule (rules/semgrep/logical-engines/is-active-without-next.yml) flags nft_is_active() calls not paired with nft_is_active_next(): it fires on nfnetlink_hook.c:141 and :173 (line numbers are for 7.1-rc1). The deactivation-without-free rule (rules/semgrep/ logical-engines/deactivation-without-free.yml) flags the exact twin: nft_setelem_catchall_deactivate() at nf_tables_api.c:7011 (also 7.1-rc1), the catchall variant of nft_setelem_deactivate(). This is the loop from Section 4 applied in practice: the rules confirm a hypothesis a human forms by reading the GC diff, they do not form it. --[ 6.4 - From Candidate to Exploit Both CVEs above produced working exploits (CVE-2023-35001: Synacktiv PoC; CVE-2024-1086: Notselwyn's "Flipping Pages"). This paper stops at candidate discovery; the exploitation reality is covered in [2][4]. --[ 6.5 - Attacker's Takeaway: Where to Hunt Next After any nf_tables patch: - Enumerate all functions operating on the same object type as the patched function. - Check for uniform generation checks (nft_is_active vs. nft_is_active_next) across every one of them. - Check for uniform locking in deactivation, rollback, and cleanup paths. - Check all user-controlled numeric fields in nf_tables_api.c, nft_byteorder.c, nft_cmp.c, nft_immediate.c. --[ 7 - Limits: What the Framework Does Not Find - False positives: a candidate path may be unreachable due to mitigations (KASLR, CFI, refcount saturation). The scorecard in Section 5 is a triage aid, not a verdict. - False negatives: bypasses from cross-family interactions, or hardware-specific behavior, will not surface from a single-family classification. - The classifier's 67.7% figure (Section 5) is validated on 62 sampled CVEs, mostly stateful kernel subsystems; a second "source-level" mode reports 85.5%, but that number is largely circular -- it maps every CVE to its subsystem's canonical file, so it measures the file, not the CVE. We have not validated it against a held-out, out-of-distribution CVE set. - Scope: this paper demonstrates the method on one Linux kernel subsystem. FreeBSD and XNU mappings are in the archive. --[ 8 - Epilogue: The Pattern is the Weapon One Architectural Scatter pair (Section 6.2) and one incomplete-fix variant (Section 6.3), inside a single subsystem: enough to show the shape of the problem without claiming more generality than one subsystem earns. A fix that closes an observed trigger without checking whether the same weakness survives elsewhere is not a fix, it is a delay. The companion extended edition of this paper applies this exact loop against eBPF's Predictive Oracles, io_uring's Concurrent Proxies, and a five-CVE Windows authentication chain, to show the same mutation surfacing across languages, vendors, and OS models with no shared code between them. Differential audit after every security patch: enumerate functions operating on the same object type, and enforce uniform invariants across every path. Do this and you are not hunting bugs anymore. You are hunting the strain before it finds its next host. All patterns, tools, and validation data are in a uuencoded archive (Section 12): the 62-CVE dataset, precision/recall scripts, and the semgrep/weggli rules referenced above. --[ 9 - Acknowledgments Julien "jvoisin" Voisin -- for the thorough technical review. Nicolas "Newsoft" Ruff -- for the original "Security is failure" talk at SSTIC and for the technical review. The Phrack staff -- for two decades of raising the bar on what offensive research looks like. In memory of my parents, without whom none of this would exist: Armel Kouhouenou (1974 - 31/03/2022), and Julienne Kouchokeho (16 February 1977 - 18 September 2026). --[ 10 - References [1] CVE-2024-1086 -- patch commit f342de4e2f33 https://nvd.nist.gov/vuln/detail/CVE-2024-1086 [2] CVE-2023-35001 -- Synacktiv (Tanguy Dubroca) writeup: https://www.synacktiv.com/publications/old-bug-shallow-bug-exploiting-ubuntu-at-pwn2own-vancouver-2023 [3] nf_tables kernel documentation https://docs.kernel.org/netlink/specs/nftables.html [4] CVE-2024-1086 -- Notselwyn "Flipping Pages" https://pwning.tech/nftables/ [5] CVE-2023-4244 -- fix series merged as 3e91b0ebd994 https://nvd.nist.gov/vuln/detail/CVE-2023-4244 [6] CVE-2024-1085 -- patch b1db244ffd04 https://nvd.nist.gov/vuln/detail/CVE-2024-1085 --[ 11 - About the Author Tedsig42 is an offensive security researcher focused on the gap between patch intent and patch coverage -- the "what else did the developer not look at?" question that turns a single CVE fix into a family of bypasses. --[ 12 - Attachments All rules, the 62-CVE dataset, and validation scripts are included as a uuencoded archive -- the decoded tree is the authoritative source for every rule path and number cited above. The archive is published with this article on the Phrack site; the full-length edition (every-patch-is-a-guess-full.txt) additionally embeds it verbatim at the end of the file, after the extended references and the extended appendix. --[ 13 - Extended Warm-Ups: Node.js and Apache MINA --[ 13.1 - Warm-Up: Node.js Path Normalization (Predictive Oracle) CVE-2025-23084 [18] -> CVE-2025-27210 [19] -> UNC bypass (unassigned) The function path.normalize() in Node.js analyzes a path string and predicts that the result is safe to use for filesystem access. This is a Predictive Oracle: analysis before execution, with a semantic gap between what the predicate checks and what the OS actually does. CVE-2025-23084 [18]: Node.js did not treat Windows drive names (C:, D:) as special. path.join("C:..\\sensitive") was parsed as a relative path. But on Windows, "C:.." is relative to the current directory on drive C. Result: escape from the intended subtree. Fix: added drive-name validation. CVE-2025-27210 [19]: the fix did not cover Windows reserved device names (CON, PRN, AUX, NUL, COM1-9, LPT1-9). path.join("C:\\", "CON", "..\\sensitive") passed the drive-name check and reached the root filesystem through the device name. The Windows kernel treats "CON" as a device file, not a directory entry. Fix: added device-name filtering. UNC bypass (unassigned): after device names were blocked, UNC paths (\\\\server\\share\\..\\sensitive) remained valid. The oracle's predicate still does not model all Windows path semantics. Two mitigations in six months -- and a third gap, still unassigned. Each fix closes one semantic gap. Each bypass is the same vulnerability, mutated into the next unvalidated path component. This is Architectural Scatter (Section 6.2) playing out across independent predicates in the same function, and it is structurally identical to CVE-2020-27194 -> CVE-2021-3490 in the eBPF verifier (Section 14): different ecosystem, identical architecture, identical mutation vector. --[ 13.2 - Warm-Up: Apache MINA Deserialization (Logical Engine) CVE-2024-52046 [20] -> CVE-2026-41409 [21] -> CVE-2026-42778 [22] Apache MINA's ObjectSerializationDecoder uses Java native deserialization to reconstruct objects from network data. The deserialization process is a state machine: allocate -> initialize -> resolve -> check. The bug is in the order of operations. CVE-2024-52046 [20]: no classname filtering at all. Attacker sends a crafted serialized stream -> arbitrary class instantiation -> RCE. Fix: added allowlist with accept(ClassNameMatcher). CVE-2026-41409 [21]: the allowlist check in AbstractIoBuffer.getObject() ran after Java deserialization had executed static initializers. A class on the allowlist could still execute arbitrary code in its static initializer before the security guard evaluated. The state machine runs initialization before the security check -- an asymmetric order of operations. Fix: moved the check before initialization, released on all three maintained branches (2.0.28, 2.1.11, 2.2.6). CVE-2026-42778 [22]: that fix was complete on 2.0.x but still incomplete on the 2.1.x and 2.2.x branches, which needed a second round (2.1.12, 2.2.7). Same root cause, same static-initializer timing flaw, different release line -- the mutation did not even need a new code path this time. It just needed an unpatched branch. This maps to the same pattern as nf_tables (Section 6): asymmetric lifecycle management where the abort/init path is not synchronized with the guard. The Windows Kerberos chain (Section 16) shows the exact same Logical Engine mutation in a completely different ecosystem: the authentication FSM (NTLM -> NEGOTIATE -> CHALLENGE -> AUTHENTICATE) has asymmetric branches where reflection-detection is enforced on some transitions (SMB loopback via mrxsmb.sys) but not others (HTTP via Windows Admin Center, arbitrary-port SMB). Each patch adds a guard to one branch without establishing the invariant globally -- the same structure as the MINA 2.0.x / 2.1.x / 2.2.x branching failure. The hunting question is identical: "Is the fix applied to all structurally equivalent state transitions, or only to the one where the bug was observed ?" --[ 14 - Extended Case Study: Predictive Oracles (eBPF/JIT) "The verifier is not wrong. It is incomplete." --[ 14.1 - Architecture: eBPF Verifier and JIT Pipeline Every eBPF program passes through a static analyzer -- the verifier -- before execution. The verifier builds an abstract model of the program state: register types, value ranges (tnum), pointer validity, and access permissions. If the model says "safe," the program runs. The gap is between the model and reality. The verifier's abstract domain is a sound over-approximation in theory, but every missed edge case creates a concrete path that the model did not anticipate. eBPF implements two verifier passes. The first pass (check_cfg) builds the control flow graph. The second pass (do_check) simulates execution and tracks state at every instruction. State pruning (regsafe) skips re-analysis of visited states if the current state is a subset of the cached state. The families converge here: the verifier is a Predictive Oracle. Its pruning decisions act as a Logical Engine (state comparison). Concurrent access to shared eBPF maps adds Concurrent Proxy risk. This is the same convergence the JVM sandbox exhibited a generation earlier (Section 18.9): a single component can be a Predictive Oracle at the top of its call stack and a Logical Engine three frames down. --[ 14.2 - Trigger A: CVE-2020-27194 [7] (ALU32 Bounds) Date: August 2020 Type: Out-of-Bounds Read/Write Fixed in: kernel 5.8.15 [TRIGGER] The 32-bit ALU operation scalar32_min_max_or() applied 64-bit bounds (smin_value/smax_value) to the result of a 32-bit OR operation. After the operation, the verifier believed the register held a 32-bit value in the range [0, 1], when the actual value (in the upper 32 bits) was uncontrolled. static void scalar32_min_max_or(struct bpf_reg_state *dst_reg, struct bpf_reg_state *src_reg) { /* Uses smin_value, smax_value, umin_value, umax_value * (64-bit) instead of s32_min_value, s32_max_value, * u32_min_value, u32_max_value (32-bit) */ dst_reg->umin_value = ...; /* 64-bit bounds leak */ } The fix switched to {u,s}32_min_value / {u,s}32_max_value for 32-bit operations [commit 5b9fbeb75b6a]. [ASSUMPTION] "Using the same bounds fields for 32-bit and 64-bit operations is safe because the verifier tracks register content by type." [BLIND SPOT] scalar32_min_max_or() is classified as a 32-bit helper but uses 64-bit fields. After the operation, the verifier thinks the register's value range is bounded by the 32-bit result. The actual register (upper 32 bits) retains uncontrolled data from prior operations. --[ 14.3 - Trigger B: CVE-2021-3490 [8] (ALU32 XOR/AND) Date: May 2021 Type: Out-of-Bounds Read/Write Fixed in: kernel 5.13 [TRIGGER] The same bug class, two more victims. scalar32_min_max_xor() and scalar32_min_max_and() both read and wrote 64-bit bounds fields inside a 32-bit helper. The original patch for CVE-2020-27194 fixed scalar32_min_max_or() but left and() and xor() untouched -- the weakness did not need a new trigger, it simply walked next door: ctags -R kernel/bpf/ && grep 'scalar32_min_max' tags scalar32_min_max_and() and scalar32_min_max_xor() were NOT modified. Both had the same bounds propagation flaw -- so did the const-subreg OR case: CVE-2021-3490 covered all three (049c4e13714e). [ASSUMPTION] "The CVE-2020-27194 fix covered all ALU32 operations." [BLIND SPOT] The patch targeted the reported function (scalar32_min_max_or) but not the structural twins (and, xor). The semgrep rule for this pattern is in the included archive at rules/semgrep/predictive-oracles/scalar32-bounds.yml. It flags any scalar32_min_max_* function that reads or writes smin_value/ smax_value (64-bit fields) without also updating s32_min_value/ s32_max_value: semgrep \ --config rules/semgrep/predictive-oracles/scalar32-bounds.yml \ path/to/linux-source/kernel/bpf/verifier.c The rule would have caught scalar32_min_max_and() and scalar32_min_max_xor() immediately after CVE-2020-27194 was fixed -- but only because a human first noticed that "32-bit helper, 64-bit field" is the structural defect, not a syntactic one. Differential audit (Section 18.3): after the CVE-2020-27194 commit, run ctags on kernel/bpf/verifier.c, grep for 'scalar32_min_max', and verify each function was updated with the 32-bit bounds fields. --[ 14.4 - Structural Symmetry Bypass: CVE-2022-23222 [9] (OR_NULL) Date: January 2022 Type: Privilege Escalation Fixed in: kernel 5.16.11 [TRIGGER] adjust_ptr_min_max_vals() performs pointer arithmetic on registers whose type is PTR_TO_ALLOC_MEM_OR_NULL. The verifier does not reject pointer arithmetic on nullable types before the null check. An attacker crafts a program where a register is marked OR_NULL but arithmetic is performed as if the pointer were valid: if (reg->type == PTR_TO_ALLOC_MEM_OR_NULL) { /* BUG: no rejection of arithmetic on nullable */ reg->smin_value += off; reg->smax_value += off; } The fix added a check: reject pointer arithmetic if reg_type_may_be_null(reg->type) is true. [ASSUMPTION] "If a register can be null, performing arithmetic on it without checking null produces a valid pointer." [BLIND SPOT] The fix only patched adjust_ptr_min_max_vals. Any other function performing arithmetic on OR_NULL types without null-check remains vulnerable. --[ 14.5 - Structural Symmetry Bypass: CVE-2023-2163 [10] (State Pruning) Date: September 2023 (discovered earlier, Google VRP) Type: Arbitrary Read/Write, Container Escape Fixed in: kernel 6.3 (backported to all stable trees) CVSS: 10.0 (Google VRP) [TRIGGER] regsafe() compares two verifier states to decide whether pruning is safe. In its SCALAR_VALUE case, if the old register is marked !precise it returns true (states equivalent) before the range_within()/tnum_in() bounds comparison -- the shortcut assumes a !precise register's bounds simply do not matter for pruning. case SCALAR_VALUE: ... if (!rold->precise) return true; /* prune: bounds "don't care" */ /* range_within()/tnum_in() run BELOW, only if * rold->precise */ return range_within(rold, rcur) && tnum_in(rold->var_off, rcur->var_off); *The above is a **pedagogical reconstruction** of the logic flaw. The actual kernel code (pre‑6.3) is more nested (involving helpers like `__reg_safe()`), but the essential bug is the premature return before bounds comparison-exactly as exploited in Google VRP's CVE‑2023‑2163.* The attacker arranges a register that is marked !precise (the verifier does not track its exact value) but whose bounds differ between the old and current state. regsafe() says "same state" and prunes the current path. The pruned path contains an out-of-bounds access that the verifier never checks. The result is arbitrary read/write in kernel memory. [ASSUMPTION] "If precision tracking left a register !precise, its bounds genuinely do not matter for pruning." [BLIND SPOT] Precision was not propagated across conditional jumps, so a register that DID matter could stay !precise -- and regsafe()'s shortcut then prunes a path whose bounds were never checked. The weggli rule for the pruning shortcut (Section 18.6, seam 2) matches the construct (brace-less, as the kernel writes it): weggli 'if (!$reg->precise) return true;' \ path/to/linux-source/kernel/bpf/verifier.c It fires on regsafe()'s SCALAR_VALUE case. The shortcut itself is not the bug -- it still exists today (v7.0). The real fix, commit 71b547f56124, is in backtrack_insn(): it propagates precision across conditional jumps (reg_mask |= sreg | dreg) so a register that influences a precise one is itself marked precise, and is no longer pruned away by the shortcut. A companion semgrep rule is at rules/semgrep/predictive-oracles/imprecise-offset.yml. --[ 14.6 - The Shock: CVSS 10.0 and the Irrelevance of KASLR CVE-2023-2163 [10] gives the attacker arbitrary read/write in kernel memory, and the CVSS 10.0 score -- the rating Google assigned as CNA (NVD's primary metric is 8.8) -- is not inflation. State pruning bypasses the bounds check, letting the attacker construct a program where a register holds a kernel pointer disguised as a scalar. When the program reads from that "scalar" offset, the kernel dereferences the pointer and exfiltrates data to user space. KASLR is irrelevant to this primitive: the vulnerability hands over arbitrary read that leaks the kernel base directly, or arbitrary write that overwrites specific structures regardless of their address. Ubuntu Security described the practical consequence plainly: an attacker can use this to crash the system or potentially run arbitrary code. The only mitigations that matter against this class are not the usual heap-hardening suspects, they are access controls: BPF disabled (CONFIG_BPF=n) or restricted to CAP_BPF, and lockdown mode (CONFIG_SECURITY_LOCKDOWN_LSM). CFI blocks the most direct function pointer overwrite paths but does nothing against data-only attacks -- cred overwriting, security pointer corruption -- which is exactly the pivot an arbitrary read/write primitive invites. This is why KASLR and CFI together are still insufficient against a CVSS 10.0 verifier bypass: both mitigations assume the attacker needs to find an address or hijack control flow. An arbitrary read/write primitive needs neither. --[ 14.7 - Attacker's Takeaway: Where to Hunt Next - ALU32 helpers not yet audited for 32-bit vs 64-bit bounds. - is_branch_taken() pruning decisions that check type but not bounds. - JIT backends for flag/truncation/overflow behavior diverging from the verifier model. - New BPF instruction classes (BPF_ATOMIC, BPF_LSM hooks, bpf_loop). Targets worth auditing today: - BPF_LSM hooks - BPF_ATOMIC operations - Loop analysis for bpf_loop() - RISC-V JIT backend (less audited than x86-64 / arm64) --[ 15 - Extended Case Study: Concurrent Proxies (io_uring) "When every path is fast, which one is safe ?" --[ 15.1 - Architecture: Submission Queue and Shared Memory io_uring is a kernel interface for async I/O. The application and kernel share two ring buffers in memory-mapped pages: the Submission Queue (SQ) and Completion Queue (CQ). The application writes SQEs (Submission Queue Entries) to the SQ; the kernel reads them, executes the corresponding operations, and writes CQEs to the CQ. The ring memory is shared between user and kernel space. The kernel reads SQE fields with READ_ONCE() to handle concurrent modification. Each READ_ONCE() of a concurrently-modifiable SQE field creates a potential TOCTOU window. As of Linux 6.6, io_uring exposes over 50 operation types, each with independent submission, execution, and cleanup paths across multiple source files. The architecture structurally guarantees that patching any single path leaves parallel paths vulnerable -- fifty opcodes sharing one architectural seam is fifty places the same bug can relocate to. --[ 15.2 - Architectural Scatter: CVE-2021-20226 -> CVE-2021-3491 CVE-2021-20226 [11] (NVD Feb 2021; fixed upstream Sep 2020): io_grab_files() assigned the async worker's files_struct pointer directly -- req->work.files = current->files -- without incrementing the files_struct refcount. With files->count == 1, fdget() takes the fast path __fget_light() which skips the atomic increment on the struct file. A concurrent IORING_OP_CLOSE via the async worker frees the file while another operation still holds a stale reference. UAF. Fix: replaced the weak assignment with get_files_struct(current), incrementing files->count so fdget() always takes the refcounting path [commit 0f2122045b94]. CVE-2021-3491 [12] (NVD Jun 2021): IORING_OP_PROVIDE_BUFFERS receives buffer addresses and lengths from userspace via the SQE. The operation stores the length directly into io_buffer.len without capping it to MAX_RW_COUNT. When the buffer is later accessed through /proc/<PID>/mem, mem_rw uses the uncapped length, producing negative values and a heap overflow. The fix added buf->len = min_t(__u32, pbuf->len, MAX_RW_COUNT) [commit d1f82808877b] [ASSUMPTION] "Fixing files_struct refcounting in io_grab_files() fixes the class." [BLIND SPOT] Both bugs are Concurrent Proxy vulnerabilities: the submission queue carries user-supplied data (file descriptor in CLOSE, buffer length in PROVIDE_BUFFERS) that crosses the trust boundary without proper validation. But the specific mechanisms differ -- one is a refcount leak, the other is a missing length cap. A fix that addresses only one mechanism within the same architectural family leaves other data paths unprotected -- the weakness simply mutated from a refcount field to a length field, inside the same ring. The semgrep rules for the Concurrent Proxy family are in rules/semgrep/concurrent-proxies/ (Section 18.2). The fdget-refcount.yml rule flags fdget()/fget() calls (CVE-2021-20226 manifests through fdget()'s fast-path behavior; the rule identifies sites where fdget/fget resource-lifecycle patterns need auditing). The multiple-user-copy.yml rule flags functions that read the same user buffer multiple times, creating a TOCTOU window. A value-range audit after CVE-2021-20226 would additionally check every user-supplied numeric field in each opcode's struct for missing bounds capping. Running: semgrep \ --config rules/semgrep/concurrent-proxies/ \ path/to/linux-source/io_uring/ flags fdget/fget sites for manual audit. The PROVIDE_BUFFERS uncapped-length pattern requires a separate range-check rule not yet in the rule set. After the io_grab_files() fix, PROVIDE_BUFFERS remained unpatched because neither a fdget audit nor a value-range audit had been applied to io_uring as a whole -- differential audit across all SQE data paths surfaces the second bug in minutes. --[ 15.3 - Structural Symmetry Bypass: CVE-2023-2598 -> CVE-2024-0582 CVE-2023-2598 [13] (May 2023): io_sqe_buffer_register() takes a userspace-provided iov and registers it as a fixed buffer. The function has a fast path for compound pages: when PageCompound() is true, it assumes physical contiguity and creates one bio_vec for the entire range. An attacker maps the same physical page at two consecutive virtual addresses, satisfying the compound-page check without physical contiguity. The result: OOB read/write when I/O assumes the buffer covers 2 pages of physical memory. Fix: removed the compound-page fast path. Every registered buffer now uses one bio_vec per physical page. BLIND SPOT: the fix only affected fixed buffer registration in rsrc.c. The Provided Buffer Ring (IORING_REGISTER_PBUF_RING) in kbuf.c has its own lifecycle and was not audited -- the same compound-buffer weakness simply relocated to a sibling subsystem that happened to be out of scope for the first patch. CVE-2024-0582 [14] (fixed December 2023, disclosed January 2024, found by Google Project Zero): a PBUF ring registered with IOU_PBUF_RING_MMAP is allocated by the kernel and mmap()'d into user space. On IORING_UNREGISTER_PBUF_RING the kernel frees those pages back to the allocator without checking that user space has unmapped them. The user mapping survives, so the application keeps read/write access to pages that can be reallocated for other kernel objects: T0: Register MMAP PBUF ring -> kernel allocs pages, app mmap()s them T1: Unregister PBUF ring -> pages freed to the allocator, but the user mapping is NOT torn down T2: Allocator reuses the freed pages for a new kernel object (spray with controlled content) T3: App reads/writes through its stale mapping -> read/write on the reallocated kernel object [ASSUMPTION] "Hardening fixed-buffer registration prevents OOB on all buffer mechanisms." [BLIND SPOT] PBUF rings use a separate code path (kbuf.c) with an independent register/unregister lifecycle; the mmap variant frees pages the user still has mapped. --[ 15.4 - The Shock: Freeing Pages the User Still Owns A use-after-free where the "use" is an ordinary userspace memory access through a mapping the kernel itself created is about as direct as a primitive gets, and CVE-2024-0582 [14] was exploited in Google kCTF and documented by Exodus Intelligence in "Mind the Patch Gap." Modern mitigations raise the bar without closing the gap: CONFIG_RANDOM_KMALLOC_CACHES (Linux 6.6+) [16] creates 16 independent copies of each kmalloc cache, selected by the allocation's code address, so the attacker cannot deterministically land their spray object in the same cache as the freed PBUF ring pages. CONFIG_SLAB_FREELIST_RANDOM [15] randomizes the order of free objects within each slab, preventing deterministic reuse. io_uring's structural advantage for an attacker undercuts both: the ring accepts thousands of concurrent SQEs that allocate and free objects of the same kmalloc bucket. Under the pigeonhole principle, with 16 independent kmalloc cache copies and more than 16 distinct call‑site addresses, a **theoretical upper bound** of 100% collision rate exists. Our simulation (`test_kmalloc_collision.c`) confirms this bound under ideal conditions. In practice, cache selection policies may reduce or skew this rate, but the flood vector remains a powerful primitive for heap spraying. gcc -O2 -o test_kmalloc_collision \ validation/scripts/test_kmalloc_collision.c -lm ./test_kmalloc_collision # expects: 100% collision rate Full results: validation/results/07-kmalloc-collision.md. The kvmalloc seed bug (pre-6.15/7.0.0) [17] made this easier historically. Whether this translates into a reliable exploitation primitive on a given target still depends on kernel version and mitigation configuration -- this is a triage observation, not a universal exploit technique (Section 5's scorecard is how you decide). What is not in question: working exploits for CVE-2024-0582 exist on recent kernels, confirming that mitigations slow but do not stop a determined attacker who treats the ring itself as the spray engine. --[ 15.5 - Attacker's Takeaway: Where to Hunt Next - io_uring multishot operations - vhost-net virtqueue handling - Intel i915 GPU command buffer submission - AF_XDP zero-copy socket buffers --[ 16 - Cross-Domain Proof: Windows Authentication Reflection Chain Sections 6, 13, 14, and 15 have already shown this chain's fingerprints inside Node.js, Apache MINA, and the Linux kernel: the three families are not Linux-specific artifacts, they are the same computer-science invariant from Section 18.9 recurring everywhere a system manages lifecycles, predicts safety, or shares memory. This section is the closing proof, in a domain with no shared code, no shared maintainers, and no shared language with the kernel subsystems above -- Active Directory authentication. --[ 16.1 - Architecture: Windows Authentication FSM Windows authentication is a state machine with multiple paths: NTLM -> NEGOTIATE -> CHALLENGE -> AUTHENTICATE Kerberos -> AS-REQ -> AS-REP -> TGS-REQ -> TGS-REP -> AP-REQ SMB -> Session Setup -> Tree Connect -> IO Each protocol can trigger the authentication state machine at different entry points. The reflection detection logic (loopback check) was added incrementally, one protocol at a time, without establishing a universal invariant -- the same incremental, per-branch patching that left MINA's 2.1.x and 2.2.x branches exposed after 2.0.x was fixed (Section 13.2). --[ 16.2 - Five CVEs, One Root Cause All five CVEs below share a root cause: Windows authentication reflection detection is incomplete across protocols. Each patch adds a guard to one component without establishing the invariant globally -- five generations of the same mutation, never the same trigger twice. | CVE | Component | Year | Description | |---------------------|------------|------|----------------------------| | CVE-2025-33073 [23] | mrxsmb.sys | 2025 | SMB client CMTI reject | | CVE-2025-58726 [24] | Ghost SPNs | 2025 | SPN re-auth on relay | | CVE-2026-24294 [25] | SMB/arb. | 2026 | Port-indep. loopback check | | | port | | | | CVE-2026-26128 [26] | SMB Server | 2026 | Homoglyph detect, DnsCache | | CVE-2026-26119 [27] | WAC/HTTP | 2026 | EPA enforcement for WAC | CVE-2025-33073: the June 2025 patch makes the SMB client (mrxsmb.sys) refuse connection attempts whose target name carries additional target information -- the marshalled CREDENTIAL_TARGET_INFORMATION (CMTI) trick that spawns a local NTLM handshake or a sanitized SPN. BLIND SPOT: killing one coercion trick does not fix the root cause -- receiving a local authentication on a controlled listener and relaying it back still works through other primitives. CVE-2025-58726: Ghost SPNs abuse domains where the SPN was deleted but the service account still exists. An attacker coerces authentication to a deleted SPN, and the KDC issues a ticket for the old SPN. CVE-2026-24294: local NTLM reflection -- a local SMB server on an arbitrary TCP port receives LSASS's NTLM handshake and relays it back to the machine's own SMB service (Synacktiv, April 2026). CVE-2026-26128: SMB Server EoP via a DNS homoglyph -- the KDC applies Unicode normalization that DnsCache does not. An attacker registers "SRV1.AD.LOCAL" with a homoglyph capital letter (punycoded SPN) -- gets a valid ticket for SRV1$ via the homoglyph SPN. CVE-2026-26119: Windows Admin Center improper authentication logic allows low-privileged users to escalate privileges by bypassing authentication context validation, gaining access as the WAC service account (often Domain Admin). --[ 16.3 - The Framework Maps Cleanly This chain maps to two families simultaneously -- proof that the families converge in Windows exactly as they converged in the eBPF verifier (Section 14.1): Concurrent Proxy failure: patching one protocol path (SMB) leaves structurally equivalent paths open (HTTP, raw Kerberos over TCP). This is the same pattern as io_uring's parallel buffer mechanisms (Section 15). Logical Engine failure: the authentication FSM has asymmetric branches where reflection-detection is enforced on some transitions (NTLM, SMB loopback) but not others (Kerberos, HTTP). Question 4 (Section 4) applied to CVE-2025-33073: "Are there other coercion paths not covered by the mrxsmb.sys patch ?" immediately surfaces Ghost SPNs, arbitrary-port SMB, homoglyph DNS, and HTTP endpoints. All four were found by researchers within twelve months of the original patch -- the same twelve-to-eighteen month mutation window observed in every Linux case study above. The strain does not care which kernel, or which company, wrote the host it is living in. --[ 17 - Extended References [7] CVE-2020-27194 -- patch commit 5b9fbeb75b6a https://nvd.nist.gov/vuln/detail/CVE-2020-27194 [8] CVE-2021-3490 -- ZDI-21-325 (Manfred Paul) https://nvd.nist.gov/vuln/detail/CVE-2021-3490 [9] CVE-2022-23222 -- patch commit 64620e0a1e71 https://nvd.nist.gov/vuln/detail/CVE-2022-23222 [10] CVE-2023-2163 -- Google VRP, patch commit 71b547f56124 https://nvd.nist.gov/vuln/detail/CVE-2023-2163 https://bughunters.google.com/blog/6303226026131456/ [11] CVE-2021-20226 -- patch commit 0f2122045b94 https://nvd.nist.gov/vuln/detail/CVE-2021-20226 [12] CVE-2021-3491 -- patch commit d1f82808877b https://nvd.nist.gov/vuln/detail/CVE-2021-3491 [13] CVE-2023-2598 -- patch commit 776617db78c6 https://nvd.nist.gov/vuln/detail/CVE-2023-2598 [14] CVE-2024-0582 -- Google Project Zero, patch commit c392cbecd8ec https://nvd.nist.gov/vuln/detail/CVE-2024-0582 https://bugs.chromium.org/p/project-zero/issues/detail?id=2504 https://blog.exodusintel.com/2024/03/27/mind-the-patch-gap/ [15] CONFIG_SLAB_FREELIST_RANDOM https://www.kernelconfig.io/CONFIG_SLAB_FREELIST_RANDOM [16] CONFIG_RANDOM_KMALLOC_CACHES https://www.kernelconfig.io/CONFIG_RANDOM_KMALLOC_CACHES [17] Linux 6.15 cache randomization seed fix https://www.phoronix.com/news/Linux-6.15-slab [18] CVE-2025-23084 -- Node.js drive name path traversal https://nvd.nist.gov/vuln/detail/CVE-2025-23084 https://nodejs.org/en/blog/vulnerability/january-2025-security-releases [19] CVE-2025-27210 -- Node.js device name path traversal bypass https://nvd.nist.gov/vuln/detail/CVE-2025-27210 https://nodejs.org/en/blog/vulnerability/july-2025-security-releases [20] CVE-2024-52046 -- Apache MINA ObjectSerializationDecoder RCE https://nvd.nist.gov/vuln/detail/CVE-2024-52046 https://lists.apache.org/thread/4wxktgjpggdbto15d515wdctohb0qmv8 [21] CVE-2026-41409 -- Apache MINA allowlist timing bypass https://nvd.nist.gov/vuln/detail/CVE-2026-41409 https://lists.apache.org/thread/9ddvsq6c4l5bhwq8l14sob4f8qjvx5c9 [22] CVE-2026-42778 -- Apache MINA cross-branch incomplete fix https://nvd.nist.gov/vuln/detail/CVE-2026-42778 https://lists.apache.org/thread/fhlx5k91hrkgyzh7yk1nghrn3k27gxy0 [23] CVE-2025-33073 -- Windows SMB Kerberos reflection (Synacktiv) https://nvd.nist.gov/vuln/detail/CVE-2025-33073 https://msrc.microsoft.com/update-guide/vulnerability/CVE-2025-33073 https://www.synacktiv.com/en/publications/ntlm-reflection-is-dead- long-live-ntlm-reflection-an-in-depth-analysis-of-cve-2025 [24] CVE-2025-58726 -- Ghost SPNs (Semperis/Andrea Pierini) https://nvd.nist.gov/vuln/detail/CVE-2025-58726 https://msrc.microsoft.com/update-guide/vulnerability/CVE-2025-58726 https://www.semperis.com/blog/ exploiting-ghost-spns-and-kerberos-reflection [25] CVE-2026-24294 -- Windows SMB arbitrary port loopback https://nvd.nist.gov/vuln/detail/CVE-2026-24294 https://msrc.microsoft.com/update-guide/vulnerability/CVE-2026-24294 https://www.synacktiv.com/en/publications/ bypassing-windows-authentication-reflection-mitigations-for-system -shells-part-1 [26] CVE-2026-26128 -- Windows SMB homoglyph DNS (Synacktiv) https://nvd.nist.gov/vuln/detail/CVE-2026-26128 https://msrc.microsoft.com/update-guide/vulnerability/CVE-2026-26128 https://www.synacktiv.com/en/publications/ bypassing-windows-authentication-reflection-mitigations-for-system -shells-part [27] CVE-2026-26119 -- Windows Admin Center EPA bypass https://nvd.nist.gov/vuln/detail/CVE-2026-26119 https://msrc.microsoft.com/update-guide/vulnerability/CVE-2026-26119 https://www.semperis.com/blog/ windows-admin-center-remote-priv-esc-cve-2026-26119 --[ 18 - Extended Appendix --[ 18.1 - Why the Parser Chokes: AST Limits in Kernel Code Before trusting a static-analysis hit -- or trusting its silence -- understand why an Abstract Syntax Tree built from kernel C is already a degraded model of the program before you write a single rule against it: Macro expansion container_of(), list_for_each_entry(), READ_ONCE(), and the netfilter generation-mask helpers are all macros. A parser that works on pre-processor output sees the expanded form and loses the semantic intent ("this is a generation check") behind syntax noise ("this is a comparison of two integer expressions"). A parser that works on raw source sees the macro invocation but cannot, without a full preprocessor model, see what it expands into. Either way, a rule written at the wrong layer either matches everything or matches nothing. Type inference collapse void * and union members erase the type information a differential audit needs most. CVE-2023-35001 (Section 6.2) is invisible to any tool that asks "does this pointer arithmetic respect the pointee's size ?" -- because the union's declared size IS sizeof(u32), and the arithmetic IS correct for that type. The bug is not a type error. It is a human architectural fact (this code intends to stride by 2 bytes sometimes) that no AST node encodes. Conditional concurrent primitives Whether mutex_lock(&x->lock) on one path and the absence of that call on a parallel path constitutes a race depends on whether both paths are reachable concurrently -- a property of the scheduler, the call graph, and the subsystem's locking discipline, none of which live in a single function's AST. A parser sees two independent functions. A researcher who has read the locking comments three call sites up sees a race. Indirect call lineage through function pointers struct file_operations, the eBPF helper dispatch table, and io_uring's per-opcode issue/cleanup function pointers all resolve their target at runtime. Static data-flow tracking that is precise for direct calls degrades to "could be any of these N functions" the moment it crosses an indirect call -- and the Concurrent Proxy family (Section 18.2) lives almost entirely on the far side of exactly this kind of dispatch table. None of this is a reason to distrust the tools. It is the reason every rule in Sections 18.2 through 18.6 is described first as a structural question a human answers, and only second as a query string. The tool's job is to take an insight that took an afternoon of reading to form and apply it to forty thousand lines of C in under a minute. That is amplification, not substitution. --[ 18.2 - Family-Specific Seams: Concurrent Proxies Structural question: Where has a performance optimization removed a barrier the security model depended on? Seams to audit: 1. Recently added "fast path" code git log --all --since="2024-01-01" \ --grep="fast path\|optimize\|avoid lock\|zero-copy" \ -- io_uring/ drivers/ | head -30 2. Functions that read from user-accessible memory more than once semgrep: rules/semgrep/concurrent-proxies/multiple-sqe-reads.yml weggli: weggli '{ READ_ONCE($sqe->_); READ_ONCE($sqe->_); }' io_uring/ 3. fget()/fdget() in async issue functions without matching fput()/fdput() semgrep: rules/semgrep/concurrent-proxies/fdget-refcount.yml weggli: weggli 'fget(_)' io_uring/ --[ 18.3 - Differential Patch Auditing (Complete Reference) Section 4 introduced the three-step loop. This section gives the full command workflow for each step. Step 1: IDENTIFY THE ADDED CODE Every added line encodes an assumption. + if (condition) return -EINVAL; -> Are there other entry points without this check? + mutex_lock(&x->lock); -> Are there other paths accessing x without this lock? + if (!nft_is_active_next(net, x)) continue; -> Are there other functions still using the current-gen check? Step 2: FIND PARALLEL CODE NOT MODIFIED Search for functions operating on the same data structure that perform similar operations (_deactivate, _destroy, _free). Practical workflow: # Extract the patch git show <fix_commit> > fix.patch grep "^+" fix.patch | grep -v "^+++" | head -40 # Find the patched structure grep "^+" fix.patch | grep "struct\|->use\|->refcount\|genmask" # Find structural twins ctags -R net/netfilter/ && grep 'nft_setelem' tags # Check if they have the same fix weggli 'nft_is_active_next(_, _)' net/netfilter/nf_tables_api.c --[ 18.4 - Quick-Reference: Tool Patterns by Family +-----------------------------------------------------------------------+ | FAMILY | Tool | Pattern / query | +==================+==========+=========================================+ | Logical Engine | semgrep | _deactivate() not followed by _free() | | | weggli | '{_deactivate($OBJ); ... free($OBJ);}' | | | cocci | rules/cocci/find-un-active.cocci | | | ctags | grep 'genmask\|_is_active\b' in tags | | | git grep | nft_is_active[^_] (without _next) | |------------------+----------+-----------------------------------------| | Predictive | semgrep | scalar32 helpers: smin_value (64-bit) | | Oracle | weggli | 'if (!reg->precise) { return true; }' | | | ctags | grep ALU32|subreg|OR_NULL in verifier | | | JIT dump | /proc/kallsyms (via | | | | /proc/sys/net/core/bpf_jit_kallsyms) | |------------------+----------+-----------------------------------------| | Concurrent Proxy | semgrep | two reads of sqe->$F in same function | | | weggli | 'READ_ONCE($sqe->_); READ_ONCE($sqe->_)'| | | ctags | grep 'fdget\|fget\|io_cancel' in tags | | | git log | --grep="optimize\|fast path\|avoid lock"| +-----------------------------------------------------------------------+ --[ 18.5 - Family-Specific Seams: Logical Engines Structural question: Is cleanup symmetric to allocation ? Seams to audit: 1. Abort and rollback handlers Least-tested paths in any transactional system. Written under time pressure after the happy path is complete. semgrep: pattern = "goto $L; ... $L: ... free($OBJ)" (check if state invariants hold before all goto targets) ctags: ctags -R net/netfilter/ && grep "nft_abort\|_rollback\|_undo" tags weggli: weggli '{ _ $abort(...); free($obj); }' net/netfilter/ 2. Deactivation functions that differ from free functions Check whether _deactivate() is always followed by _free(). semgrep: pattern-not: "_free($OBJ)" following "_deactivate($OBJ)" in all paths 3. Generation bitmask checks: current vs next nft_is_active() vs nft_is_active_next() weggli: (weggli cannot exclude function calls natively; use grep -v as workaround) weggli 'nft_is_active(_, _)' net/netfilter/ \ | grep -v nft_is_active_next semgrep: rules/semgrep/logical-engines/is-active-without-next.yml coccinelle: rules/cocci/find-un-active.cocci semgrep: rules/semgrep/logical-engines/union-pointer-stride.yml 4. Union pointer stride mismatch union { T1; T2; } *ptr where sizeof(T1) != sizeof(T2) semgrep: rules/semgrep/logical-engines/union-pointer-stride.yml --[ 18.6 - Family-Specific Seams: Predictive Oracles Structural question: Does the Oracle track all dimensions of the domain it claims to verify? Seams to audit: 1. ALU32 helpers using 64-bit bounds semgrep: flag functions named scalar32_* that do not update both s32_{min,max}_value and u32_{min,max}_value weggli: weggli 'scalar32_min_max_and(_, _)' kernel/bpf/verifier.c 2. State pruning shortcuts semgrep: flag paths where !precise skips bounds comparison weggli: weggli 'if (!reg->precise) return true;' kernel/bpf/verifier.c 3. JIT-verifier divergence Compile BPF program, extract JIT output via /proc/kallsyms (enabled via /proc/sys/net/core/bpf_jit_kallsyms), and compare native code against the verifier's model for specific sequences. --[ 18.7 - The Hunting Checklist Question 1: What family is this ? Classify as Logical Engine, Predictive Oracle, or Concurrent Proxy. If you cannot classify it, you do not yet understand the architecture well enough to hunt in it. Question 2: What is the nominal path, and what are the exceptional paths ? For each exceptional path (abort, cancellation, unusual encoding, concurrent modification), does the code maintain the same invariants as the nominal path ? Question 3: What does each function assume about the state it receives ? Read the first ten lines of each function in the audited path. Write down the implicit preconditions. Is there any caller that does not satisfy them ? Question 4: Is the fix symmetric ? If the patch adds a check to function X, list all functions in the same file with the same structural role. Were they also patched? This is the single most productive question for CVE bypass discovery. It found CVE-2024-1085 and CVE-2024-0582. Question 5: What changed recently for performance? For Concurrent Proxies: commits mentioning "optimize", "fast path", "avoid lock", "amortize", "batch" -- each removed a barrier. For Predictive Oracles: commits adding new instruction classes. For Logical Engines: commits adding new operation or object types. --[ 18.8 - Triage: Separating Candidates from Exploits Bypass candidates are not exploits. Use this scorecard to prioritize -- and to decide which strain is worth following into its mutation, detailed live in Sections 6, 14, and 15: +1 if: Subsystem had kernel CVE in last 2 years +1 if: >20 commits/month in subsystem +1 if: Fix is in last 3 releases +2 if: Twin is in a different file (harder for reviewer to notice) +1 if: semgrep severity=WARNING +2 if: semgrep severity=ERROR +1 if: Bug had working PoC Score >= 6: investigate Score >= 8: prioritize Before building a PoC, verify: - Can you trace a path from userspace to the candidate function ? - Is it reachable with unprivileged permissions ? - Does the unpatched path exist in the same kernel build ? - Is there a CONFIG option disabling the vulnerable path ? In practice, exhaustive twin auditing is rarely feasible. We recommend the following **triage shortcut** : - Prioritize functions reachable from **unprivileged userspace** (io_uring, bpf, netlink). - Ignore static functions called from a single site unless they handle attacker‑controlled data. - For Logical Engines, prioritize **abort and rollback** paths before commit paths-they are historically undertested. - If a twin is in a different file (e.g., rsrc.c vs kbuf.c), escalate its priority (+2 in the scorecard). Abandon if: - Candidate is static, called from one site only - Patch explicitly mentions the candidate - Candidate is in __init code (runs only at boot) - The candidate path is disabled by default The detailed Mitigation Reality analyses for each family are now integrated directly into the corresponding case study sections: see Sections 6.5 (nf_tables), 14.7 (eBPF), and 15.5 (io_uring). --[ 18.9 - The Historical Invariant This is not a 2020s novelty. The same three failure modes-asymmetric state, incomplete static analysis, and racing proxies-have driven sandbox escapes for two decades (Java SecurityManager, bytecode verifier, ClassLoader confusion). Distributed state desynchronization is not an io_uring invention; it is computer science's oldest architectural seam, wearing a kernel module. This is why the framework in Sections 2 through 4 is not a taxonomy of CVEs. It is a taxonomy of the boundary conditions every stateful system has always had, going back to the first multi-user operating system that ever had to decide whether a half-finished operation could be safely rolled back. --[ 18.10 - The Hunting Questions Before reading a single line of kernel code, ask: Q1: Does it manage objects with explicit lifecycles? -> Logical Engine Q2: Does it analyze or transform input before execution? -> Predictive Oracle Q3: Does it rely on shared memory with minimal synchronization? -> Concurrent Proxy These three questions successfully map every CVE **in our sampled dataset** (62 CVEs, primarily from stateful kernel subsystems) to one of the three families. We treat this as a **search heuristic**, not a formal partition-some bugs (e.g., simple NULL dereferences or information leaks) may not fit cleanly. If you cannot classify a component, you do not yet understand its architecture well enough to hunt in it. --[ 18.11 - What the Framework Does Not Find The framework identifies structural bypass candidates, not confirmed exploitable vulnerabilities. Honest limitations: - False positives: a candidate path may be unreachable due to mitigations (KASLR, CFI, refcount saturation). Section 5 provides a triage scorecard. - False negatives: bypasses from complex cross-family interactions or hardware-specific behavior may not surface from single-family classification. - Taxonomy overlap: the families are not mutually exclusive. The classification is a search heuristic, not a formal partition. - Scope: the framework was developed against Linux kernel subsystems. The Windows Kerberos chain (Section 16) provides cross-domain validation. FreeBSD (dhclient CVE-2026-42511, NFS CVE-2026-4747) and XNU (IOKit IOUserClient, Mach IPC) mappings are documented in cross-kernel/ in the included archive. --[ 18.12 - Fix Analysis The patch replaces the old gc_batch mechanism with a new GC transaction API: a gc_seq sequence counter lets the work queue detect when the control plane has touched an element, and the per-element "busy mark" is removed. [ASSUMPTION] "The new gc_seq transaction API closes the GC vs control-plane double-free window." [BLIND SPOT] Removing the busy mark exposed a latent twin -- nft_setelem_catchall_deactivate() -- still using the current-gen check. The vulnerability did not close. It mutated sideways into the one function nobody re-audited after the GC rework. --[ 18.13 - Why the Fix Was Incomplete nft_setelem_catchall_deactivate() performs semantically equivalent operations on the same object type as the patched paths, but used nft_is_active() (current generation) instead of nft_is_active_next(). The CVE-2023-4244 GC rework that removed the busy mark is what exposed it. Textbook structural twin: find the patched function, locate its twins, audit them for the same missing invariant. The semgrep rule for this exact pattern is rules/semgrep/logical-engines/is-active-without-next.yml in the archive at Section 12. Running it against net/netfilter/ flags every call site using nft_is_active() (current-generation check) that is not paired with nft_is_active_next() -- which is how nft_setelem_catchall_deactivate() was found as a structural twin: semgrep \ --config \ rules/semgrep/logical-engines/is-active-without-next.yml \ path/to/linux-source/net/netfilter/ The rule fires on nft_setelem_catchall_deactivate() and returns a severity=WARNING hit with message: "Using nft_is_active() instead of nft_is_active_next(). This was the root cause of CVE-2024-1085." This is the differential audit loop (Section 4) applied in practice: nft_is_active_next() is the predicate the transaction paths already use, and the rule finds every site still on the bare current-generation check. The rule confirms the hypothesis a human already formed by reading the GC rework diff -- it does not form the hypothesis itself (Section 18.1). --[ 18.14 - The Shock: From Bug to SLUB Corruption A structural twin is an architecture diagram. A double-free in a live kernel allocator is a weapon, and the two CVEs above are not theoretical: both produced working exploits (CVE-2023-35001 [2]: Synacktiv PoC; CVE-2024-1086 [1][4]: Notselwyn "Flipping Pages"). KASLR is not a barrier for either. The OOB read in nft_byteorder (CVE-2023-35001) can leak return addresses or heap pointers straight out of kernel stack memory, handing the attacker runtime addresses directly -- there is no base address to hide when the bug reads past the register into whatever happens to sit next to it. The double-free in CVE-2024-1086 corrupts the SLUB allocator's freelist pointer directly, and a corrupted freelist is not a crash, it is a redirect: the next allocation of that size class lands wherever the attacker's forged freelist pointer says it should. CONFIG_SLAB_FREELIST_RANDOM [15] and CONFIG_RANDOM_KMALLOC_CACHES (Linux 6.6+) [16] make deterministic heap spray harder by randomizing freelist order and creating 16 independent kmalloc cache copies. These mitigations raise complexity but do not eliminate it -- the attacker can still spray via nf_tables transactions that allocate elements of the same kmem_cache, because the ring of transactions itself is the spray primitive. CFI (CONFIG_CFI_CLANG) blocks the most direct control-flow hijack paths, which only narrows the exit: an attacker who achieves a write primitive pivots to data-only targets -- cred structures, security pointers -- that CFI was never designed to protect, because CFI checks where you jump, not what you overwrite on the way there. The bottom line: these are not "in vitro" vulnerabilities. Every CVE in this case study has a known exploit or writeup demonstrating practical impact, and the mitigation stack slows the strain without sterilizing it. begin 777 every-patch-is-a-guess.tar.gz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email protected] M9'8\[email protected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email protected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email protected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email protected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email protected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email protected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email protected]<.O X6;,S3X)R,CLX7^"@GI)D&_"-D<R$J)DS]Q M:^MG+B1JN.[<6V*'0Y%AX.'JA!PGGL0@09(PVK7]W,1^GM1CB,UC,W0B!X=: M=V6PU2B+,KR7R/KE;\SE1#O@?M>YWHT%,A5B:NAC^].%NN3HAEPCJX%VAH_B M:"(3QE5H%V[+ BE95$>>E@G46A-ZG"'NB( KG&!!V4P%8(2G>>'> >6LAKR^ MRV._QQ0"&X\H(>T=YB7J#(BPRD8&D!=J%+Q708=G5,)QO7_@Z-TN)L8-X<V] M549=&@DQQ<:XY]P+YHT!!G]B08C:@HO)U=LU/BR=?(T?X0;*[email protected] M[BSZ% WA8X)6'C$M 2[<_U;3_\&XN=$HQP5$[EV@_T,CWO-_IHO_>6_I_V#6 M?7D*AW<[email protected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email protected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ͼ&^5['[email protected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email protected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end |=[ EOF ]=---------------------------------------------------------------=|
![..[ Phrack Magazine ]..](/images/phrack-new.webp)