EN PT-BR
← Voltar ao feed
critical cloud technology

Linux kernel CVE-2026-98432: IPv4 TCP use-after-free (CVSS 9.8) with Anthropic AI-written remote root exploit; Ubuntu and RHEL still unpatched

5 fonte(s) · 5 atualização(ões) · publicado 29/10/2026 12:57 · coletado há 11 horas

Matéria consolidada. Reúne 2 matérias que o Tifon tinha publicado separadamente sobre a mesma história. A IA reescreveu o relatório com o que as 5 fontes trouxeram em 03/11/2026 00:11; os links antigos trazem para cá.

Fontes: fortiguard.fortinet.com ↗ aikido.dev ↗ bleepingcomputer.com ↗ gbhackers.com ↗ thehackernews.com ↗

CVE-2026-98432 is a race-condition use-after-free in the Linux kernel IPv4 TCP connection setup path, present since kernel 4.4. The kernel CVE team scores it 9.8 CVSS, while vendors score it lower (thehackernews.com). Anthropic disclosed on October 27 that an unreleased Claude model wrote an exploit giving unauthenticated remote root on Linux servers with a multipath TCP (MPTCP) listener, bypassing KASLR (thehackernews.com, bleepingcomputer.com). The fix landed upstream on October 5, credited to Anthropic researcher Xinyang Ge (thehackernews.com). Debian, Oracle Linux, SUSE and Amazon Linux have shipped at least some fixed kernels, but Ubuntu and RHEL were still unpatched as of October 30 (bleepingcomputer.com). No in-the-wild exploitation has been reported. Fortinet PSIRT published advisory FG-IR-26-188 on November 2, but the source does not say which Fortinet products are affected (fortiguard.fortinet.com).

Entenda a falha

A vulnerabilidade

CVE-2026-98432 is a race-condition use-after-free in the Linux kernel IPv4 TCP connection setup code, present since kernel 4.4. An unauthenticated remote attacker who can reach a TCP service can try to win the race by sending a connection request carrying IPv4 options and retransmitting it. The result can be a memory leak, memory corruption or a crash. The kernel CVE team scores it 9.8, while Red Hat (5.9 on its CVE page, 8.1 in its advisory file), SUSE, Oracle (8.1) and Amazon (7.0) score it lower because they rate the race as hard to win. Anthropic reports remote root on servers with an MPTCP listener. Fixed upstream in 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113, 6.18.56, 7.2.10 and 7.3. Debian, Oracle Linux, SUSE and Amazon Linux have shipped some fixes; Ubuntu and RHEL had not as of October 30. IPv6 is not affected by the main race.

Como funciona

During a TCP handshake, the server keeps the IP options from the client's SYN with the pending request. When the connection completes, tcp_v4_syn_recv_sock() hands those options to the new child socket without copying them. If the client retransmits its SYN, another CPU can handle it for the same request and build a SYNACK through tcp_check_req(), tcp_v4_send_synack() and inet_csk_route_req(), reading the options under rcu_read_lock() only. ip_build_and_send_pkt() and ip_options_build() then read the options length twice. If the child socket is destroyed meanwhile and its options freed, the SYNACK builder reads freed memory. A similar race exists when request sockets are migrated and the old and cloned requests share the same options.

Por que é vulnerável

Since commit 079096f103fa, request sockets are processed without the listener lock, so nothing stops the child socket from being freed while a SYNACK is still being built. TCP child sockets lack SOCK_RCU_FREE, and inet_sock_destruct() frees inet_opt with a plain kfree() even though readers rely on RCU. In migration (net.ipv4.tcp_migrate_req=1 or a BPF_SK_REUSEPORT_SELECT_OR_MIGRATE program), the clone made by inet_reqsk_clone() shares ireq_opt and tcp_v4_reqsk_destructor() also uses a plain kfree(). The fix uses kfree_rcu() in both destructors, so memory is reclaimed only after current RCU readers finish.

Onde está

Linux kernel IPv4 TCP: tcp_v4_syn_recv_sock(), tcp_check_req(), tcp_v4_send_synack(), ip_options_build(), reqsk_timer_handler(), inet_sock_destruct() and tcp_v4_reqsk_destructor(); the overlap became possible after commit 079096f103fa.

Como foi descoberta

The kernel fix credits Anthropic researcher Xinyang Ge with reporting the bug; the articles do not say how it was found. Developers applied the fix upstream on October 5, the CVE was published on October 15 (Ubuntu issued an advisory the same day), and the kernel CVE team added the 9.8 score on October 24. On October 27 Anthropic published a post saying an unreleased Claude model, after initial prompting, wrote a remote root exploit for servers with an MPTCP listener. Neither the CVE record nor the fix mentions MPTCP. No vendor advisory or report shows exploitation in the wild.

Explicação gerada por IA a partir da matéria e dos artigos, em 03/11/2026 00:26.

Detalhes técnicos

  • In tcp_v4_syn_recv_sock(), ownership of ireq->ireq_opt passes to the child socket (newinet->inet_opt) without copying. Another CPU may be handling a retransmitted SYN and building a SYNACK via tcp_check_req(), tcp_v4_send_synack() and inet_csk_route_req() under rcu_read_lock() only. ip_build_and_send_pkt() and ip_options_build() then read opt->optlen twice (bleepingcomputer.com, fortiguard.fortinet.com).
  • inet_sock_destruct() frees inet_opt with a plain kfree(), and TCP child sockets lack SOCK_RCU_FREE. This produces a use-after-free in ip_options_build(). The overlap became possible after commit 079096f103fa, which stopped handling request sockets under the listener lock (gbhackers.com, fortiguard.fortinet.com).
  • A related race exists in request socket migration with net.ipv4.tcp_migrate_req=1 or a BPF_SK_REUSEPORT_SELECT_OR_MIGRATE program. reqsk_timer_handler() clones the request socket via inet_reqsk_clone(), so the old and cloned sockets share ireq_opt, and tcp_v4_reqsk_destructor() also uses a plain kfree() (fortiguard.fortinet.com).
  • The kernel CVE team's scenario sends a connection request carrying IPv4 Record Route or Timestamp options and retransmits it to win the race. The result can be memory disclosure, corruption or a crash (thehackernews.com).
  • IPv6 is unaffected by the main race because options are copied. A similar IPv6 variant tied to CALIPSO will be fixed in a separate patch (fortiguard.fortinet.com, aikido.dev). macOS (XNU) and RHEL 7 are not affected (bleepingcomputer.com). AWS Lambda is not affected (aikido.dev).
  • Anthropic's two-stage exploit first leaks a kernel address to defeat KASLR, then overflows a buffer to overwrite a callback pointer with an attacker-staged payload, running a shell command as root. A lost race fails only that connection, so attempts can be repeated. In testing, the leak took 48 minutes and code execution 5 more, about 53 minutes in total (thehackernews.com, bleepingcomputer.com). Anthropic published no exploit code, kernel version or test configuration.
  • MPTCP is not normally enabled by default. Since Go 1.24, however, the Go runtime enables MPTCP on Linux TCP listeners when the kernel supports it, so some Go services may expose MPTCP listeners unknowingly (bleepingcomputer.com, aikido.dev). AlmaLinux and CloudLinux ship with MPTCP off by default (aikido.dev).

Remediação

  • Install vendor-fixed kernels and reboot. AWS says this is the only complete fix and that a livepatch also resolves it (aikido.dev).
  • Self-built kernels should move to 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113, 6.18.56, 7.2.10 or 7.3. The kernel CVE team recommends the latest stable kernel rather than cherry-picking (thehackernews.com).
  • The upstream fix replaces kfree() with kfree_rcu() in inet_sock_destruct() and tcp_v4_reqsk_destructor() (gbhackers.com, fortiguard.fortinet.com).
  • Debian fixed the flaw in 6.1.190-1 (Bookworm) and 6.12.113-1 (Trixie) (bleepingcomputer.com).
  • Amazon Linux 2 and 2023 have patched kernels (4.14 to 6.18). Update EKS nodes to AMI release v20261031 or later and ECS on EC2 to ECS-optimized AMI 20261031. Fargate tasks launched after November 1, 2026 6:00 AM PDT use the patched platform. Bottlerocket AMIs were expected from November 2 (aikido.dev).
  • AWS recommends updating instances within 30 days and gave an iptables rule dropping IPv4 packets with IP options. It also issued exposure guidance for front-end services (ALB, NLB, CloudFront) (aikido.dev).
  • Ubuntu 22.04, 24.04, 26.04 LTS and 26.10 generic kernels, plus several cloud variants, remain vulnerable with fixes in progress. 20.04, 18.04 and 16.04 are listed as vulnerable and unfixed. Do not treat listed target builds as released security updates (gbhackers.com, bleepingcomputer.com, aikido.dev).
  • RHEL 8, 9 and 10 and Rocky Linux 8, 9 and 10 are affected with fixes pending. Red Hat says fixes are deferred to a future update (bleepingcomputer.com, thehackernews.com).
  • Interim mitigations: restrict access to trusted networks or IPs. CIQ recommends nftables rules dropping IPv4 TCP packets that carry IP options (bleepingcomputer.com). Disabling MPTCP alone is not sufficient because ordinary TCP can still trigger the bug (bleepingcomputer.com). The Go Security Team calls per-program MPTCP disabling fragile (aikido.dev).

Relatos divergentes

  • Severity: The kernel CVE team scores it 9.8 (thehackernews.com). Red Hat scores it 5.9 Moderate on its CVE page and 8.1/Important in its machine-readable advisory, citing a hard-to-win race and potential denial of service. SUSE scores it 8.1, Amazon 7.0, and Oracle also lower (thehackernews.com, bleepingcomputer.com). The 9.8 figure rests on both sides of the race being attacker-induced.
  • Exploitability: Anthropic says its model produced a working remote root exploit on MPTCP listeners, and AWS confirms RCE is possible under specific conditions (aikido.dev). Ubuntu's advisory describes crashes and service disruption and does not confirm a reliable remote exploit (gbhackers.com). CloudLinux's kernel team says it could not reproduce Anthropic's claim, and no exploit or configuration was published (aikido.dev). RCE has not been publicly demonstrated without an MPTCP listener (bleepingcomputer.com), though the Go Security Team notes that systems without MPTCP listeners may be open to similar attacks (aikido.dev).
  • Mitigations: Red Hat says no mitigation meets its deployability standards, and advisories did not say whether disabling MPTCP or blocking IP-option packets reduces risk (thehackernews.com). The later reports from CIQ and AWS give packet-filtering rules for IP options, and CIQ says disabling MPTCP alone is insufficient (bleepingcomputer.com, aikido.dev).
  • Exploitation status: As of October 29, no vendor advisory reported in-the-wild attacks (aikido.dev), and none was confirmed as of October 30 (bleepingcomputer.com). The Fortinet advisory states no active exploitation (fortiguard.fortinet.com).

Linha do tempo

  1. fortiguard.fortinet.com ↗ · pub. 02/11/2026 05:00 · coletado 02/11 14:01

    Fortinet PSIRT advisory FG-IR-26-188 describes Linux kernel IPv4 TCP use-after-free (CVE-2026-98432)

    ler a análise desta fonte

    Fortinet PSIRT published advisory FG-IR-26-188 on 2026-11-02 for CVE-2026-98432, a use-after-free in the Linux kernel IPv4 TCP stack (inet_opt / ireq_opt handling). The article does not specify which Fortinet products are affected. It does not state active exploitation, a CVSS score, fixed versions or attribution. IPv6 is not affected by the first issue, and the IPv6 migration variant tied to CALIPSO will be fixed in a separate patch.

    Technical Details

    • tcp_v4_syn_recv_sock() transfers ownership of ireq->ireq_opt to the child socket without copying it. Another CPU can concurrently handle a retransmitted SYN for the same request socket and send a SYNACK from tcp_check_req().
    • tcp_v4_send_synack() and inet_csk_route_req() read ireq->ireq_opt under rcu_read_lock() only, and ip_build_and_send_pkt() / ip_options_build() read opt->optlen twice.
    • Since commit 079096f103fa, request sockets are processed without the listener lock, so the child socket can be freed while the SYNACK is still being built. TCP child sockets lack SOCK_RCU_FREE, and inet_sock_destruct() frees inet_opt with a plain kfree(), causing a use-after-free in ip_options_build().
    • A similar issue exists with request socket migration (net.ipv4.tcp_migrate_req=1 or a BPF_SK_REUSEPORT_SELECT_OR_MIGRATE program). reqsk_timer_handler() clones the request socket via inet_reqsk_clone(), so the old and cloned sockets share ireq_opt, and tcp_v4_reqsk_destructor() also uses a plain kfree().
    • The fix described uses kfree_rcu() in inet_sock_destruct() and tcp_v4_reqsk_destructor(), consistent with other paths that replace inet_opt.

    Remediation

    The article gives no fixed versions and does not specify which Fortinet products are affected. Defenders should consult the FG-IR-26-188 advisory for affected products and apply the fix described (kfree_rcu() in the two destructors) once it is available in their kernel.

  2. aikido.dev ↗ · pub. 02/11/2026 06:00 · coletado 02/11 08:01

    CVE-2026-98432: Linux kernel IPv4 TCP use-after-free (9.8) fix guide, AWS patches and interim mitigations

    Aikido adds AWS-specific remediation (Amazon Linux, EKS AMI v20261031, ECS AMI 20261031, Fargate platform after November 1, Bottlerocket), an AWS iptables rule dropping IPv4 packets with IP options, exposure guidance for AWS front-end services (ALB, NLB, CloudFront; AWS recommends updating instances within 30 days), and notes that AWS Lambda is not affected. It also reports CloudLinux could not reproduce Anthropic's claim, that AlmaLinux and CloudLinux ship MPTCP off by default, and that the Go Security Team calls per-program MPTCP disabling fragile.

    ler a análise desta fonte

    Aikido adds AWS-specific remediation (Amazon Linux, EKS AMI v20261031, ECS AMI 20261031, Fargate platform after November 1, Bottlerocket), an AWS iptables rule dropping IPv4 packets with IP options, exposure guidance for AWS front-end services (ALB, NLB, CloudFront; AWS recommends updating instances within 30 days), and notes that AWS Lambda is not affected. It also reports CloudLinux could not reproduce Anthropic's claim, that AlmaLinux and CloudLinux ship MPTCP off by default, and that the Go Security Team calls per-program MPTCP disabling fragile.

    CVE-2026-98432 is a use-after-free race in how the Linux kernel handles IPv4 TCP connection setup, present since kernel 4.4. The kernel CVE team rates it 9.8. Anthropic reports that an unreleased Claude research model wrote an exploit giving unauthenticated remote root on Linux servers with an MPTCP listener. AWS confirms that under specific conditions a remote, unauthenticated attacker could achieve remote code execution. As of October 29, no vendor advisory reported in-the-wild attacks. This Aikido guide focuses on remediation and exposure by environment.

    Technical Details

    During the TCP handshake the kernel keeps the SYN's IP options with the request socket. In tcp_v4_syn_recv_sock() the child socket takes ownership of those options without copying them. If another CPU is building a SYNACK for a retransmitted SYN, the options can be freed while still in use, causing a use-after-free in ip_options_build(). Request socket migration (net.ipv4.tcp_migrate_req=1 or a BPF reuseport migrate program) opens a similar path. IPv6 is not affected by the main race because options are copied; a similar CALIPSO-related IPv6 issue is being fixed in a separate patch. The upstream fix switches the frees to kfree_rcu(). Anthropic's test took 48 minutes to leak a kernel address and 5 more to reach command execution; a lost race fails only that handshake, so retries are possible. Anthropic did not publish the exploit, kernel version or configuration, and CloudLinux's kernel team says it could not reproduce the claim. The Go Security Team notes that systems without MPTCP listeners may be open to similar attacks, and that Go 1.24+ requests MPTCP listeners by default where the kernel allows it. Some distributions such as AlmaLinux and CloudLinux ship with MPTCP off.

    Remediation

    Install a patched kernel and reboot; AWS says this is the only complete fix and that a livepatch also resolves it. Amazon Linux 2 and 2023 have patched kernels (4.14 to 6.18). EKS nodes should move to AMI release v20261031 or later, ECS-optimized AMI release 20261031 applies to ECS on EC2, Fargate tasks launched after November 1, 2026 6:00 AM PDT use the patched platform, and Bottlerocket AMIs were expected from November 2. Ubuntu fixes were in progress for 22.04, 24.04, 26.04 and 26.10, with 20.04, 18.04 and 16.04 listed as vulnerable and unfixed; RHEL 8, 9 and 10 had no fixed kernel. Debian, Oracle Linux and SUSE have shipped fixes for at least some kernels. Upstream fixed versions are 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113, 6.18.56, 7.2.10 and 7.3. Interim measures: disable MPTCP (net.mptcp.enabled=0, GODEBUG=multipathtcp=0), a weak control; drop IPv4 packets with IP options (iptables -I INPUT 1 -m u32 --u32 '0>>24&0x0f=6:15' -j DROP, or an AWS Network Firewall equivalent); or place the host behind a separate load balancer that terminates TCP. A proxy on the same host does not help. Verify the running kernel after reboot against the vendor's fixed version.

  3. bleepingcomputer.com ↗ · pub. 30/10/2026 18:09 · coletado 30/10 19:00

    Linux kernel CVE-2026-98432 TCP use-after-free: Anthropic AI exploit gives remote root on MPTCP servers; Ubuntu and RHEL unpatched

    BleepingComputer adds CIQ's interim mitigation (nftables rules blocking IPv4 TCP packets with IP options, with a warning that disabling MPTCP alone is insufficient), the fact that Go 1.24+ enables MPTCP listeners by default, the confirmed Debian fixed versions, the affected status of Rocky Linux 8/9/10 and RHEL 8/9/10 (RHEL 7 unaffected), and that Ubuntu remained vulnerable as of October 30.

    ler a análise desta fonte

    BleepingComputer adds CIQ's interim mitigation (nftables rules blocking IPv4 TCP packets with IP options, with a warning that disabling MPTCP alone is insufficient), the fact that Go 1.24+ enables MPTCP listeners by default, the confirmed Debian fixed versions, the affected status of Rocky Linux 8/9/10 and RHEL 8/9/10 (RHEL 7 unaffected), and that Ubuntu remained vulnerable as of October 30.

    A critical use-after-free in the Linux kernel's IPv4 TCP code, tracked as CVE-2026-98432 (9.8 from the Linux kernel CVE team), allows unauthenticated remote code execution as root on servers with a Multipath TCP (MPTCP) listener. Researchers at Anthropic reported that an unreleased Claude model built an exploit that bypassed KASLR and ran arbitrary commands as root in about 53 minutes. The flaw also affects ordinary TCP, although RCE has not been publicly demonstrated without an MPTCP listener. Red Hat rates it only 5.9 (potential denial of service) because the race is hard to win. Fixes exist upstream and in some distributions, but Ubuntu and RHEL remained unpatched as of October 30, and no in-the-wild exploitation has been confirmed.

    Technical Details

    The bug is a race in which tcp_v4_syn_recv_sock() hands ireq->ireq_opt to the child socket (newinet->inet_opt) without copying it. Another CPU can process a retransmitted SYN and send a SYNACK via tcp_check_req(), tcp_v4_send_synack() and inet_csk_route_req() under rcu_read_lock() only, after which ip_build_and_send_pkt() and ip_options_build() read opt->optlen twice. A similar issue occurs during request socket migration when net.ipv4.tcp_migrate_req is enabled or a BPF_SK_REUSEPORT_SELECT_OR_MIGRATE program is used. Affected kernels date back to 4.4. MPTCP is not normally enabled by default, but since Go 1.24 the runtime enables MPTCP on Linux TCP listeners when the kernel supports it, so some Go services may expose MPTCP listeners unknowingly. macOS (XNU) is not affected, and RHEL 7 is not affected because it lacks the vulnerable code.

    Remediation

    Install updated kernel packages as distributions release them. Debian fixed the flaw in 6.1.190-1 (Bookworm) and 6.12.113-1 (Trixie), and Oracle Linux, Amazon Linux and SUSE have fixed some kernel versions. Ubuntu 22.04, 24.04, 26.04 LTS and 26.10 remain vulnerable with fixes in progress, and RHEL 8, 9 and 10 plus Rocky Linux 8, 9 and 10 are affected with fixes pending. Where no patch exists, restrict access to trusted networks or IPs. CIQ recommends temporarily dropping IPv4 TCP packets carrying IP options with nftables rules, and warns that disabling MPTCP alone is insufficient because ordinary TCP can still trigger the bug.

  4. gbhackers.com ↗ · pub. 30/10/2026 03:59 · coletado 30/10 04:00

    Ubuntu advisory details CVE-2026-98432 Linux kernel IPv4 TCP use-after-free; generic kernels still unpatched

    Adds Ubuntu's own advisory timeline (published October 15, updated October 29, 2026) and precise code-level detail: the faulty ownership transfer in tcp_v4_syn_recv_sock(), the kfree() in inet_sock_destruct() racing RCU readers in ip_options_build(), the role of commit 079096f103fa, and a second race in request socket migration with net.ipv4.tcp_migrate_req=1 or BPF_SK_REUSEPORT_SELECT_OR_MIGRATE. It also specifies the upstream fix (kfree_rcu() in inet_sock_destruct() and tcp_v4_reqsk_destructor()) and names the Ubuntu releases still awaiting fixes (26.10, 26.04 LTS, 24.04 LTS, 22.04 LTS generic plus cloud variants). Ubuntu's framing emphasizes crash/DoS impact and does not confirm a reliable remote exploit.

    ler a análise desta fonte

    Adds Ubuntu's own advisory timeline (published October 15, updated October 29, 2026) and precise code-level detail: the faulty ownership transfer in tcp_v4_syn_recv_sock(), the kfree() in inet_sock_destruct() racing RCU readers in ip_options_build(), the role of commit 079096f103fa, and a second race in request socket migration with net.ipv4.tcp_migrate_req=1 or BPF_SK_REUSEPORT_SELECT_OR_MIGRATE. It also specifies the upstream fix (kfree_rcu() in inet_sock_destruct() and tcp_v4_reqsk_destructor()) and names the Ubuntu releases still awaiting fixes (26.10, 26.04 LTS, 24.04 LTS, 22.04 LTS generic plus cloud variants). Ubuntu's framing emphasizes crash/DoS impact and does not confirm a reliable remote exploit.

    Ubuntu published and then updated (October 15 and October 29, 2026) an advisory for CVE-2026-98432, a use-after-free in the Linux kernel's IPv4 TCP connection handling rated 9.8 CVSS. The flaw stems from tcp_v4_syn_recv_sock() transferring ownership of IP options from a request socket's ireq->ireq_opt to a child socket without copying, while another CPU may still read those options when building a SYNACK for a retransmitted SYN. Ubuntu's description notes crashes and service disruption but does not claim a reliable remote exploit or active exploitation. Ubuntu's tracker lists generic kernels for 26.10, 26.04 LTS, 24.04 LTS and 22.04 LTS as vulnerable with fixes still in progress, along with older releases and several cloud kernel variants.

    Technical Details

    inet_sock_destruct() frees inet_opt with plain kfree(), so concurrent RCU readers in ip_options_build() — reached via tcp_check_req(), tcp_v4_send_synack() and inet_csk_route_req() — can touch freed memory. The overlap became possible once request sockets stopped being handled under the listener lock (commit 079096f103fa). A related race affects request socket migration when net.ipv4.tcp_migrate_req=1 or a BPF_SK_REUSEPORT_SELECT_OR_MIGRATE program is used, since cloned request sockets briefly share ireq_opt.

    Remediation

    The upstream fix replaces immediate deallocation with kfree_rcu() in inet_sock_destruct() and tcp_v4_reqsk_destructor(), deferring reclamation until RCU readers finish. Ubuntu users should track the advisory and not mistake listed target builds for released security updates.

  5. thehackernews.com ↗ · pub. 29/10/2026 12:57 · coletado 29/10 14:00 primeiro registro

    CVE-2026-98432: Linux kernel IPv4 TCP use-after-free rated 9.8; AI-written root exploit demoed, Ubuntu and RHEL unpatched

    ler a análise desta fonte

    A race condition in the Linux kernel's IPv4 TCP connection setup path, tracked as CVE-2026-98432, allows a use-after-free when IP options stored with a pending request are handed to a new socket without being copied. The kernel CVE team scored it 9.8 CVSS (low attack complexity, both sides of the race attacker-induced), while Red Hat (5.9 Moderate on its CVE page, 8.1/Important in its machine-readable advisory), Oracle, SUSE (8.1) and Amazon (7.0) score it lower. Anthropic disclosed on October 27 that its Claude model wrote a working exploit achieving unauthenticated remote root on Linux servers running a multipath TCP (MPTCP) listener. The bug was introduced in Linux 4.4, fixed upstream on October 5 and credited to Anthropic researcher Xinyang Ge. As of October 29, Ubuntu and RHEL had not shipped fixed kernels and no in-the-wild exploitation was reported.

    Technical Details

    During the TCP handshake, when a connection completes the kernel passes the IP options from the pending request to the new socket by reference. If that socket is destroyed while another CPU is still using those options to answer a retransmitted SYN, the kernel reads freed memory. IPv6 is unaffected because options are copied. The kernel CVE team's scenario involves sending a connection request carrying IPv4 Record Route or Timestamp options and retransmitting it to win the race, leading to kernel memory disclosure, corruption or crash. Anthropic's two-stage exploit first leaks a kernel address to defeat KASLR, then overflows a buffer to overwrite a callback pointer with an attacker-staged payload, executing a shell command as root. A lost race only fails that connection, allowing repeated attempts; in testing the leak took 48 minutes and code execution 5 more. No exploit code, kernel version or test configuration was published.

    Remediation

    Install fixed vendor kernels where available (Debian, Oracle Linux, SUSE, Amazon Linux have shipped at least some fixed packages). Self-built kernels should move to 5.10.272, 5.15.223, 6.1.190, 6.6.159, 6.12.113, 6.18.56, 7.2.10 or 7.3; the kernel CVE team recommends updating to the latest stable kernel rather than cherry-picking. Ubuntu (including cloud, FIPS and real-time kernels) and RHEL have no fix yet, with Red Hat stating fixes are deferred to a future update and that no mitigation meets its deployability standards. Advisories do not state whether disabling MPTCP or blocking inbound packets carrying IP options reduces risk.