{"id":"CVE-2026-74742","state":"PUBLISHED","title":"veth: fix queue index used to wake the peer txq in veth_poll","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nveth: fix queue index used to wake the peer txq in veth_poll\n\nveth_poll() derives the index of the peer TX queue to wake from\nrq->xdp_rxq.queue_index. That field is only initialized by\nxdp_rxq_info_reg() in veth_enable_xdp_range(), which runs only when an\nXDP program is attached. On the plain GRO/NAPI path\n(veth_napi_enable_range()) xdp_rxq_info_reg() is never called, so\nqueue_index stays 0 for every queue, as priv->rq is zero-allocated.\n\nSo in a multi-queue setup with GRO enabled and no XDP program attached,\nevery NAPI instance looks at the peer's TX queue 0. If veth_xmit() stops\npeer TX queue 1 because the ptr_ring is full (NETDEV_TX_BUSY), nothing\never wakes it again: the poller draining queue 1 wakes queue 0 instead.\nveth implements no ndo_tx_timeout, so the netdev watchdog does not kick\nin either, and the queue stays stopped indefinitely.\n\nDerive the index from the position of the rq within priv->rq instead,\nwhich is correct regardless of whether XDP was ever enabled.\n\nScripts to reproduce the stall are available at\nhttps://github.com/netoptimizer/veth-backpressure-performance-testing","publishedAt":"2026-08-26T14:36:53.580Z","updatedAt":"2026-08-27T05:01:02.886Z","severity":{"score":7.5,"label":"High","version":"3.1","vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","source":"Linux"},"assigner":"Linux","products":["Linux Linux"],"affected":[{"vendor":"Linux","product":"Linux","defaultStatus":"unaffected","versions":[{"version":"9fe31b3f314534e238aa6d0b6fb492134cbcf8be","status":"affected","lessThan":"b662a1fb4f3a5ea19bac24eea8315b1d05be51e7","versionType":"git"},{"version":"dc82a33297fc2c58cb0b2b008d728668d45c0f6a","status":"affected","lessThan":"73f8dd22b1e533a99ecc3f9b5de6c6daccaecace","versionType":"git"},{"version":"dc82a33297fc2c58cb0b2b008d728668d45c0f6a","status":"affected","lessThan":"90bb11fb29d3c55a2c46dc7c386d096b286e7fcf","versionType":"git"},{"version":"dc82a33297fc2c58cb0b2b008d728668d45c0f6a","status":"affected","lessThan":"60db47f02bfa2aa688938aa199117ec4f8e31d23","versionType":"git"},{"version":"6.12.61","status":"affected","lessThan":"6.12.105","versionType":"semver"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["drivers/net/veth.c"]},{"vendor":"Linux","product":"Linux","defaultStatus":"affected","versions":[{"version":"6.16","status":"affected"},{"version":"0","status":"unaffected","lessThan":"6.16","versionType":"semver"},{"version":"6.12.105","status":"unaffected","lessThanOrEqual":"6.12.*","versionType":"semver"},{"version":"6.18.46","status":"unaffected","lessThanOrEqual":"6.18.*","versionType":"semver"},{"version":"7.1.10","status":"unaffected","lessThanOrEqual":"7.1.*","versionType":"semver"},{"version":"7.2","status":"unaffected","lessThanOrEqual":"*","versionType":"original_commit_for_fix"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["drivers/net/veth.c"]}],"references":["https://git.kernel.org/stable/c/b662a1fb4f3a5ea19bac24eea8315b1d05be51e7","https://git.kernel.org/stable/c/73f8dd22b1e533a99ecc3f9b5de6c6daccaecace","https://git.kernel.org/stable/c/90bb11fb29d3c55a2c46dc7c386d096b286e7fcf","https://git.kernel.org/stable/c/60db47f02bfa2aa688938aa199117ec4f8e31d23"],"weaknesses":[],"ingestedAt":"2026-10-29T20:12:58.252Z","source":{"repository":"CVEProject/cvelistV5","path":"cves/2026/74xxx/CVE-2026-74742.json","commit":"556c9d3a8cd17e218bb8e452ba7cd619a384ff47"},"raw":{"dataType":"CVE_RECORD","dataVersion":"5.2","cveMetadata":{"cveId":"CVE-2026-74742","state":"PUBLISHED","datePublished":"2026-08-26T14:36:53.580Z","dateUpdated":"2026-08-27T05:01:02.886Z","assignerShortName":"Linux","assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","dateReserved":"2026-08-15T05:44:03.930Z"},"containers":{"cna":{"title":"veth: fix queue index used to wake the peer txq in veth_poll","descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nveth: fix queue index used to wake the peer txq in veth_poll\n\nveth_poll() derives the index of the peer TX queue to wake from\nrq->xdp_rxq.queue_index. That field is only initialized by\nxdp_rxq_info_reg() in veth_enable_xdp_range(), which runs only when an\nXDP program is attached. On the plain GRO/NAPI path\n(veth_napi_enable_range()) xdp_rxq_info_reg() is never called, so\nqueue_index stays 0 for every queue, as priv->rq is zero-allocated.\n\nSo in a multi-queue setup with GRO enabled and no XDP program attached,\nevery NAPI instance looks at the peer's TX queue 0. If veth_xmit() stops\npeer TX queue 1 because the ptr_ring is full (NETDEV_TX_BUSY), nothing\never wakes it again: the poller draining queue 1 wakes queue 0 instead.\nveth implements no ndo_tx_timeout, so the netdev watchdog does not kick\nin either, and the queue stays stopped indefinitely.\n\nDerive the index from the position of the rq within priv->rq instead,\nwhich is correct regardless of whether XDP was ever enabled.\n\nScripts to reproduce the stall are available at\nhttps://github.com/netoptimizer/veth-backpressure-performance-testing"}],"affected":[{"vendor":"Linux","product":"Linux","defaultStatus":"unaffected","versions":[{"version":"9fe31b3f314534e238aa6d0b6fb492134cbcf8be","status":"affected","lessThan":"b662a1fb4f3a5ea19bac24eea8315b1d05be51e7","versionType":"git"},{"version":"dc82a33297fc2c58cb0b2b008d728668d45c0f6a","status":"affected","lessThan":"73f8dd22b1e533a99ecc3f9b5de6c6daccaecace","versionType":"git"},{"version":"dc82a33297fc2c58cb0b2b008d728668d45c0f6a","status":"affected","lessThan":"90bb11fb29d3c55a2c46dc7c386d096b286e7fcf","versionType":"git"},{"version":"dc82a33297fc2c58cb0b2b008d728668d45c0f6a","status":"affected","lessThan":"60db47f02bfa2aa688938aa199117ec4f8e31d23","versionType":"git"},{"version":"6.12.61","status":"affected","lessThan":"6.12.105","versionType":"semver"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["drivers/net/veth.c"]},{"vendor":"Linux","product":"Linux","defaultStatus":"affected","versions":[{"version":"6.16","status":"affected"},{"version":"0","status":"unaffected","lessThan":"6.16","versionType":"semver"},{"version":"6.12.105","status":"unaffected","lessThanOrEqual":"6.12.*","versionType":"semver"},{"version":"6.18.46","status":"unaffected","lessThanOrEqual":"6.18.*","versionType":"semver"},{"version":"7.1.10","status":"unaffected","lessThanOrEqual":"7.1.*","versionType":"semver"},{"version":"7.2","status":"unaffected","lessThanOrEqual":"*","versionType":"original_commit_for_fix"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["drivers/net/veth.c"]}],"references":[{"url":"https://git.kernel.org/stable/c/b662a1fb4f3a5ea19bac24eea8315b1d05be51e7"},{"url":"https://git.kernel.org/stable/c/73f8dd22b1e533a99ecc3f9b5de6c6daccaecace"},{"url":"https://git.kernel.org/stable/c/90bb11fb29d3c55a2c46dc7c386d096b286e7fcf"},{"url":"https://git.kernel.org/stable/c/60db47f02bfa2aa688938aa199117ec4f8e31d23"}],"metrics":[{"cvssV3_1":{"version":"3.1","vectorString":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","baseScore":7.5,"baseSeverity":"HIGH"},"scenarios":[{"lang":"en","value":"AV:N - Remote packets routed, NATed, or bridged into a suitably configured container veth reach veth_xmit() and veth_poll() through normal netdev transmit and NAPI softirq paths; no local syscall or physical access is required.\nAC:L - Sustained multi-flow traffic can deterministically fill the ptr_ring on a non-zero RX queue and drive its NAPI poller; waking queue 0 instead of the stopped queue is a logic bug, not a race or layout-dependent condition the attacker cannot control.\nPR:N - Multi-queue veth, GRO, and a non-noqueue qdisc are infrastructure prerequisites, not attacker privileges; forwarded packets traverse veth_xmit/veth_poll with no capability or authentication check, as with CVE-2025-68232 in the same backpressure code.\nUI:N - Exploitation requires only attacker-generated network traffic to drive backpressure on the veth pair; no victim action such as mounting filesystems or opening files is necessary.\nS:U - The defect leaves transmit queues in DRV_XOFF within the host kernel veth driver and does not cross VM, IOMMU, or sandbox security boundaries.\nC:N - The bug only mis-associates NAPI poll instances with peer TX queue indices during wakeup; there is no out-of-bounds access, use-after-free, or other memory read primitive.\nI:N - No memory corruption, arbitrary write, packet tampering, or control-flow influence occurs; the failure is limited to incorrect queue state management on the transmit path.\nA:H - When a non-zero peer TX queue is stopped for a full ptr_ring, veth_poll() never wakes it again and veth has no ndo_tx_timeout recovery, causing an indefinite transmit stall and persistent loss of connectivity on affected queues."}]}],"providerMetadata":{"shortName":"Linux","orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","dateUpdated":"2026-08-27T05:01:02.886Z"},"cpeApplicability":[{"nodes":[{"operator":"OR","negate":false,"cpeMatch":[{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.12.61","versionEndExcluding":"6.12.105"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.16","versionEndExcluding":"6.18.46"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.16","versionEndExcluding":"7.1.10"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.16","versionEndExcluding":"7.2"}]}]}],"x_generator":{"engine":"bippy-1.2.0"}}}}}