{ "dataType": "CVE_RECORD", "dataVersion": "5.2", "cveMetadata": { "cveId": "CVE-2025-39677", "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "state": "PUBLISHED", "assignerShortName": "Linux", "dateReserved": "2025-04-16T07:20:57.112Z", "datePublished": "2025-09-05T17:20:43.145Z", "dateUpdated": "2026-08-05T12:04:29.256Z" }, "containers": { "cna": { "providerMetadata": { "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "shortName": "Linux", "dateUpdated": "2026-08-05T12:04:29.256Z" }, "descriptions": [ { "lang": "en", "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nnet/sched: Fix backlog accounting in qdisc_dequeue_internal\n\nThis issue applies for the following qdiscs: hhf, fq, fq_codel, and\nfq_pie, and occurs in their change handlers when adjusting to the new\nlimit. The problem is the following in the values passed to the\nsubsequent qdisc_tree_reduce_backlog call given a tbf parent:\n\n When the tbf parent runs out of tokens, skbs of these qdiscs will\n be placed in gso_skb. Their peek handlers are qdisc_peek_dequeued,\n which accounts for both qlen and backlog. However, in the case of\n qdisc_dequeue_internal, ONLY qlen is accounted for when pulling\n from gso_skb. This means that these qdiscs are missing a\n qdisc_qstats_backlog_dec when dropping packets to satisfy the\n new limit in their change handlers.\n\n One can observe this issue with the following (with tc patched to\n support a limit of 0):\n\n export TARGET=fq\n tc qdisc del dev lo root\n tc qdisc add dev lo root handle 1: tbf rate 8bit burst 100b latency 1ms\n tc qdisc replace dev lo handle 3: parent 1:1 $TARGET limit 1000\n echo ''; echo 'add child'; tc -s -d qdisc show dev lo\n ping -I lo -f -c2 -s32 -W0.001 127.0.0.1 2>&1 >/dev/null\n echo ''; echo 'after ping'; tc -s -d qdisc show dev lo\n tc qdisc change dev lo handle 3: parent 1:1 $TARGET limit 0\n echo ''; echo 'after limit drop'; tc -s -d qdisc show dev lo\n tc qdisc replace dev lo handle 2: parent 1:1 sfq\n echo ''; echo 'post graft'; tc -s -d qdisc show dev lo\n\n The second to last show command shows 0 packets but a positive\n number (74) of backlog bytes. The problem becomes clearer in the\n last show command, where qdisc_purge_queue triggers\n qdisc_tree_reduce_backlog with the positive backlog and causes an\n underflow in the tbf parent's backlog (4096 Mb instead of 0).\n\nTo fix this issue, the codepath for all clients of qdisc_dequeue_internal\nhas been simplified: codel, pie, hhf, fq, fq_pie, and fq_codel.\nqdisc_dequeue_internal handles the backlog adjustments for all cases that\ndo not directly use the dequeue handler.\n\nThe old fq_codel_change limit adjustment loop accumulated the arguments to\nthe subsequent qdisc_tree_reduce_backlog call through the cstats field.\nHowever, this is confusing and error prone as fq_codel_dequeue could also\npotentially mutate this field (which qdisc_dequeue_internal calls in the\nnon gso_skb case), so we have unified the code here with other qdiscs." } ], "metrics": [ { "cvssV3_1": { "version": "3.1", "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H", "baseScore": 7.8, "baseSeverity": "HIGH" }, "scenarios": [ { "lang": "en", "value": "AV:L - The vulnerability is reached exclusively through RTM_NEWQDISC/tc netlink messages that invoke the qdisc ->change() handlers on a local netlink socket. Per kernel scoring guidance, tc/netlink qdisc configuration is a local attack vector.\nAC:L - The commit message contains a fully deterministic reproducer (tbf parent + fq child, ping to fill the queue, then change the limit and graft) with no race, no memory-layout dependence, and no non-default kernel config — fq and fq_codel are built in on virtually all distributions and fq_codel is the default qdisc.\nPR:L - tc_modify_qdisc is gated only by netlink_net_capable(skb, CAP_NET_ADMIN), which is evaluated against net->user_ns, so an unprivileged user obtains it with `unshare -Urn` and configures qdiscs on the loopback device in the new namespace. No real root in the init namespace is needed.\nUI:N - The attacker performs the entire sequence — creating the qdisc hierarchy, generating traffic into the queue, changing the limit, and grafting — with no victim action required.\nS:U - The corrupted state and its consequences are confined to the kernel's qdisc tree within the same security authority; there is no VM, IOMMU, or sandbox boundary crossed.\nC:H - The qdisc queue-accounting desync is the same primitive that has repeatedly escalated to use-after-free on qdisc class objects via wrongly-driven cops->qlen_notify() (CVE-2025-21703, CVE-2025-38000/38001), and a UAF on those objects yields attacker-controlled reads of freed kernel memory.\nI:H - The bug produces an attacker-controlled u32 underflow of qstats.backlog that propagates through qdisc_tree_reduce_backlog into every ancestor qdisc, corrupting kernel state that drives class active-list maintenance and enqueue/drop decisions; the resulting accounting desync is exploitable for heap corruption in the same manner as the prior net/sched accounting CVEs.\nA:H - The underflowed backlog permanently blackholes the device (bfifo/plug limit checks and sch_red's qavg see a ~4 GiB queue and drop everything), and the unguarded pre-fix trim loops give an outright kernel oops via qdisc_pkt_len(NULL) in codel/pie/fq_pie/fq_codel or an infinite loop in hhf while holding sch_tree_lock and RTNL." } ] } ], "affected": [ { "product": "Linux", "vendor": "Linux", "defaultStatus": "unaffected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "include/net/sch_generic.h", "net/sched/sch_codel.c", "net/sched/sch_fq.c", "net/sched/sch_fq_codel.c", "net/sched/sch_fq_pie.c", "net/sched/sch_hhf.c", "net/sched/sch_pie.c" ], "versions": [ { "version": "4b549a2ef4bef9965d97cbd992ba67930cd3e0fe", "lessThan": "a225f44d84b8900d679c5f5a9ea46fe9c0cc7802", "status": "affected", "versionType": "git" }, { "version": "4b549a2ef4bef9965d97cbd992ba67930cd3e0fe", "lessThan": "52bf272636bda69587952b35ae97690b8dc89941", "status": "affected", "versionType": "git" } ] }, { "product": "Linux", "vendor": "Linux", "defaultStatus": "affected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "include/net/sch_generic.h", "net/sched/sch_codel.c", "net/sched/sch_fq.c", "net/sched/sch_fq_codel.c", "net/sched/sch_fq_pie.c", "net/sched/sch_hhf.c", "net/sched/sch_pie.c" ], "versions": [ { "version": "3.5", "status": "affected" }, { "version": "0", "lessThan": "3.5", "status": "unaffected", "versionType": "semver" }, { "version": "6.16.4", "lessThanOrEqual": "6.16.*", "status": "unaffected", "versionType": "semver" }, { "version": "6.17", "lessThanOrEqual": "*", "status": "unaffected", "versionType": "original_commit_for_fix" } ] } ], "cpeApplicability": [ { "nodes": [ { "operator": "OR", "negate": false, "cpeMatch": [ { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "3.5", "versionEndExcluding": "6.16.4" }, { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "3.5", "versionEndExcluding": "6.17" } ] } ] } ], "references": [ { "url": "https://git.kernel.org/stable/c/a225f44d84b8900d679c5f5a9ea46fe9c0cc7802" }, { "url": "https://git.kernel.org/stable/c/52bf272636bda69587952b35ae97690b8dc89941" } ], "title": "net/sched: Fix backlog accounting in qdisc_dequeue_internal", "x_generator": { "engine": "bippy-1.2.0" } } } }