GHSA-P6RX-8WRQ-4GC7

Vulnerability from github – Published: 2026-10-18 18:31 – Updated: 2026-10-18 18:31
VLAI
Details

In the Linux kernel, the following vulnerability has been resolved:

bpf: Avoid soft lockup in __htab_map_lookup_and_delete_batch()

__htab_map_lookup_and_delete_batch() has no rescheduling point. The batch count bounds how many entries are copied out, not how many buckets are visited, so one BPF_MAP_LOOKUP_BATCH call can walk the map end to end. The empty-bucket fast path is worse: it stays inside a single rcu_read_lock() / bpf_disable_instrumentation() section for any run of consecutive empty buckets.

That holds up on small maps, but it falls apart at scale. On a 144-CPU arm64 host running a CONFIG_PREEMPT_NONE kernel, periodic BPF_MAP_LOOKUP_BATCH calls against an LRU hash map with 16,777,216 buckets held a CPU inside the batch op for 77+ seconds and triggered the soft lockup watchdog.

Commit 75134f16e7dd ("bpf: Add schedule points in batch ops") fixed this same problem in the generic batch ops, but not in this htab-native path, which every htab-based hash map variant uses for its lookup[_and_delete] batch ops.

Complete that fix here. Leave the critical section after 64 consecutive empty buckets, call cond_resched_tasks_rcu_qs(), and resume at the saved bucket cursor. No locks are held at that point, and resuming from the cursor is already the function's behavior for non-empty buckets. Add the same call to the per-bucket loop after copy_to_user(), where every lock has been dropped. cond_resched_rcu() is not enough here: sleeping with bpf_prog_active elevated makes tracing programs on that CPU silently skip their invocations.

Plain cond_resched() is not enough either. It is a no-op under PREEMPT and PREEMPT_LAZY, the only models arm64 and x86 have offered since commit 7dadeaa6e851 ("sched: Further restrict the preemption modes"). It is also never a Tasks RCU quiescent state, in any model: the reschedule counts as a preemption. The walking task stays a holdout and stalls every synchronize_rcu_tasks() caller, ftrace and BPF trampoline teardown included, until the syscall returns [1]. cond_resched_tasks_rcu_qs() is the usual tool for that [2]. It reports the quiescent state at each yield and still reschedules as cond_resched() does on PREEMPT_NONE and PREEMPT_VOLUNTARY kernels.

Show details on source website
To clipboard 

{
  "affected": [],
  "aliases": [
    "CVE-2026-98644"
  ],
  "database_specific": {
    "cwe_ids": [],
    "github_reviewed": false,
    "github_reviewed_at": null,
    "nvd_published_at": "2026-10-18T16:17:17Z",
    "severity": null
  },
  "details": "In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Avoid soft lockup in __htab_map_lookup_and_delete_batch()\n\n__htab_map_lookup_and_delete_batch() has no rescheduling point. The\nbatch count bounds how many entries are copied out, not how many\nbuckets are visited, so one BPF_MAP_LOOKUP_BATCH call can walk the\nmap end to end. The empty-bucket fast path is worse: it stays inside\na single rcu_read_lock() / bpf_disable_instrumentation() section for\nany run of consecutive empty buckets.\n\nThat holds up on small maps, but it falls apart at scale. On a\n144-CPU arm64 host running a CONFIG_PREEMPT_NONE kernel, periodic\nBPF_MAP_LOOKUP_BATCH calls against an LRU hash map with 16,777,216\nbuckets held a CPU inside the batch op for 77+ seconds and triggered\nthe soft lockup watchdog.\n\nCommit 75134f16e7dd (\"bpf: Add schedule points in batch ops\") fixed this\nsame problem in the generic batch ops, but not in this htab-native path,\nwhich every htab-based hash map variant uses for its lookup[_and_delete]\nbatch ops.\n\nComplete that fix here. Leave the critical section after 64 consecutive\nempty buckets, call cond_resched_tasks_rcu_qs(), and resume at the saved\nbucket cursor. No locks are held at that point, and resuming from the\ncursor is already the function\u0027s behavior for non-empty buckets. Add the\nsame call to the per-bucket loop after copy_to_user(), where every lock\nhas been dropped. cond_resched_rcu() is not enough here: sleeping with\nbpf_prog_active elevated makes tracing programs on that CPU silently\nskip their invocations.\n\nPlain cond_resched() is not enough either. It is a no-op under PREEMPT\nand PREEMPT_LAZY, the only models arm64 and x86 have offered since\ncommit 7dadeaa6e851 (\"sched: Further restrict the preemption modes\").\nIt is also never a Tasks RCU quiescent state, in any model: the\nreschedule counts as a preemption. The walking task stays a holdout and\nstalls every synchronize_rcu_tasks() caller, ftrace and BPF trampoline\nteardown included, until the syscall returns [1].\ncond_resched_tasks_rcu_qs() is the usual tool for that [2]. It reports\nthe quiescent state at each yield and still reschedules as\ncond_resched() does on PREEMPT_NONE and PREEMPT_VOLUNTARY kernels.",
  "id": "GHSA-p6rx-8wrq-4gc7",
  "modified": "2026-10-18T18:31:57Z",
  "published": "2026-10-18T18:31:57Z",
  "references": [
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-98644"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/44f9bcb5ab2b268bc763ed3ebec83308b4972936"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/4e48fe9b1c77966bede0a73feb3059d7595f642d"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/5c8b1869cb99a32a962388848cd19604fb41263e"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/77a3b52b4455e6377e549539e757bc55eedefb51"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/85136bf22404474a815fc0ed26ec0d1cbc1bc3f9"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/ca5bd55fc508fbb5c08c62206d02237a5b6fc7e8"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/f0c09de08ec077098b7e1563f2285f1c136cd816"
    }
  ],
  "schema_version": "1.4.0",
  "severity": []
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.




Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.

Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline

Sightings and recorded dates, grouped by day · UTC

JSON

Loading timeline…




    Related by attack behaviour

    Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.