CVE-2026-98644
bpf: Avoid soft lockup in __htab_map_lookup_and_delete_batch()
Published Oct 18, 2026
CVSS 3.1 · 4.4 Medium
EPSS < 1%
KEV · Not listed
Description
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.
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.
Affected Products
-
CVE record
-
NVD configurations
No data
-
Red Hat packages
No data
-
OpenCVE enrichment
Default status is the baseline for the product, each version can override it (e.g. patched versions marked unaffected).
| Vendor | Product | Default status | Versions | ||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Linux | Linux | unaffected |
|
||||||||||||||||||||||||||||||
| Linux | Linux | affected |
|
No data.
No data.
| Vendor | Product | Versions | ||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Linux | Linux Kernel |
|
||||||||||||||||||||||||||||||||||||||||
| Linux | Linux Kernel |
|
Remediation
Tracking
Advisories
| Source | ID | Title |
|---|---|---|
 Debian DLA |
DLA-4850-1 | linux-6.12 security update |
 Ubuntu USN |
USN-8961-1 | Linux kernel vulnerabilities |
| Ubuntu USN |
USN-9019-1 | Linux kernel (Azure) vulnerabilities |
| Ubuntu USN |
USN-9020-1 | Linux kernel vulnerabilities |
| Ubuntu USN |
USN-9022-1 | Linux kernel vulnerabilities |
| Ubuntu USN |
USN-9023-1 | Linux kernel vulnerabilities |
| Ubuntu USN |
USN-9024-1 | Linux kernel vulnerabilities |
| Ubuntu USN |
USN-9025-1 | Linux kernel vulnerabilities |
| Ubuntu USN |
USN-9026-1 | Linux kernel (Azure) vulnerabilities |
| Ubuntu USN |
USN-9029-1 | Linux kernel (Raspberry Pi) vulnerabilities |
| Ubuntu USN |
USN-9039-1 | Linux kernel (Azure CVM) vulnerabilities |
| Ubuntu USN |
USN-9040-1 | Linux kernel (Low Latency NVIDIA) vulnerabilities |
| Ubuntu USN |
USN-9066-1 | Linux kernel (HWE) vulnerabilities |
| Ubuntu USN |
USN-9091-1 | Linux kernel (NVIDIA Tegra) vulnerabilities |
Metrics
No CVSS v4.0
Attack Vector Local
Attack Complexity Low
Privileges Required High
Scope Unchanged
Confidentiality Impact None
Integrity Impact None
Availability Impact High
User Interaction None
No CVSS v3.0
No CVSS v2
This CVE is not in the KEV list.
Exploit prediction score: < 1% (0.206% probability within 30 days)
Key SSVC decision points have not yet been added.
References
History
Oct 23, 2026 12:45 AM
Weaknesses updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Weaknesses | CWE-12 |
Oct 22, 2026 12:30 PM
Weaknesses updated. Added references. Scoring metrics updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Weaknesses | CWE-1050 | |
| References |
| |
| Metrics |
threat_severity
|
cvssV3_1
|
Oct 18, 2026 7:30 PM
Weaknesses updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Weaknesses | CWE-12 |
Oct 18, 2026 3:30 PM
Description updated. Title updated. First appearance in the feed. Affected versions updated. Affected vendors updated. Added references
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Description | 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. | |
| Title | bpf: Avoid soft lockup in __htab_map_lookup_and_delete_batch() | |
| First Time appeared |
Linux
Linux linux Kernel |
|
| CPEs | cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* | |
| Vendors & Products |
Linux
Linux linux Kernel |
|
| References |
|
|
Subscriptions
Weaknesses
