DF-4953

aibs_attach ENXIO path leaks every kmalloc'd sc_asens_* sensor array — newbus never calls detach after a failed attach

Summary

Arrays allocated+published before element validation; malformed-but-well-typed elements skip attachment, and the zero-sensor ENXIO exit never frees - DFly newbus never detaches a failed attach, so the leak is permanent (re-attach orphans again). Crafted-ACPI precondition (pre-boot control). Free-on-ENXIO fix.

Discussion (0)

No comments yet.

Severity: Low
Status: new
Bucket: Base system — drivers
Confidence: certain
Pass: 2 · GLM 5.3 second pass
CVSS 3.1: CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:U/C:N/I:N/A:L
CWE: CWE-401
File: sys/dev/acpica/aibs/atk0110.c
Lines: 126-218
Area: dev
Discovered: 2026-09-25
Updated: 2026-09-25

Team triage

untriaged

priority: — · owner: —

PoC verification

untested

Status: untested
Reproduced: no
Impact: leak
Confidence: certain
Tested: 2026-09-25 10:09:29
Attempts: 0
Runtime: —
Code hash: —
Guest: —
Build cmd: —
Run cmd: —

jump to evidence pack ↓

Fix verification

not_testable

Fix status: not_testable
Baseline reproduced: —
Patched reproduced: —
Confidence: certain
Tested: 2026-09-25 10:09:29
Runtime: —
Kernel: per-fix-DF-4953
fix.diff: findings/poc/DF-4953/fix.diff
Runner: df-bsd-reviewer

jump to fix verification ↓

PoC verification

Evidence pack

findings/poc/DF-4953 · 1 files

File	Type	Description	Size
verdict.json	file		1.1 KB	view raw

Fix verification

not_testable

per-fix-DF-4953

Confirmed kernel references

Detail

Exploit chain

crafted ?SIF packages (valid integer header, all elements malformed) -> arrays allocated+published -> zero sensors attached -> ENXIO without free -> permanent leak

Evidence (decisive lines)

findings/poc/DF-4953/ (this verdict)

Verdict

Firmware-gated (crafted ATK0110 DSDT; no ASUS node on guest): aibs_attach_sif kmallocs and publishes the per-family sensor arrays before element validation; malformed-but-well-typed elements 'continue' so no sensors attach, and when all families yield zero the ENXIO exit never frees the arrays - and DFly newbus never calls detach after a failed attach (DS_NOTPRESENT), so the allocations leak permanently (a re-attach would orphan them again). One-shot ~112B/entry leak per family; pre-boot control precondition. Free-on-ENXIO fix.