DF-5267

Five linear rate lookups store MTW_RIDX_MAX (36) on a miss; the stale ridx then indexes rt2860_rates[36] one element out of bounds (mtw_tx/mtw_set_tx_desc/tx_mgt/radiotap wt_rate)

Summary

rt2860_rates[36] but loops exit with ridx==36 on a miss (media_change/iter_func/newassoc/sendprot/tx_param) feeding rt2860_rates[ridx] one-past-end reads in tx/set_tx_desc/tx_mgt + radiotap wt_rate (bpf-visible). Requires an off-table rate (odd net80211 state) - speculative. Twin of DF-0984. Clamp-after-miss fix at all 5 sites.

Discussion (0)

No comments yet.

Severity: Low
Status: new
Bucket: Kernel memory leaks / uninitialized disclosure
Confidence: speculative
Pass: 2 · GLM 5.3 second pass
CVSS 3.1: CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:L/I:N/A:N
CWE: CWE-125
File: sys/bus/u4b/wlan/if_mtw.c
Lines: 1720-3430
Area: hw
Discovered: 2026-10-01
Updated: 2026-10-01

Team triage

untriaged

priority: — · owner: —

PoC verification

untested

Status: untested
Reproduced: no
Impact: none
Confidence: speculative
Tested: 2026-10-01 21:11:25
Attempts: 1
Runtime: 3s
Code hash: n/a
Guest: DragonFly dfbsd 6.5-DEVELOPMENT DragonFly 6.5-DEVELOPMENT #0: Thu Jul 2 06:02:54 UTC 2026 root@dfbsd:/usr/obj/usr/src/sys/X86_64_GENERIC x86_64
Build cmd: n/a
Run cmd: n/a

jump to evidence pack ↓

Fix verification

not_testable

Fix status: not_testable
Baseline reproduced: no
Patched reproduced: no
Confidence: speculative
Tested: 2026-10-01 21:11:25
Runtime: —
Kernel: per-fix-DF-5267
fix.diff: findings/poc/DF-5267/fix.diff
Runner: df-bsd-reviewer

jump to fix verification ↓

PoC verification

Evidence pack

findings/poc/DF-5267 · 3 files

File	Type	Size
env.txt	—	183 B	view raw
fix.diff	—	1.3 KB	view raw
verdict.json	—	2.9 KB	view raw

Fix verification

not_testable

baseline no→ patch + rebuild →patched clean

fix.diff authored and verified with 'git apply --check' against the pristine sys/ tree, but cannot be kernel-tested: the guest has no USB controller and no MT7601U device model, so neither baseline nor patched kernel behavior is exercisable. Hardware-gated: not_testable.

fix.diff (git-apply clean); VERDICT.md

↓ fix.diffper-fix-DF-5267

Confirmed kernel references

Detail

Exploit chain

node with ni_txrate not in rt2860_rates -> amrr_ridx=36 -> data->ridx=36 -> rt2860_rates[36] OOB read -> value lands in TXWI MCS field and radiotap wt_rate (bpf listeners)

Evidence (decisive lines)

Code-cited: 5 loops + 4 use sites; array size counted (36 entries). untested (hardware-gated)

Verified recommended fix

Clamp ridx to MTW_RIDX_CCK1 after each miss

Verdict

rt2860_rates[] (if_mtwreg.h:963-998) has exactly 36 entries (indices 0..35) and MTW_RIDX_MAX=36. Five lookup loops exit with ridx==MTW_RIDX_MAX when the searched rate is not in the table - mtw_media_change (1720-1723, rn->fix_ridx), mtw_iter_func (2320-2327, rn->amrr_ridx, e.g. an ni_txrate not among the 36 codes), mtw_newassoc (2394-2396, mn->mgt_ridx), mtw_sendprot (3349-3351) and mtw_tx_param (3428-3430) (data->ridx). The stored ridx is later used to index rt2860_rates[ridx] one element past the end: mtw_tx (3115-3122, ctl_ridx), mtw_set_tx_desc (2998-3034, .mcs/.phy), mtw_tx_mgt (3258, .rate) and the tx callback's radiotap tap->wt_rate = rt2860_rates[data->ridx].rate (2846) which leaks the adjacent static bytes to any radiotap listener. One-element OOB read of a const array (adjacent rodata), impact is a 1-2 byte info leak of static data plus wrong hardware fields; a miss requires an off-table rate (e.g. odd ni_txrate), hence speculative. Twin of DF-0984 (urtwn ridx2rate).