β¬’ DragonFlyBSD Kernel Audit
← triage Β· dashboard
DF-1135

Heap OOB read in hdaa_eld_dump: computed length never bounds field/SAD access

Summary

hdaa_eld_dump at hdaa.c:636-714: only bounds guard is w->eld_len<4 at :644. len=min(eld_len, eld[2]*4) computed at :651 but NEVER used to bound subsequent accesses. Malicious HDMI/DP codec reports eld_len=4 but sets MNL(eld[4]&0x1f)=31 and SAD_Count(eld[5]>>4)=15 -> unconditional reads eld[4]..eld[19] (already past 4-byte allocation), monitor name at eld[20] for 31 chars, SAD loop up to eld[20+31+14*3+2]=eld[95] -> 91 bytes past 4-byte heap allocation. Gated behind HDA_BOOTVERBOSE (boot -v or hw.snd.verbose>3). Attacker: malicious HDA codec (QEMU VM host or compromised firmware). Impact: kernel panic (guard page) or heap info leak to console/msgbuf (world-readable). Fix: bound all accesses by len.

Discussion (0)

No comments yet.

PoC verification

Evidence pack

findings/poc/DF-1135 Β· 3 files
FileTypeDescriptionSize
VERDICT.md verdict Source-confirmation verdict for DF-1135 617 B ↓ raw
fix.diff suggested-fix Bound eld accesses with minimum length check 369 B view raw
../fix_build.log build-log Batch kernel build log (all fixes, rc=0) 5.6 MB ↓ download
VERDICT.md verdict Source-confirmation verdict for DF-1135
↓ download raw

DF-1135 Verification Verdict

Severity: Low Impact class: oob_read Verification method: Source-only confirmation (HW-gated, not triggerable on QEMU guest)

Verdict: REPRODUCED (source-confirmed)

The bug is confirmed in the audited source at the cited path:line. Triggerable but requires specific driver/config.

Fix: Bound eld accesses with minimum length check

Fix applied and validated in batch kernel build (rc=0, -Werror).

Fix validation

All 41-fix patches batched into single make -j6 nativekernel KERNCONF=X86_64_GENERIC build. Build result: rc=0, 0 errors (full -Werror clean).

Fix verification

fixed
baseline reproduced→ patch + rebuild →patched clean

VALIDATED: fix.diff applies cleanly + batch kernel build rc=0 -Werror; bug HW/module/runtime-gated, no runtime PoC re-test possible on guest.

VALIDATED: fix.diff applies cleanly + batch kernel build rc=0 -Werror; bug HW/module/runtime-gated, no runtime PoC re-test possible on guest.
↓ fix.diffcombined build rc=0

Confirmed kernel references

β€”

Detail

Exploit chain

none

Evidence (decisive lines)

REPRODUCED (source-only): hdaa_eld_dump bounds check (w->eld_len<4) is insufficient; computed len = min(eld_len, eld[2]*4) is never used to bound subsequent accesses. Malicious codec ELD -> OOB read.

Verified recommended fix

REPRODUCED (source-only): hdaa_eld_dump bounds check (w->eld_len<4) is insufficient; computed len = min(eld_len, eld[2]*4) is never used to bound subsequent accesses. Malicious codec ELD -> OOB read.

Verdict

REPRODUCED (source-only): hdaa_eld_dump bounds check (w->eld_len<4) is insufficient; computed len = min(eld_len, eld[2]*4) is never used to bound subsequent accesses. Malicious codec ELD -> OOB read.