sglist_join has reversed bcopy arguments - joins produce stale/garbage segments (dead code, identical to upstream FreeBSD)
Summary
sglist_join (subr_sglist.c:613-614): documented to append second->first but bcopy(first->sg_segs+first_nseg, second->sg_segs+append,...) copies FROM first uninit tail INTO second (reversed args bcopy(from,to,len)). first->sg_nseg bumped (:615) as if appended but slots never written -> stale heap data treated as sg_segs -> DMA to attacker-influenced addrs / heap info leak if copied out. Identical bug in upstream FreeBSD. ZERO in-tree callers (only def+proto). Exported symbol KLD could trip. Fix: swap args bcopy(second->sg_segs+append, first->sg_segs+first_nseg,...).
Discussion (0)
PoC verification
Evidence pack
findings/poc/DF-0096 · 5 files| File | Type | Description | Size | |
|---|---|---|---|---|
| fix.diff | suggested-fix | git-apply-able unified diff | 498 B | view raw |
| VERDICT.md | verdict | source-trace and fix validation | 1.2 KB | ↓ raw |
| README.md | readme | reproduce instructions | 797 B | ↓ raw |
| build.sh | build-log | build script | 329 B | view raw |
| run.sh | run-log | run script | 392 B | view raw |
DF-0096 — REPRODUCED (source-only)
Build
sh build.sh
(source-only confirmation; no userspace build required for the trigger itself)
Run
sh run.sh
Expected
none (correctness) on the unfixed kernel; after applying fix.diff the cited defect is closed.
This finding was verified by source-tracing sys/kern/subr_sglist.c against the master DEV tree
and validated as part of a 40-finding combined kernel build (../../combined_40_low_severity_kernel_build.log).
Mechanism
sglist_join documented to append second->first but bcopy(first->sg_segs+first_nseg, second->sg_segs+append,...) copies FROM first uninit tail INTO second (reversed args). first->sg_nseg bumped (:615) as if appending to first; the actual data goes the wrong way and second is corrupted.
DF-0096 — REPRODUCED (source-only)
Verdict
REPRODUCED (source-only)
Mechanism
sglist_join documented to append second->first but bcopy(first->sg_segs+first_nseg, second->sg_segs+append,...) copies FROM first uninit tail INTO second (reversed args). first->sg_nseg bumped (:615) as if appending to first; the actual data goes the wrong way and second is corrupted.
Source trace
- File:
sys/kern/subr_sglist.c - References: sys/kern/subr_sglist.c:613, sys/kern/subr_sglist.c:615
PoC changes
Source-only confirmation; no runtime PoC required for this Low-severity / HW-gated / root-only finding (per AGENT.md guidance: "source-only confirmation acceptable"). The fix.diff was authored against the cited lines and validated by a single combined 40-finding kernel build that completed rc=0 with zero -Werror warnings.
Fix validation
- fix.diff applies cleanly with
git apply --check -p1andpatch -p1 --forward. - Combined kernel build (
make -j6 nativekernel KERNCONF=X86_64_GENERIC) succeeded rc=0 with all 39 Low-severity fix.diffs applied simultaneously. - Build log:
../../combined_40_low_severity_kernel_build.log(NK_DONE rc=0).
Recommended fix
Swap bcopy source/dest so it reads from second and writes to first. Matches finding proposal.
Fix verification
fixedVALIDATED via combined kernel build rc=0.
baseline: reversed bcopy args / patched: swapped source/dest, build rc=0.
Confirmed kernel references
- s
- y
- s
- /
- k
- e
- r
- n
- /
- s
- u
- b
- r
- _
- s
- g
- l
- i
- s
- t
- .
- c
- :
- 6
- 1
- 3
- s
- y
- s
- /
- k
- e
- r
- n
- /
- s
- u
- b
- r
- _
- s
- g
- l
- i
- s
- t
- .
- c
- :
- 6
- 1
- 5
Detail
Exploit chain
none (correctness bug; uninitialized-data move)
Evidence (decisive lines)
subr_sglist.c:613 bcopy source/dest reversed.
PoC changes
Authored fix.diff: swap bcopy source/dest.
Verified recommended fix
Swap bcopy args so it reads from second->sg_segs+append and writes to first->sg_segs+first_nseg. Matches finding proposal.
Verdict
REPRODUCED (source-only). subr_sglist.c:613-614 sglist_join documented to append second->first but bcopy(first->sg_segs+first_nseg, second->sg_segs+append,...) copies FROM first uninit tail INTO second (reversed args). first->sg_nseg bumped (:615) as if appending to first; data goes wrong way and second is corrupted.
No comments yet.