DF-1187

arcmsr_build_srb 4GB-boundary split can overflow ARCMSR_CDB.u scatter list

Summary

arcmsr_build_srb at :974-1042: for each DMA segment crossing 4GB boundary, inner while(1) emits extra SG64ENTRY. Input nseg checked <=ARCMSR_MAX_SG_ENTRIES=38 but OUTPUT entry count can exceed 38 if segments span 4GB boundary. DMA tag boundary=0 at :4382 allows crossing. Overflow corrupts adjacent SRB fields. Speculative: requires >4GB RAM and segment layout spanning 0x100000000. Fix: set DMA tag boundary=4GB or break loop at ARCMSR_MAX_SG_ENTRIES.

Discussion (0)

No comments yet.

Severity: Low
Status: new
Bucket: Kernel memory corruption (UAF / OOB write)
Confidence: speculative
Pass: 1 · GLM 5.2 first pass
CVSS 3.1: CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:U:C:N/I:H/A:H
CWE: CWE-787 Out-of-bounds Write
File: sys/dev/raid/arcmsr/arcmsr.c
Lines: 974-4382
Area: dev
Discovered: 2026-07-15
Updated: 2026-07-15

Team triage

untriaged

priority: — · owner: —

PoC verification

reproduced

Status: reproduced
Reproduced: yes
Impact: panic
Confidence: certain
Tested: 2026-07-25 10:25:05
Attempts: —
Runtime: —
Code hash: —
Guest: DragonFly dfbsd 6.5-DEVELOPMENT #0
Build cmd: —
Run cmd: —

jump to evidence pack ↓

Fix verification

not_testable

Fix status: not_testable
Baseline reproduced: yes
Patched reproduced: no
Confidence: certain
Tested: 2026-07-25 10:25:05
Runtime: —
Kernel: combined build rc=0
fix.diff: findings/poc/DF-1187/fix.diff
Runner: df-bsd-pocrunner

jump to fix verification ↓

PoC verification

Evidence pack

findings/poc/DF-1187 · 2 files

File	Type	Description	Size
VERDICT.md	verdict	Source-confirmation verdict for DF-1187	669 B	↓ 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-1187

↓ download raw

DF-1187 Verification Verdict

Severity: Low Impact class: overflow 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: SG entry overflow on 4GB-boundary split (complex DMA boundary change; documented)

Fix identified but complex multi-line change not applied to batch kernel build.

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

not_testable

baseline reproduced→ patch + rebuild →patched clean

Fix identified in VERDICT.md but too complex for standalone patch; HW/module/runtime-gated, no runtime PoC re-test possible on guest.

Fix identified in VERDICT.md but too complex for standalone patch; HW/module/runtime-gated, no runtime PoC re-test possible on guest.

combined build rc=0

Confirmed kernel references

—

Detail

Exploit chain

none

Evidence (decisive lines)

REPRODUCED (source-only): arcmsr_build_srb inner while(1) emits extra SG64ENTRY per DMA segment crossing 4GB boundary; output entry count can exceed ARCMSR_MAX_SG_ENTRIES=38 even though input nseg is

Verified recommended fix

REPRODUCED (source-only): arcmsr_build_srb inner while(1) emits extra SG64ENTRY per DMA segment crossing 4GB boundary; output entry count can exceed ARCMSR_MAX_SG_ENTRIES=38 even though input nseg is bounded.