⬢ DragonFlyBSD Kernel Audit
DF-4888 / verdict.json
← back to finding ↓ download raw 
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
{
  "finding_id": "DF-4888",
  "status": "untested",
  "reproduced": 0,
  "impact": "Permanent BAR/rman resource leak on every failed attach; late attach failure (bus_setup_intr) leaks all DMA state and /dev/iirN and leaves XPT-registered CAM SIMs with no interrupt handler - CAM scan CCBs hang forever (iir_timeout empty), wedging the rescan thread; hardware-gated local DoS/resource exhaustion (CWE-404)",
  "confidence": "certain",
  "verdict": "Source-verified, hardware-gated: every iir_pci_attach error path leaks the BAR0 resource (release commented out at sys/dev/raid/iir/iir_pci.c:349-352, and the saved rid was clobbered to 0 at :195), and a bus_setup_intr failure after successful iir_init()/iir_attach() (:336-341) releases only the IRQ - no iir_free(), no XPT/sim/shutdown-handler teardown - leaving leaked pools plus live registered SIMs whose completions can never be delivered (CCBs hang; iir_timeout is empty per DF-4176). Not runnable on the audit guest: no Intel 0x8086:0x0600 / ICP controller exists in pciconf (env.txt); driver is compiled in (kldstat -v: 100 pci/iir) so the code path is live on any host with the card.",
  "exploit_chain": "",
  "evidence": "findings/poc/DF-4888/VERDICT.md (full path:line trace); findings/poc/DF-4888/fix.diff; guest hardware-absence proof in findings/poc/DF-4888/env.txt",
  "kernel_refs": [
    "sys/dev/raid/iir/iir_pci.c:186",
    "sys/dev/raid/iir/iir_pci.c:195",
    "sys/dev/raid/iir/iir_pci.c:326",
    "sys/dev/raid/iir/iir_pci.c:333",
    "sys/dev/raid/iir/iir_pci.c:336",
    "sys/dev/raid/iir/iir_pci.c:346",
    "sys/dev/raid/iir/iir.c:199",
    "sys/dev/raid/iir/iir.c:272",
    "sys/dev/raid/iir/iir.c:494",
    "sys/dev/raid/iir/iir.c:514"
  ],
  "poc_changes": "no runnable PoC (hardware absent); evidence pack documents the trace and fix",
  "attempts": 0,
  "guest_uname": "DragonFly dfbsd 6.5-DEVELOPMENT #0: Thu Jul 2 06:02:54 UTC 2026 root@dfbsd:/usr/obj/usr/src/sys/X86_64_GENERIC x86_64",
  "runtime_sec": 0,
  "guest_dirty": 0,
  "build_cmd": "",
  "run_cmd": "",
  "code_hash": "",
  "notes": "Residual after fix.diff: iir_attach() itself leaves sims[i] dangling (non-NULL) when xpt_create_path() fails at bus i (iir.c:503-510); the fixs cleanup keys on paths[i] != NULL so it stays correct, but a follow-up should NULL sims[i] there. Severity Low consistent with sibling iir findings DF-4173/DF-4175/DF-4176.",
  "recommended_fix": "Save the memory rid and release io on every error path; on bus_setup_intr failure deregister XPT paths/SIMs, drop the shutdown handler and call iir_free() (see fix.diff)",
  "fix_status": "not_testable",
  "fix_baseline_reproduced": 0,
  "fix_patched_reproduced": 0,
  "fix_verdict": "fix.diff authored post-verification against the read-only sys/ tree; not build-validated (no controller hardware to attach the patched kernel to)",
  "fix_evidence": "findings/poc/DF-4888/fix.diff"
}