{
  "finding_id": "DF-4233",
  "status": "inconclusive",
  "reproduced": 0,
  "impact": "none",
  "confidence": "likely",
  "verdict": "Source-certain control-flow bug with a likely-corrupting consequence: txp_init() ignores txp_rxring_fill()s return (if_txp.c:1127) and enables TX/RX anyway (:1129-1130). A prior txp_stop freed all 256 rx clusters (txp_rxring_empty, :1081-1090) without clearing the DMA-shared rb_paddrlo ring or hv_rx_buf_write_idx; a mid-fill MCLGET(M_NOWAIT) failure (memory pressure, remotely seedable) leaves a stale prefix that the firmware still believes is available: inbound frames are DMAd into freed mbuf clusters (heap use-after-free WRITE with attacker-supplied frame content), and the completion descriptor hits m = sd->sd_mbuf == NULL -> NULL write at :676 (panic). Confidence likely only because the firmwares exact stale-write_idx behavior is inferred from the hostvar design, not observed. Runtime inconclusive: no 3c990 on the guest (run.log).",
  "exploit_chain": "memory pressure during ifconfig txp0 up (or a peer flooding while a re-init runs) -> half-filled ring -> inbound frame DMAd into a freed 2KB cluster with attacker-controlled bytes -> mbuf-cluster-zone UAF write -> groom victim object adjacent in the cluster zone -> kernel R/W; independently, the NULL-deref at :676 panics the kernel.",
  "evidence": "run.log (driver in kernel, NO_3COM_PCI_DEVICE, md5 parity), VERDICT.md (init/stop/fill traced path:line incl. the success-only write_idx at :1066), trigger.c, fix.diff",
  "kernel_refs": [
    "sys/dev/netif/txp/if_txp.c:1110",
    "sys/dev/netif/txp/if_txp.c:1127",
    "sys/dev/netif/txp/if_txp.c:1129",
    "sys/dev/netif/txp/if_txp.c:1130",
    "sys/dev/netif/txp/if_txp.c:1051",
    "sys/dev/netif/txp/if_txp.c:1066",
    "sys/dev/netif/txp/if_txp.c:1081",
    "sys/dev/netif/txp/if_txp.c:1089"
  ],
  "poc_changes": "No seed (pass-2 new); annotated precondition probe.",
  "attempts": 2,
  "guest_uname": "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",
  "runtime_sec": 90,
  "guest_dirty": 0,
  "build_cmd": "cc -o trig_4233 t4233.c (guest)",
  "run_cmd": "./trig_4233 && (pciconf -l | grep -i 10b7 || echo NO_3COM_PCI_DEVICE)",
  "code_hash": "87577089a12baeb38bf418ca0347af77133dde81b0130a29ca1bc511b52dff58",
  "notes": "The MGET/MCLGET failure precondition is stageable in-kernel, but with no device consuming the ring there is nothing to observe on the guest. First-up-after-attach variant additionally posts rb_paddrlo=0 entries (firmware DMA to physical 0) -- noted in the finding.",
  "recommended_fix": "Abort txp_init() when txp_rxring_fill() fails (leave the interface down), mirroring every other DFly NIC drivers init error path; see fix.diff.",
  "fix_status": "not_testable",
  "fix_baseline_reproduced": 0,
  "fix_patched_reproduced": 0,
  "fix_verdict": "fix.diff is a 7-line guarded early-return; not runtime-validable without the device.",
  "fix_evidence": "fix.diff; run.log absence proof"
}