cd9660_vptofh truncates ino_t into int ifid_ino; cd9660_vget_internal's int lbn truncation makes the volume-space bounds check bypassable — NFS file-handle inode aliasing and negative-offset bread on crafted/large images
Summary
vptofh stores ip->i_number (ino_t, up to ~2^43 since isodirino = (extent+ext_attr)<<bshift) into int ifid_ino (:924); fhtovp passes it back sign-extended (:677). In vget_internal the int lbn = lblkno(imp, ino) (:774) truncates, so for ino >= 2^42 (extent 0x80000000 in a crafted image, or legitimate >4GB-path extents) lbn is negative and the guard 'lbn >= imp->volume_space_size' (:775) is skipped (signed compare against a negative value false). The subsequent bread at lblktooff(imp, lbn) < 0 is contained to EIO by the device strategy's dscheck bounds, so no corruption demonstrable; the observable defect is NFS file-handle aliasing (truncated handles resolve to the wrong vnode / ESTALE churn) on exports of such images. NFS-export-gated (cd9660 fhtovp is the only VFS_VGET caller). Not run (NFS-export setup root-gated; memory-safety ceiling traced to EIO). Fix: widen the guard arithmetic to off_t and reject ino >= volume_space_size<<bshift in vptofh. [VERIFIED pass2: NFS-export gated; memory-safety ceiling traced to EIO containment; fix compile-validated]
Discussion (0)
PoC verification
Evidence pack
findings/poc/DF-3126 · 9 files| File | Type | Description | Size | |
|---|---|---|---|---|
| fix.diff | — | 542 B | view raw | |
| README.md | — | 408 B | ↓ raw | |
| VERDICT.md | — | 1.6 KB | ↓ raw | |
| build.sh | — | 106 B | view raw | |
| run.sh | — | 120 B | view raw | |
| manifest.json | — | 498 B | view raw | |
| verdict.json | — | 1.3 KB | view raw | |
| fix.diff | — | 542 B | view raw | |
| env.txt | — | 272 B | view raw |
DF-3126 -- verification pack
Finding: cd9660 vptofh int truncation and vget_internal int lbn guard bypass
Status: not_reproduced / impact denial_of_service / confidence likely
What was run
n/a
How to reproduce
n/a (requires NFS-exported cd9660 mount)
Expected
n/a
See VERDICT.md for the full narrative and verdict.json for the persisted machine verdict. Artifacts: fix.diff.
DF-3126 verification
Kernel: 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
Bottom line
Not dynamically exercised: cd9660_vptofh/vget reachability requires an NFS export of a cd9660 mount (VFS_VGET is called only from the NFS file-handle path) and the memory-safety ceiling is traced to EIO: for an ino above 2^42 the int lbn truncation makes the lbn >= volume_space_size guard pass on negative values, but the subsequent bread at the negative lblktooff offset is contained by the device strategy bounds check. The observable defect is NFS file-handle aliasing (truncated ifid_ino binds handles to the wrong vnode). Fix widens the guard to off_t and the compile-validated patch is included.
Kernel references
Exploit chain
(none: availability/correctness finding; no privilege chain)
Fix
off_t lbn guard + reject oversized ino
See fix.diff in this directory (git-apply-able against the read-only sys/ tree; applied only inside the disposable verification guest).
Reproduction
build: n/a
run: n/a (requires NFS-exported cd9660 mount)
expected: n/a
Notes on the PoC (what changed vs the finding sketch)
(none)
Fix status (batch validation)
Applied with the 36 sibling fix.diffs of this batch in findings/poc/_batch41/all_fixes.patch; single in-guest make -j6 nativekernel KERNCONF=X86_64_GENERIC completed RC=0 with zero errors under -Werror (full log: findings/poc/_batch41/fix_build8.log). Behavior A/B intentionally skipped for this Info-tier item.
Fix verification
not_testableConfirmed kernel references
Detail
Evidence (decisive lines)
VERDICT.md line walk over vptofh (:924), fhtovp (:677) and vget_internal (:774-775)
Verified recommended fix
off_t lbn guard + reject oversized ino
Verdict
Not dynamically exercised: cd9660_vptofh/vget reachability requires an NFS export of a cd9660 mount (VFS_VGET is called only from the NFS file-handle path) and the memory-safety ceiling is traced to EIO: for an ino above 2^42 the int lbn truncation makes the lbn >= volume_space_size guard pass on negative values, but the subsequent bread at the negative lblktooff offset is contained by the device strategy bounds check. The observable defect is NFS file-handle aliasing (truncated ifid_ino binds handles to the wrong vnode). Fix widens the guard to off_t and the compile-validated patch is included.
No comments yet.