DF-3784 — verdict

REPRODUCED | reproduced=1 | impact=leak | confidence=certain

REPRODUCED at source level (decision-logic harness): with iov_len=0x80000000 the (int) cast yields -2147483648; neither guard rejects it; uiomove's n = 0xffffffff80000000 and its loop copies MIN(2GiB, n) = 2147483648 bytes from vbibuffer. The 2 GiB kernel disclosure itself needs the device (VBI data must exist) and cannot run on this guest.

Root cause (citations)

vbi_read() computes readsize = (int)uio->uio_iov->iov_len (bktr_core.c:1159). DragonFly's uio_resid/iov_len are 64-bit size_t and read() accepts any positive ssize_t length, so read(fd, buf, 0x80000000) (2 GiB) is legal and truncates to INT_MIN. The clamp if (readsize > bktr->vbisize) (1161) and the wrap check (1165) both fail to catch a negative value, and uiomove(vbibuffer + vbistart, readsize, uio) (1177) receives (size_t)(-2147483648) = 0xffffffff80000000. DF's uiomove (sys/kern/kern_subr.c) copies MIN(iov_len, n) = 2 GiB from the kernel VA of vbibuffer into the user buffer until the first unmapped kernel page faults (copyout EFAULT, partial data already delivered). vbibuffer is a contigmalloc'd kernel_map allocation, so everything mapped above it - kernel heap, data, pointers - is disclosed. Additionally bktr->vbisize/vbistart are corrupted by int overflow (1183-1187).

Unprivileged on any machine with the driver: /dev/vbi0 is 0444; the read blocks until VBI data is produced (attacker's second fd runs continuous PAL capture), then over-reads.

Guest reality (why no runtime kernel trigger)

The audit guest is DragonFly 6.5-DEVELOPMENT x86_64 (X86_64_GENERIC). The bktr driver is NOT compiled into the GENERIC kernel (sys/conf/files: optional bktr pci), there is no /boot/modules tree, no /dev/bktr, /dev/tuner, /dev/vbi* nodes, no Brooktree PCI device (QEMU has no Bt848/Bt878 device model; pciconf -lv shows no vendor 109e). Therefore the kernel code path is unreachable ON THIS GUEST. Verification is source-level: faithful harnesses of the exact kernel arithmetic (the same methodology as the accepted DF-1216 evidence pack), run on the DragonFly guest with its native cc, plus fix.diff compile-validation against the guest's /usr/src. All PoC triggers compile cleanly against the guest's kernel headers (poc_build_and_device_check.log).

Exploit chain

open /dev/vbi0 (0444) + continuous PAL capture on /dev/bktr0 -> read(/dev/vbi0, buf, 0x80000000) -> (int) truncation -> uiomove with n=0xffffffff80000000 -> 2 GiB of kernel VA copied to userspace until EFAULT; vbisize/vbistart corrupted.

Decisive evidence

run.log: step-by-step guard evaluation showing every check bypassed and the final 2 GiB copy length; post-state corruption vbisize=-2147418112.

Fix validation

fix.diff (in this directory) applies cleanly against the pristine sys/ tree (git apply --check) and compiles as part of the bktr KLD module on the guest's /usr/src with -Werror together with the seven other pass-2 bktr fixes (fix_build.log, MAKE_RC=0, bktr.ko + bktr_mem.ko produced). Runtime patched-behavior validation is not possible on this guest (no device), hence fix_status=not_testable (compile-validated + logic-reviewed).

Kernel references

• sys/dev/video/bktr/bktr_core.c:1159
• sys/dev/video/bktr/bktr_core.c:1161
• sys/dev/video/bktr/bktr_core.c:1165
• sys/dev/video/bktr/bktr_core.c:1177
• sys/kern/kern_subr.c:uiomove
• sys/kern/sys_generic.c:136

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.