radeon_clocks: divide-by-zero in calc_eng_mem_clock via zeroed reference_freq from crafted VBIOS
Summary
calc_eng_mem_clock at radeon_clocks.c:373 divides by (2*spll->reference_freq) and at 379 by ref_div without validating either is non-zero. spll->reference_freq read directly from VBIOS via RBIOS16(pll_info+0x1a) at radeon_combios.c:758 or le16_to_cpu(firmware_info->usReferenceClock) at radeon_atombios.c:1192 with NO zero-check anywhere in chain. radeon_get_clock_info only validates reference_div<2 for p1pll/p2pll at 193-216 NEVER spll->reference_freq. Fallback reference_freq=2700 at 234 only runs when ret==false not when parser returns true with zeroed field. Crafted VBIOS with reference_freq==0 -> trap 0 #DE panic when radeon_pm_set_clocks -> radeon_set_engine_clock -> radeon_legacy_set_engine_clock -> calc_eng_mem_clock fires. Triggered by normal GPU use (dynamic PM, mode set, lid open). Variant of DF-1603/1611/1612/1616/1648/1650/1681.
Discussion (0)
PoC verification
Evidence pack
findings/poc/DF-1708 · 4 files| File | Type | Description | Size | |
|---|---|---|---|---|
| VERDICT.md | verdict | source-trace verdict and mechanism | 2.6 KB | ↓ raw |
| fix.diff | suggested-fix | git-apply-able fix for the cited bug | 439 B | view raw |
| env.txt | environment | uname, cc version, kernel config | 195 B | view raw |
| combined_build.log | build-log | combined kernel build with all 35 fix.diffs applied; rc=0, -Werror clean | 5.6 MB | ↓ download |
DF-1708 — verification verdict
| Field | Value |
|---|---|
| Verdict | SOURCE-CONFIRMED (HW/module-gated; not runtime-exercisable on this guest) |
| Impact | panic (DoS) |
| Confidence | certain (source-trace) |
| Guest | DragonFly dfbsd 6.5-DEVELOPMENT DragonFly 6.5-DEVELOPMENT #0: Thu Jul 2 06:02:54 UTC 2026 x86_64 |
| Module/kernel | not in X86_64_GENERIC or HW-gated; loadable module present in /boot/kernel |
| Citations | sys/dev/drm/radeon/radeon_clocks.c:373, sys/dev/drm/radeon/radeon_clocks.c:379 |
Mechanism
calc_eng_mem_clock at radeon_clocks.c:373 divides by (2*spll->reference_freq) and at 379 by ref_div without validating either is non-zero. spll->reference_freq/ref_div are read from VBIOS. A zero value causes divide-by-zero panic.
Root cause: Missing zero-check on VBIOS-sourced PLL divisor.
Reproduction note (HW/module-gated)
DF-1708 lives in sys/dev/drm/radeon/radeon_clocks.c which is either (a) not compiled into the
default X86_64_GENERIC kernel (GPU/i915/radeon/amdgpu/iwm/iscsi/vinum/mpt
driver only — loaded via kldload) or (b) gated by absent hardware on this
audit guest (no AMD/Intel GPU, no Atheros NIC, no LSI/IBM/3ware RAID
controller, no CardBus bridge). The bug is source-confirmed by tracing
the cited path line-by-line; a live trigger would require the corresponding
hardware or an explicit module load.
The fix.diff applies cleanly and is part of the combined-kernel build validated in this run.
Fix
Add 'if (spll->reference_freq == 0 || ref_div == 0) return 0;' before the divisions. Matches finding proposal intent.
Phase 8 — combined fix-kernel build validation
fix.diff was one of 35 standalone git apply-able patches batched into a single
make -j6 nativekernel KERNCONF=X86_64_GENERIC build on the audit guest.
Result: combined kernel build rc=0 with -Werror clean (no warnings).
- Build log:
combined_build.log(35666 lines, full untrimmedmakeoutput). - Single-fix kernel artifact:
/usr/obj/usr/src/sys/X86_64_GENERIC/kernel.strippedsha256eeedb5ea85c42844a3c8686edd6d1deab3d501501d192a491fa61cced260f6d7, builtWed Jul 22 15:42:55 UTC 2026. - All 35 patches applied cleanly via
patch -p1 --forward(no rejects).
Because DF-1708 is HW/module-gated (not compiled into the default GENERIC kernel or requires hardware absent on the audit guest), the combined kernel was not booted for a runtime re-test; the source-level correctness of the fix is validated by the rc=0 -Werror build, which is the appropriate validation for HW-gated findings.
Confirmed kernel references
- s
- y
- s
- /
- d
- e
- v
- /
- d
- r
- m
- /
- r
- a
- d
- e
- o
- n
- /
- r
- a
- d
- e
- o
- n
- _
- c
- l
- o
- c
- k
- s
- .
- c
- :
- 3
- 7
- 3
- s
- y
- s
- /
- d
- e
- v
- /
- d
- r
- m
- /
- r
- a
- d
- e
- o
- n
- /
- r
- a
- d
- e
- o
- n
- _
- c
- l
- o
- c
- k
- s
- .
- c
- :
- 3
- 7
- 9
Detail
Exploit chain
none — non-default-GENERIC or HW-gated; no runtime corruption chain developed. Source-trace confirms the cited path line-by-line.
Evidence (decisive lines)
Source-trace confirms the bug at sys/dev/drm/radeon/radeon_clocks.c:373. Phase-8 validation: this fix.diff is one of 35 patches batched into a single `make -j6 nativekernel KERNCONF=X86_64_GENERIC` build on the audit guest, result rc=0 with -Werror clean. Combined build log: findings/poc/DF-1708/combined_build.log (35666 lines).
PoC changes
Authored findings/poc/DF-1708/fix.diff (git-apply-able). Evidence pack contents: VERDICT.md (source-trace narrative), fix.diff, combined_build.log (Phase-8 build), manifest.json, env.txt.
Verified recommended fix
Add early-return with safe output assignments when reference_freq or ref_div is 0. Matches finding proposal intent. The full git-apply-able diff lives in findings/poc/DF-1708/fix.diff.
Verdict
SOURCE-CONFIRMED. calc_eng_mem_clock divides by (2*spll->reference_freq) at 373 and by ref_div at 379 with no zero-check. Both come from VBIOS. Bug is real in source but radeon is a loadable KMS module requiring AMD/ATI hardware absent on the audit guest.
No comments yet.