DF-3741 — Zero-stride VBIOS record walks cause infinite kernel loops (radeon_atom_init_mc_reg_table, atom_lookup_voltage_object_v3)

Severity: Medium · CWE: CWE-835 Loop with Unreachable Exit Condition · File: sys/dev/drm/radeon/radeon_atombios.c:4023-4045, 3423-3429

What this is

Two BIOS-gated infinite loops (kernel hang at driver init / first DPM use): (1) radeon_atom_init_mc_reg_table():4023-4045 walks reg_data in usRegDataBlkSize strides; with a BIOS stride of 0 and mem-id mismatch, neither the END sentinel nor the num_ranges<VBIOS_MAX_AC_TIMING_ENTRIES cap can ever trigger — non-terminating loop; with a nonzero stride the walk scans unbounded past the VBIOS image heap until it stumbles on 0xFFFFFFFF (OOB read). (2) atom_lookup_voltage_object_v3():3423-3429 advances by the object's self-declared usSize; usSize==0 with a type/mode mismatch never advances — non-terminating. Harness on guest: both replicas fail to terminate within 2^24 iterations (the cap exists only in the harness); a stride-4 control with an END marker terminates in 3. Same source certainty as pass-1 class but distinct functions; no AMD hardware on this guest.

Reproduce

./build.sh    # cc -O1 -g -Wall -o <bin> harness_hang.c   (DragonFly guest, cc 8.3)
./run.sh      # expected: both zero-stride walks hit the 2^24-iteration harness cap (non-termination); control terminates in 3; exit_code=0

The harness replicates the kernel parse arithmetic VERBATIM (cited path:line) against a synthetic crafted-VBIOS buffer with canary-guarded destinations.

Guest / environment

Guest has NO AMD/radeon GPU (virtio-only); /boot/kernel/radeon.ko exists (code shipped in the default install) but cannot attach, so the kernel code path is unreachable at runtime on this guest. Evidence is source-level certainty (path:line traced) plus a verbatim-arithmetic userland harness compiled and executed on the guest (build.log/run.log). See env.txt (uname, cc, kldstat, /boot/kernel/radeon.ko presence).

Fix

reject zero strides — see fix.diff (authored after verification; never applied to sys/).

DF-3741 — Zero-stride VBIOS record walks cause infinite kernel loops (radeon_atom_init_mc_reg_table, atom_lookup_voltage_object_v3)

Severity: Medium · CWE: CWE-835 Loop with Unreachable Exit Condition · File: sys/dev/drm/radeon/radeon_atombios.c:4023-4045, 3423-3429

What this is

Two BIOS-gated infinite loops (kernel hang at driver init / first DPM use): (1) radeon_atom_init_mc_reg_table():4023-4045 walks reg_data in usRegDataBlkSize strides; with a BIOS stride of 0 and mem-id mismatch, neither the END sentinel nor the num_ranges<VBIOS_MAX_AC_TIMING_ENTRIES cap can ever trigger — non-terminating loop; with a nonzero stride the walk scans unbounded past the VBIOS image heap until it stumbles on 0xFFFFFFFF (OOB read). (2) atom_lookup_voltage_object_v3():3423-3429 advances by the object's self-declared usSize; usSize==0 with a type/mode mismatch never advances — non-terminating. Harness on guest: both replicas fail to terminate within 2^24 iterations (the cap exists only in the harness); a stride-4 control with an END marker terminates in 3. Same source certainty as pass-1 class but distinct functions; no AMD hardware on this guest.

Reproduce

./build.sh    # cc -O1 -g -Wall -o <bin> harness_hang.c   (DragonFly guest, cc 8.3)
./run.sh      # expected: both zero-stride walks hit the 2^24-iteration harness cap (non-termination); control terminates in 3; exit_code=0

The harness replicates the kernel parse arithmetic VERBATIM (cited path:line) against a synthetic crafted-VBIOS buffer with canary-guarded destinations.

Guest / environment

Guest has NO AMD/radeon GPU (virtio-only); /boot/kernel/radeon.ko exists (code shipped in the default install) but cannot attach, so the kernel code path is unreachable at runtime on this guest. Evidence is source-level certainty (path:line traced) plus a verbatim-arithmetic userland harness compiled and executed on the guest (build.log/run.log). See env.txt (uname, cc, kldstat, /boot/kernel/radeon.ko presence).

Fix

reject zero strides — see fix.diff (authored after verification; never applied to sys/).

VERDICT — DF-3741

Status: inconclusive (source-certain; harness-reproduced arithmetic; kernel path unreachable on this guest) reproduced: 0 (in-kernel) — harness demonstration: YES Impact on real hardware: Medium — CWE-835 Loop with Unreachable Exit Condition

Finding

Two BIOS-gated infinite loops (kernel hang at driver init / first DPM use): (1) radeon_atom_init_mc_reg_table():4023-4045 walks reg_data in usRegDataBlkSize strides; with a BIOS stride of 0 and mem-id mismatch, neither the END sentinel nor the num_ranges<VBIOS_MAX_AC_TIMING_ENTRIES cap can ever trigger — non-terminating loop; with a nonzero stride the walk scans unbounded past the VBIOS image heap until it stumbles on 0xFFFFFFFF (OOB read). (2) atom_lookup_voltage_object_v3():3423-3429 advances by the object's self-declared usSize; usSize==0 with a type/mode mismatch never advances — non-terminating. Harness on guest: both replicas fail to terminate within 2^24 iterations (the cap exists only in the harness); a stride-4 control with an END marker terminates in 3. Same source certainty as pass-1 class but distinct functions; no AMD hardware on this guest.

Why not reproduced in-kernel

The QEMU guest is virtio-only (no AMD GPU): pciconf -l shows no 1002 device; kldstat shows radeon not loaded. /boot/kernel/radeon.ko exists (the vulnerable code ships in the default install) but cannot attach, so radeon_atombios.c parse paths never execute here. All conclusions are source-level (path:line above) plus the on-guest harness (harness_hang.c).

Harness result

Command: ./run.sh (after ./build.sh) Expected/observed: both zero-stride walks hit the 2^24-iteration harness cap (non-termination); control terminates in 3; exit_code=0 Full logs: build.log, run.log.

Impact / exploit chain

crafted VBIOS (VRAM_Info frev2/crev1 with usRegDataBlkSize=0, or VoltageObjectInfo v3 object with usSize=0) -> kldload/boot hangs the kernel in a tight loop -> local DoS (requires VBIOS control)

Fix

reject zero strides — fix.diff.

Honest notes

• DF-3741 does NOT re-report pass-1 findings DF-1198/1199/1200 (power-table state_index mismatch, asic_ss_info underflow, power-table indices) — different functions/tables/fields.
• fix validation not possible on this guest (no hardware); fix.diff is source-authored only.

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.

VERDICT — DF-3741

Status: inconclusive (source-certain; harness-reproduced arithmetic; kernel path unreachable on this guest) reproduced: 0 (in-kernel) — harness demonstration: YES Impact on real hardware: Medium — CWE-835 Loop with Unreachable Exit Condition

Finding

Two BIOS-gated infinite loops (kernel hang at driver init / first DPM use): (1) radeon_atom_init_mc_reg_table():4023-4045 walks reg_data in usRegDataBlkSize strides; with a BIOS stride of 0 and mem-id mismatch, neither the END sentinel nor the num_ranges<VBIOS_MAX_AC_TIMING_ENTRIES cap can ever trigger — non-terminating loop; with a nonzero stride the walk scans unbounded past the VBIOS image heap until it stumbles on 0xFFFFFFFF (OOB read). (2) atom_lookup_voltage_object_v3():3423-3429 advances by the object's self-declared usSize; usSize==0 with a type/mode mismatch never advances — non-terminating. Harness on guest: both replicas fail to terminate within 2^24 iterations (the cap exists only in the harness); a stride-4 control with an END marker terminates in 3. Same source certainty as pass-1 class but distinct functions; no AMD hardware on this guest.

Why not reproduced in-kernel

The QEMU guest is virtio-only (no AMD GPU): pciconf -l shows no 1002 device; kldstat shows radeon not loaded. /boot/kernel/radeon.ko exists (the vulnerable code ships in the default install) but cannot attach, so radeon_atombios.c parse paths never execute here. All conclusions are source-level (path:line above) plus the on-guest harness (harness_hang.c).

Harness result

Command: ./run.sh (after ./build.sh) Expected/observed: both zero-stride walks hit the 2^24-iteration harness cap (non-termination); control terminates in 3; exit_code=0 Full logs: build.log, run.log.

Impact / exploit chain

crafted VBIOS (VRAM_Info frev2/crev1 with usRegDataBlkSize=0, or VoltageObjectInfo v3 object with usSize=0) -> kldload/boot hangs the kernel in a tight loop -> local DoS (requires VBIOS control)

Fix

reject zero strides — fix.diff.

Honest notes

• DF-3741 does NOT re-report pass-1 findings DF-1198/1199/1200 (power-table state_index mismatch, asic_ss_info underflow, power-table indices) — different functions/tables/fields.
• fix validation not possible on this guest (no hardware); fix.diff is source-authored only.

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.