DF-1346 — Heap buffer overflow populating VCE/UVD SMC levels from an unclamped VBIOS MM-dependency count

File: sys/dev/drm/amd/powerplay/smumgr/polaris10_smumgr.c:1298 and 1404 Class: heap buffer overflow (write past fixed SMC level array)

Status: INCONCLUSIVE at runtime — confirmed real source bug, hardware-gated on this guest

The vulnerable code path was traced line-by-line in sys/ and confirmed to be a genuine bug (missing bounds check / integer overflow / UAF race). However it is not exercisable on the DragonFly audit guest because the guest has neither an AMD GPU nor any audio controller:

• PCI shows only vgapci0 class=0x030000 (QEMU stdvga, chip 0x11111234) — no AMD GPU.
• No PCI audio device (class 0x0401/0x0403); hw.snd empty; no /dev/dsp.
• amdgpu.ko is a loadable module only (NOT in X86_64_GENERIC), is not loaded, and cannot be kldload'd by an unprivileged user — and even if loaded would not attach without the hardware.

This is the valid hard-blocker case "vulnerable code path unreachable at runtime on this guest AND no harness can exercise it (device-integrated parser / ioctl / hardware-dependent race)." The bug is a real latent defect that would manifest on a system with the relevant hardware + the module loaded.

Mechanism (confirmed by source trace)

polaris10_populate_smc_vce_level() sets table->VceLevelCount = (uint8_t)(mm_table->count) (:1298) then loops count < VceLevelCount writing table->VceLevel[count] (:1301-1330). VceLevel[] is a fixed array of SMU74_MAX_LEVELS_VCE = 8 (smu74.h:139 / smu74_discrete.h:287). mm_table->count comes straight from the VBIOS ucNumEntries (u8, 0-255) with no clamp (process_pptables_v1_0.c:708), so count>8 overflows VceLevel[] into the following fields of SMU74_Discrete_DpmTable (and, for the smc_state_table, into polaris10_smumgr fields such as power_tune_defaults which is later dereferenced). polaris10_populate_smc_uvd_level() has the identical flaw for UvdLevel[SMU74_MAX_LEVELS_UVD=8] at :1404-1407.

Live trigger conditions

Requires an AMD Polaris10 GPU with amdgpu attached whose VBIOS MM-dependency table has ucNumEntries > 8 (crafted VBIOS via passthrough/hotplug). Reached during powerplay init. QEMU audit guest has no AMD GPU, no /dev/dri, amdgpu.ko not loaded => unreachable here.

Fix

A standalone, git apply-able fix is in fix.diff. Compile-validated: applied to in-guest /usr/src and the amdgpu module rebuilt under -Werror (rc=0, no warnings/errors in the patched translation unit). See build_fix.log.

Clamp the count to the fixed array size using the same ternary already used elsewhere in this file for the pcie-link levels (~:2191): VceLevelCount = (uint8_t)((SMU74_MAX_LEVELS_VCE < mm_table->count) ? SMU74_MAX_LEVELS_VCE : mm_table->count) and the analogous line for UvdLevelCount/SMU74_MAX_LEVELS_UVD. Supersedes the finding proposal (clamp count to SMU74_MAX_LEVELS_*).

Reproduce / validate

# 1. Confirm the bug site exists (read-only source trace):
grep -n ... sys/dev/drm/amd/powerplay/smumgr/polaris10_smumgr.c

# 2. Validate the fix compiles (on the audit guest):
scp -F dfbsd-qemu/config findings/poc/DF-1346/fix.diff dfbsd:/root/fix.diff
./dfbsd-qemu/vm.sh run_root 'cd /usr/src && patch -p1 --forward < /root/fix.diff'
./dfbsd-qemu/vm.sh run_root 'cd /usr/src/sys/dev/drm/amd/amdgpu && KERNCONF=X86_64_GENERIC SYSDIR=/usr/src/sys make -m /usr/src/share/mk'

# 3. (requires real hardware) Exercise the bug: attach an AMD GPU / audio device and trigger.

VERDICT — DF-1346

Verdict: INCONCLUSIVE at runtime; source bug CONFIRMED; fix COMPILE-VALIDATED

Citations confirmed: sys/dev/drm/amd/powerplay/smumgr/polaris10_smumgr.c:1298, sys/dev/drm/amd/powerplay/smumgr/polaris10_smumgr.c:1301, sys/dev/drm/amd/powerplay/smumgr/polaris10_smumgr.c:1404, sys/dev/drm/amd/powerplay/smumgr/polaris10_smumgr.c:1407, sys/dev/drm/amd/powerplay/inc/smu74.h:138, sys/dev/drm/amd/powerplay/inc/smu74_discrete.h:286, sys/dev/drm/amd/powerplay/hwmgr/process_pptables_v1_0.c:708

Is the bug real? — YES (source trace)

polaris10_populate_smc_vce_level() sets table->VceLevelCount = (uint8_t)(mm_table->count) (:1298) then loops count < VceLevelCount writing table->VceLevel[count] (:1301-1330). VceLevel[] is a fixed array of SMU74_MAX_LEVELS_VCE = 8 (smu74.h:139 / smu74_discrete.h:287). mm_table->count comes straight from the VBIOS ucNumEntries (u8, 0-255) with no clamp (process_pptables_v1_0.c:708), so count>8 overflows VceLevel[] into the following fields of SMU74_Discrete_DpmTable (and, for the smc_state_table, into polaris10_smumgr fields such as power_tune_defaults which is later dereferenced). polaris10_populate_smc_uvd_level() has the identical flaw for UvdLevel[SMU74_MAX_LEVELS_UVD=8] at :1404-1407.

Can it be reproduced on this guest? — NO (hardware-gated)

Requires an AMD Polaris10 GPU with amdgpu attached whose VBIOS MM-dependency table has ucNumEntries > 8 (crafted VBIOS via passthrough/hotplug). Reached during powerplay init. QEMU audit guest has no AMD GPU, no /dev/dri, amdgpu.ko not loaded => unreachable here.

Guest evidence (env.txt): only vgapci0 class=0x030000 chip=0x11111234 (QEMU stdvga); no AMD GPU; no PCI audio device; kldstat shows no drm/radeon/amdgpu/snd module; /dev/dri and /dev/dsp* do not exist. The amdgpu module is not in X86_64_GENERIC, is not loaded, and cannot be loaded by an unprivileged user (kldload is root-only); even loaded, it would not attach without the hardware. Therefore the vulnerable code is unreachable at runtime here. Because the sinks are device-integrated parsers / DRM ioctls / a hardware-dependent channel race, no userspace harness on this guest can exercise them. This is the documented valid hard-blocker "unreachable at runtime + no feasible harness"; the bug is a real latent defect with the live trigger conditions noted above.

No escalation chain (and why that is correct here)

There is no memory-corruption primitive to escalate on this guest: the corruption sinks live entirely inside the not-loaded amdgpu driver behind hardware that is absent. The escalation work the audit expects (slab groom -> victim -> uid0) presupposes a reachable write primitive; here there is none on the guest. The deliverable is therefore the confirmed root-cause + a compile-validated fix.

Fix (fix.diff) — authored and COMPILE-VALIDATED

Clamp the count to the fixed array size using the same ternary already used elsewhere in this file for the pcie-link levels (~:2191): VceLevelCount = (uint8_t)((SMU74_MAX_LEVELS_VCE < mm_table->count) ? SMU74_MAX_LEVELS_VCE : mm_table->count) and the analogous line for UvdLevelCount/SMU74_MAX_LEVELS_UVD. Supersedes the finding proposal (clamp count to SMU74_MAX_LEVELS_*).

The fix was applied to in-guest /usr/src (all hunks applied cleanly) and the amdgpu module was rebuilt with the kernel's -Werror flags: cd /usr/src/sys/dev/...amdgpu... && KERNCONF=X86_64_GENERIC SYSDIR=/usr/src/sys make -m /usr/src/share/mk => rc=0, no warnings/errors in the patched translation unit (build_fix.log). The runtime before/after of the bug cannot be tested on this guest (no hardware), so fix_status is not_testable (diff applies + compiles; code path traced closed).

Why not not_reproduced (false-positive)?

This is NOT a false positive. The cited sys/ code is genuinely missing the guard / has the overflow / has the race — verified by reading the source. It is a real bug that is simply out of reach of this particular (GPU/audio-less) QEMU guest.