# DF-1391 -- rv6xx_parse_power_table derefs attacker-controlled VBIOS pointers

**File:** `sys/dev/drm/radeon/rv6xx_dpm.c:1889`
**Class:** memory corruption (heap OOB / OOB write / OOB read)

## Status: INCONCLUSIVE at runtime -- CONFIRMED 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 / unvalidated HBA-supplied
index). However it is **not exercisable on the DragonFly audit guest** because the
guest has no radeon hardware:

- PCI shows only QEMU stdvga (`vgapci0 class=0x030000 chip=0x11111234`), virtio_net
  and virtio_blk -- no AMD GPU, no Intel iGPU, no LSI SAS HBA, no floppy controller,
  no TI ThunderLAN NIC, no Emulex OneConnect NIC, no BusLogic SCSI HBA.
- The radeon driver (whether a loadable .ko or compiled-in) never attaches.
- `/dev/fd0`, `/dev/dri`, `/dev/dsp*` do not exist on this guest.

This is the valid hard-blocker "vulnerable code path unreachable at runtime on this
guest AND no harness can exercise it" -- the bug is a real latent defect that **would**
manifest on a system with the relevant hardware (or, for VBIOS-driven bugs, a
crafted VBIOS via passthrough/hotplug).

## Mechanism (confirmed by source trace)

rv6xx_parse_power_table() computes power_state/non_clock_info/clock_info from
    VBIOS u16/u8 fields (usStateArrayOffset, usNonClockInfoArrayOffset,
usClockInfoArrayOffset, ucStateEntrySize, ucNonClockSize, ucClockInfoSize,
ucNonClockStateIndex, idx[j]) with NO bounds check vs the BIOS allocation.
atom_context has no bios_size field, so the parser cannot detect OOB. Crafted
VBIOS -> OOB heap read. Sibling of DF-1333 (rv770 same pattern).

## Live trigger conditions
Requires the radeon hardware (and the driver loaded). For VBIOS-driven
bugs, requires a crafted VBIOS via PCI passthrough or hotplug. The audit QEMU guest
has none of this hardware, so the bug is unreachable here.

## Fix
A standalone, `git apply`-able fix is in `fix.diff`. **Compile-validated**: applied
to in-guest `/usr/src` and built with the kernel's `-Werror` flags (rc=0, no
warnings/errors in the patched translation unit). See `build_fix.log`.

Three-file change: (1) added `uint32_t bios_size` to struct radeon_device
(radeon.h); (2) populated it at every rdev->bios allocation site in
radeon_bios.c (5 sites); (3) added bounds checks against rdev->bios_size at
every pointer arithmetic in rv6xx_parse_power_table (state, non_clock_info,
idx array, clock_info) with an `inval:` error path. (supersedes the finding
proposal which suggested adding bios_size to atom_context; this is equivalent
but keeps the change inside the radeon driver.)

## Reproduce / validate
```
# 1. Confirm the bug site (read-only source trace):
grep -n ... sys/dev/drm/radeon/rv6xx_dpm.c

# 2. Compile-validate the fix on the audit guest:
scp -F dfbsd-qemu/config findings/poc/DF-1391/fix.diff dfbsd:/root/DF-1391.fix.diff
./dfbsd-qemu/vm.sh run_root 'cd /usr/src && patch -p1 --forward < /root/DF-1391.fix.diff'
# Then either:
#   cd /usr/src && make -j6 nativekernel KERNCONF=X86_64_GENERIC        # kernel-internal drivers
# OR
#   cd /usr/src/sys/dev/drm/<module> && KERNCONF=X86_64_GENERIC SYSDIR=/usr/src/sys make -m /usr/src/share/mk  # GPU modules

# 3. (requires real hardware) Exercise the bug: attach the HW and trigger.
```