DF-5230 fix validation build (guest, after `vm.sh reset with-src` then `patch -p1 < fix.diff` in /usr/src):

$ cd /usr/src && patch -p1 < /root/fix.diff
Patching file sys/dev/drm/radeon/uvd_v4_2.c using Plan A...
Hunk #1 succeeded at 40.
done
PATCHED
51:	if (rdev->uvd_fw == NULL)

$ cd /usr/src/sys/dev/drm/radeon && rm -f uvd_v4_2.o radeon.ko && make -j6
(incremental: only uvd_v4_2.c recompiled + radeon.ko relinked; full .o list identical
 to the baseline link line — see build.sh / the baseline build in /tmp/build_radeon.log
 on the pre-reset guest)
FIXED_RADEON_BUILT
-rw-r--r-- 1 root wheel 1000 2029128 Oct 1 09:48 radeon.ko

NOTE: guest was reset after validation (single-fix kernel was a kmod rebuild of
radeon.ko loaded on the stock X86_64_GENERIC kernel: uname matches baseline —
"DragonFly dfbsd 6.5-DEVELOPMENT #0: Thu Jul 2 06:02:54 UTC 2026"), so /tmp logs
from the fixed build are from the session transcript above, not a live file.