========================================================================
DF-1149 fix compile-validation  (radeon ci_dpm.c, SMC level tables)
Guest: DragonFly 6.5-DEVELOPMENT #0, cc 8.3, with-src snapshot (unpatched)
========================================================================

[1] BASELINE: full radeon.ko built clean from /usr/src under -Werror
    (includes ci_dpm.o baseline = 55424 bytes)
    $ cd /usr/src/sys/dev/drm/radeon && make
    rc=0 ; radeon.ko produced (2030856 bytes), no error:/warning:

[2] APPLY fix.diff:
    $ cd /usr/src && patch -p1 --forward < /root/df1149.diff
    Patching file sys/dev/drm/radeon/ci_dpm.c using Plan A...
    Hunk #1 succeeded at 2687.
    Hunk #2 succeeded at 2731.
    Hunk #3 succeeded at 2764.
    Hunk #4 succeeded at 2798.
    done

[3] INCREMENTAL REBUILD of patched ci_dpm.c:
    $ cd sys/dev/drm/radeon && rm -f ci_dpm.o && make ci_dpm.o
    cc -O2 -pipe ... -std=gnu11 -Werror -DKLD_MODULE ... \
       -c /usr/src/sys/dev/drm/radeon/ci_dpm.c
    (no error:/warning: emitted under -Werror)

[4] RESULT:
    /usr/obj/usr/src/sys/dev/drm/radeon/ci_dpm.o
      baseline  = 55424 bytes
      patched   = 55456 bytes   (+32 bytes from added min_t clamps)
    => fix.diff APPLIES + COMPILES clean.

[5] REVERT (source returned to unpatched baseline):
    $ cd /usr/src && patch -p1 -R < /root/df1149.diff
    Hunk #4 succeeded at 2795. ... done
    grep -c "min_t(u8, rdev" ci_dpm.c  ->  0  (clean)

RUNTIME: not exercised -- no AMD GPU on guest (latent). See run.sh / VERDICT.md.