# DF-4634 fix validation: sound.ko module rebuilds (guest /usr/src, native cc)
#
# FIXED module build (clone.c with fix.diff applied via /tmp/fix.sh awk patch):
#   cd /usr/src/sys/dev/sound/sound && make obj && make
# final link line:
cc -Wl,--build-id=sha1 -nostdlib -Wl,--hash-style=sysv   -r -Wl,-d -o sound.ko ac97_if.o channel_if.o feeder_if.o mixer_if.o feeder.o feeder_rate.o feeder_volume.o feeder_chain.o feeder_eq.o feeder_format.o feeder_matrix.o feeder_mixer.o mpu_if.o mpufoi_if.c/synth_if.o ac97.o ac97_patch.o buffer.o channel.o clone.o dsp.o mixer.o sndstat.o sound.o unit.o vchan.o midi.o mpu401.o sequencer.o
install -o root -g wheel -m 555   sound.ko /boot/kernel
# md5 (fixed): cf8c3fcacda2760826cc3126a64a7ba2
#
# VULNERABLE baseline rebuild (clone.c restored from clone.c.orig, rm clone.o):
#   cd /usr/src/sys/dev/sound/sound && rm -f clone.o sound.ko && make && make install && sync
# md5 (vulnerable): e869fcdebe0137ee26a9de85429c0baf
#
# NOTE: full -v compiler output was not retained for every pass; the decisive
# artifacts are the two module md5s above and the A/B run logs (run.log
# vulnerable vs fix_run.log fixed). One operational hazard encountered and
# worked around: an unclean QEMU kill after `make install` truncated the
# installed module (md5 d41d8cd9... = empty file) - always `sync` and shut
# down gracefully after installing into /boot.