DF-5053 kernel-build evidence (EXCERPT — session-captured lines; the full
/tmp/kbuild.log was cleaned from the guest before harvesting, my error).
Every line below is verbatim from the run; the decisive facts:

  1) fix.diff applied to guest /usr/src (patched qsphy.c md5
     2308bacbd5684e93b604334dd245bb97; pristine 6c17ac33ff09cf4e26e4f5131b6adf7a),
  2) `make -j6 nativekernel KERNCONF=X86_64_GENERIC` (the DF-5009 route;
     plain buildkernel refuses: "You must buildworld before buildkernel")
       BUILD_RC=0
       >>> Kernel build for X86_64_GENERIC completed on Sat Sep 26 16:16:10 UTC 2026
  3) qsphy.c compiled under the full kernel -Werror flag set
     (kbuild.log:18760-18761):
       --- qsphy.o ---
       cc -c -O2 -pipe -Wall -Wredundant-decls -Wnested-externs
         -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith
         -Wcast-qual -Wold-style-definition -Wmissing-include-dirs
         -Wno-pointer-sign -Winit-self -Wundef -Wno-unused-but-set-variable
         -std=gnu11 -g -nostdinc ... -D_KERNEL ... -Werror
         /usr/src/sys/dev/netif/mii_layer/qsphy.c
  4) `make installkernel KERNCONF=X86_64_GENERIC`
       INSTALL_RC=0
  5) rebooted into the patched kernel (uname: "#1: Sat Sep 26 16:13:05
     UTC 2026") and re-ran the exact baseline trigger -> survived (run.3.log)
  6) sources reverted, md5 back to pristine, guest reset to with-src
     snapshot afterwards.