This traces back to the build using /usr/include instead of /usr/src/sys and so picking up an old vm/uma_int.h file. The detailed evidence follows. svnlite update -r295453 /usr/src (for an ^/head context) then a buildworld buildkernel got: --- all_subdir_libmemstat --- --- memstat_uma.So --- cc -fpic -DPIC -g -O2 -pipe -MD -MP -MF.depend.memstat_uma.So -MTmemstat_uma.So -std=gnu99 -fstack-protector-strong -Wsystem-headers -Wall -Wno-format-y2k -W -Wno-unused-parameter -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Wno-uninitialized -Wno-pointer-sign -c /usr/src/lib/libmemstat/memstat_uma.c -o memstat_uma.So . . . In file included from /usr/src/lib/libmemstat/memstat_uma.c:37: . . . --- all_subdir_libmemstat --- /usr/include/vm/uma_int.h:310: error: field 'uz_maxaction' has incomplete type Note the "/usr/include" based path instead of a /usr/src/sys/vm/uma_int.h path. Note the lack of any -I's in the cc command. Stop reading here unless you want more context details. src.conf content: TO_TYPE=powerpc KERNCONF=GENERICvtsc-NODEBUG TARGET=powerpc .if ${.MAKE.LEVEL} == 0 TARGET_ARCH=${TO_TYPE} .export TARGET_ARCH .endif WITH_FAST_DEPEND= WITH_LIB32= WITH_BOOT= COMPILER_TYPE=gcc WITHOUT_GCC_BOOTSTRAP= WITH_GCC= WITH_GCC_IS_CC= WITH_GNUCXX= WITHOUT_LIBCPLUSPLUS= WITHOUT_CLANG_BOOTSTRAP= WITHOUT_CLANG= WITHOUT_CLANG_IS_CC= WITHOUT_CLANG_FULL= WITHOUT_CLANG_EXTRAS= NO_WERROR= MALLOC_PRODUCTION= WITH_DEBUG_FILES= # freebsd-version -ku; uname -aKU 11.0-CURRENT 11.0-CURRENT FreeBSD FBSDG4C0 11.0-CURRENT FreeBSD 11.0-CURRENT #6 r295350M: Sat Feb 6 13:54:57 PST 2016 root@FBSDG4C0:/usr/obj/gcc421/powerpc.powerpc/usr/src/sys/GENERICvtsc-NODEBUG powerpc 1100097 1100097 Command used to buildworld buildkernel: env __MAKE_CONF=/root/src.configs/make.conf SRC_ENV_CONF=/root/src.configs/src.conf.powerpc-gcc421.powerpc-host MAKEOBJDIRPREFIX=/usr/obj/gcc421/powerpc.powerpc make -j 6 buildworld buildkernel (This is a classic gcc 4.2.1 TARGET_ARCH=powerpc context, not a clang 3.8.0 based experiment.)
I got some detailed references wrong in what I originally wrote. Correcting. . . The right place for finding the new uma_int.h was: /usr/obj/gcc421/powerpc.powerpc/usr/src/tmp/usr/include/vm/uma_int.h not directly in /usr/src/sys/vm/uma_int.h . The -I reference might have been over specific: There may be alternatives for causing the #include <vm/uma_int.h> to find ${WORLDTMP}/usr/include/vm/uma_int.h , such as -isysroot usage. At least the general direction of my note still applies.
(In reply to Mark Millard from comment #1) If I add X_COMPILER_TYPE=gcc .if ${.MAKE.LEVEL} == 0 CC=/usr/bin/gcc CXX=/usr/bin/g++ CPP=/usr/bin/cpp .export CC .export CXX .export CPP .endif to the src.conf content that I showed in the description then it turns out that buidlworld adds --sysroot and -isystem and other things globally. The result seems to include avoidance of /usr/include for what I was reporting. [XCC ends up being a copy of the CC and its being a full path is part of what enables this. X_COMPILER_TYPE=gcc enables more.] It still may be that CFLAGS having something for include paths is supposed to be present. I'm not sure.
(In reply to Mark Millard from comment #2) My original src.conf had a bad "optimization" where I had been thinking that since I had a working compiler that was not being updated with new source code that I could avoid the time to rebuild it during the bootstrap: WITHOUT_GCC_BOOTSTRAP= But this makes the default include/library paths wrong for buildworld and the like and so requires them to be enforced on the command line. The build environment is not set up to automatically do that in such cases. So I had wondered out of bounds.