3ac58492b3
With this patch, it is now possible to generate coverage information for MINIX3 system services with LLVM. In particular, the system can be built with MKCOVERAGE=yes, either with a native "make build" or with crosscompilation. Either way, MKCOVERAGE=yes will build the MINIX3 system services with coverage profiling support, generating a .gcno file for each source module. After a reboot it is possible to obtain runtime coverage data (.gcda files) for individual system services using gcov-pull(8). The combination of the .gcno and .gcda files can then be inspected with llvm-cov(1). For reasons documented in minix.gcov.mk, only system service program modules are supported for now; system service libraries (libsys etc.) are not included. Userland programs are not affected by MKCOVERAGE. The heart of this patch is the libsys code that writes data generated by the LLVM coverage hooks into a serialized format using the routines we already had for GCC GCOV. Unfortunately, the new llvm_gcov.c code is LLVM ABI dependent, and may therefore have to be updated later when we upgrade LLVM. The current implementation should support all LLVM versions 3.x with x >= 4. The rest of this patch is mostly a light cleanup of our existing GCOV infrastructure, with as most visible change that gcov-pull(8) now takes a service label string rather than a PID number. Change-Id: I6de055359d3d2b3f53e426f3fffb17af7877261f
114 lines
3.2 KiB
Makefile
114 lines
3.2 KiB
Makefile
# Makefile for kernel
|
|
.include <bsd.own.mk>
|
|
|
|
PROG= kernel
|
|
BINDIR= /usr/sbin
|
|
MAN=
|
|
|
|
.if ${MACHINE_ARCH} == "earm" && ${MKLLVM:Uno} == "yes"
|
|
# BJG - problems with optimisation of the kernel by llvm
|
|
DBG=-O0
|
|
.endif
|
|
|
|
.include "arch/${MACHINE_ARCH}/Makefile.inc"
|
|
|
|
SRCS+= clock.c cpulocals.c interrupt.c main.c proc.c system.c \
|
|
table.c utility.c usermapped_data.c
|
|
|
|
LDADD+= -ltimers -lsys -lexec
|
|
|
|
LINKERSCRIPT= ${.CURDIR}/arch/${MACHINE_ARCH}/kernel.lds
|
|
|
|
.if ${HAVE_GOLD:U} != ""
|
|
CFLAGS+= -fno-common
|
|
.endif
|
|
LDFLAGS+= -T ${LINKERSCRIPT}
|
|
LDFLAGS+= -nostdlib -L${DESTDIR}/usr/lib
|
|
CFLAGS += -fno-stack-protector
|
|
|
|
CPPFLAGS+= -D__kernel__
|
|
|
|
# kernel headers are always called through kernel/*.h
|
|
CPPFLAGS+= -I${NETBSDSRCDIR}/minix
|
|
|
|
# kernel headers are always called through kernel/*.h, this
|
|
# time for generated headers, during cross compilation
|
|
CPPFLAGS+= -I${.OBJDIR}/..
|
|
|
|
# Machine-dependent headers, order is important!
|
|
CPPFLAGS+= -I${.CURDIR}/arch/${MACHINE_ARCH}
|
|
CPPFLAGS+= -I${.CURDIR}/arch/${MACHINE_ARCH}/include
|
|
CPPFLAGS+= -I${.CURDIR}/arch/${MACHINE_ARCH}/bsp/include
|
|
CPPFLAGS+= -I${NETBSDSRCDIR}/minix/include/arch/${MACHINE_ARCH}/include
|
|
|
|
.include "system/Makefile.inc"
|
|
|
|
.if ${MKPAE:Uno} != "no"
|
|
CPPFLAGS+= -DPAE=1
|
|
.endif
|
|
|
|
.ifdef CONFIG_SMP
|
|
SRCS+= smp.c
|
|
.endif
|
|
|
|
.if ${USE_WATCHDOG} != "no"
|
|
SRCS+= watchdog.c
|
|
CPPFLAGS+= -DUSE_WATCHDOG=1
|
|
.endif
|
|
|
|
# Extra debugging routines
|
|
.if ${USE_SYSDEBUG} != "no"
|
|
SRCS+= debug.c
|
|
CPPFLAGS+= -DUSE_SYSDEBUG=1
|
|
.endif
|
|
|
|
# These come last, so the profiling buffer is at the end of the data segment
|
|
SRCS+= profile.c do_sprofile.c
|
|
|
|
.if ${USE_LIVEUPDATE} != "no"
|
|
CPPFLAGS+= -DUSE_UPDATE=1
|
|
.endif
|
|
|
|
CLEANFILES+=extracted-errno.h extracted-mfield.h extracted-mtype.h procoffsets.h
|
|
|
|
debug.o debug.d: extracted-errno.h extracted-mfield.h extracted-mtype.h
|
|
|
|
extracted-errno.h: extract-errno.sh ../../include/errno.h
|
|
${_MKTARGET_CREATE}
|
|
cd ${.CURDIR} ; ${HOST_SH} extract-errno.sh > ${.OBJDIR}/extracted-errno.h
|
|
|
|
extracted-mfield.h: extract-mfield.sh ../lib/libc/sys/*.c ../lib/libsys/*.c
|
|
${_MKTARGET_CREATE}
|
|
cd ${.CURDIR} ; ${HOST_SH} extract-mfield.sh > ${.OBJDIR}/extracted-mfield.h
|
|
|
|
extracted-mtype.h: extract-mtype.sh ../include/minix/com.h
|
|
${_MKTARGET_CREATE}
|
|
cd ${.CURDIR} ; ${HOST_SH} extract-mtype.sh > ${.OBJDIR}/extracted-mtype.h
|
|
|
|
.if ${USE_BITCODE:Uno} == "yes"
|
|
# dcvmoole: this is a copy of the "${_P}: ${_P}.bcl.o" block from bsd.prog.mk,
|
|
# with two changes: 1) ${OBJS} is added so as to link in objects that have not
|
|
# been compiled with bitcode, and 2) we are directly loading the gold plugin
|
|
# rather than through ${BITCODE_LD_FLAGS_2ND.kernel}, because LLVMgold will
|
|
# not load libLTO when no LTO can be performed (due to the non- bitcode
|
|
# objects), causing it to fail on unrecognized -disable-opt/-disable-inlining
|
|
# options. At least I think that's what's going on? I'm no expert here..
|
|
kernel: kernel.bcl.o
|
|
${_MKTARGET_LINK}
|
|
${_CCLINK.kernel} \
|
|
${_LDFLAGS.kernel} \
|
|
-L${DESTDIR}/usr/lib \
|
|
${_LDSTATIC.kernel} -o ${.TARGET} \
|
|
${.TARGET}.bcl.o ${OBJS} ${_PROGLDOPTS} ${_LDADD.kernel} \
|
|
-Wl,-plugin=${GOLD_PLUGIN} \
|
|
-Wl,--allow-multiple-definition
|
|
.endif
|
|
|
|
# Disable magic and ASR passes for the kernel.
|
|
USE_MAGIC=no
|
|
|
|
# Disable coverage profiling for the kernel, at least for now.
|
|
MKCOVERAGE=no
|
|
|
|
.include <minix.service.mk>
|