Add a FAST_DEPEND option, off by default, which speeds up the build significantly.

This speeds up buildworld by 16% on my system and buildkernel by 35%.

Rather than calling mkdep(1), which is just a wrapper around 'cc -E',
use the modern -MD -MT -MF flags to gather and generate dependencies during
compilation.  This flag was introduced in GCC "a long time ago", in GCC 3.0,
and is also supported by Clang.  (It appears that ICC also supports this but I
do not have access to test it).  This avoids running the preprocessor *twice*
for every build, in both 'make depend' and 'make all'.  This is especially
noticeable when using ccache since it does not cache preprocessor results from
mkdep(1) / 'cc -E', but still speeds up compilation with the -MD flags.

For 'make depend' a tree-walk is still done to ensure that all DPSRCS
are generated when expected, and that beforedepend/afterdepend and
_EXTRADEPEND are all still respected.  In time this may change but for now
I've been conservative.  The time for a tree-walk with -j combined with
SUBDIR_PARALLEL is not significant.  For example, it takes about 9 seconds
with -j15 to walk all of src/ for 'make depend' now on my system.

A .depend file is still generated with the various rules that apply to
the final target, or custom rules.  Otherwise there are now
per-built-object-file .depend files, such as .depend.filename.o.  These
are included directly by make rather than populating .depend with a loop
and .depend lines, which only added overhead to the now almost-NOP 'make
depend' phase.

Before this I experimented with having mkdep(1) called in parallel per-file.
While this improved the kernel and lib/libc 'make depend' phase, it resulted
in slower build times overall.

The -M flags are removed from CFLAGS when linking since they have no effect.

Enabling this by default, for src or out-of-src, can be done once more testing
has been done, such as a ports exp-run, and with more compilers.

The system I used for testing was:
  WITNESS
  Build options: -j20 WITH_LLDB=yes WITH_DEBUG_FILES=yes WITH_FAST_DEPEND=yes
  DISK: ZFS 3-way mirror with very slow disks using SSD l2arc/log.
        The arc was fully populated with src tree files.
  RAM: 76GiB
  CPU: Intel(R) Xeon(R) CPU L5520 @2.27GHz
       2 package(s) x 4 core(s) x 2 SMT threads = hw.ncpu=16

buildworld:
  x buildworld-before
  + buildworld-fastdep
  +-------------------------------------------------------------------------------+
  |+                                                                              |
  |+                                                                              |
  |+                                                                       xx    x|
  |                                                                       |_MA___||
  |A                                                                              |
  +-------------------------------------------------------------------------------+
      N           Min           Max        Median           Avg        Stddev
  x   3       3744.13       3794.31       3752.25     3763.5633     26.935139
  +   3       3153.34       3155.16        3154.2     3154.2333    0.91045776
  Difference at 95.0% confidence
          -609.33 +/- 43.1943
          -16.1902% +/- 1.1477%
          (Student's t, pooled s = 19.0569)

buildkernel:
  x buildkernel-before
  + buildkernel-fastdep
  +-------------------------------------------------------------------------------+
  |+                                                                            x |
  |++                                                                           xx|
  |                                                                             A||
  |A|                                                                             |
  +-------------------------------------------------------------------------------+
      N           Min           Max        Median           Avg        Stddev
  x   3        571.57        573.94        571.79     572.43333     1.3094401
  +   3        369.12        370.57         369.3     369.66333    0.79033748
  Difference at 95.0% confidence
          -202.77 +/- 2.45131
          -35.4225% +/- 0.428227%
          (Student's t, pooled s = 1.0815)

Sponsored by:	EMC / Isilon Storage Division
MFC after:	3 weeks
Relnotes:	yes
This commit is contained in:
bdrewery 2015-11-06 04:45:29 +00:00
parent 0731342a69
commit a175970362
7 changed files with 63 additions and 16 deletions

View File

@ -54,6 +54,18 @@ MKDEPCMD?= CC='${CC} ${DEPFLAGS}' mkdep
MKDEPCMD?= mkdep
.endif
DEPENDFILE?= .depend
DEPENDFILES= ${DEPENDFILE}
.if ${MK_FAST_DEPEND} == "yes"
DEPENDFILES+= ${DEPENDFILE}.*
DEPEND_CFLAGS+= -MD -MP -MF${DEPENDFILE}.${.TARGET}
DEPEND_CFLAGS+= -MT${.TARGET}
CFLAGS+= ${DEPEND_CFLAGS}
DEPENDOBJS+= ${OBJS} ${POBJS} ${SOBJS}
.for __obj in ${DEPENDOBJS:O:u}
.sinclude "${DEPENDFILE}.${__obj}"
DEPENDFILES_OBJS+= ${DEPENDFILE}.${__obj}
.endfor
.endif # ${MK_FAST_DEPEND} == "yes"
# Keep `tags' here, before SRCS are mangled below for `depend'.
.if !target(tags) && defined(SRCS) && !defined(NO_TAGS)
@ -161,7 +173,7 @@ afterdepend: beforedepend
depend: beforedepend ${DEPENDFILE} afterdepend
# Tell bmake not to look for generated files via .PATH
.NOPATH: ${DEPENDFILE}
.NOPATH: ${DEPENDFILE} ${DEPENDFILES_OBJS}
# Different types of sources are compiled with slightly different flags.
# Split up the sources, and filter out headers and non-applicable flags.
@ -172,6 +184,7 @@ MKDEP_CXXFLAGS= ${CXXFLAGS:M-nostdinc*} ${CXXFLAGS:M-[BIDU]*} \
DPSRCS+= ${SRCS}
${DEPENDFILE}: ${DPSRCS}
.if ${MK_FAST_DEPEND} == "no"
rm -f ${DEPENDFILE}
.if !empty(DPSRCS:M*.[cS])
${MKDEPCMD} -f ${DEPENDFILE} -a ${MKDEP} \
@ -182,7 +195,11 @@ ${DEPENDFILE}: ${DPSRCS}
${MKDEPCMD} -f ${DEPENDFILE} -a ${MKDEP} \
${MKDEP_CXXFLAGS} \
${.ALLSRC:M*.cc} ${.ALLSRC:M*.C} ${.ALLSRC:M*.cpp} ${.ALLSRC:M*.cxx}
.else
.endif
.else
: > ${.TARGET}
.endif # ${MK_FAST_DEPEND} == "no"
.if target(_EXTRADEPEND)
_EXTRADEPEND: .USE
${DEPENDFILE}: _EXTRADEPEND
@ -207,12 +224,12 @@ afterdepend:
cleandepend:
.if defined(SRCS)
.if ${CTAGS:T} == "gtags"
rm -f ${DEPENDFILE} GPATH GRTAGS GSYMS GTAGS
rm -f ${DEPENDFILES} GPATH GRTAGS GSYMS GTAGS
.if defined(HTML)
rm -rf HTML
.endif
.else
rm -f ${DEPENDFILE} tags
rm -f ${DEPENDFILES} tags
.endif
.endif
.endif

View File

@ -305,10 +305,12 @@ all: _manpages
.endif
_EXTRADEPEND:
.if ${MK_FAST_DEPEND} == "no"
@TMP=_depend$$$$; \
sed -e 's/^\([^\.]*\).o[ ]*:/\1.o \1.po \1.So:/' < ${DEPENDFILE} \
> $$TMP; \
mv $$TMP ${DEPENDFILE}
.endif
.if !defined(NO_EXTRADEPEND) && defined(SHLIB_NAME)
.if defined(DPADD) && !empty(DPADD)
echo ${SHLIB_NAME_FULL}: ${DPADD} >> ${DEPENDFILE}

View File

@ -66,6 +66,7 @@ __DEFAULT_YES_OPTIONS = \
WARNS
__DEFAULT_NO_OPTIONS = \
FAST_DEPEND \
CTF \
DEBUG_FILES \
INSTALL_AS_USER \

View File

@ -82,9 +82,9 @@ ${PROG_FULL}: beforelinking
.endif
${PROG_FULL}: ${OBJS}
.if defined(PROG_CXX)
${CXX} ${CXXFLAGS} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
${CXX} ${CXXFLAGS:N-M*} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
.else
${CC} ${CFLAGS} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
${CC} ${CFLAGS:N-M*} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
.endif
.if ${MK_CTF} != "no"
${CTFMERGE} ${CTFFLAGS} -o ${.TARGET} ${OBJS}
@ -112,9 +112,9 @@ ${PROG_FULL}: beforelinking
.endif
${PROG_FULL}: ${OBJS}
.if defined(PROG_CXX)
${CXX} ${CXXFLAGS} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
${CXX} ${CXXFLAGS:N-M*} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
.else
${CC} ${CFLAGS} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
${CC} ${CFLAGS:N-M*} ${LDFLAGS} -o ${.TARGET} ${OBJS} ${LDADD}
.endif
.if ${MK_CTF} != "no"
${CTFMERGE} ${CTFFLAGS} -o ${.TARGET} ${OBJS}

View File

@ -45,6 +45,7 @@ __DEFAULT_YES_OPTIONS = \
__DEFAULT_NO_OPTIONS = \
EISA \
FAST_DEPEND \
NAND \
OFED

View File

@ -198,18 +198,37 @@ kernel-depend: .depend
SRCS= assym.s vnode_if.h ${BEFORE_DEPEND} ${CFILES} \
${SYSTEM_CFILES} ${GEN_CFILES} ${SFILES} \
${MFILES:T:S/.m$/.h/}
DEPENDFILES= .depend
.if ${MK_FAST_DEPEND} == "yes"
DEPENDFILES+= .depend.*
DEPEND_CFLAGS+= -MD -MP -MF.depend.${.TARGET}
DEPEND_CFLAGS+= -MT${.TARGET}
CFLAGS+= ${DEPEND_CFLAGS}
DEPENDOBJS+= ${SYSTEM_OBJS}
.for __obj in ${DEPENDOBJS:O:u}
.sinclude ".depend.${__obj}"
DEPENDFILES_OBJS+= .depend.${__obj}
.endfor
.endif # ${MK_FAST_DEPEND} == "yes"
.NOPATH: .depend ${DEPENDFILES_OBJS}
.depend: .PRECIOUS ${SRCS}
rm -f .newdep
.if ${MK_FAST_DEPEND} == "no"
rm -f ${.TARGET}.tmp
${MAKE} -V CFILES_NOCDDL -V SYSTEM_CFILES -V GEN_CFILES | \
MKDEP_CPP="${CC} -E" CC="${CC}" xargs mkdep -a -f .newdep ${CFLAGS}
CC="${CC}" xargs mkdep -a -f ${.TARGET}.tmp ${CFLAGS}
${MAKE} -V CFILES_CDDL | \
MKDEP_CPP="${CC} -E" CC="${CC}" xargs mkdep -a -f .newdep ${ZFS_CFLAGS} ${FBT_CFLAGS} ${DTRACE_CFLAGS}
CC="${CC}" xargs mkdep -a -f ${.TARGET}.tmp ${ZFS_CFLAGS} \
${FBT_CFLAGS} ${DTRACE_CFLAGS}
${MAKE} -V SFILES_NOCDDL | \
MKDEP_CPP="${CC} -E" xargs mkdep -a -f .newdep ${ASM_CFLAGS}
CC="${CC}" xargs mkdep -a -f ${.TARGET}.tmp ${ASM_CFLAGS}
${MAKE} -V SFILES_CDDL | \
MKDEP_CPP="${CC} -E" xargs mkdep -a -f .newdep ${ZFS_ASM_CFLAGS}
rm -f .depend
mv .newdep .depend
CC="${CC}" xargs mkdep -a -f ${.TARGET}.tmp ${ZFS_ASM_CFLAGS}
mv ${.TARGET}.tmp ${.TARGET}
.else
: > ${.TARGET}
.endif
_ILINKS= machine
.if ${MACHINE} != ${MACHINE_CPUARCH} && ${MACHINE} != "arm64"
@ -237,8 +256,8 @@ ${_ILINKS}:
ln -s $$path ${.TARGET}
# .depend needs include links so we remove them only together.
kernel-cleandepend:
rm -f .depend ${_ILINKS}
kernel-cleandepend: .PHONY
rm -f ${DEPENDFILES} ${_ILINKS}
kernel-tags:
@[ -f .depend ] || { echo "you must make depend first"; exit 1; }

View File

@ -0,0 +1,7 @@
.\" $FreeBSD$
Set to generate
.Sy .depend
files in the build during compilation instead of the
historial
.Xr mkdep 1
call during the "make depend" phase.