Build time of my C++ project is very big. It consists of a bunch of subprojects (libraries), structured in file system folders. I want to speed up it with parallel build with --jobs (-j) parameter of make. What is the correct way of doing it? Documentation says that there are some tricks of doing parallel build with subdirs.
Makefile in root directory looks like:
...
all:
$(MAKE) -C DirA
$(MAKE) -C DirB
...
Makefile in DirA:
all:
$(MAKE) -C SubDirA
$(MAKE) -C SubDirB
$(MAKE) -C SubDirC
In DirB:
all:
$(MAKE) -C SubDirD
$(MAKE) -C SubDirE
$(MAKE) -C SubDirF
And so on. Makefiles in leaf folders are quite simple, contain only build instruction without any tricks.
You can just add -j to make using these makefiles as-is and you'll get some parallelism, but you won't get maximum parallelism. The problem is that make only parallelizes targets: clearly it won't work (in general) to run multiple commands in the same target in parallel!
So, in the top level make will run $(MAKE) -C DirA then $(MAKE) -C DirB, serially. When make builds DirA, it will first run $(MAKE) -C SubDirA, then $(MAKE) -C SubDirB, etc. serially. Then finally when make gets into SubDirA, it will build the targets there in parallel. This is fine, and maybe what you need to do if the order of building directories is important, except that there will be times when make could start working on SubDirB targets but won't, until all targets in SubDirA are complete.
A better way to handle subdirectories is to use make rules:
SUBDIRS := DirA DirB
all: $(SUBDIRS)
$(SUBDIRS):
$(MAKE) -C $#
.PHONY: $(SUBDIRS)
and ditto for the subdirectories. Note you must include the .PHONY.
Since each directory is a separate target, now make can invoke them in parallel if you use -j.
Oh, and if some of the contents of some subdirectories depends on others, you can declare that dependency explicitly in the makefile so make know about it:
SUBDIRS := SubDirA SubDirB SubDirC SubDirD
all: $(SUBDIRS)
$(SUBDIRS):
$(MAKE) -C $#
.PHONY: $(SUBDIRS)
SubDirA: SubDirD
etc.
I've seen a few approaches to making output directories in Make.
These include making all directories ahead of time outside of any rule, and
making an object's destination directory as part of the object's rule.
Both of these approaches involve making directories that likely already exist.
Am I missing any gotchas or drawbacks that explain why I haven't seen the below approach?
.SECONDEXPANSION:
$(OBJDIR)%.o: %.c | $$(#D)/
# Compile command
.PRECIOUS: %/
%/:
# mkdir Command
make is very good at dealing with files. make is not very good at dealing with directories.
So treating directories as implementation detail internal to the target rule makes sense, because then make never has to consider the directory at all:
MKDIR_P = mkdir -p
$(objdir)%.o: %.c
#$(MKDIR_P) $(#D)
$(COMPILE.c) -o $# -c $<
Note that the processing and IO required for the mkdir -p can be neglected next to the processing and IO required for the compilation.
The problem with directories is that (contrary to any other target) you don't care for their timestamp, you only need them to exist. Many Makefiles get directories somehow wrong, and creating them over and over again is what you observe, so make will never detect "Nothing to be done for ...".
In fact, the only thing you need for correct handling of directories with GNU make is an "order only dependency", like shown in your example. The trailing slash normally isn't needed (you seem to use it in order to have a pattern rule, I'm not sure whether this works), and you don't need .PRECIOUS either. Your trick with .SECONDEXPANSION looks quite neat, I guess this will work, given the pattern rule indeed works that way (didn't try).
For an alternative, most Makefiles that handle directories correctly take a simpler approach by concatenating all needed output directories for a rule in a single variable and use this variable as a target for another rule, e.g. like in this simplified example:
MODULES:=src/main
OBJDIR?=obj
OBJS:=$(addprefix $(OBJDIR)/,$(addsuffix .c,$(MODULES)))
DIRS:=$(sort $(addprefix $(OBJDIR)/,$(dir $(OBJS))))
TARGET:= myprogram
all: $(TARGET)
myprogram: $(OBJS)
$(CC) -o$# $^
$(DIRS):
mkdir -p $(DIRS)
$(OBJDIR)/%.o: %.c Makefile | $(DIRS)
$(CC) -c -o$# $<
clean:
rm -fr $(OBJDIR)
.PHONY: all clean
I have
$(foreach ___project___, $(UNIT_TEST_STUBS),$(MAKE) -C ../../$(___project___) $(UT_CMD) || exit 1;)
I want make to be parallel INSIDE each submake but I don't want the submakes to be executed in parallel.
How do I do this?
As #Michael Livshin points out, you already seem to have answered your own question.
Might be neater to rely on make rather than shell syntax to tie a load of commands together though.
Basically you want make to see something like this:
.PHONY: all
all:
${MAKE} -C ../../foo/ -j9 ut-cmd
${MAKE} -C ../../bar/ -j9 ut-cmd
${MAKE} -C ../../bum/ -j9 ut-cmd
Auto generation from a list is fairly straight-forward.
submakes := foo bar bum
define generate-submake
${MAKE} -C ../../$1/ ut-cmd
endef
.PHONY: all
all: ; $(foreach _,${submakes},$(call generate-submake,$1))
Note the blank line in the definition of generate-submake. It's important.
Run this with make -j9.
If you really want everything in this makefile to run serially (but the sub-makes to be parallel), then just introduce a .NOTPARALLEL target.
.NOTPARALLEL:
In this project, a top level rule looks like this:
all:
for l in $(LIBDIRS); do $(MAKE) -C $$l all; done
If I run make -jx, the files in each directory are compiled in parallel, but because of this bash loop, make doesn't move onto files in the next library until everything is done in the current one. What's the best way to change the loop so that it call truly be done in parallel? I only use gnumake, so gnu extensions are fine.
* Edit * as an aside, I was trying something like this
ALL_DEPENDENCY = $(foreach l, $(LIBDIRS), $(l).PHONY_LIB_RULE)
.PHONY: $(ALL_DEPENDENCY)
%.PHONY_LIB_RULE:
$(MAKE) --no-print-directory -C $(patsubst %.PHONY_LIB_RULE,%,$#) all
all: $(ALL_DEPENDENCY)
but these rules aren't tripped.
You can check this answer for an example of how to do this more correctly: makefile: foreach "make -C" call
I have a directory (root_dir), that contains a number of sub-directories (subdir1, subdir2, ...).
I want to run the make in each directory in root_dir, using a Makefile placed in it.
(Obviously supposed that each of subdir... has inside its own Makefile).
So there are essentially two questions:
How to get a list of directories in Makefile (automatically)?
How to run make for each of the directories inside a make file?
As I know in order to run make in a specific directory I need to do the following:
$(MAKE) -C subdir
There are various problems with doing the sub-make inside a for loop in a single recipe. The best way to do multiple subdirectories is like this:
SUBDIRS := $(wildcard */.)
all: $(SUBDIRS)
$(SUBDIRS):
$(MAKE) -C $#
.PHONY: all $(SUBDIRS)
(Just to point out this is GNU make specific; you didn't mention any restrictions on the version of make you're using).
ETA Here's a version which supports multiple top-level targets.
TOPTARGETS := all clean
SUBDIRS := $(wildcard */.)
$(TOPTARGETS): $(SUBDIRS)
$(SUBDIRS):
$(MAKE) -C $# $(MAKECMDGOALS)
.PHONY: $(TOPTARGETS) $(SUBDIRS)
Try this :
SUBDIRS = foo bar baz
subdirs:
for dir in $(SUBDIRS); do \
$(MAKE) -C $$dir; \
done
This may help you link
Edit : you can also do :
The simplest way is to do:
CODE_DIR = code
.PHONY: project_code
project_code:
$(MAKE) -C $(CODE_DIR)
The .PHONY rule means that project_code is not a file that needs to be
built, and the -C flag indicates a change in directory (equivalent to
running cd code before calling make). You can use the same approach
for calling other targets in the code Makefile.
For example:
clean:
$(MAKE) -C $(CODE_DIR) clean
Source
This is another approach to MadScientist's answer. .PHONY is a GNU-specific feature that can be used to force make into recursing into each subdirectory. However, some non-GNU versions of make do not support .PHONY, so an alternative is a force target.
4.7 Rules without Recipes or Prerequisites
If a rule has no prerequisites or recipe, and the target of the rule
is a nonexistent file, then make imagines this target to have been
updated whenever its rule is run. This implies that all targets
depending on this one will always have their recipe run.
An example will illustrate this:
clean: FORCE
rm $(objects)
FORCE:
Here the target ‘FORCE’ satisfies the special conditions, so the
target clean that depends on it is forced to run its recipe. There is
nothing special about the name ‘FORCE’, but that is one name commonly
used this way.
As you can see, using ‘FORCE’ this way has the same results as using
‘.PHONY: clean’.
Using ‘.PHONY’ is more explicit and more efficient. However, other
versions of make do not support ‘.PHONY’; thus ‘FORCE’ appears in many
makefiles. See Phony Targets.
The following is a minimal example that recurses make into each subdirectory, each of which presumably contains a Makefile. If you simply run make, only the first subdirectory, which is non-deterministic, is processed. You may also run make subdir1 subdir2 ....
# Register all subdirectories in the project's root directory.
SUBDIRS := $(wildcard */.)
# Recurse `make` into each subdirectory.
$(SUBDIRS): FORCE
$(MAKE) -C $#
# A target without prerequisites and a recipe, and there is no file named `FORCE`.
# `make` will always run this and any other target that depends on it.
FORCE:
Here is another example with top-level phony targets: all and clean. Note that the all and clean targets, passed from command-line via $(MAKECMDGOALS), are handled by each subdirectory's all and clean targets respectively.
# Register all subdirectories in the project's root directory.
SUBDIRS := $(wildcard */.)
# Top-level phony targets.
all clean: $(SUBDIRS) FORCE
# Similar to:
# .PHONY: all clean
# all clean: $(SUBDIRS)
# GNU's .PHONY target is more efficient in that it explicitly declares non-files.
# Recurse `make` into each subdirectory
# Pass along targets specified at command-line (if any).
$(SUBDIRS): FORCE
$(MAKE) -C $# $(MAKECMDGOALS)
# Force targets.
FORCE:
You can also define a function in the Makefile (also you of course need an additional makefile in each subdirectory). This is shell-dependent, but can be useful:
define FOREACH
for DIR in packages/*; do \
$(MAKE) -C $$DIR $(1); \
done
endef
.PHONY: build
build:
$(call FOREACH,build)
.PHONY: clean
clean:
$(call FOREACH,clean)
.PHONY: test
test:
$(call FOREACH,test)
Only a small icing on the cake after MadScientist's answer in order to make all the individual targets in the sub-directories available from the top level (you will need to have the SUBDIRS variable defined in order to use the following snippet – you can use MadScientist's answer for that):
# Make all the individual targets in the sub-directories available from the top
# level; as in, for instance, `make foo/my_program` or `make bar/clean`
$(foreach __dir__,$(SUBDIRS),$(__dir__)/%):
#$(MAKE) -C '$(#D)' '$(#F)'
With the code above you can run, for instance,
make foo/my_program
or
make bar/clean
Furthermore, by pasting the code above you can even use an individual target from a sub-directory as a prerequisite for a target in the top level. For example:
my_target: my_subdirectory/my_prerequisite
'my_subdirectory/my_prerequisite' > 'my_target'
…With the example above, launching make my_target from the top level will first build the my_subdirectory/my_prerequisite program, then the latter will be run for building the my_target file.
Since I was not aware of the MAKECMDGOALS variable and overlooked that MadScientist has its own implementation of multiple top-level targets, I wrote an alternative implementation. Maybe someone find it useful.
SUBDIRS := $(wildcard */.)
define submake
for d in $(SUBDIRS); \
do \
$(MAKE) $(1) --directory=$$d; \
done
endef
all:
$(call submake,$#)
install:
$(call submake,$#)
.PHONY: all install $(SUBDIRS)
There is a library called prorab for GNU make which supports inclusion of standalone makefiles in subdirectories.
Some info on github: https://github.com/cppfw/prorab/blob/master/wiki/HomePage.adoc
Basically, with prorab invoking all makefiles in subdirectories looks like this:
include prorab.mk
$(eval $(prorab-build-subdirs))
In reference to https://stackoverflow.com/posts/17845120/revisions
This is what I learned from that post.
Top Level Makefile
# set the default goal.
# I want the default to really just dump contents of dirs
# as a stub. For instance, I don't want it to
# push code or
.DEFAULT_GOAL := deploy
TOPTARGETS := all clean
SUBDIRS := docs src
$(TOPTARGETS): $(SUBDIRS)
$(SUBDIRS):
echo "make arg is" $(MAKECMDGOALS)
$(MAKE) -C $# $(MAKECMDGOALS)
SUBCLEAN = $(addsuffix .clean,$(SUBDIRS))
clean: $(SUBCLEAN)
$(SUBCLEAN): %.clean:
$(MAKE) -C $* clean
deploy:
echo do deploy stub
The src/ and docs/ common to this Makefile directory, all have a corresponding Makefile.
Here is an example of the docs setup:
# set the default goal.
.DEFAULT_GOAL := list_docs
list_docs:
ls -l
clean:
echo "docs: make clean"
-rm "*.backup"
I did this a little different than any of the answers because I didn't want to have to define each possible make target
SUBDIRS := $(patsubst %/,%,$(wildcard */))
.PHONY: all $(MAKECMDGOALS) $(SUBDIRS)
$(MAKECMDGOALS) all: $(SUBDIRS)
$(SUBDIRS):
$(MAKE) -C $# $(MAKECMDGOALS)