Overriding implicit rule of certain targets without interfering with -MMD - makefile

Say I have a simple Makefile like so:
CXXFLAGS+=-O3 -Wall -pednatic -MMD -Isrc
SRCS=$(shell find src -name '*.cpp')
OBJS=${SRCS:.cpp=.o}
TEST_SRCS=$(shell find test -name '*.cpp')
TEST_OBJS=${TEST_SRCS:.cpp=.o}
all: bin/product
bin/product: $(OBJS)
$(CXX) ($CXX_FLAGS) -o $# $^
test: test/runner
./test/runner
test/runner: $(TEST_OBJS) $(OBJS)
$(CXX) ($CXX_FLAGS) -o $# $^
-include ${OBJS:.o=.d}
-include ${TEST_OBJS:.o=.d}
.PHONY: all test
The important bits to note here are that I have two separate binary targets (one production with objects from src/ and one test runner with objects from both src/ and test/). I want to change the CXXFLAGS just for targets in test (test/runner, but also test/foo_test.o) but not in src. I could of course define my own rule to override the implicit one (note the added -Itest, which is one of many differences in how I'd like to compile tests):
test/%.o: test/%.cpp
$(CXX) $(CXXFLAGS) -Itest -c -o $# $^
But this has the unfortunate consequence of overriding the dependencies from the included *.d files created by -MMD. For example test/foo_test.d:
test/foo_test.o: test/foo_test.cpp src/foo.o
Is there any way to override the implicit rule for objects in test/ without losing the dependencies?

You can define or modify make variables on a per target basis (see GNU make documentation):
$(TEST_OBJS): CXX_FLAGS += -Itest
Notes:
You shouldn't use the same CXX_FLAGS for compilation and linking. The options are different. For instance, you do not pass -I... options to the linker. Use LDFLAGS for linking, instead.
Your Makefile uses CXXFLAGS and CXX_FLAGS. Typos in your question? The standard make variable to pass options to the C++ compiler is CXXFLAGS, not CXX_FLAGS.

Related

Crafting generic Make targets with dynamic dependencies

I have a need to generate a few object files with different load addresses. I also have a make target for invoking objdump to show me the respective disassembly.
code0:
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE0_RELOC) $#.S -o $#.o
objcopy -O binary $#.o --only-section=.text $#.bin
code1:
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE1_RELOC) $#.S -o $#.o
objcopy -O binary $#.o --only-section=.text $#.bin
dump: dump_code0 dump_code1
dump_code0: code0.bin
objdump $< -D > $<.decomp
dump_code1: code1.bin
objdump $< -D > $<.decomp
This will fail because there is no target for code.bin, but target dump_code depends on code.bin being present. To prevent it from being confusing (i.e. using dependency code0 such that code0.bin is then produced) I thought it would be easier to create a %.bin target that will create the respective binary file associated with each object file (since there is also duplication of this code across all code* targets anyway. This target will then be a dependency for each code* target. However, this necessitates (or at least I think it does) the need for dynamically assigning dependencies. I tried something like this that did not work:
%.bin: $(basename $#)
objcopy -O binary $(basename $#).o --only-section=.text $#.bin
code0: code0.bin
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE0_RELOC) $#.S -o $#.o
With the logic being that a dependency such as code0.bin would cause this target to execute with the single effective dependency of code0 to first build the object file.
I may be misunderstanding how the % character may be used to wildcard make targets. The first usesof basename was evaluating as empty (the second worked), so the target had no dependencies. It does make sense that the dependency tree be evaluated once at invocation, but I was hoping for a more dynamic capability.
Can this be done in Make?
What you are looking for is probably Static Pattern Rules with which you could rework your first attempt:
code0:
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE0_RELOC) $#.S -o $#.o
objcopy -O binary $#.o --only-section=.text $#.bin
as:
code0.o code1.o: code%.o: code%.S
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE$*_RELOC) $< -o $#
code0.bin code1.bin: %.bin: %.o
objcopy -O binary $< --only-section=.text $#
See the principle? $* in the recipe of the first rule is substituted by the matching stem of the static pattern rule (the digit in our case). This works but pattern-specific variables would probably be easier to understand and maintain:
# Default code reloc option
CODE_RELOC := default_code_reloc_option
# code1-specific code reloc option, if different from default
code1.o: CODE_RELOC := code1_code_reloc_option
code0.o code1.o: %.o: %.S
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE_RELOC) $< -o $#
The last part of your first makefile could also use static pattern rules, plus phony targets. Something like:
.PHONY: dump dump_code0 dump_code1
dump: dump_code0 dump_code1
dump_code0 dump_code1: dump_%: %.bin.decomp
code0.bin.decomp code1.bin.decomp: %.bin.decomp: %.bin
objdump $< -D > $#
Finally, we can use some more make tricks (discovery of source files, pattern substitutions) to automate a bit further:
CODES := $(wildcard *.S)
OBJS := $(patsubst %.S,%.o,$(CODES))
BINS := $(patsubst %.S,%.bin,$(CODES))
DECS := $(patsubst %.S,%.bin.decomp,$(CODES))
DUMPS := $(patsubst %.S,dump_%,$(CODES))
# Default code reloc option
CODE_RELOC := default_code_reloc_option
# code1-specific code reloc option, if different from default
code1.o: CODE_RELOC := code1_code_reloc_option
.PHONY: dump $(DUMPS)
dump: $(DUMPS)
$(DUMPS): dump_%: %.bin.decomp
$(DECS): %.bin.decomp: %.bin
objdump $< -D > $#
$(BINS): %.bin: %.o
objcopy -O binary $< --only-section=.text $#
$(OBJS): %.o: %.S
gcc -m32 $(CFLAGS) -Wl,$(LDFLAGS),-Ttext=$(CODE_RELOC) $< -o $#
.PHONY: clean
clean:
rm -f $(OBJS) $(BINS) $(DECS)
The clean target is a bonus gift.
All this has a significant advantage over what you were trying to do with targets that were not real files. Here, files depend on other files, except the phony targets that are just a kind of shorthand for real target files. And this is 100% in line with make's philosophy: express inter-files dependences such that make compares files timestamps and decides what is up-to-date and what must be rebuilt. Significant savings when a lot of targets are already up-to-date.
A second advantage is that expressing all inter-files dependences is parallel safe. If you run make on this makefile with:
make -j 8
(assuming you have about 8 cores on your computer), you can expect a speed-up factor of 8. Not much if you have only two source files, but quite interesting if you have hundreds of them...

Forcing the order of implicit rule/pattern rule evaluation in GNU Make

I have a domain specific language compiler (homemade) which takes a file x.inflow and generates two files: x.c and x.h. The C file is compiled in the conventional manner and the generated header file has to be included into any file that calls the functions defined within it.
The header files therefore have to be generated before any C files that use them are compiled. My current Makefile, below, works fine except for the first build from clean where it can try and compile main.c before the header file that it includes has been created.
NAME = simplest
OBJ = $(patsubst %.c,%.o,$(wildcard *.c)) \
$(patsubst %.inflow,%.o,$(wildcard *.inflow))
CC = gcc
CFLAGS = -g -Wall
$(NAME): $(OBJ)
$(CC) $(CFLAGS) -o $# $^ $(CLIBS)
# Dependencies for existing .o files.
-include $(OBJ:.o=.d)
# Compile an inflow file into both a .c and .h file.
# Note that this rule has two targets.
%.c %.h: %.inflow
inflow $<
# Compile object files and generate dependency information.
%.o: %.c
$(CC) -MD -MP -c $(CFLAGS) -o $# $<
Obviously, I can fix this for specific cases by adding, for example (where simplest.h is a generated header):
main.o: simplest.h
But is there a general way to force one type of pattern rule (%.c %.h: %.inflow) to be run before any invokations of another (%.o: %.c)?
Well, you can force any target to be run before any other target with order-only prerequisites. So for example, you can write:
%.o : %.c | simplest.h
$(CC) -MD -MP -c $(CFLAGS) -o $# $<
which will ensure that no target that uses this pattern rule to build will be invoked before the target simplest.h is created. However, I don't think you can put patterns in an order-only prerequisite. To be honest, I've never tried it so it's possible that it works, I'm not sure.
If not, you could just list all the order-only prerequisites in the %.o pattern rule; this would ensure that all the inflow files are generated before any of the object files are built. That's probably OK.
It seems the problem is twofold:
Make doesn't know that it needs to generate simplest.h before compiling main.c.
You don't want to have to explicitly tell Make about the dependency (and remember to update it when it changes).
Rather than force Make to evaluate rules in a set order, you can solve your problem by letting Make create the dependencies for you. Check out this section of the Gnu Make manual: http://www.gnu.org/software/make/manual/make.html#Automatic-Prerequisites
When you run Make, it will scan your source files and gather their dependencies for you (and you won't have to explicitly list that main.o depends on simplest.h).

Understanding a makefile

I am talking about this question where the person has updated his final solution with a makefile for the task. I am having a hard time understanding how it's done.
There is a rule:
$(OBJECTS): $(OBJDIR)/%.o : $(SRCDIR)/%.c
#$(CC) $(CFLAGS) -c $< -o $#
which I am unable to understand, but by intuition I know what it will be doing. Almost everything else is pretty much clear. Thanks!
This is a static pattern rule. The first field is a list of targets, the second is a target pattern which Make uses to isolate a target's "stem", the third is the prerequisite pattern which Make uses to construct the list of prerequisites.
Suppose you have
SRCDIR = src
OBJDIR = obj
OBJECTS = obj/foo.o obj/bar.o obj/baz.o
$(OBJECTS): $(OBJDIR)/%.o : $(SRCDIR)/%.c
#$(CC) $(CFLAGS) -c $< -o $#
If you make obj/foo.o, Make first identifies this rule as the one to use (since obj/foo.o is in the target list $(OBJECTS)), matches it against the target pattern obj/%.o and finds that the stem (the part matched by the wildcard %) is foo, then plugs that into the prereq pattern src/%.c and finds that the prerequisite is src/foo.c.
If you've also defined the variables
CC = gcc
CFLAGS = -thisflag -thatflag=something
Then the command in the rule becomes
#gcc -thisflag -thatflag=something -c src/foo.c -o obj/foo.o
(Note that $< is the first prerequisite and $# is the target name.)
In answer to your other question: Yes, a makefile can handle a dependency on a header file (x.h) so that if the header has been modified, Make will rebuild the target. No, this makefile doesn't do that. You can modify the makefile by hand, adding rules like
a.o: x.h
assuming you know what the inclusions actually are, or you can have the makefile do it automatically, which is an advanced technique you probably shouldn't attempt yet.
This line is explaining how to obtain the object files (.o) from the source (.c), it avoids having to repeat the line for each .c file.
The objects will be in OBJDIR and the sources in SRCDIR
$(CC) will contain the compiler, CFLAGS will contain the options for the compiler and -c tells gcc to compile the source into objects.
For example:
CC = gcc
CFLAGS = -g -Wall
can be converted into
gcc -g -Wall -c test.c -o test.o

How to write a simpler makefile for a lot of single-c-file programmes?

I want to write a lot of tiny example programmes for one same library, each needs gcc $(OtherOpt) -o xxx -lthelibname xxx.c.
How to write a Makefile without dozens of tagret lines ?
Pattern rules are your friend for these situations. As long as your targets all match a predictable pattern -- and they do in this case, as they are all of the form "create foo from foo.c" -- you can write a single pattern rule that will be used for all of the targets:
OtherOpt=-Wall -g
all: $(patsubst %.c,%,$(wildcard *.c))
%: %.c
gcc $(OtherOpt) -o $# -lthelibname $<
Now you can either run simply make to build all your apps, or make appname to build a specific app. Here I've created a single pattern rule that will be used anytime you want to create something from something.c. I used the $# automatic variable, which will expand to the name of the output, and the $< variable, which will expand to the name of the first prerequisite, so that the command-line is correct regardless of the specific app being built. Technically you don't need the all line, but I figured you probably didn't want to always have to type in the name(s) of the apps you want to build.
Also, technically you can probably get away without having any of this makefile, because GNU make already has a built-in pattern rule for the %: %.c relationship! I mention this option only for completeness; personally, I prefer doing things the way I've shown here because it's a little bit more explicit what's going on.
%.o: %.c
gcc $(OtherOpt) -c -o $# -lthelibname $<
That compiles all .c files to their .o files (object code) of the same base name. Then in your actual target(s), you would include all necessary .o files as dependencies and use gcc $(OtherOpt) -o $# $^ -lthelibname, assuming I'm not misunderstanding how your build is set up.
Some versions of make also support the suffix rule .c.o to be ALMOST the same thing as %.o: %.c, but the suffix rules can't have any dependencies. Writing .c.o: foo.h tells make to compile "foo.h" to "foo.c.o" rather than requiring "foo.h" as a dependency of any file with a .c suffix as %.o: %.c foo.h would correctly do.
I learnd from http://sourceforge.net/projects/gcmakefile/
LDLIB = -lpthread
LDFLAGS = -Wl,-O1 -Wl,--sort-common -Wl,--enable-new-dtags -Wl,--hash-style=both $(LDLIB)
SRCDIRS =
SRCEXTS = .c .C .cc .cpp .CPP .c++ .cxx .cp
CFLAGS = -pipe -march=core2 -mtune=generic -Wfloat-equal \
#-Wall -pedantic
ifeq ($(SRCDIRS),)
SRCDIRS = .
endif
SOURCES = $(foreach d,$(SRCDIRS),$(wildcard $(addprefix $(d)/*,$(SRCEXTS))))
TARGET = $(addprefix bin/,$(basename $(SOURCES)))
all: $(TARGET)
ls -l $(TARGET)
bin/%: %.c dir
gcc $(CFLAGS) $(LDFLAGS) -o $# $<
dir:
#-mkdir bin
.PHONY : clean
clean:
-rm $(TARGET)
-rmdir bin

What is the best approach to use different CFLAGS for the same source files?

i need to build the same source tree twice,
1 - with normal cflags to build the project binary
2 - with cflags plus -fPIC to build a static library that would be some sort of SDK to develop project dynamic modules.
Using only one Makefile, what is the best approach to accomplish this?
It would be nice to do some sort of :
all: $(OBJECTS)
lib_rule: $(OBJECTS)
CFLAGS += -fPIC
.cpp.o:
$(CC) -c $< -o $# $(CFLAGS)
But obviously it can't be done.
Thanks
One thing I've used in the past is a different extension:
.cpp.o:
$(CC) -c $< -o $# $(CFLAGS)
.cpp.lo:
$(CC) -c $< -o $# $(CFLAGS) $(EXTRA_CFLAGS)
You then build your static library from the .lo files and you binary from the .o files:
prog: a.o b.o
libsdk.a: a.lo b.lo
Assuming you are using GNU Make, you can use some built in functions to only have to maintain the list of objects one time:
OBJS = a.o b.o
LOBJS = $(patsubst %.o, %.lo, $(OBJS))
GNU make offers also "Target-specific Variable Values". Consider the following Makefile:
# Makefile
CFLAGS := My Cflags
all: $(OBJECTS)
#echo "$# CFLAGS is: " $(CFLAGS)
lib_rule: CFLAGS += extended by -fPIC
lib_rule: $(OBJECTS)
#echo "$# CFLAGS is: " $(CFLAGS)
# Makefile - end.
$ make all
all CFLAGS is: My Cflags
$ make lib_rule
lib_rule CFLAGS is: My Cflags extended by -fPIC
$
(Please note: if you copy and paste the example, remember to re-add the tabstops in front of the command lines. I always get caught by that.)
Instead of placing the compiled .o files in the same directory as the source, I create them in labeled sub-directories. In your case, you can have the static library files created as source_dir/lib/*.o and your normal files as source_dir/bin/*.o. In your different build targets after you set up your unique CFLAGS, simply generate a DIR_NAME value holding the name of the appropriate sub-folder. You can use this variable when you create the paths for the compiler to use when building and when linking.
In a different make tool such as CMake, you can express something like that much more easily.
For instance, you could well do
set(sources ABC.cpp DEF.cpp XYZ.cpp)
ADD_LIBRARY(mylib STATIC ${sources})
add_executable(myExecutable ${sources} main.cpp)
Or, you could repeatedly build the same directory with different flags by including it several times from the directory's logical parent, i.e.
set(MyTweakFlag 2)
add_subdirectory("libDir" "libDir2")
set(MyTweakFlag 3)
add_subdirectory("libDir" "libDir3")
...and then use if() or whatnot in the child directory to set the right flags.
Particularly if you have many such configurations, using make becomes quite fragile; make won't correctly find the transitive closure of recursive make dependancies (and certainly won't correctly find the dependancy on the makefile itself - if you change flags, say) so if you're going to do complicated makefile magic: do it with a better tool!
(CMake simply happens to be what I replaced make with, but there are various other replacements possible, of course)

Resources