Fortran: makefile with already compiled modules - makefile

I have a project structure like this
-Project
--Common
---types.f90
---global.f90
---common_routines.f90
--Program 1
---program1.f90
---module1.f90
---module2.f90
---etc...
--Program 2
--etc...
Where, Common is folder that contains some modules that are shared across all programs. How do I include this modules on my makefile?
I tried this:
FC = gfortran
FCFLAGS = -fbounds-check -O3
FCFLAGS += -I ../Common
all: program1
program1: module1.o module2.o module3.o
program1.o: module1.o module2.o module3.o
module2.o: module1.o
module3.o: module2.o module1.o
%: %.o
$(FC) $(FCFLAGS) -o $# $^
%.o: %.f90
$(FC) $(FCFLAGS) -c $<
clean:
rm -rf *.o *.mod
but I get an undefined reference error to the common modules variables.

I tried FCFLAGS += -I../Common types.o global.o common_routines.o
This will not work because -I is an option to the GNU Fortran preprocessor
to specify a path that the preprocessor shall search for files to be INCLUDE-ed
prior to compilation. You cannot use it to specify a path where object files (*.o)
will be searched for, after compilation, by the linker. It means nothing to
the linker and is not passed to the linker.
For simplicity let's assume that the object files you need to to link for
program1 are just program1/program1.o plus the pre-existing common/types.o,
common/global.o and common/common_routines.o
Then the following Makefile, placed in directory program1, will build it:
OBJS = program1.o ../common/types.o ../common/global.o ../common/common_routines.o
.phony: all
all: program1
program1: program1.o
$(FC) -o $# $(FCFLAGS) $(OBJS)
clean:
rm -f *.o program1
Just list all the required object files to the linker, in this case via $(OBJS)
You might wish to take the precaution of making sure that the common modules
are up to date before you build program1, and you now might think that you can do that
simply be replacing:
program1: program1.o
with:
program1: $(OBJS)
thus prompting make to recompile any of the four object files that is out
of date with respect to the corresponding source file, as a prerequisite of
building program1
make will certainly endeavour to do that, but take care. That way, it will
recompile, say, ../common/types.o from ../common/types.f90 just by its
implicit default recipe for making an .o from an .f90, since this makefile is
not telling it to do any different. But that may not be the way in which
../common/types.f90 is meant to be compiled, if you also have is a makefile in common
that stipulates how to do it in some non-default manner.
In that case, the common object files should always be compiled as per the
makefile in common. Better leave the prerequisites of program1 alone but change the recipe to:
program1: program1.o
$(MAKE) -C ../common
$(FC) -o $# $(FCFLAGS) $(OBJS)
Now, any time program1 needs to be rebuilt, the recipe will preemptively run make in ../common
before it links the four object files. (It's a small inelegance that this $(MAKE) -C ../common
will be invoked even if there's nothing for it to do: this is avoidable by more advanced make usage).
Lastly you might also find a need (if not in this case, then in another) to distinguish
between flags passed to preprocessing and/or flags passed to compilation and/or flags passed to linkage.
Conventionally, these are assigned to distinct make variables, e.g. FPPFLAGS (preprocessor),
FCFLAGS (compiler), LDFLAGS (linker).

The GNU syntax to define additional include directory is -Idir not -I dir (extra space)
Also make sure that common modules are already compiled and include search path points to the directory where you have compiled modules, not source files:
This path is also used to search for .mod files when previously compiled modules are required by a USE statement.

Related

How to write Make rules for *.h files as dependencies

I'm trying to understand how to handle header file dependencies in Make rules. Let me give you a specific example.
I'm building application called myap using GNU Make. It consists of various *.h and *.c files.
Directory inc/ contains defs.h and util.h header files.
Directory src/ contains main.c, cmd.c and win.c files.
Directory obj/ contains all generated object files.
I have multiple applications that need different build options. So I don't want to rely on any implicit rules and would like to specify my own rules for all object files, etc.
I would like to specify the following rules:
Object files depend on specific *.h and *.c files. If any of them change, all object files must be regenerated. However, even though *.h files are part of the prerequisites list, I don't want to pass them to the compiler. I only want to compile *.c files.
Executable myapp depends on specific *.o files. If any of them change, executable file must be regenerated.
So far, the following Makefile with a static pattern rule seems to work correctly:
myapp_inc := inc/defs.h inc/util.h
myapp_src := src/main.c src/cmd.c src/win.c
myapp_obj := $(patsubst src/%.c,obj/%.o,$(myapp_src))
myapp_bin := obj/myapp
.PHONY: all
all:
# Create obj/main.o obj/cmd.o and obj/win.o from various *.c files
# If any *.h files in $(myapp_inc) list change, all objects are regenerated.
# If any *.c files in $(myapp_src) list change, all objects are regenerated.
$(myapp_obj): obj/%.o: src/%.c $(myapp_inc) $(myapp_src)
gcc -c -o $# $<
# Create obj/myapp from various *.o files
# If any *.o files in $(myapp_obj) list change, executable is regenerated.
$(myapp_bin): $(myapp_obj)
gcc -o $# $^
all: $(myapp_bin)
.PHONY: clean
clean:
rm -f obj/*
I don't quite understand how Make rules should be written correctly in order to handle such use case. Is the above static pattern rule, the only way that works correctly?
Specifically, I have tried the following combinations, as given in various simple examples on the Internet, and they all failed for various reasons.
This rule causes $< to always pass the name of the first prerequisite, which doesn't work with multiple *.c files:
$(myapp_obj): $(myapp_src) $(myapp_inc)
gcc -c -o $# $<
$ make
gcc -c -o obj/main.o src/main.c
gcc -c -o obj/cmd.o src/main.c
gcc -c -o obj/win.o src/main.c
gcc -o obj/myapp obj/main.o obj/cmd.o obj/win.o
/bin/ld: obj/cmd.o: in function `main':
main.c:(.text+0x0): multiple definition of `main'; obj/main.o:main.c:(.text+0x0): first defined here
/bin/ld: obj/win.o: in function `main':
main.c:(.text+0x0): multiple definition of `main'; obj/main.o:main.c:(.text+0x0): first defined here
collect2: error: ld returned 1 exit status
make: *** [Makefile:18: obj/myapp] Error 1
This rule causes $^ to always pass the names of all prerequisites, which fails:
$(myapp_obj): $(myapp_src) $(myapp_inc)
gcc -c -o $# $^
$ make
gcc -c -o obj/main.o src/main.c src/cmd.c src/win.c inc/defs.h inc/util.h
gcc: fatal error: cannot specify ‘-o’ with ‘-c’, ‘-S’ or ‘-E’ with multiple files
compilation terminated.
make: *** [Makefile:13: obj/main.o] Error 1
Now I understand the difference between $< and $^ variables, but a lot of documentation is not clear on how they should be used when dealing with a list of multiple *.c and *.h files as prerequisites.
What are the recommended usage pattern for this?
Why is it that when using $< only *.c files get passed to the recipe, but not *.h files? Is Make doing some internal filtering? Is this documented anywhere? Is it possible to modify this behavior for custom suffixes?
Is the above static pattern rule, the only way to make objects depend on *.h and *.c files, but exclude *.h files during compilation?
I don't understand the goal of trying to avoid implicit rules. But in any event, it doesn't matter to the recipe you write whether the rule was implicit or explicit: the same automatic variables are set either way. The $< automatic variable is always the first prerequisite, so if you write your rules such that the first prerequisite is the appropriate .c file then you can always use $< in your recipe to mean the .c file and no other files. All the following will work:
%.o : %.c $(headers)
gcc -c -o $# $<
foo.o: foo.c $(headers)
gcc -c -o $# $<
foo.o : %.o : %.c $(headers)
gcc -c -o $# $<
%.o : %.c
gcc -c -o $# $<
$(srcs) : $(headers)
and others.
Does this mean that all of the prerequisites apply, but only those that match the pattern get passed to the recipe?
I don't understand the question, really. The value of variables and the expansion of the recipe happens only AFTER make has decided to run the rule and is not really related (except for some special automatic variables like $?). Once make has decided that the target is out of date and the recipe needs to be run, it will assign the appropriate automatic variables, expand the recipe, then pass the recipe to the shell to be run.
The automatic variables are assigned as described in the manual: $# is the target, $< is the first prerequisite, $^ is all the prerequisites, etc.
ETA
You still haven't really explained why you don't want to use static pattern rules. They are a perfectly fine and reasonable way to do things.
If you explain what you don't like about static pattern rules, or what you wish you could do differently, then we can probably suggest alternatives that meet those requirements.
Specifically, I have tried the following combinations, as given in various simple examples on the Internet,
$(myapp_obj): $(myapp_src) $(myapp_inc)
Wherever you found this as a recommended example on the Internet, you should immediately delete from any bookmarks as that site doesn't know anything about make.
We see this paradigm at least once a week on SO. I've never really understand why people think it will work: I guess they think make is much more "magical" than it is. Consider, what does the above expand to? Suppose myapp_obj contained foo.o bar.o biz.o and myapp_src contained foo.c bar.c biz.c and myapp_inc contained foo.h bar.h, then make sees:
foo.o bar.o biz.o: foo.c bar.c biz.c foo.h bar.h
I suppose some people think make will intuit that the ".o" files should somehow match up with the ".c" files and will generate a bunch of rules that make that true. That's not what make does. The above line is exactly identical to writing this:
foo.o: foo.c bar.c biz.c foo.h bar.h
bar.o: foo.c bar.c biz.c foo.h bar.h
biz.o: foo.c bar.c biz.c foo.h bar.h
That is, if you have multiple targets make creates one copy of the rule for each target, with the same prerequisites and recipe.
This is obviously not what you want, and that's why none of the examples that try to do things this way can ever work properly.
Why is it that when using $< only *.c files get passed to the recipe, but not *.h files? Is Make doing some internal filtering? Is this documented anywhere? Is it possible to modify this behavior for custom suffixes?
None of that is the case. As I described above, the $< expands to the first prerequisite. That's all. It doesn't matter whether the first prerequisite is a .c file, a .h file, or some other file; whatever it is, $< will be that value. If you write your rule as:
foo.o : foo.c foo.h ; $(CC) -c -o $# $<
then your compiler will be invoked with foo.c. If you write your rule as:
foo.o : foo.h foo.c ; $(CC) -c -o $# $<
then your compiler will be invoked with foo.h. There's no magic here.

How to have the proper order of .mod in makefile

I'm trying to do a Makefile for my project in Fortran and make it reusable for now projects. The Mkefile I have come up with after many tries is the following. The thing is that it works OK in few scenarios but right now I have this files:
main.f90, initial.f90, parameters.f90, functions.f90
The first one has my program and the others are modules where I have subroutines. The thing is that initial needs parameters to work but the Makefile decides to try compiling initial before parameters. Is there a way I can make it to choose them in the right order?
vpath %.o obj
vpath %.mod mod
TOP_DIR = .
MOD_DIR = $(TOP_DIR)/mod
OBJ_DIR = $(TOP_DIR)/obj
SOURCES = $(wildcard *.f90)
OBJECTS := $(patsubst %.f90, $(OBJ_DIR)/%.o, $(SOURCES))
ALL_MODULES = $(patsubst %.f90, %.mod, $(SOURCES))
MODULES = $(filter-out main.mod, $(ALL_MODULES))
FFLAGS = -O3 -J$(MOD_DIR) -march=native -fimplicit-none -Wall -Wline-truncation -fwhole-file
.PHONY: main clean
main: main.out
./main.out
main.out: $(MODULES) $(OBJECTS)
gfortran $(FFLAGS) $(OBJECTS) -o main.out
$(OBJ_DIR)/%.o : %.f90
gfortran $(FFLAGS) $^ -c $#
%.mod: %.f90
gfortran $(FFLAGS) -c $^
clean:
rm -f $(OBJ_DIR)/*.o *.out $(MOD_DIR)/*.mod
List the source files in the dependency order instead of automatically: SOURCES = parameters.f90 initial.f90 functions.f90.
Separate the source for the modules and for the program: main.f90 does not, in general, generate .mod files and that will break the simplistic Makefile you have.
I also advise to use another build system (cmake, for instance) but for small projects you can get by with a Makefile if you prefer to.
With modern Fortran compilation order is very important. When using makefiles, for anything non-trivial it's too tedious to keep track of the dependencies manually. Enter various tools to generate these dependency lists. I have successfully used makedepf90 which is available as a package in EPEL (for RHEL/CentOS) and Debian/Ubuntu.
You can even make a makefile rule for regenerating the dependency list, if you keep the list in a separate file, like:
include depend.mk
# Dependencies generated by "makedepf90" tool.
depend depend.mk:
cd $(srcdir) && makedepf90 $(yourprogramname_SOURCES) >depend.mk
The above BTW works both with plain makefiles and automake (in which case you put the above snippet to your Makefile.am)

Why is makefile exhibiting non-deterministic behaviour?

I have a makefile that is trying to do the following: identify all files under the current directory (all sub-directories included) with .c and .s extensions, for each one compile a non-linked object file and put it into a directory. All C files end up in objects/c, all assembly files end up in objects/ass.
The makefile always works as expected on the first execution (all commands are called in the right order) and no errors are produced.
However if I call make again, half of the time i get "nothing to be done for 'all'.". Which is what you would expect, since no files have been modified. But the other half of the time, make is selecting a random assembly file and compiling that file. That is to say,if I keep doing "make" I sometimes compile file1.s sometimes file2.s. and it keeps randomly swapping between the assembly files add infinitum (it never reaches a "nothing to be done") state.
How is make exhibitting non deterministic behaviour?
This is the smallest makefile I could make that reproduces the error:
SRC_C = $(wildcard *.c) $(wildcard **/*.c)
SRC_ASS = $(wildcard *.s) $(wildcard **/*.s)
OBJECTS_C = $(addprefix $(OBJECT_DIR)c/, $(notdir $(SRC_C:.c=.o)))
OBJECTS_ASS = $(addprefix $(OBJECT_DIR)ass/, $(notdir $(SRC_ASS:.s=.o)))
OBJECTS = $(OBJECTS_C) $(OBJECTS_ASS)
OBJECT_DIR = objects/
all: $(OBJECTS)
%/:
mkdir $#
$(OBJECTS_C): $(OBJECT_DIR) $(OBJECT_DIR)c/
arm-none-eabi-gcc -O0 -march=armv8-a $(wildcard */$(#F:.o=.c)) -nostartfiles -c -o $#
$(OBJECTS_ASS): $(OBJECT_DIR) $(OBJECT_DIR)ass/
arm-none-eabi-as -march=armv8-a $(wildcard */$(#F:.o=.s)) -c -o $#
.PHONY: clean
clean:
rm -rf $(OBJECT_DIR)
You have many errors here.
The biggest is a conceptual one: By flattening all your object files into one directory, there's no way to express proper dependencies using pattern rules, so your object files do not really depend on their respective source files. I'd say: just don't do that! Having object directories is fine, but they should mirror the directory structure of the source tree.
Further errors:
directly depending on directories. This will not work as expected, directories should always be order-only dependencies, as already stated in the comments
Make doesn't support recursive wildcards -- if you really need that, you could write your own function or, assuming you're always building on *nix, just call find instead
Pattern rules for creating directories are not the best idea either -- I'd suggest to collect all needed directories in a variable and loop over that.
Stylistic improvements:
Assign variables that don't need deferred evaluation with :=
Assign variables influencing the build process with ?=, so the user can override them at the command line
Use "standard" variables like CC, AS, CROSS_COMPILE
declare all phony targets in .PHONY.
Your Makefile with these changes applied would look like this:
OBJECT_DIR ?= objects
C_OBJECT_DIR ?= $(OBJECT_DIR)/c
AS_OBJECT_DIR ?= $(OBJECT_DIR)/ass
SRC_C:= $(shell find -name \*.c)
SRC_ASS:= $(shell find -name \*.s)
OBJECTS_C:= $(addprefix $(C_OBJECT_DIR)/, $(SRC_C:.c=.o))
OBJECTS_ASS:= $(addprefix $(AS_OBJECT_DIR)/, $(SRC_ASS:.s=.o))
OBJECTS:= $(OBJECTS_C) $(OBJECTS_ASS)
OUTDIRS:= $(sort $(dir $(OBJECTS)))
CROSS_COMPILE ?= arm-none-eabi-
CC ?= gcc
AS ?= as
CFLAGS ?= -O0 -march=armv8-a -nostartfiles
ASFLAGS ?= -march=armv8-a
all: $(OBJECTS)
$(OUTDIRS):
$(foreach _dir,$#,mkdir -p $(_dir);)
$(C_OBJECT_DIR)/%.o: %.c | $(OUTDIRS)
$(CROSS_COMPILE)$(CC) -c -o $# $(CFLAGS) $<
$(AS_OBJECT_DIR)/%.o: %.s | $(OUTDIRS)
$(CROSS_COMPILE)$(AS) -c -o $# $(ASFLAGS) $<
clean:
rm -rf $(OBJECT_DIR)
.PHONY: all clean
Note there is one important thing missing: automatic dependencies. With this Makefile, each object file depends on its respective source file, but completely misses any headers included. For anything other than a simple toy, you should add that, google for "gnu make gcc automatic dependencies" or something similar (not the scope of this question).

Different dependency locations in Makefile based on target

I'm trying to set up a Makefile to handle two different targets from one set of sources, and I'm a little out of my element. The vast majority of it works fine, but my dependency structure is hosed and thus I'm forced to do a full recompile each time. A pared down sample is as follows:
first: OBJDIR = obj
second: OBJDIR = obj-2
SRCS = $(wildcard src/*.cc)
OBJECTS = $(patsubst %.cc,$(OBJDIR)/%.o,$(SRCS))
first: CFLAGS = -g -Wall -Wextra -std=c++11 -MMD
second: CFLAGS = -g -Wall -Wextra -std=c++11 -MMD -DCOMPILE_FLAG
$(OBJDIR)/%.o: %.cc
#mkdir -p $(OBJDIR)/src
clang++ -c $(CFLAGS) -o $(OBJDIR)$# $<
#DEPENDENCIES AREN'T WORKING PROPERLY
-include $(OBJECTS:.o=.d)
first: $(OBJECTS)
clang++ -o gen/first $(OBJECTS)
second: $(OBJECTS)
clang++ -o gen/second $(OBJECTS)
If I #echo $(OBJECTS:.o=.d) under my first: executable generation (or as it's used in the compilation step), it properly expands to /obj/src/filename.d. However, when it's in the include line, it shows up simply as /src/filename.d. Obviously the .d files don't exist in that location, so it doesn't find them and does a full recompile.
Makefiles are not something I'm heavily experienced with, so if there are better ways to do anything up above, I'm all ears. The key point, though, is being able to have two sets of object files and two sets of dependencies from the same sources.
--
To clarify on the eventual goals, what I have is a set of source files that are used to build two separate executables, with the differences handled via #ifdefs.
What I want to get out of the makefile structure is a single makefile with two targets. Each target generates its own .o/.d files, so that when a change is made to the source, I can run make first and make second to generate the two new executables without having to recompile everything from scratch. I've handled this previously by having two separate makefiles, but that just seems wrong.
You've missed a critical sentence in the GNU make manual related to target-specific variables:
As with automatic variables, these values are only available within the context of a target's recipe
This means that you can't use target-specific variables in either the target or prerequisite lists: any variables used there will ALWAYS have the global value and never the target-specific value. Similarly, include lines are parsed as the makefile is read in: there's no target context at all here either so the global value is used.
Further, any time you see ANY rule in a makefile that is creating a file which is not exactly $# but is instead some modification of it, that's a red flag and you know you have a problem: -o $(OBJDIR)$# is wrong.
In general, there's a lot wrong with this makefile and it's not entirely clear exactly what you're trying to do here. Maybe if you stepped back and described the goal you want to achieve we can give you some pointers.
I believe the answer is simply to rethink the way I was doing it. I've rewritten the makefile to be as follows (trimming out anything unrelated), after reading a lot of Makefile documentation and taking into consideration some comments from MadScientist:
CC = clang++
SRCS = $(wildcard src/*.cc)
OBJECTS = $(patsubst %.cc,$(OBJDIR)/%.o,$(SRCS))
CFLAGS = -g -Wall -Wextra -std=c++11 -MMD
.PHONY: all clean run
all: $(EXECUTABLE)
-include $(OBJECTS:.o=.d)
$(EXECUTABLE): $(OBJECTS)
#mkdir -p gen
$(CC) -o gen/$(EXECUTABLE) $(OBJECTS)
$(OBJDIR)/%.o: %.cc
#mkdir -p $(#D)
$(CC) -c $(CFLAGS) $(CMDFLAGS) -o $# $<
clean:
rm -rf obj obj-2 gen
run:
cd gen && ./$(EXECUTABLE)
From there, I made a couple aliases in my .bash_profile:
alias mfirst="make OBJDIR=obj EXECUTABLE=first"
alias msecond="make OBJDIR=obj-2 CMDFLAGS=-DCOMPILE_FLAG EXECUTABLE=second"
Because the variables are now set outside of any target specification, everything plays nicely. It keeps the object files and the dependencies separate, and the aliases still allow quick usage (including mfirst run, for example).
Feel free to point out any flaws here, but I'm fairly happy with the result.

Why doesn't my make file leave behind object files?

I am new to make files and I put this together with a bit of trial & error. This code is used to compile my c++ program.
My main.cpp file in the same folder as the makefile. I have a lib/ folder that contains the headers main depends on.
The following makefile results in a correct and complete compilation of my code. But I was expecting that I would find *.o objects left behind. (Note that I've tried to make both with and without the "clean" rule, and I get the same results both times.)
#
# Makefile
#
CXX = g++
CCFLAGS = -O3 -I/sw/include -L/sw/lib
## /sw/include and /sw/lib contain dependencies for files in my lib/
LDFLAGS = -lpng
OPTS = $(CCFLAGS) $(LDFLAGS)
SOURCES = $(wildcard lib/*.cpp) main.cpp
OBJECTS = $(SOURCES: .cpp = .o)
TARGET = spirals
$(TARGET): $(OBJECTS)
$(CXX) $(OPTS) $^ -o $#
.PHONY: depend
depend:
g++ -MM $(SOURCES) > depend
## generate dependencies list
include depend
.PHONY: clean
clean:
rm -f *.o lib/*.o $(TARGET)
Also, in case it matters, I'm on MacOSX and my program is designed in xcode. (I know that xcode has its own build flow, but I'm designing a command-line program for a linux system and I'd like to test compilation & linking in a bash environment instead of only going through xcode.)
Questions:
Am I correct to expect makefiles to produce *.o files that stick around once the main target has been created?
If so, why doesn't my makefile do this?
If you observe what command your $(TARGET) rule causes to be run:
g++ -O3 -I/sw/include -L/sw/lib -lpng lib/bar.cpp lib/foo.cpp main.cpp -o spirals
you'll see that $(OBJECTS) in fact contains *.cpp files, and there are no *.o files sticking around because you haven't asked for any.
The problem is here:
OBJECTS = $(SOURCES:.cpp=.o)
In your GNU makefile as written, this substitution reference is written with excess spaces, so never matches anything and $(OBJECTS) ends up the same as $(SOURCES). Rewrite it as above and it'll do what you expect.
(Other notes: -lpng needs to go at the end of the link command to work in general, so you should introduce another make variable (traditionally called $(LDLIBS)) so as to arrange that. Especially as someone new to makefiles, you would do better to spell out your dependencies explicitly rather than playing games with $(wildcard) and a computed $(OBJECTS). -I options are needed during compilation while -L options are used during linking, so it would be good to arrange separate $(CXXFLAGS)/$(LDFLAGS) variables used in separate rules so they are only added when required.)

Resources