8

I have a build file on OSX Lion

VPATH = src include
CFLAGS ="-I include -std=gnu99"

hello: hello.o
    gcc $^ -o $@

hello.o: hello.h hello.c
    gcc $(CFLAGS) -c $< -o $@

But when I try and run this make file I get the following error

    ld: warning: ignoring file hello.o, file was built for unsupported file format which is not the architecture being linked (x86_64)
Undefined symbols for architecture x86_64:
  "_main", referenced from:
      start in crt1.10.6.o
ld: symbol(s) not found for architecture x86_64
collect2: ld returned 1 exit status

I have tried using the flag -arch x86_64 but still get the same error.

Running the arch command gives: i386.

uname -a tells me: Darwin Kernel Version 11.3.0: Thu Jan 12 18:47:41 PST 2012; root:xnu-1699.24.23~1/RELEASE_X86_64 x86_64

I've also tried to add the switch -march=x86-64 as described in this answer file was built for i386 which is not the architecture being linked (x86_64) while compiling OpenCV2.2 for iOS 4.2 on Mac OSX 10.6 but that hasn't worked for me.

The output from the command line is:

gcc -I include -std=gnu99 -m64  -c include/hello.h -o hello.o  
gcc -I include -std=gnu99 -m64  hello.o -o hello
ld: warning: ignoring file hello.o, file was built for unsupported file format which is not the architecture being linked (x86_64)
Undefined symbols for architecture x86_64:
  "_main", referenced from:
      start in crt1.10.6.o
ld: symbol(s) not found for architecture x86_64
collect2: ld returned 1 exit status
make: *** [hello] Error 1
Community
  • 1
  • 1

3 Answers3

4
  1. Remove all the object files.

  2. Revise the makefile more like:

    VPATH   = src include
CFLAGS  = -I include -std=gnu99 -m64
CC      = gcc
LDLIBS  =
LDFLAGS =

hello: hello.o
    $(CC) $(CFLAGS) $^ -o $@

hello.o: hello.c hello.h
    $(CC) $(CFLAGS) -c $< -o $@ $(LDFLAGS) $(LDLIBS)

Note that I've macro-ized everything on the command lines. The CFLAGS are used in all compilations. They are not enclosed in double quotes. The -m64 option requests a 64-bit build; it shouldn't be necessary, but it makes it explicit. You don't yet need the LDFLAGS or LDLIBS macros (so you could omit them without causing yourself problems), but they show how you might proceed when you do need some libraries at link time.

For my own makefiles, I do things like:

IFLAGS = -Iinclude
WFLAG1 = -Wall
WFLAG2 = -Werror
WFLAG3 = -Wextra
WFLAGS = $(WFLAG1) $(WFLAG2) $(WFLAG3)
OFLAGS = -g -O3
SFLAG1 = -std=c99
SFLAG2 = -m64
SFLAGS = $(SFLAG1) $(SFLAG2)
DFLAGS = # -Doptions
UFLAGS = # Set on make command line only
CFLAGS = $(SFLAGS) $(DFLAGS) $(IFLAGS) $(OFLAGS) $(WFLAGS) $(UFLAGS)

That way I can adjust just about any single argument to the C compiler on the command line. For example, to do 32-bit builds, I can run:

make SFLAG2=-m32

Etc. The downside is I can never remember which xFLAGn option affects which. However, a quick look at the makefile rectifies that, and I can change the compilation without modifying the makefile at all.

(I also often use CC="gcc -m64" to force 64-bit compilations on other people's software.)

Jonathan Leffler
  • 730,956
  • 141
  • 904
  • 1,278
  • Thanks, but I'm still getting the exact same error: "file was built for unsupported file format which is not the architecture being linked (x86_64)". The makefile is now the exact same as your first example –  Apr 17 '12 at 06:10
  • That is surprising. Can you show the command lines executed? There should be two, of course: one producing the `.o` file, and one attempting to produce the executable. What happens if you try: `gcc -o hello hello.c`, producing the object file direct from the source code? What happens if you drop the VPATH line, or comment it out? Where is the file `hello.c`? What about `hello.h`? – Jonathan Leffler Apr 17 '12 at 06:26
  • I updated the question with the command output so that I can format it –  Apr 17 '12 at 06:29
  • I see the update. Given that `-m64` isn't working, what happens when you try `-m32` instead? Does that also fail? – Jonathan Leffler Apr 17 '12 at 06:30
  • when I use -m32 I get: file was built for unsupported file format which is not the architecture being linked (i386) –  Apr 17 '12 at 06:32
  • 1
    Hold on: the first command line - `gcc -I include -std=gnu99 -m64 -c include/hello.h -o hello.o`. It isn't compiling `hello.c` at all. Have a look at what `$^` is supposed to be; you probably need to use something like `$*` or `$<` or something to specify the C file. When you run `file hello.o`, what does it tell you? File is empty? You might need to list `hello.c` before `hello.h` in the dependencies for `hello.o`. – Jonathan Leffler Apr 17 '12 at 06:32
  • 1
    Phew! Chapter 10 of the GNU Make manual covers the macros such as `$<` and `$^`. – Jonathan Leffler Apr 17 '12 at 06:43
  • Thanks I'll go and download it –  Apr 17 '12 at 06:45
1

I had this problem when I accidentally included a .h file in an archive...

Sohail
  • 3,020
  • 1
  • 25
  • 23
0

In my case, the -M option was creating this issue. I added this option to project dependencies but somehow it was causing issues.

Curt
  • 5,518
  • 1
  • 21
  • 35
user2657341
  • 1