In this link, what is an inline function and what is the inline keyword is explained. I'm reading through it because I realized I've never understood the meaning of these two concepts and how they should be used in practice. I'm quoting and commenting from the link I provided
An inline function or inline variable (since C++17) is a function or variable (since C++17) with the following properties:
1) There may be more than one definition of an inline function or variable (since C++17) in the program as long as each definition appears in a different translation unit. For example, an inline function or an inline variable (since C++17) may be defined in a header file that is include'd in multiple source files.
Here I already have understanding problems, declaration is the specification of new identifiers like
void func(void);
while a definition is the actual implementation, including the body
void func(void) {
//some code...
}
The point 1) means that I can give different implementation as long as they're in different translation units (i.e. one implementation per header e per source files), but I'm puzzled in the case I have a source file source.cc
with a declaration for func
and an header file with another declaration of func
the translation unit is the pair source.cc+header.h
and in such a case having declared two times func
doesn't make any sense, is that right?
2) The definition of an inline function or variable (since C++17) must be present in the translation unit where it is accessed (not necessarily before the point of access).
This is the usual case where I separate definition from declaration, the first in an header file, the second one is in the source file, if I need to use the function I have to include only the header right? The access point would be provided by the source during the linking phase, correct?
3) An inline function or variable (since C++17) with external linkage (e.g. not declared static) has the following additional properties: 1) It must be declared inline in every translation unit. 2) It has the same address in every translation unit.
Could you provide a simple example of what this means? I can't picture a practical case of such a case. The case 3) states that the keyword inline
is mandatory unless the function to be declared is static.
Is everything I said so far correct?
In practice a function should be inline when such a function is very small, but not always the compiler would inline the function declared as inline, for example if it has loops inside or recursion (Effective C++ states so). In general then it's compiler dependent, I the wonder now...
Say I have two functions the first one is self-contained (it doesn't internally call any-other function), the second one call's the first one (you can assume they're both 10 lines for sake of argument). Should both of them declared inline? should they be declared in an header file? or should I separate definition in an header file and the implementation in an source file? What would be better?
Edit 1:
Following one of the answer is better if I work by examples, with related assembly code analysis.
I removed the previous code because it was meaningless (the -O3
flag optimization wasn't set).
I start again... I have 5 files header.h
,src.cc
, src1.cc
, src2.cc
and main.cc
. For each translation unit the related assembly code is posted.
I've manipulated such files in three different ways and later observed the assembly code generated, this helped me to understand how the inline keyword works.
Example 1:
header.h
#ifndef HEADER_H_
#define HEADER_H_
int func(int a, int b);
int test_1();
int test_2();
#endif /* HEADER_H_ */
src.cc
#include "header.h"
int func(int a, int b)
{
return a + b;
}
src1.cc
#include "header.h"
int test_1()
{
int a, b, c;
a = 3;
b = 7;
c = func(a, b);
return c;
}
src2.cc
#include "header.h"
int test_2()
{
int a, b, c;
a = 7;
b = 8;
c = func(a, b);
return c;
}
main.cc
int main(int argc, char** argv)
{
test_1();
test_2();
test_1();
test_2();
}
Assembly 1:
src.s
GAS LISTING /tmp/cc0j97WY.s page 1
1 .file "src.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl _Z4funcii
6 .type _Z4funcii, @function
7 _Z4funcii:
8 .LFB2:
9 0000 8D043E leal (%rsi,%rdi), %eax
10 0003 C3 ret
11 .LFE2:
12 .size _Z4funcii, .-_Z4funcii
13 .globl __gxx_personality_v0
14 .section .eh_frame,"a",@progbits
15 .Lframe1:
16 0000 1C000000 .long .LECIE1-.LSCIE1
17 .LSCIE1:
18 0004 00000000 .long 0x0
19 0008 01 .byte 0x1
20 0009 7A505200 .string "zPR"
21 000d 01 .uleb128 0x1
22 000e 78 .sleb128 -8
23 000f 10 .byte 0x10
24 0010 06 .uleb128 0x6
25 0011 03 .byte 0x3
26 0012 00000000 .long __gxx_personality_v0
27 0016 03 .byte 0x3
28 0017 0C .byte 0xc
29 0018 07 .uleb128 0x7
30 0019 08 .uleb128 0x8
31 001a 90 .byte 0x90
32 001b 01 .uleb128 0x1
33 001c 00000000 .align 8
34 .LECIE1:
35 .LSFDE1:
36 0020 14000000 .long .LEFDE1-.LASFDE1
37 .LASFDE1:
38 0024 24000000 .long .LASFDE1-.Lframe1
39 0028 00000000 .long .LFB2
40 002c 04000000 .long .LFE2-.LFB2
41 0030 00 .uleb128 0x0
42 0031 00000000 .align 8
42 000000
43 .LEFDE1:
44 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
45 .section .note.GNU-stack,"",@progbits
src1.s
GAS LISTING /tmp/cchSilt1.s page 1
1 .file "src1.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl _Z6test_1v
6 .type _Z6test_1v, @function
7 _Z6test_1v:
8 .LFB2:
9 0000 BE070000 movl $7, %esi
9 00
10 0005 BF030000 movl $3, %edi
10 00
11 000a E9000000 jmp _Z4funcii
11 00
12 .LFE2:
13 .size _Z6test_1v, .-_Z6test_1v
14 .globl __gxx_personality_v0
15 .section .eh_frame,"a",@progbits
16 .Lframe1:
17 0000 1C000000 .long .LECIE1-.LSCIE1
18 .LSCIE1:
19 0004 00000000 .long 0x0
20 0008 01 .byte 0x1
21 0009 7A505200 .string "zPR"
22 000d 01 .uleb128 0x1
23 000e 78 .sleb128 -8
24 000f 10 .byte 0x10
25 0010 06 .uleb128 0x6
26 0011 03 .byte 0x3
27 0012 00000000 .long __gxx_personality_v0
28 0016 03 .byte 0x3
29 0017 0C .byte 0xc
30 0018 07 .uleb128 0x7
31 0019 08 .uleb128 0x8
32 001a 90 .byte 0x90
33 001b 01 .uleb128 0x1
34 001c 00000000 .align 8
35 .LECIE1:
36 .LSFDE1:
37 0020 14000000 .long .LEFDE1-.LASFDE1
38 .LASFDE1:
39 0024 24000000 .long .LASFDE1-.Lframe1
40 0028 00000000 .long .LFB2
41 002c 0F000000 .long .LFE2-.LFB2
42 0030 00 .uleb128 0x0
43 0031 00000000 .align 8
43 000000
44 .LEFDE1:
45 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
46 .section .note.GNU-stack,"",@progbits
src2.s
GAS LISTING /tmp/cc2JMtt3.s page 1
1 .file "src2.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl _Z6test_2v
6 .type _Z6test_2v, @function
7 _Z6test_2v:
8 .LFB2:
9 0000 BE080000 movl $8, %esi
9 00
10 0005 BF070000 movl $7, %edi
10 00
11 000a E9000000 jmp _Z4funcii
11 00
12 .LFE2:
13 .size _Z6test_2v, .-_Z6test_2v
14 .globl __gxx_personality_v0
15 .section .eh_frame,"a",@progbits
16 .Lframe1:
17 0000 1C000000 .long .LECIE1-.LSCIE1
18 .LSCIE1:
19 0004 00000000 .long 0x0
20 0008 01 .byte 0x1
21 0009 7A505200 .string "zPR"
22 000d 01 .uleb128 0x1
23 000e 78 .sleb128 -8
24 000f 10 .byte 0x10
25 0010 06 .uleb128 0x6
26 0011 03 .byte 0x3
27 0012 00000000 .long __gxx_personality_v0
28 0016 03 .byte 0x3
29 0017 0C .byte 0xc
30 0018 07 .uleb128 0x7
31 0019 08 .uleb128 0x8
32 001a 90 .byte 0x90
33 001b 01 .uleb128 0x1
34 001c 00000000 .align 8
35 .LECIE1:
36 .LSFDE1:
37 0020 14000000 .long .LEFDE1-.LASFDE1
38 .LASFDE1:
39 0024 24000000 .long .LASFDE1-.Lframe1
40 0028 00000000 .long .LFB2
41 002c 0F000000 .long .LFE2-.LFB2
42 0030 00 .uleb128 0x0
43 0031 00000000 .align 8
43 000000
44 .LEFDE1:
45 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
46 .section .note.GNU-stack,"",@progbits
main.s
GAS LISTING /tmp/cc5CfYBW.s page 1
1 .file "main.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl main
6 .type main, @function
7 main:
8 .LFB2:
9 0000 4883EC08 subq $8, %rsp
10 .LCFI0:
11 0004 E8000000 call _Z6test_1v
11 00
12 0009 E8000000 call _Z6test_2v
12 00
13 000e E8000000 call _Z6test_1v
13 00
14 .p2align 4,,5
15 0013 E8000000 call _Z6test_2v
15 00
16 0018 31C0 xorl %eax, %eax
17 001a 4883C408 addq $8, %rsp
18 .p2align 4,,1
19 001e C3 ret
20 .LFE2:
21 .size main, .-main
22 .globl __gxx_personality_v0
23 .section .eh_frame,"a",@progbits
24 .Lframe1:
25 0000 1C000000 .long .LECIE1-.LSCIE1
26 .LSCIE1:
27 0004 00000000 .long 0x0
28 0008 01 .byte 0x1
29 0009 7A505200 .string "zPR"
30 000d 01 .uleb128 0x1
31 000e 78 .sleb128 -8
32 000f 10 .byte 0x10
33 0010 06 .uleb128 0x6
34 0011 03 .byte 0x3
35 0012 00000000 .long __gxx_personality_v0
36 0016 03 .byte 0x3
37 0017 0C .byte 0xc
38 0018 07 .uleb128 0x7
39 0019 08 .uleb128 0x8
40 001a 90 .byte 0x90
41 001b 01 .uleb128 0x1
42 001c 00000000 .align 8
43 .LECIE1:
44 .LSFDE1:
45 0020 14000000 .long .LEFDE1-.LASFDE1
46 .LASFDE1:
47 0024 24000000 .long .LASFDE1-.Lframe1
48 0028 00000000 .long .LFB2
49 002c 1F000000 .long .LFE2-.LFB2
50 0030 00 .uleb128 0x0
51 0031 44 .byte 0x4
52 .long .LCFI0-.LFB2
53 0032 0E .byte 0xe
GAS LISTING /tmp/cc5CfYBW.s page 2
54 0033 10 .uleb128 0x10
55 0034 00000000 .align 8
56 .LEFDE1:
57 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
58 .section .note.GNU-stack,"",@progbits
Example 2:
header.h
#ifndef HEADER_H_
#define HEADER_H_
inline int func(int a, int b)
{
return a + b;
}
int test_1();
int test_2();
#endif /* HEADER_H_ */
src.cc
#include "header.h"
/*
int func(int a, int b)
{
return a + b;
}*/
src1.cc
#include "header.h"
int test_1()
{
int a, b, c;
a = 3;
b = 7;
c = func(a, b);
return c;
}
src2.cc
#include "header.h"
int test_2()
{
int a, b, c;
a = 7;
b = 8;
c = func(a, b);
return c;
}
main.cc
int main(int argc, char** argv)
{
test_1();
test_2();
test_1();
test_2();
}
Assembly 2:
src.s
GAS LISTING /tmp/cczLx8os.s page 1
1 .file "src.cc"
2 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
3 .section .note.GNU-stack,"",@progbits
src1.s
GAS LISTING /tmp/ccMFMy9s.s page 1
1 .file "src1.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl _Z6test_1v
6 .type _Z6test_1v, @function
7 _Z6test_1v:
8 .LFB3:
9 0000 B80A0000 movl $10, %eax
9 00
10 0005 C3 ret
11 .LFE3:
12 .size _Z6test_1v, .-_Z6test_1v
13 .globl __gxx_personality_v0
14 .section .eh_frame,"a",@progbits
15 .Lframe1:
16 0000 1C000000 .long .LECIE1-.LSCIE1
17 .LSCIE1:
18 0004 00000000 .long 0x0
19 0008 01 .byte 0x1
20 0009 7A505200 .string "zPR"
21 000d 01 .uleb128 0x1
22 000e 78 .sleb128 -8
23 000f 10 .byte 0x10
24 0010 06 .uleb128 0x6
25 0011 03 .byte 0x3
26 0012 00000000 .long __gxx_personality_v0
27 0016 03 .byte 0x3
28 0017 0C .byte 0xc
29 0018 07 .uleb128 0x7
30 0019 08 .uleb128 0x8
31 001a 90 .byte 0x90
32 001b 01 .uleb128 0x1
33 001c 00000000 .align 8
34 .LECIE1:
35 .LSFDE1:
36 0020 14000000 .long .LEFDE1-.LASFDE1
37 .LASFDE1:
38 0024 24000000 .long .LASFDE1-.Lframe1
39 0028 00000000 .long .LFB3
40 002c 06000000 .long .LFE3-.LFB3
41 0030 00 .uleb128 0x0
42 0031 00000000 .align 8
42 000000
43 .LEFDE1:
44 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
45 .section .note.GNU-stack,"",@progbits
src2.s
GAS LISTING /tmp/ccNXXmLv.s page 1
1 .file "src2.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl _Z6test_2v
6 .type _Z6test_2v, @function
7 _Z6test_2v:
8 .LFB3:
9 0000 B80F0000 movl $15, %eax
9 00
10 0005 C3 ret
11 .LFE3:
12 .size _Z6test_2v, .-_Z6test_2v
13 .globl __gxx_personality_v0
14 .section .eh_frame,"a",@progbits
15 .Lframe1:
16 0000 1C000000 .long .LECIE1-.LSCIE1
17 .LSCIE1:
18 0004 00000000 .long 0x0
19 0008 01 .byte 0x1
20 0009 7A505200 .string "zPR"
21 000d 01 .uleb128 0x1
22 000e 78 .sleb128 -8
23 000f 10 .byte 0x10
24 0010 06 .uleb128 0x6
25 0011 03 .byte 0x3
26 0012 00000000 .long __gxx_personality_v0
27 0016 03 .byte 0x3
28 0017 0C .byte 0xc
29 0018 07 .uleb128 0x7
30 0019 08 .uleb128 0x8
31 001a 90 .byte 0x90
32 001b 01 .uleb128 0x1
33 001c 00000000 .align 8
34 .LECIE1:
35 .LSFDE1:
36 0020 14000000 .long .LEFDE1-.LASFDE1
37 .LASFDE1:
38 0024 24000000 .long .LASFDE1-.Lframe1
39 0028 00000000 .long .LFB3
40 002c 06000000 .long .LFE3-.LFB3
41 0030 00 .uleb128 0x0
42 0031 00000000 .align 8
42 000000
43 .LEFDE1:
44 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
45 .section .note.GNU-stack,"",@progbits
main.s
GAS LISTING /tmp/cc2cc5rp.s page 1
1 .file "main.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl main
6 .type main, @function
7 main:
8 .LFB3:
9 0000 4883EC08 subq $8, %rsp
10 .LCFI0:
11 0004 E8000000 call _Z6test_1v
11 00
12 0009 E8000000 call _Z6test_2v
12 00
13 000e E8000000 call _Z6test_1v
13 00
14 .p2align 4,,5
15 0013 E8000000 call _Z6test_2v
15 00
16 0018 31C0 xorl %eax, %eax
17 001a 4883C408 addq $8, %rsp
18 .p2align 4,,1
19 001e C3 ret
20 .LFE3:
21 .size main, .-main
22 .globl __gxx_personality_v0
23 .section .eh_frame,"a",@progbits
24 .Lframe1:
25 0000 1C000000 .long .LECIE1-.LSCIE1
26 .LSCIE1:
27 0004 00000000 .long 0x0
28 0008 01 .byte 0x1
29 0009 7A505200 .string "zPR"
30 000d 01 .uleb128 0x1
31 000e 78 .sleb128 -8
32 000f 10 .byte 0x10
33 0010 06 .uleb128 0x6
34 0011 03 .byte 0x3
35 0012 00000000 .long __gxx_personality_v0
36 0016 03 .byte 0x3
37 0017 0C .byte 0xc
38 0018 07 .uleb128 0x7
39 0019 08 .uleb128 0x8
40 001a 90 .byte 0x90
41 001b 01 .uleb128 0x1
42 001c 00000000 .align 8
43 .LECIE1:
44 .LSFDE1:
45 0020 14000000 .long .LEFDE1-.LASFDE1
46 .LASFDE1:
47 0024 24000000 .long .LASFDE1-.Lframe1
48 0028 00000000 .long .LFB3
49 002c 1F000000 .long .LFE3-.LFB3
50 0030 00 .uleb128 0x0
51 0031 44 .byte 0x4
52 .long .LCFI0-.LFB3
53 0032 0E .byte 0xe
GAS LISTING /tmp/cc2cc5rp.s page 2
54 0033 10 .uleb128 0x10
55 0034 00000000 .align 8
56 .LEFDE1:
57 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
58 .section .note.GNU-stack,"",@progbits
Example 3:
header.h
#ifndef HEADER_H_
#define HEADER_H_
inline int func(int a, int b)
{
return a + b;
}
inline int test_1()
{
int a, b, c;
a = 3;
b = 7;
c = func(a, b);
return c;
}
inline int test_2()
{
int a, b, c;
a = 7;
b = 8;
c = func(a, b);
return c;
}
#endif /* HEADER_H_ */
src.cc
#include "header.h"
/*
int func(int a, int b)
{
return a + b;
}*/
src1.cc
#include "header.h"
/*int test_1()
{
int a, b, c;
a = 3;
b = 7;
c = func(a, b);
return c;
}*/
src2.cc
#include "header.h"
/*int test_2()
{
int a, b, c;
a = 7;
b = 8;
c = func(a, b);
return c;
}*/
main.cc
int main(int argc, char** argv)
{
test_1();
test_2();
test_1();
test_2();
}
Assembly 3:
src.s
GAS LISTING /tmp/ccfPkzMC.s page 1
1 .file "src.cc"
2 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
3 .section .note.GNU-stack,"",@progbits
src1.s
GAS LISTING /tmp/cckRkoWG.s page 1
1 .file "src1.cc"
2 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
3 .section .note.GNU-stack,"",@progbits
src2.s
GAS LISTING /tmp/ccfmb3gI.s page 1
1 .file "src2.cc"
2 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
3 .section .note.GNU-stack,"",@progbits
main.s
GAS LISTING /tmp/ccGBsR8z.s page 1
1 .file "main.cc"
2 .text
3 .align 2
4 .p2align 4,,15
5 .globl main
6 .type main, @function
7 main:
8 .LFB5:
9 0000 31C0 xorl %eax, %eax
10 0002 C3 ret
11 .LFE5:
12 .size main, .-main
13 .globl __gxx_personality_v0
14 .section .eh_frame,"a",@progbits
15 .Lframe1:
16 0000 1C000000 .long .LECIE1-.LSCIE1
17 .LSCIE1:
18 0004 00000000 .long 0x0
19 0008 01 .byte 0x1
20 0009 7A505200 .string "zPR"
21 000d 01 .uleb128 0x1
22 000e 78 .sleb128 -8
23 000f 10 .byte 0x10
24 0010 06 .uleb128 0x6
25 0011 03 .byte 0x3
26 0012 00000000 .long __gxx_personality_v0
27 0016 03 .byte 0x3
28 0017 0C .byte 0xc
29 0018 07 .uleb128 0x7
30 0019 08 .uleb128 0x8
31 001a 90 .byte 0x90
32 001b 01 .uleb128 0x1
33 001c 00000000 .align 8
34 .LECIE1:
35 .LSFDE1:
36 0020 14000000 .long .LEFDE1-.LASFDE1
37 .LASFDE1:
38 0024 24000000 .long .LASFDE1-.Lframe1
39 0028 00000000 .long .LFB5
40 002c 03000000 .long .LFE5-.LFB5
41 0030 00 .uleb128 0x0
42 0031 00000000 .align 8
42 000000
43 .LEFDE1:
44 .ident "GCC: (GNU) 4.1.2 20080704 (Red Hat 4.1.2-54)"
45 .section .note.GNU-stack,"",@progbits
Example 1 and example 3 are the ones I'm particularly interested in, because they should highlight somehow what is the difference between an inline function and a not inline function (following the points 1,2 and 3 of the link I posted above), I don't see any lack of properties in the not inline functions compared to the inline version. Can someone highlight the difference for me (again in terms of the points 1,2 and 3)?
Let's put the issue of whether
inline
is forced or not, aside for now (there are lot's of discussions on the topic).Inlining a function is equivalent of pasting the contents of the function at the location of the function call (invocation).
So given the following:
When the
Hello
function is inlined, you will get the equivalent of:The compiler is allowed to inline functions that are not marked as being inlined. This feature is often triggered by an optimization setting.
Edit 1: Common reason for inlining
A common reason for inlining a function is when the content is smaller or equal than the overhead to call the function.
There is a protocol associated with call a function, such as moving parameters to the stack or registers. The protocol exists regardless of the size of the function. So, inlining will remove the calling protocol (thus reducing program code size and increasing performance).
Another reason to inline is to reduce the quantity of function calls. In some processors, a branch instruction (function call), causes the instruction cache (or pipeline) to be reloaded. This takes time. Inlining reduces the function calls and improves execution time.
Edit 2: Code Bloat
One reason to create functions is to reduce code size. Inlining of large functions may result in code bloat or the increasing of the program's size.
Code bloat and function inlining are under the Time versus Space trade off. Inlining of large functions may speed up execution, but you are trading space for it. Place common code into functions may decrease the size of your program, but take more time to execute.