forked from hardaker/net-snmp
-
Notifications
You must be signed in to change notification settings - Fork 2
/
PORTING
104 lines (79 loc) · 3.48 KB
/
PORTING
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
--- INTRODUCTION
Just a quick note on porting and sending me patches:
First off, you probably should subscribe to
[email protected] by sending a message to
[email protected] with a subject line of
subscribe. This is a mailing list to discuss all oft the coding
aspects of the project.
Additionally, you should probably be developing against the latest
snapshot of the source code, which can be obtained through the
net-snmp cvs server. Details can be found at
http://www.net-snmp.org/cvs/.
If you send patches to us, it would greatly help us if you sent them
to us based on the current checked out copy from CVS. To do this,
send us the output of "cvs diff -u" run in the top level net-snmp
source tree after you have modified the files that will fix the
problem or add the feature you're submitting the patch for.
Quite a while back I started using the GNU autoconf testing suite to
greatly enhance portability. Because of this porting to new
architectures is much easier than before. However, new people porting
the package to new architectures rarely take advantage of this setup
and send me patches with lots of '#ifdef ARCH' type C code in it. Let
me say up front, I *hate* this type of coding now (even though I used
to use it a lot). What is better is to check for the necessary
functionality using the configure script and then use the results of
those tests.
To do this, you need to install the GNU 'autoconf' package which also
requires the GNU 'm4' (gm4) package as well. This double installation
is extremely easy and shouldn't take you more than 15 minutes max.
After that, modify the configure.in and acconfig.h files as needed
instead of modifying the config.h or configure files directly. The
Makefile will re-produce these files from the first two.
Worst case: Don't put in #ifdef architecture style statements.
Rather, create a new define in the s/ and m/ system specific header
files and use those defines to test against in the C code. This
should only be done for things that can't be checked using configure
though.
Some autoconf examples:
--- HEADER FILES
In configure.in:
AC_CHECK_HEADERS(headdir/header.h)
Then in your source code:
#ifdef HAVE_HEADDIR_HEADER_H
#include <headdir/header.h>
#ENDIF
--- LIBRARY ROUTIENS
In configure.in:
AC_CHECK_LIB(libexample, example_function)
Thats it. The Makefiles will automatically link against -llibexample
if example_function is found in the library.
--- FUNCTION CHECKS
In configure.in:
AC_CHECK_FUNCS(example_function)
In source code:
#ifdef HAVE_EXAMPLE_FUNCTION
/* use it */
#endif
--- STRUCTURE MEMBER CHECKS
In configure.in:
AC_CHECK_MEMBERS([struct STRUCTURE.MEMBER],,,[[
#include lines
]])
^^^^^^^^^ ^^^^^^ (change)
In source code:
#ifdef HAVE_STRUCT_STRUCTURE_MEMBER
/* use it */
#endif
--- READ THE MANUAL
The GNU autoconf info files are extremely well written and easy to
follow. Please check them out.
I'd be happy to help you through anything you don't understand or
through more complex examples (eg, checking for structure parts or
existance). I'd be far less happy to get patches ignoring the above
request. If you simple can't abide by this, please send the patches
anyway, but it'll just take me longer to get them applied.
Submit the patch to http://www.net-snmp.org/patches/.
Please include what version of the net-snmp package it was applied to
and state the arcitectures you have tested it on.
Thanks a lot for the consideration,
Wes