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
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
|
/* co-gcc.lnt: This is the seed file for configuring Lint for use with
GCC versions 2.95.3 and later.
Like all compiler options files this file is intended to be used
as follows:
lint co-gcc.lnt source-files-to-be-linted
Some of the information that co-gcc.lnt requires needs to be
furnished with the help of the gcc system itself.
If any of these directions are unclear, you may gain a better insight
into what is happening by checking the file gcc-readme.txt
For C, first create an empty file named empty.c and then
run the command (options are case sensitive):
gcc -E -dM empty.c >lint_cmac.h
This will capture macro definitions in a file that will be included
automatically at the beginning of each module by use of the -header
option within co-gcc.lnt. The macros may change as you change
compiler options so that ultimately you may want to incorporate
this step into a make facility.
Next we need to determine the search directories. If you run
gcc -c -v empty.c
you will see among other things this coveted list. For example you
might get:
...
#include "..." search starts here
#include <...> search starts here
/usr/local/include
/usr/lib/gcc/i686-apple-darwin8/4.0.1/include
/usr/include
/System/Library/Frameworks
/Library/Frameworks
End of search list.
...
For each directory shown (there are five in the list above) prefix
the directory name by a "--i" and place it in a file whose name is,
say, include.lnt. You may then begin linting programs by using the
command
lint co-gcc.lnt include.lnt source-files
Note: it is conventional to place both .lnt files into a single .lnt
file called std.lnt
For C++, run the command (options are again case sensitive):
g++ -E -dM empty.c >lint_cppmac.h
This will capture C++ macro definitions in a file that will be
included automatically at the beginning of each C++ module
at the request of co-gcc.lnt.
Next we need to determine C++ search directories. If you run
g++ -c -v empty.c
As in the case of C you should prepend a --i onto each directory
displayed and place these options into a file such as include.lnt.
Again, there is nothing sacred about the name and if you intend to
do mixed C and C++ programming it will be necessary for you to use
a differently named file. The rest proceeds as before.
Note, some options in this file (such as the size options, i.e. -sp4
indicating that pointers are four bytes wide) may need to be changed.
See "System Dependent Options" below.
*/
-cgnu // Notifies FlexeLint that gcc is being used.
// Begin: System Dependent Options
// -------------------------------
-a#machine(i386) // #assert's machine(i386) (SVR4 facility).
+fdi // Use the directory of the including file
-si4 // size of int
-sp4 // size of pointer
// -----------------------------
// End: System Dependent Options
+cpp(.cc,.c) // extensions for C++ that are commonly used in addition
// to the default extensions of .cpp and .cxx
-header(scripts/pclint/co-gcc.h) // Includes header generated by GCC.
+libh(co-gcc.h) // Marks that header as library code.
// =========================================================
// +rw and -d options to cope with GNU syntax:
+ppw(ident) // Tolerate #ident keyword definitions for SCCS/RCS
+ppw(warning)
// GCC provides alternative spellings of certain keywords:
-rw_asgn(__inline,inline)
-rw_asgn(__inline__,inline)
-rw_asgn(__signed__,signed)
-rw_asgn( __volatile__, volatile )
-rw_asgn( __volatile, volatile )
++d__const=const // gconv.h uses __const rather than const
++dconst=const // ensure const expands to const.
-rw_asgn( asm, _up_to_brackets )
-rw_asgn( __asm, _up_to_brackets )
-rw_asgn( __asm__, _up_to_brackets )
// This re-definition of the various spellings of the asm keyword enables
// Lint to pass gracefully over expression-statements like:
// __asm __volatile ("fsqrt" : "=t" (__result) : "0" (__x));
++d__attribute__()= // ignore this keyword and following parenthetical
++d__attribute()= // variant spelling of "__attribute__"
// "__extension__" is GCC's way of allowing the use of non-standard
// constructs in a strict Standard-conforming mode. We don't currently
// have explicit support for it, but we can use local suppressions. For
// example, we can use -e(160) so that we will not see any Errors about
// GNU statement-expressions wrapped in __extension__().
++d"__extension__=/*lint -e(160) */"
++d__builtin_va_list=void* // used by stdarg.h
++d__builtin_stdarg_start()=_to_semi // ditto
++d__builtin_va_end()=_to_semi // ditto
++d"__builtin_va_arg(a,b)=(*( (b *) ( ((a) += sizeof(b)) - sizeof(b) )))"
++d__null=0
+rw(_to_semi) // needed for the two macros above.
+rw(__typeof__) // activate __typeof__ keyword
-d__typeof=__typeof__ // an alternative to using __typeof__
+rw( __restrict )
+rw( __restrict__ )
-rw(__except) // This MS reserved word is used as an identifier
+rw( __complex__, __real__, __imag__ ) // reserved words that can be ignored.
++d__builtin_strchr=(char*) // permits the inline definition ...
++d__builtin_strpbrk=(char*) // of these functions to be linted ...
++d__builtin_strrchr=(char*) // without drawing a complaint
++d__builtin_strstr=(char*) // about the use of a non-standard name
++d__PRETTY_FUNCTION__=___function___ // lint defines ___function___ internally
++d__FUNCTION__=___function___ // lint defines ___function___ internally
// =========================================================
// Other options supporting GNU C/C++ syntax:
+fld // enables the processing of _L_abel _D_esignators E.g.:
// union { double d; int i; } u = { d: 3.141 };
// +fwc // wchar_t might be builtin; if so, uncomment
// this option.
// =========================================================
// Generally useful suppressions:
-wlib(1) // sets the warning level within library headers to 1
// (no warnings, just syntax errors). Comment out if you
// are actually linting library headers.
-elib(123) // 123 is really a warning, but it's in the "Error" range.
-elib(93) // allow newlines within quoted string arguments to macros
-elibsym(628) // Suppress 628 for __builtin symbols.
-esym(528,__huge_val,__nan,__qnan,__qnanf,__snan,__snanf)
// We don't care if we don't reference some GNU functions
-esym(528,__gnu_malloc,__gnu_calloc)
// The following functions exhibit variable return modes.
// That is, they may equally-usefully be called for a value
// as called just for their effects. Accordingly we inhibit
// Warning 534 for these functions.
// Feel free to add to or subtract from this list.
-esym(534,close,creat,fclose,fprintf,fputc, nanosleep, time)
-esym(534,fputs,fscanf,fseek,fwrite,lseek,memcpy,memmove,memset)
-esym(534,printf,puts,scanf,sprintf,sscanf,strcat,strcpy)
-esym(534,strncat,strncpy,unlink,write, snprintf, dprintf)
// For non-ANSI compilers we suppress messages 515 and 516
// for functions known to have variable argument lists.
// For ANSI compilers, header files should take care of this.
-esym(515,fprintf,printf,sprintf,fscanf,scanf,sscanf)
-esym(516,fprintf,printf,sprintf,fscanf,scanf,sscanf)
-esym(1702,*operator<<,*operator>>)
-esym(534,*operator<<,*operator>>)
-esym(1055,*__builtin*)
-esym(718,*__builtin*) // The compiler does not need these ...
-esym(746,*__builtin*) // declared and it knows their prototypes.
|
