1 |
20 |
jlechner |
\input texinfo
|
2 |
|
|
|
3 |
|
|
@c ---------------------------------------------------------------------
|
4 |
|
|
@c Prologue
|
5 |
|
|
@c ---------------------------------------------------------------------
|
6 |
|
|
|
7 |
|
|
@setfilename porting.info
|
8 |
|
|
@settitle Porting libstdc++-v3
|
9 |
|
|
@setchapternewpage odd
|
10 |
|
|
|
11 |
|
|
@copying
|
12 |
|
|
Copyright @copyright{} 2000, 2001, 2002, 2003, 2005
|
13 |
|
|
Free Software Foundation, Inc.
|
14 |
|
|
|
15 |
|
|
Permission is granted to copy, distribute and/or modify this document
|
16 |
|
|
under the terms of the GNU Free Documentation License, Version 1.2 or
|
17 |
|
|
any later version published by the Free Software Foundation; with the
|
18 |
|
|
Invariant Sections being ``GNU General Public License'', the Front-Cover
|
19 |
|
|
texts being (a) (see below), and with the Back-Cover Texts being (b)
|
20 |
|
|
(see below). A copy of the license is included in the section entitled
|
21 |
|
|
``GNU Free Documentation License''.
|
22 |
|
|
|
23 |
|
|
(a) The FSF's Front-Cover Text is:
|
24 |
|
|
|
25 |
|
|
A GNU Manual
|
26 |
|
|
|
27 |
|
|
(b) The FSF's Back-Cover Text is:
|
28 |
|
|
|
29 |
|
|
You have freedom to copy and modify this GNU Manual, like GNU
|
30 |
|
|
software. Copies published by the Free Software Foundation raise
|
31 |
|
|
funds for GNU development.
|
32 |
|
|
@end copying
|
33 |
|
|
|
34 |
|
|
@ifinfo
|
35 |
|
|
This file explains how to port libstdc++-v3 (the GNU C++ library) to
|
36 |
|
|
a new target.
|
37 |
|
|
|
38 |
|
|
@insertcopying
|
39 |
|
|
@end ifinfo
|
40 |
|
|
|
41 |
|
|
@c ---------------------------------------------------------------------
|
42 |
|
|
@c Titlepage
|
43 |
|
|
@c ---------------------------------------------------------------------
|
44 |
|
|
|
45 |
|
|
@titlepage
|
46 |
|
|
@title Porting libstdc++-v3
|
47 |
|
|
@author Mark Mitchell
|
48 |
|
|
@page
|
49 |
|
|
@vskip 0pt plus 1filll
|
50 |
|
|
@insertcopying
|
51 |
|
|
@end titlepage
|
52 |
|
|
|
53 |
|
|
@c ---------------------------------------------------------------------
|
54 |
|
|
@c Top
|
55 |
|
|
@c ---------------------------------------------------------------------
|
56 |
|
|
|
57 |
|
|
@node Top
|
58 |
|
|
@top Porting libstdc++-v3
|
59 |
|
|
|
60 |
|
|
This document explains how to port libstdc++-v3 (the GNU C++ library) to
|
61 |
|
|
a new target.
|
62 |
|
|
|
63 |
|
|
In order to make the GNU C++ library (libstdc++-v3) work with a new
|
64 |
|
|
target, you must edit some configuration files and provide some new
|
65 |
|
|
header files. Unless this is done, libstdc++-v3 will use generic
|
66 |
|
|
settings which may not be correct for your target; even if they are
|
67 |
|
|
correct, they will likely be inefficient.
|
68 |
|
|
|
69 |
|
|
Before you get started, make sure that you have a working C library on
|
70 |
|
|
your target. The C library need not precisely comply with any
|
71 |
|
|
particular standard, but should generally conform to the requirements
|
72 |
|
|
imposed by the ANSI/ISO standard.
|
73 |
|
|
|
74 |
|
|
In addition, you should try to verify that the C++ compiler generally
|
75 |
|
|
works. It is difficult to test the C++ compiler without a working
|
76 |
|
|
library, but you should at least try some minimal test cases.
|
77 |
|
|
|
78 |
|
|
(Note that what we think of as a ``target,'' the library refers to as
|
79 |
|
|
a ``host.'' The comment at the top of @file{configure.ac} explains why.)
|
80 |
|
|
|
81 |
|
|
Here are the primary steps required to port the library:
|
82 |
|
|
|
83 |
|
|
@menu
|
84 |
|
|
* Operating system:: Configuring for your operating system.
|
85 |
|
|
* CPU:: Configuring for your processor chip.
|
86 |
|
|
* Character types:: Implementing character classification.
|
87 |
|
|
* Thread safety:: Implementing atomic operations.
|
88 |
|
|
* Numeric limits:: Implementing numeric limits.
|
89 |
|
|
* Libtool:: Using libtool.
|
90 |
|
|
* GNU Free Documentation License:: How you can copy and share this manual.
|
91 |
|
|
@end menu
|
92 |
|
|
|
93 |
|
|
@c ---------------------------------------------------------------------
|
94 |
|
|
@c Operating system
|
95 |
|
|
@c ---------------------------------------------------------------------
|
96 |
|
|
|
97 |
|
|
@node Operating system
|
98 |
|
|
@chapter Operating system
|
99 |
|
|
|
100 |
|
|
If you are porting to a new operating system (as opposed to a new chip
|
101 |
|
|
using an existing operating system), you will need to create a new
|
102 |
|
|
directory in the @file{config/os} hierarchy. For example, the IRIX
|
103 |
|
|
configuration files are all in @file{config/os/irix}. There is no set
|
104 |
|
|
way to organize the OS configuration directory. For example,
|
105 |
|
|
@file{config/os/solaris/solaris-2.6} and
|
106 |
|
|
@file{config/os/solaris/solaris-2.7} are used as configuration
|
107 |
|
|
directories for these two versions of Solaris. On the other hand, both
|
108 |
|
|
Solaris 2.7 and Solaris 2.8 use the @file{config/os/solaris/solaris-2.7}
|
109 |
|
|
directory. The important information is that there needs to be a
|
110 |
|
|
directory under @file{config/os} to store the files for your operating
|
111 |
|
|
system.
|
112 |
|
|
|
113 |
|
|
You might have to change the @file{configure.host} file to ensure that
|
114 |
|
|
your new directory is activated. Look for the switch statement that sets
|
115 |
|
|
@code{os_include_dir}, and add a pattern to handle your operating system
|
116 |
|
|
if the default will not suffice. The switch statement switches on only
|
117 |
|
|
the OS portion of the standard target triplet; e.g., the @code{solaris2.8}
|
118 |
|
|
in @code{sparc-sun-solaris2.8}. If the new directory is named after the
|
119 |
|
|
OS portion of the triplet (the default), then nothing needs to be changed.
|
120 |
|
|
|
121 |
|
|
The first file to create in this directory, should be called
|
122 |
|
|
@file{os_defines.h}. This file contains basic macro definitions
|
123 |
|
|
that are required to allow the C++ library to work with your C library.
|
124 |
|
|
|
125 |
|
|
Several libstdc++-v3 source files unconditionally define the macro
|
126 |
|
|
@code{_POSIX_SOURCE}. On many systems, defining this macro causes
|
127 |
|
|
large portions of the C library header files to be eliminated
|
128 |
|
|
at preprocessing time. Therefore, you may have to @code{#undef} this
|
129 |
|
|
macro, or define other macros (like @code{_LARGEFILE_SOURCE} or
|
130 |
|
|
@code{__EXTENSIONS__}). You won't know what macros to define or
|
131 |
|
|
undefine at this point; you'll have to try compiling the library and
|
132 |
|
|
seeing what goes wrong. If you see errors about calling functions
|
133 |
|
|
that have not been declared, look in your C library headers to see if
|
134 |
|
|
the functions are declared there, and then figure out what macros you
|
135 |
|
|
need to define. You will need to add them to the
|
136 |
|
|
@code{CPLUSPLUS_CPP_SPEC} macro in the GCC configuration file for your
|
137 |
|
|
target. It will not work to simply define these macros in
|
138 |
|
|
@file{os_defines.h}.
|
139 |
|
|
|
140 |
|
|
At this time, there are a few libstdc++-v3-specific macros which may be
|
141 |
|
|
defined:
|
142 |
|
|
|
143 |
|
|
@code{_GLIBCXX_USE_C99_CHECK} may be defined to 1 to check C99
|
144 |
|
|
function declarations (which are not covered by specialization below)
|
145 |
|
|
found in system headers against versions found in the library headers
|
146 |
|
|
derived from the standard.
|
147 |
|
|
|
148 |
|
|
@code{_GLIBCXX_USE_C99_DYNAMIC} may be defined to an expression that
|
149 |
|
|
yields 0 if and only if the system headers are exposing proper support
|
150 |
|
|
for C99 functions (which are not covered by specialization below). If
|
151 |
|
|
defined, it must be 0 while bootstrapping the compiler/rebuilding the
|
152 |
|
|
library.
|
153 |
|
|
|
154 |
|
|
@code{_GLIBCXX_USE_C99_LONG_LONG_CHECK} may be defined to 1 to check
|
155 |
|
|
the set of C99 long long function declarations found in system headers
|
156 |
|
|
against versions found in the library headers derived from the
|
157 |
|
|
standard.
|
158 |
|
|
|
159 |
|
|
@code{_GLIBCXX_USE_C99_LONG_LONG_DYNAMIC} may be defined to an
|
160 |
|
|
expression that yields 0 if and only if the system headers are
|
161 |
|
|
exposing proper support for the set of C99 long long functions. If
|
162 |
|
|
defined, it must be 0 while bootstrapping the compiler/rebuilding the
|
163 |
|
|
library.
|
164 |
|
|
|
165 |
|
|
@code{_GLIBCXX_USE_C99_FP_MACROS_DYNAMIC} may be defined to an
|
166 |
|
|
expression that yields 0 if and only if the system headers
|
167 |
|
|
are exposing proper support for the related set of macros. If defined,
|
168 |
|
|
it must be 0 while bootstrapping the compiler/rebuilding the library.
|
169 |
|
|
|
170 |
|
|
@code{_GLIBCXX_USE_C99_FLOAT_TRANSCENDENTALS_CHECK} may be defined
|
171 |
|
|
to 1 to check the related set of function declarations found in system
|
172 |
|
|
headers against versions found in the library headers derived from
|
173 |
|
|
the standard.
|
174 |
|
|
|
175 |
|
|
@code{_GLIBCXX_USE_C99_FLOAT_TRANSCENDENTALS_DYNAMIC} may be defined
|
176 |
|
|
to an expression that yields 0 if and only if the system headers
|
177 |
|
|
are exposing proper support for the related set of functions. If defined,
|
178 |
|
|
it must be 0 while bootstrapping the compiler/rebuilding the library.
|
179 |
|
|
|
180 |
|
|
Finally, you should bracket the entire file in an include-guard, like
|
181 |
|
|
this:
|
182 |
|
|
|
183 |
|
|
@example
|
184 |
|
|
#ifndef _GLIBCXX_OS_DEFINES
|
185 |
|
|
#define _GLIBCXX_OS_DEFINES
|
186 |
|
|
...
|
187 |
|
|
#endif
|
188 |
|
|
@end example
|
189 |
|
|
|
190 |
|
|
We recommend copying an existing @file{os_defines.h} to use as a
|
191 |
|
|
starting point.
|
192 |
|
|
|
193 |
|
|
@c ---------------------------------------------------------------------
|
194 |
|
|
@c CPU
|
195 |
|
|
@c ---------------------------------------------------------------------
|
196 |
|
|
|
197 |
|
|
@node CPU
|
198 |
|
|
@chapter CPU
|
199 |
|
|
|
200 |
|
|
If you are porting to a new chip (as opposed to a new operating system
|
201 |
|
|
running on an existing chip), you will need to create a new directory in the
|
202 |
|
|
@file{config/cpu} hierarchy. Much like the @ref{Operating system} setup,
|
203 |
|
|
there are no strict rules on how to organize the CPU configuration
|
204 |
|
|
directory, but careful naming choices will allow the configury to find your
|
205 |
|
|
setup files without explicit help.
|
206 |
|
|
|
207 |
|
|
We recommend that for a target triplet @code{<CPU>-<vendor>-<OS>}, you
|
208 |
|
|
name your configuration directory @file{config/cpu/<CPU>}. If you do this,
|
209 |
|
|
the configury will find the directory by itself. Otherwise you will need to
|
210 |
|
|
edit the @file{configure.host} file and, in the switch statement that sets
|
211 |
|
|
@code{cpu_include_dir}, add a pattern to handle your chip.
|
212 |
|
|
|
213 |
|
|
Note that some chip families share a single configuration directory, for
|
214 |
|
|
example, @code{alpha}, @code{alphaev5}, and @code{alphaev6} all use the
|
215 |
|
|
@file{config/cpu/alpha} directory, and there is an entry in the
|
216 |
|
|
@file{configure.host} switch statement to handle this.
|
217 |
|
|
|
218 |
|
|
The @code{cpu_include_dir} sets default locations for the files controlling
|
219 |
|
|
@ref{Thread safety} and @ref{Numeric limits}, if the defaults are not
|
220 |
|
|
appropriate for your chip.
|
221 |
|
|
|
222 |
|
|
|
223 |
|
|
@c ---------------------------------------------------------------------
|
224 |
|
|
@c Character types
|
225 |
|
|
@c ---------------------------------------------------------------------
|
226 |
|
|
|
227 |
|
|
@node Character types
|
228 |
|
|
@chapter Character types
|
229 |
|
|
|
230 |
|
|
The library requires that you provide three header files to implement
|
231 |
|
|
character classification, analogous to that provided by the C libraries
|
232 |
|
|
@file{<ctype.h>} header. You can model these on the files provided in
|
233 |
|
|
@file{config/os/generic}. However, these files will almost
|
234 |
|
|
certainly need some modification.
|
235 |
|
|
|
236 |
|
|
The first file to write is @file{ctype_base.h}. This file provides
|
237 |
|
|
some very basic information about character classification. The libstdc++-v3
|
238 |
|
|
library assumes that your C library implements @file{<ctype.h>} by using
|
239 |
|
|
a table (indexed by character code) containing integers, where each of
|
240 |
|
|
these integers is a bit-mask indicating whether the character is
|
241 |
|
|
upper-case, lower-case, alphabetic, etc. The @file{ctype_base.h}
|
242 |
|
|
file gives the type of the integer, and the values of the various bit
|
243 |
|
|
masks. You will have to peer at your own @file{<ctype.h>} to figure out
|
244 |
|
|
how to define the values required by this file.
|
245 |
|
|
|
246 |
|
|
The @file{ctype_base.h} header file does not need include guards.
|
247 |
|
|
It should contain a single @code{struct} definition called
|
248 |
|
|
@code{ctype_base}. This @code{struct} should contain two type
|
249 |
|
|
declarations, and one enumeration declaration, like this example, taken
|
250 |
|
|
from the IRIX configuration:
|
251 |
|
|
|
252 |
|
|
@example
|
253 |
|
|
struct ctype_base
|
254 |
|
|
@{
|
255 |
|
|
typedef unsigned int mask;
|
256 |
|
|
typedef int* __to_type;
|
257 |
|
|
|
258 |
|
|
enum
|
259 |
|
|
@{
|
260 |
|
|
space = _ISspace,
|
261 |
|
|
print = _ISprint,
|
262 |
|
|
cntrl = _IScntrl,
|
263 |
|
|
upper = _ISupper,
|
264 |
|
|
lower = _ISlower,
|
265 |
|
|
alpha = _ISalpha,
|
266 |
|
|
digit = _ISdigit,
|
267 |
|
|
punct = _ISpunct,
|
268 |
|
|
xdigit = _ISxdigit,
|
269 |
|
|
alnum = _ISalnum,
|
270 |
|
|
graph = _ISgraph
|
271 |
|
|
@};
|
272 |
|
|
@};
|
273 |
|
|
@end example
|
274 |
|
|
|
275 |
|
|
@noindent
|
276 |
|
|
The @code{mask} type is the type of the elements in the table. If your
|
277 |
|
|
C library uses a table to map lower-case numbers to upper-case numbers,
|
278 |
|
|
and vice versa, you should define @code{__to_type} to be the type of the
|
279 |
|
|
elements in that table. If you don't mind taking a minor performance
|
280 |
|
|
penalty, or if your library doesn't implement @code{toupper} and
|
281 |
|
|
@code{tolower} in this way, you can pick any pointer-to-integer type,
|
282 |
|
|
but you must still define the type.
|
283 |
|
|
|
284 |
|
|
The enumeration should give definitions for all the values in the above
|
285 |
|
|
example, using the values from your native @file{<ctype.h>}. They can
|
286 |
|
|
be given symbolically (as above), or numerically, if you prefer. You do
|
287 |
|
|
not have to include @file{<ctype.h>} in this header; it will always be
|
288 |
|
|
included before @file{ctype_base.h} is included.
|
289 |
|
|
|
290 |
|
|
The next file to write is @file{ctype_noninline.h}, which also does
|
291 |
|
|
not require include guards. This file defines a few member functions
|
292 |
|
|
that will be included in @file{include/bits/locale_facets.h}. The first
|
293 |
|
|
function that must be written is the @code{ctype<char>::ctype}
|
294 |
|
|
constructor. Here is the IRIX example:
|
295 |
|
|
|
296 |
|
|
@example
|
297 |
|
|
ctype<char>::ctype(const mask* __table = 0, bool __del = false,
|
298 |
|
|
size_t __refs = 0)
|
299 |
|
|
: _Ctype_nois<char>(__refs), _M_del(__table != 0 && __del),
|
300 |
|
|
_M_toupper(NULL),
|
301 |
|
|
_M_tolower(NULL),
|
302 |
|
|
_M_ctable(NULL),
|
303 |
|
|
_M_table(!__table
|
304 |
|
|
? (const mask*) (__libc_attr._ctype_tbl->_class + 1)
|
305 |
|
|
: __table)
|
306 |
|
|
@{ @}
|
307 |
|
|
@end example
|
308 |
|
|
|
309 |
|
|
@noindent
|
310 |
|
|
There are two parts of this that you might choose to alter. The first,
|
311 |
|
|
and most important, is the line involving @code{__libc_attr}. That is
|
312 |
|
|
IRIX system-dependent code that gets the base of the table mapping
|
313 |
|
|
character codes to attributes. You need to substitute code that obtains
|
314 |
|
|
the address of this table on your system. If you want to use your
|
315 |
|
|
operating system's tables to map upper-case letters to lower-case, and
|
316 |
|
|
vice versa, you should initialize @code{_M_toupper} and
|
317 |
|
|
@code{_M_tolower} with those tables, in similar fashion.
|
318 |
|
|
|
319 |
|
|
Now, you have to write two functions to convert from upper-case to
|
320 |
|
|
lower-case, and vice versa. Here are the IRIX versions:
|
321 |
|
|
|
322 |
|
|
@example
|
323 |
|
|
char
|
324 |
|
|
ctype<char>::do_toupper(char __c) const
|
325 |
|
|
@{ return _toupper(__c); @}
|
326 |
|
|
|
327 |
|
|
char
|
328 |
|
|
ctype<char>::do_tolower(char __c) const
|
329 |
|
|
@{ return _tolower(__c); @}
|
330 |
|
|
@end example
|
331 |
|
|
|
332 |
|
|
@noindent
|
333 |
|
|
Your C library provides equivalents to IRIX's @code{_toupper} and
|
334 |
|
|
@code{_tolower}. If you initialized @code{_M_toupper} and
|
335 |
|
|
@code{_M_tolower} above, then you could use those tables instead.
|
336 |
|
|
|
337 |
|
|
Finally, you have to provide two utility functions that convert strings
|
338 |
|
|
of characters. The versions provided here will always work -- but you
|
339 |
|
|
could use specialized routines for greater performance if you have
|
340 |
|
|
machinery to do that on your system:
|
341 |
|
|
|
342 |
|
|
@example
|
343 |
|
|
const char*
|
344 |
|
|
ctype<char>::do_toupper(char* __low, const char* __high) const
|
345 |
|
|
@{
|
346 |
|
|
while (__low < __high)
|
347 |
|
|
@{
|
348 |
|
|
*__low = do_toupper(*__low);
|
349 |
|
|
++__low;
|
350 |
|
|
@}
|
351 |
|
|
return __high;
|
352 |
|
|
@}
|
353 |
|
|
|
354 |
|
|
const char*
|
355 |
|
|
ctype<char>::do_tolower(char* __low, const char* __high) const
|
356 |
|
|
@{
|
357 |
|
|
while (__low < __high)
|
358 |
|
|
@{
|
359 |
|
|
*__low = do_tolower(*__low);
|
360 |
|
|
++__low;
|
361 |
|
|
@}
|
362 |
|
|
return __high;
|
363 |
|
|
@}
|
364 |
|
|
@end example
|
365 |
|
|
|
366 |
|
|
You must also provide the @file{ctype_inline.h} file, which
|
367 |
|
|
contains a few more functions. On most systems, you can just copy
|
368 |
|
|
@file{config/os/generic/ctype_inline.h} and use it on your system.
|
369 |
|
|
|
370 |
|
|
In detail, the functions provided test characters for particular
|
371 |
|
|
properties; they are analogous to the functions like @code{isalpha} and
|
372 |
|
|
@code{islower} provided by the C library.
|
373 |
|
|
|
374 |
|
|
The first function is implemented like this on IRIX:
|
375 |
|
|
|
376 |
|
|
@example
|
377 |
|
|
bool
|
378 |
|
|
ctype<char>::
|
379 |
|
|
is(mask __m, char __c) const throw()
|
380 |
|
|
@{ return (_M_table)[(unsigned char)(__c)] & __m; @}
|
381 |
|
|
@end example
|
382 |
|
|
|
383 |
|
|
@noindent
|
384 |
|
|
The @code{_M_table} is the table passed in above, in the constructor.
|
385 |
|
|
This is the table that contains the bitmasks for each character. The
|
386 |
|
|
implementation here should work on all systems.
|
387 |
|
|
|
388 |
|
|
The next function is:
|
389 |
|
|
|
390 |
|
|
@example
|
391 |
|
|
const char*
|
392 |
|
|
ctype<char>::
|
393 |
|
|
is(const char* __low, const char* __high, mask* __vec) const throw()
|
394 |
|
|
@{
|
395 |
|
|
while (__low < __high)
|
396 |
|
|
*__vec++ = (_M_table)[(unsigned char)(*__low++)];
|
397 |
|
|
return __high;
|
398 |
|
|
@}
|
399 |
|
|
@end example
|
400 |
|
|
|
401 |
|
|
@noindent
|
402 |
|
|
This function is similar; it copies the masks for all the characters
|
403 |
|
|
from @code{__low} up until @code{__high} into the vector given by
|
404 |
|
|
@code{__vec}.
|
405 |
|
|
|
406 |
|
|
The last two functions again are entirely generic:
|
407 |
|
|
|
408 |
|
|
@example
|
409 |
|
|
const char*
|
410 |
|
|
ctype<char>::
|
411 |
|
|
scan_is(mask __m, const char* __low, const char* __high) const throw()
|
412 |
|
|
@{
|
413 |
|
|
while (__low < __high && !this->is(__m, *__low))
|
414 |
|
|
++__low;
|
415 |
|
|
return __low;
|
416 |
|
|
@}
|
417 |
|
|
|
418 |
|
|
const char*
|
419 |
|
|
ctype<char>::
|
420 |
|
|
scan_not(mask __m, const char* __low, const char* __high) const throw()
|
421 |
|
|
@{
|
422 |
|
|
while (__low < __high && this->is(__m, *__low))
|
423 |
|
|
++__low;
|
424 |
|
|
return __low;
|
425 |
|
|
@}
|
426 |
|
|
@end example
|
427 |
|
|
|
428 |
|
|
@c ---------------------------------------------------------------------
|
429 |
|
|
@c Thread safety
|
430 |
|
|
@c ---------------------------------------------------------------------
|
431 |
|
|
|
432 |
|
|
@node Thread safety
|
433 |
|
|
@chapter Thread safety
|
434 |
|
|
|
435 |
|
|
The C++ library string functionality requires a couple of atomic
|
436 |
|
|
operations to provide thread-safety. If you don't take any special
|
437 |
|
|
action, the library will use stub versions of these functions that are
|
438 |
|
|
not thread-safe. They will work fine, unless your applications are
|
439 |
|
|
multi-threaded.
|
440 |
|
|
|
441 |
|
|
If you want to provide custom, safe, versions of these functions, there
|
442 |
|
|
are two distinct approaches. One is to provide a version for your CPU,
|
443 |
|
|
using assembly language constructs. The other is to use the
|
444 |
|
|
thread-safety primitives in your operating system. In either case, you
|
445 |
|
|
make a file called @file{atomicity.h}, and the variable
|
446 |
|
|
@code{ATOMICITYH} must point to this file.
|
447 |
|
|
|
448 |
|
|
If you are using the assembly-language approach, put this code in
|
449 |
|
|
@file{config/cpu/<chip>/atomicity.h}, where chip is the name of
|
450 |
|
|
your processor (@pxref{CPU}). No additional changes are necessary to
|
451 |
|
|
locate the file in this case; @code{ATOMICITYH} will be set by default.
|
452 |
|
|
|
453 |
|
|
If you are using the operating system thread-safety primitives approach,
|
454 |
|
|
you can also put this code in the same CPU directory, in which case no more
|
455 |
|
|
work is needed to locate the file. For examples of this approach,
|
456 |
|
|
see the @file{atomicity.h} file for IRIX or IA64.
|
457 |
|
|
|
458 |
|
|
Alternatively, if the primitives are more closely related to the OS
|
459 |
|
|
than they are to the CPU, you can put the @file{atomicity.h} file in
|
460 |
|
|
the @ref{Operating system} directory instead. In this case, you must
|
461 |
|
|
edit @file{configure.host}, and in the switch statement that handles
|
462 |
|
|
operating systems, override the @code{ATOMICITYH} variable to point to
|
463 |
|
|
the appropriate @code{os_include_dir}. For examples of this approach,
|
464 |
|
|
see the @file{atomicity.h} file for AIX.
|
465 |
|
|
|
466 |
|
|
With those bits out of the way, you have to actually write
|
467 |
|
|
@file{atomicity.h} itself. This file should be wrapped in an
|
468 |
|
|
include guard named @code{_GLIBCXX_ATOMICITY_H}. It should define one
|
469 |
|
|
type, and two functions.
|
470 |
|
|
|
471 |
|
|
The type is @code{_Atomic_word}. Here is the version used on IRIX:
|
472 |
|
|
|
473 |
|
|
@example
|
474 |
|
|
typedef long _Atomic_word;
|
475 |
|
|
@end example
|
476 |
|
|
|
477 |
|
|
@noindent
|
478 |
|
|
This type must be a signed integral type supporting atomic operations.
|
479 |
|
|
If you're using the OS approach, use the same type used by your system's
|
480 |
|
|
primitives. Otherwise, use the type for which your CPU provides atomic
|
481 |
|
|
primitives.
|
482 |
|
|
|
483 |
|
|
Then, you must provide two functions. The bodies of these functions
|
484 |
|
|
must be equivalent to those provided here, but using atomic operations:
|
485 |
|
|
|
486 |
|
|
@example
|
487 |
|
|
static inline _Atomic_word
|
488 |
|
|
__attribute__ ((__unused__))
|
489 |
|
|
__exchange_and_add (_Atomic_word* __mem, int __val)
|
490 |
|
|
@{
|
491 |
|
|
_Atomic_word __result = *__mem;
|
492 |
|
|
*__mem += __val;
|
493 |
|
|
return __result;
|
494 |
|
|
@}
|
495 |
|
|
|
496 |
|
|
static inline void
|
497 |
|
|
__attribute__ ((__unused__))
|
498 |
|
|
__atomic_add (_Atomic_word* __mem, int __val)
|
499 |
|
|
@{
|
500 |
|
|
*__mem += __val;
|
501 |
|
|
@}
|
502 |
|
|
@end example
|
503 |
|
|
|
504 |
|
|
@c ---------------------------------------------------------------------
|
505 |
|
|
@c Numeric limits
|
506 |
|
|
@c ---------------------------------------------------------------------
|
507 |
|
|
|
508 |
|
|
@node Numeric limits
|
509 |
|
|
@chapter Numeric limits
|
510 |
|
|
|
511 |
|
|
The C++ library requires information about the fundamental data types,
|
512 |
|
|
such as the minimum and maximum representable values of each type.
|
513 |
|
|
You can define each of these values individually, but it is usually
|
514 |
|
|
easiest just to indicate how many bits are used in each of the data
|
515 |
|
|
types and let the library do the rest. For information about the
|
516 |
|
|
macros to define, see the top of @file{include/bits/std_limits.h}.
|
517 |
|
|
|
518 |
|
|
If you need to define any macros, you can do so in @file{os_defines.h}.
|
519 |
|
|
However, if all operating systems for your CPU are likely to use the
|
520 |
|
|
same values, you can provide a CPU-specific file instead so that you
|
521 |
|
|
do not have to provide the same definitions for each operating system.
|
522 |
|
|
To take that approach, create a new file called @file{cpu_limits.h} in
|
523 |
|
|
your CPU configuration directory (@pxref{CPU}).
|
524 |
|
|
|
525 |
|
|
@c ---------------------------------------------------------------------
|
526 |
|
|
@c Libtool
|
527 |
|
|
@c ---------------------------------------------------------------------
|
528 |
|
|
|
529 |
|
|
@node Libtool
|
530 |
|
|
@chapter Libtool
|
531 |
|
|
|
532 |
|
|
The C++ library is compiled, archived and linked with libtool.
|
533 |
|
|
Explaining the full workings of libtool is beyond the scope of this
|
534 |
|
|
document, but there are a few, particular bits that are necessary for
|
535 |
|
|
porting.
|
536 |
|
|
|
537 |
|
|
Some parts of the libstdc++-v3 library are compiled with the libtool
|
538 |
|
|
@code{--tags CXX} option (the C++ definitions for libtool). Therefore,
|
539 |
|
|
@file{ltcf-cxx.sh} in the top-level directory needs to have the correct
|
540 |
|
|
logic to compile and archive objects equivalent to the C version of libtool,
|
541 |
|
|
@file{ltcf-c.sh}. Some libtool targets have definitions for C but not
|
542 |
|
|
for C++, or C++ definitions which have not been kept up to date.
|
543 |
|
|
|
544 |
|
|
The C++ run-time library contains initialization code that needs to be
|
545 |
|
|
run as the library is loaded. Often, that requires linking in special
|
546 |
|
|
object files when the C++ library is built as a shared library, or
|
547 |
|
|
taking other system-specific actions.
|
548 |
|
|
|
549 |
|
|
The libstdc++-v3 library is linked with the C version of libtool, even
|
550 |
|
|
though it is a C++ library. Therefore, the C version of libtool needs to
|
551 |
|
|
ensure that the run-time library initializers are run. The usual way to
|
552 |
|
|
do this is to build the library using @code{gcc -shared}.
|
553 |
|
|
|
554 |
|
|
If you need to change how the library is linked, look at
|
555 |
|
|
@file{ltcf-c.sh} in the top-level directory. Find the switch statement
|
556 |
|
|
that sets @code{archive_cmds}. Here, adjust the setting for your
|
557 |
|
|
operating system.
|
558 |
|
|
|
559 |
|
|
@c ---------------------------------------------------------------------
|
560 |
|
|
@c GFDL
|
561 |
|
|
@c ---------------------------------------------------------------------
|
562 |
|
|
|
563 |
|
|
@include fdl.texi
|
564 |
|
|
|
565 |
|
|
@c ---------------------------------------------------------------------
|
566 |
|
|
@c Epilogue
|
567 |
|
|
@c ---------------------------------------------------------------------
|
568 |
|
|
|
569 |
|
|
@contents
|
570 |
|
|
@bye
|