specific.html revision 1.19 1 <!DOCTYPE html>
2 <html>
3 <!-- Created by GNU Texinfo 7.2, https://www.gnu.org/software/texinfo/ -->
4 <head>
5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
6 <!-- Copyright 1988-2024 Free Software Foundation, Inc.
7
8 Permission is granted to copy, distribute and/or modify this document
9 under the terms of the GNU Free Documentation License, Version 1.3 or
10 any later version published by the Free Software Foundation; with no
11 Invariant Sections, the Front-Cover texts being (a) (see below), and
12 with the Back-Cover Texts being (b) (see below). A copy of the
13 license is included in the section entitled "GNU
14 Free Documentation License".
15
16 (a) The FSF's Front-Cover Text is:
17
18 A GNU Manual
19
20 (b) The FSF's Back-Cover Text is:
21
22 You have freedom to copy and modify this GNU Manual, like GNU
23 software. Copies published by the Free Software Foundation raise
24 funds for GNU development. -->
25 <title>Host/Target specific installation notes for GCC</title>
26
27 <meta name="description" content="Host/Target specific installation notes for GCC">
28 <meta name="keywords" content="Host/Target specific installation notes for GCC">
29 <meta name="resource-type" content="document">
30 <meta name="distribution" content="global">
31 <meta name="Generator" content="makeinfo">
32 <meta name="viewport" content="width=device-width,initial-scale=1">
33
34 <style type="text/css">
35 <!--
36 a.copiable-link {visibility: hidden; text-decoration: none; line-height: 0em}
37 div.example {margin-left: 3.2em}
38 span:hover a.copiable-link {visibility: visible}
39 ul.mark-bullet {list-style-type: disc}
40 -->
41 </style>
42
43
44 </head>
45
46 <body lang="en">
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 <a class="index-entry-id" id="index-Specific"></a>
75 <a class="index-entry-id" id="index-Specific-installation-notes"></a>
76 <a class="index-entry-id" id="index-Target-specific-installation"></a>
77 <a class="index-entry-id" id="index-Host-specific-installation"></a>
78 <a class="index-entry-id" id="index-Target-specific-installation-notes"></a>
79
80 <p>Please read this document carefully <em class="emph">before</em> installing the
81 GNU Compiler Collection on your machine.
82 </p>
83 <p>Note that this list of install notes is <em class="emph">not</em> a list of supported
84 hosts or targets. Not all supported hosts and targets are listed
85 here, only the ones that require host-specific or target-specific
86 information have to.
87 </p>
88 <ul class="itemize mark-bullet">
89 <li><a class="uref" href="#aarch64-x-x">aarch64*-*-*</a>
90 </li><li><a class="uref" href="#amdgcn-x-amdhsa">amdgcn-*-amdhsa</a>
91 </li><li><a class="uref" href="#amd64-x-solaris2">amd64-*-solaris2*</a>
92 </li><li><a class="uref" href="#arc-x-elf32">arc-*-elf32</a>
93 </li><li><a class="uref" href="#arc-linux-uclibc">arc-linux-uclibc</a>
94 </li><li><a class="uref" href="#arm-x-eabi">arm-*-eabi</a>
95 </li><li><a class="uref" href="#avr">avr</a>
96 </li><li><a class="uref" href="#bfin">Blackfin</a>
97 </li><li><a class="uref" href="#cris">cris</a>
98 </li><li><a class="uref" href="#dos">DOS</a>
99 </li><li><a class="uref" href="#epiphany-x-elf">epiphany-*-elf</a>
100 </li><li><a class="uref" href="#ft32-x-elf">ft32-*-elf</a>
101 </li><li><a class="uref" href="#x-x-freebsd">*-*-freebsd*</a>
102 </li><li><a class="uref" href="#h8300-hms">h8300-hms</a>
103 </li><li><a class="uref" href="#hppa-hp-hpux">hppa*-hp-hpux*</a>
104 </li><li><a class="uref" href="#hppa-hp-hpux11">hppa*-hp-hpux11</a>
105 </li><li><a class="uref" href="#x-x-linux-gnu">*-*-linux-gnu</a>
106 </li><li><a class="uref" href="#ix86-x-linux">i?86-*-linux*</a>
107 </li><li><a class="uref" href="#ix86-x-solaris2">i?86-*-solaris2*</a>
108 </li><li><a class="uref" href="#ia64-x-linux">ia64-*-linux</a>
109 </li><li><a class="uref" href="#ia64-x-hpux">ia64-*-hpux*</a>
110 </li><li><a class="uref" href="#x-ibm-aix">*-ibm-aix*</a>
111 </li><li><a class="uref" href="#iq2000-x-elf">iq2000-*-elf</a>
112 </li><li><a class="uref" href="#loongarch">loongarch</a>
113 </li><li><a class="uref" href="#lm32-x-elf">lm32-*-elf</a>
114 </li><li><a class="uref" href="#lm32-x-uclinux">lm32-*-uclinux</a>
115 </li><li><a class="uref" href="#m32c-x-elf">m32c-*-elf</a>
116 </li><li><a class="uref" href="#m32r-x-elf">m32r-*-elf</a>
117 </li><li><a class="uref" href="#m68k-x-x">m68k-*-*</a>
118 </li><li><a class="uref" href="#m68k-x-uclinux">m68k-*-uclinux</a>
119 </li><li><a class="uref" href="#microblaze-x-elf">microblaze-*-elf</a>
120 </li><li><a class="uref" href="#mips-x-x">mips-*-*</a>
121 </li><li><a class="uref" href="#moxie-x-elf">moxie-*-elf</a>
122 </li><li><a class="uref" href="#msp430-x-elf">msp430-*-elf</a>
123 </li><li><a class="uref" href="#nds32le-x-elf">nds32le-*-elf</a>
124 </li><li><a class="uref" href="#nds32be-x-elf">nds32be-*-elf</a>
125 </li><li><a class="uref" href="#nvptx-x-none">nvptx-*-none</a>
126 </li><li><a class="uref" href="#or1k-x-elf">or1k-*-elf</a>
127 </li><li><a class="uref" href="#or1k-x-linux">or1k-*-linux</a>
128 </li><li><a class="uref" href="#powerpc-x-x">powerpc*-*-*</a>
129 </li><li><a class="uref" href="#powerpc-x-darwin">powerpc-*-darwin*</a>
130 </li><li><a class="uref" href="#powerpc-x-elf">powerpc-*-elf</a>
131 </li><li><a class="uref" href="#powerpc-x-linux-gnu">powerpc*-*-linux-gnu*</a>
132 </li><li><a class="uref" href="#powerpc-x-netbsd">powerpc-*-netbsd*</a>
133 </li><li><a class="uref" href="#powerpc-x-eabisim">powerpc-*-eabisim</a>
134 </li><li><a class="uref" href="#powerpc-x-eabi">powerpc-*-eabi</a>
135 </li><li><a class="uref" href="#powerpcle-x-elf">powerpcle-*-elf</a>
136 </li><li><a class="uref" href="#powerpcle-x-eabisim">powerpcle-*-eabisim</a>
137 </li><li><a class="uref" href="#powerpcle-x-eabi">powerpcle-*-eabi</a>
138 </li><li><a class="uref" href="#riscv32-x-elf">riscv32-*-elf</a>
139 </li><li><a class="uref" href="#riscv32-x-linux">riscv32-*-linux</a>
140 </li><li><a class="uref" href="#riscv64-x-elf">riscv64-*-elf</a>
141 </li><li><a class="uref" href="#riscv64-x-linux">riscv64-*-linux</a>
142 </li><li><a class="uref" href="#rl78-x-elf">rl78-*-elf</a>
143 </li><li><a class="uref" href="#rx-x-elf">rx-*-elf</a>
144 </li><li><a class="uref" href="#s390-x-linux">s390-*-linux*</a>
145 </li><li><a class="uref" href="#s390x-x-linux">s390x-*-linux*</a>
146 </li><li><a class="uref" href="#s390x-ibm-tpf">s390x-ibm-tpf*</a>
147 </li><li><a class="uref" href="#x-x-solaris2">*-*-solaris2*</a>
148 </li><li><a class="uref" href="#sparc-x-x">sparc*-*-*</a>
149 </li><li><a class="uref" href="#sparc-sun-solaris2">sparc-sun-solaris2*</a>
150 </li><li><a class="uref" href="#sparc-x-linux">sparc-*-linux*</a>
151 </li><li><a class="uref" href="#sparc64-x-solaris2">sparc64-*-solaris2*</a>
152 </li><li><a class="uref" href="#sparcv9-x-solaris2">sparcv9-*-solaris2*</a>
153 </li><li><a class="uref" href="#c6x-x-x">c6x-*-*</a>
154 </li><li><a class="uref" href="#visium-x-elf">visium-*-elf</a>
155 </li><li><a class="uref" href="#x-x-vxworks">*-*-vxworks*</a>
156 </li><li><a class="uref" href="#x86-64-x-x">x86_64-*-*, amd64-*-*</a>
157 </li><li><a class="uref" href="#x86-64-x-solaris2">x86_64-*-solaris2*</a>
158 </li><li><a class="uref" href="#xtensa-x-elf">xtensa*-*-elf</a>
159 </li><li><a class="uref" href="#xtensa-x-linux">xtensa*-*-linux*</a>
160 </li><li><a class="uref" href="#windows">Microsoft Windows</a>
161 </li><li><a class="uref" href="#x-x-cygwin">*-*-cygwin</a>
162 </li><li><a class="uref" href="#x-x-mingw32">*-*-mingw32</a>
163 </li><li><a class="uref" href="#os2">OS/2</a>
164 </li><li><a class="uref" href="#older">Older systems</a>
165 </li></ul>
166
167 <ul class="itemize mark-bullet">
168 <li><a class="uref" href="#elf">all ELF targets</a> (SVR4, Solaris, etc.)
169 </li></ul>
170
171
172 <!-- -------- host/target specific issues start here ---------------- -->
173 <hr />
174 <a class="anchor" id="aarch64-x-x"></a><h3 class="heading" id="aarch64*-*-*"><span>aarch64*-*-*<a class="copiable-link" href="#aarch64*-*-*"> ¶</a></span></h3>
175 <p>Binutils pre 2.24 does not have support for selecting <samp class="option">-mabi</samp> and
176 does not support ILP32. If it is used to build GCC 4.9 or later, GCC will
177 not support option <samp class="option">-mabi=ilp32</samp>.
178 </p>
179 <p>To enable a workaround for the Cortex-A53 erratum number 835769 by default
180 (for all CPUs regardless of -mcpu option given) at configure time use the
181 <samp class="option">--enable-fix-cortex-a53-835769</samp> option. This will enable the fix by
182 default and can be explicitly disabled during compilation by passing the
183 <samp class="option">-mno-fix-cortex-a53-835769</samp> option. Conversely,
184 <samp class="option">--disable-fix-cortex-a53-835769</samp> will disable the workaround by
185 default. The workaround is disabled by default if neither of
186 <samp class="option">--enable-fix-cortex-a53-835769</samp> or
187 <samp class="option">--disable-fix-cortex-a53-835769</samp> is given at configure time.
188 </p>
189 <p>To enable a workaround for the Cortex-A53 erratum number 843419 by default
190 (for all CPUs regardless of -mcpu option given) at configure time use the
191 <samp class="option">--enable-fix-cortex-a53-843419</samp> option. This workaround is applied at
192 link time. Enabling the workaround will cause GCC to pass the relevant option
193 to the linker. It can be explicitly disabled during compilation by passing the
194 <samp class="option">-mno-fix-cortex-a53-843419</samp> option. Conversely,
195 <samp class="option">--disable-fix-cortex-a53-843419</samp> will disable the workaround by default.
196 The workaround is disabled by default if neither of
197 <samp class="option">--enable-fix-cortex-a53-843419</samp> or
198 <samp class="option">--disable-fix-cortex-a53-843419</samp> is given at configure time.
199 </p>
200 <p>To enable Branch Target Identification Mechanism and Return Address Signing by
201 default at configure time use the <samp class="option">--enable-standard-branch-protection</samp>
202 option. This is equivalent to having <samp class="option">-mbranch-protection=standard</samp>
203 during compilation. This can be explicitly disabled during compilation by
204 passing the <samp class="option">-mbranch-protection=none</samp> option which turns off all
205 types of branch protections. Conversely,
206 <samp class="option">--disable-standard-branch-protection</samp> will disable both the
207 protections by default. This mechanism is turned off by default if neither
208 of the options are given at configure time.
209 </p>
210 <hr />
211 <a class="anchor" id="amd64-x-solaris2"></a><h3 class="heading" id="amd64-*-solaris2*"><span>amd64-*-solaris2*<a class="copiable-link" href="#amd64-*-solaris2*"> ¶</a></span></h3>
212 <p>This is a synonym for ‘<samp class="samp">x86_64-*-solaris2*</samp>’.
213 </p>
214 <hr />
215 <a class="anchor" id="amdgcn-x-amdhsa"></a><h3 class="heading" id="amdgcn-*-amdhsa"><span>amdgcn-*-amdhsa<a class="copiable-link" href="#amdgcn-*-amdhsa"> ¶</a></span></h3>
216 <p>AMD GCN GPU target.
217 </p>
218 <p>Instead of GNU Binutils, you will need to install LLVM 15, or later, and copy
219 <samp class="file">bin/llvm-mc</samp> to <samp class="file">amdgcn-amdhsa/bin/as</samp>,
220 <samp class="file">bin/lld</samp> to <samp class="file">amdgcn-amdhsa/bin/ld</samp>,
221 <samp class="file">bin/llvm-nm</samp> to <samp class="file">amdgcn-amdhsa/bin/nm</samp>, and
222 <samp class="file">bin/llvm-ar</samp> to both <samp class="file">bin/amdgcn-amdhsa-ar</samp> and
223 <samp class="file">bin/amdgcn-amdhsa-ranlib</samp>. Note that LLVM 13.0.1 or LLVM 14 can be used
224 by specifying a <code class="code">--with-multilib-list=</code> that does not list <code class="code">gfx1100</code>
225 and <code class="code">gfx1103</code>.
226 </p>
227 <p>Use Newlib (4.3.0 or newer; 4.4.0 contains some improvements and git commit
228 7dd4eb1db (2025-03-25, post-4.4.0) fixes device console output for GFX10 and
229 GFX11 devices).
230 </p>
231 <p>To run the binaries, install the HSA Runtime from the
232 <a class="uref" href="https://rocm.docs.amd.com/">ROCm Platform</a>, and use
233 <samp class="file">libexec/gcc/amdhsa-amdhsa/<var class="var">version</var>/gcn-run</samp> to launch them
234 on the GPU.
235 </p>
236 <p>To enable support for GCN3 Fiji devices (gfx803), GCC has to be configured with
237 <samp class="option">--with-arch=<code class="code">fiji</code></samp> or
238 <samp class="option">--with-multilib-list=<code class="code">fiji</code>,...</samp>. Note that support for Fiji
239 devices has been removed in ROCm 4.0 and support in LLVM was deprecated and has
240 been removed in LLVM 18.
241 </p>
242 <hr />
243 <a class="anchor" id="arc-x-elf32"></a><h3 class="heading" id="arc-*-elf32"><span>arc-*-elf32<a class="copiable-link" href="#arc-*-elf32"> ¶</a></span></h3>
244
245 <p>Use ‘<samp class="samp">configure --target=arc-elf32 --with-cpu=<var class="var">cpu</var> --enable-languages="c,c++"</samp>’
246 to configure GCC, with <var class="var">cpu</var> being one of ‘<samp class="samp">arc600</samp>’, ‘<samp class="samp">arc601</samp>’,
247 or ‘<samp class="samp">arc700</samp>’.
248 </p>
249 <hr />
250 <a class="anchor" id="arc-linux-uclibc"></a><h3 class="heading" id="arc-linux-uclibc-1"><span>arc-linux-uclibc<a class="copiable-link" href="#arc-linux-uclibc-1"> ¶</a></span></h3>
251
252 <p>Use ‘<samp class="samp">configure --target=arc-linux-uclibc --with-cpu=arc700 --enable-languages="c,c++"</samp>’ to configure GCC.
253 </p>
254 <hr />
255 <a class="anchor" id="arm-x-eabi"></a><h3 class="heading" id="arm-*-eabi"><span>arm-*-eabi<a class="copiable-link" href="#arm-*-eabi"> ¶</a></span></h3>
256 <p>ARM-family processors.
257 </p>
258 <p>Building the Ada frontend commonly fails (an infinite loop executing
259 <code class="code">xsinfo</code>) if the host compiler is GNAT 4.8. Host compilers built from the
260 GNAT 4.6, 4.9 or 5 release branches are known to succeed.
261 </p>
262 <hr />
263 <a class="anchor" id="avr"></a><h3 class="heading" id="avr-1"><span>avr<a class="copiable-link" href="#avr-1"> ¶</a></span></h3>
264 <p>ATMEL AVR-family micro controllers. These are used in embedded
265 applications. There are no standard Unix configurations.
266 See “AVR Options” in the main manual
267 for the list of supported MCU types.
268 </p>
269 <p>Use ‘<samp class="samp">configure --target=avr --enable-languages="c"</samp>’ to configure GCC.
270 </p>
271 <p>Further installation notes and other useful information about AVR tools
272 can also be obtained from:
273 </p>
274 <ul class="itemize mark-bullet">
275 <li><a class="uref" href="http://www.nongnu.org/avr/">http://www.nongnu.org/avr/</a>
276 </li><li><a class="uref" href="http://www.amelek.gda.pl/avr/">http://www.amelek.gda.pl/avr/</a>
277 </li></ul>
278
279 <p>The following error:
280 </p><div class="example smallexample">
281 <pre class="example-preformatted">Error: register required
282 </pre></div>
283
284 <p>indicates that you should upgrade to a newer version of the binutils.
285 </p>
286 <hr />
287 <a class="anchor" id="bfin"></a><h3 class="heading" id="Blackfin"><span>Blackfin<a class="copiable-link" href="#Blackfin"> ¶</a></span></h3>
288 <p>The Blackfin processor, an Analog Devices DSP.
289 See “Blackfin Options” in the main manual
290 </p>
291 <p>More information, and a version of binutils with support for this processor,
292 are available at <a class="uref" href="https://sourceforge.net/projects/adi-toolchain/">https://sourceforge.net/projects/adi-toolchain/</a>.
293 </p>
294 <hr />
295 <a class="anchor" id="cris"></a><h3 class="heading" id="CRIS"><span>CRIS<a class="copiable-link" href="#CRIS"> ¶</a></span></h3>
296 <p>CRIS is a CPU architecture in Axis Communications systems-on-a-chip, for
297 example the ETRAX series. These are used in embedded applications.
298 </p>
299 <p>See “CRIS Options” in the main manual
300 for a list of CRIS-specific options.
301 </p>
302 <p>Use ‘<samp class="samp">configure --target=cris-elf</samp>’ to configure GCC for building
303 a cross-compiler for CRIS.
304 <hr />
305 <a class="anchor" id="dos"></a></p><h3 class="heading" id="DOS"><span>DOS<a class="copiable-link" href="#DOS"> ¶</a></span></h3>
306 <p>Please have a look at the <a class="uref" href="binaries.html">binaries page</a>.
307 </p>
308 <p>You cannot install GCC by itself on MSDOS; it will not compile under
309 any MSDOS compiler except itself. You need to get the complete
310 compilation package DJGPP, which includes binaries as well as sources,
311 and includes all the necessary compilation tools and libraries.
312 </p>
313 <hr />
314 <a class="anchor" id="epiphany-x-elf"></a><h3 class="heading" id="epiphany-*-elf"><span>epiphany-*-elf<a class="copiable-link" href="#epiphany-*-elf"> ¶</a></span></h3>
315 <p>Adapteva Epiphany.
316 This configuration is intended for embedded systems.
317 </p>
318 <hr />
319 <a class="anchor" id="x-x-freebsd"></a><h3 class="heading" id="g_t*-*-freebsd*"><span>*-*-freebsd*<a class="copiable-link" href="#g_t*-*-freebsd*"> ¶</a></span></h3>
320 <p>We support FreeBSD using the ELF file format with DWARF 2 debugging
321 for all CPU architectures. There are
322 no known issues with mixing object files and libraries with different
323 debugging formats.
324 </p>
325 <p>We recommend bootstrapping against the latest GNU binutils or the
326 version found in the <samp class="file">devel/binutils</samp> port. This also has been
327 known to enable additional features and improve overall testsuite
328 results.
329 </p>
330 <p>Ada and D (or rather their respective libraries) are broken on
331 FreeBSD/i386. This also affects building 32-bit libraries on
332 FreeBSD/amd64, so configure with <samp class="option">--disable-multilib</samp>
333 there in case you are building one of these front ends.
334 </p>
335 <p>Go (or rather libgo) is generally broken on FreeBSD.
336 </p>
337 <hr />
338 <a class="anchor" id="ft32-x-elf"></a><h3 class="heading" id="ft32-*-elf"><span>ft32-*-elf<a class="copiable-link" href="#ft32-*-elf"> ¶</a></span></h3>
339 <p>The FT32 processor.
340 This configuration is intended for embedded systems.
341 </p>
342 <hr />
343 <a class="anchor" id="h8300-hms"></a><h3 class="heading" id="h8300-hms-1"><span>h8300-hms<a class="copiable-link" href="#h8300-hms-1"> ¶</a></span></h3>
344 <p>Renesas H8/300 series of processors.
345 </p>
346 <p>Please have a look at the <a class="uref" href="binaries.html">binaries page</a>.
347 </p>
348 <p>The calling convention and structure layout has changed in release 2.6.
349 All code must be recompiled. The calling convention now passes the
350 first three arguments in function calls in registers. Structures are no
351 longer a multiple of 2 bytes.
352 </p>
353 <hr />
354 <a class="anchor" id="hppa-hp-hpux"></a><h3 class="heading" id="hppa*-hp-hpux*"><span>hppa*-hp-hpux*<a class="copiable-link" href="#hppa*-hp-hpux*"> ¶</a></span></h3>
355 <p>We require using gas on all hppa platforms. Version 2.19 or
356 later is recommended.
357 </p>
358 <p>It may be helpful to configure GCC with the
359 <a class="uref" href="./configure.html#with-gnu-as"><samp class="option">--with-gnu-as</samp></a> and
360 <samp class="option">--with-as=…</samp> options to ensure that GCC can find GAS.
361 </p>
362 <p>There are two default scheduling models for instructions. These are
363 PROCESSOR_7100LC and PROCESSOR_8000. They are selected from the pa-risc
364 architecture specified for the target machine when configuring.
365 PROCESSOR_8000 is the default. PROCESSOR_7100LC is selected when
366 the target is a ‘<samp class="samp">hppa1*</samp>’ machine.
367 </p>
368 <p>The PROCESSOR_8000 model is not well suited to older processors. Thus,
369 it is important to completely specify the machine architecture when
370 configuring if you want a model other than PROCESSOR_8000. The macro
371 TARGET_SCHED_DEFAULT can be defined in BOOT_CFLAGS if a different
372 default scheduling model is desired.
373 </p>
374 <p>As of GCC 4.0, GCC uses the UNIX 95 namespace for HP-UX 10.10
375 through 11.00, and the UNIX 98 namespace for HP-UX 11.11 and later.
376 This namespace change might cause problems when bootstrapping with
377 an earlier version of GCC or the HP compiler as essentially the same
378 namespace is required for an entire build. This problem can be avoided
379 in a number of ways. With HP cc, <code class="env">UNIX_STD</code> can be set to ‘<samp class="samp">95</samp>’
380 or ‘<samp class="samp">98</samp>’. Another way is to add an appropriate set of predefines
381 to <code class="env">CC</code>. The description for the <samp class="option">munix=</samp> option contains
382 a list of the predefines used with each standard.
383 </p>
384 <p>More specific information to ‘<samp class="samp">hppa*-hp-hpux*</samp>’ targets follows.
385 </p>
386 <hr />
387 <a class="anchor" id="hppa-hp-hpux11"></a><h3 class="heading" id="hppa*-hp-hpux11"><span>hppa*-hp-hpux11<a class="copiable-link" href="#hppa*-hp-hpux11"> ¶</a></span></h3>
388 <p>Refer to <a class="uref" href="binaries.html">binaries</a> for information about obtaining
389 precompiled GCC binaries for HP-UX. Precompiled binaries must be obtained
390 to build the Ada language as it cannot be bootstrapped using C. Ada is
391 only available for the 32-bit PA-RISC runtime.
392 </p>
393 <p>Starting with GCC 3.4 an ISO C compiler is required to bootstrap. The
394 bundled compiler supports only traditional C; you will need either HP’s
395 unbundled compiler, or a binary distribution of GCC.
396 </p>
397 <p>It is possible to build GCC 3.3 starting with the bundled HP compiler,
398 but the process requires several steps. GCC 3.3 can then be used to
399 build later versions.
400 </p>
401 <p>There are several possible approaches to building the distribution.
402 Binutils can be built first using the HP tools. Then, the GCC
403 distribution can be built. The second approach is to build GCC
404 first using the HP tools, then build binutils, then rebuild GCC.
405 There have been problems with various binary distributions, so it
406 is best not to start from a binary distribution.
407 </p>
408 <p>On 64-bit capable systems, there are two distinct targets. Different
409 installation prefixes must be used if both are to be installed on
410 the same system. The ‘<samp class="samp">hppa[1-2]*-hp-hpux11*</samp>’ target generates code
411 for the 32-bit PA-RISC runtime architecture and uses the HP linker.
412 The ‘<samp class="samp">hppa64-hp-hpux11*</samp>’ target generates 64-bit code for the
413 PA-RISC 2.0 architecture.
414 </p>
415 <p>The script config.guess now selects the target type based on the compiler
416 detected during configuration. You must define <code class="env">PATH</code> or <code class="env">CC</code> so
417 that configure finds an appropriate compiler for the initial bootstrap.
418 When <code class="env">CC</code> is used, the definition should contain the options that are
419 needed whenever <code class="env">CC</code> is used.
420 </p>
421 <p>Specifically, options that determine the runtime architecture must be
422 in <code class="env">CC</code> to correctly select the target for the build. It is also
423 convenient to place many other compiler options in <code class="env">CC</code>. For example,
424 <code class="env">CC="cc -Ac +DA2.0W -Wp,-H16376 -D_CLASSIC_TYPES -D_HPUX_SOURCE"</code>
425 can be used to bootstrap the GCC 3.3 branch with the HP compiler in
426 64-bit K&R/bundled mode. The <samp class="option">+DA2.0W</samp> option will result in
427 the automatic selection of the ‘<samp class="samp">hppa64-hp-hpux11*</samp>’ target. The
428 macro definition table of cpp needs to be increased for a successful
429 build with the HP compiler. _CLASSIC_TYPES and _HPUX_SOURCE need to
430 be defined when building with the bundled compiler, or when using the
431 <samp class="option">-Ac</samp> option. These defines aren’t necessary with <samp class="option">-Ae</samp>.
432 </p>
433 <p>It is best to explicitly configure the ‘<samp class="samp">hppa64-hp-hpux11*</samp>’ target
434 with the <samp class="option">--with-ld=…</samp> option. This overrides the standard
435 search for ld. The two linkers supported on this target require different
436 commands. The default linker is determined during configuration. As a
437 result, it’s not possible to switch linkers in the middle of a GCC build.
438 This has been reported to sometimes occur in unified builds of binutils
439 and GCC.
440 </p>
441 <p>A recent linker patch must be installed for the correct operation of
442 GCC 3.3 and later. <code class="code">PHSS_26559</code> and <code class="code">PHSS_24304</code> are the
443 oldest linker patches that are known to work. They are for HP-UX
444 11.00 and 11.11, respectively. <code class="code">PHSS_24303</code>, the companion to
445 <code class="code">PHSS_24304</code>, might be usable but it hasn’t been tested. These
446 patches have been superseded. Consult the HP patch database to obtain
447 the currently recommended linker patch for your system.
448 </p>
449 <p>The patches are necessary for the support of weak symbols on the
450 32-bit port, and for the running of initializers and finalizers. Weak
451 symbols are implemented using SOM secondary definition symbols. Prior
452 to HP-UX 11, there are bugs in the linker support for secondary symbols.
453 The patches correct a problem of linker core dumps creating shared
454 libraries containing secondary symbols, as well as various other
455 linking issues involving secondary symbols.
456 </p>
457 <p>GCC 3.3 uses the ELF DT_INIT_ARRAY and DT_FINI_ARRAY capabilities to
458 run initializers and finalizers on the 64-bit port. The 32-bit port
459 uses the linker <samp class="option">+init</samp> and <samp class="option">+fini</samp> options for the same
460 purpose. The patches correct various problems with the +init/+fini
461 options, including program core dumps. Binutils 2.14 corrects a
462 problem on the 64-bit port resulting from HP’s non-standard use of
463 the .init and .fini sections for array initializers and finalizers.
464 </p>
465 <p>Only the HP linker is supported for the ‘<samp class="samp">hppa64-hp-hpux11*</samp>’ target.
466 </p>
467 <p>At this time, the GNU linker does not support the creation of long
468 branch stubs. As a result, it cannot successfully link binaries
469 containing branch offsets larger than 8 megabytes. In addition,
470 there are problems linking shared libraries, linking executables
471 with <samp class="option">-static</samp>, and with dwarf2 unwind and exception support.
472 It also doesn’t provide stubs for internal calls to global functions
473 in shared libraries, so these calls cannot be overloaded.
474 </p>
475 <p>The HP dynamic loader does not support GNU symbol versioning, so symbol
476 versioning is not supported. It may be necessary to disable symbol
477 versioning with <samp class="option">--disable-symvers</samp> when using GNU ld.
478 </p>
479 <p>POSIX threads are the default. The optional DCE thread library is not
480 supported, so <samp class="option">--enable-threads=dce</samp> does not work.
481 </p>
482 <hr />
483 <a class="anchor" id="x-x-linux-gnu"></a><h3 class="heading" id="g_t*-*-linux-gnu"><span>*-*-linux-gnu<a class="copiable-link" href="#g_t*-*-linux-gnu"> ¶</a></span></h3>
484 <p>The <code class="code">.init_array</code> and <code class="code">.fini_array</code> sections are enabled
485 unconditionally which requires at least glibc 2.1 and binutils 2.12.
486 </p>
487 <p>Versions of libstdc++-v3 starting with 3.2.1 require bug fixes present
488 in glibc 2.2.5 and later. More information is available in the
489 libstdc++-v3 documentation.
490 </p>
491 <hr />
492 <a class="anchor" id="ix86-x-linux"></a><h3 class="heading" id="i_003f86-*-linux*"><span>i?86-*-linux*<a class="copiable-link" href="#i_003f86-*-linux*"> ¶</a></span></h3>
493 <p>As of GCC 3.3, binutils 2.13.1 or later is required for this platform.
494 See <a class="uref" href="https://gcc.gnu.org/PR10877">bug 10877</a> for more information.
495 </p>
496 <p>If you receive Signal 11 errors when building on GNU/Linux, then it is
497 possible you have a hardware problem. Further information on this can be
498 found on <a class="uref" href="https://www.bitwizard.nl/sig11/">www.bitwizard.nl</a>.
499 </p>
500 <hr />
501 <a class="anchor" id="ix86-x-solaris2"></a><h3 class="heading" id="i_003f86-*-solaris2*"><span>i?86-*-solaris2*<a class="copiable-link" href="#i_003f86-*-solaris2*"> ¶</a></span></h3>
502 <p>Use this for Solaris 11.3 or later on x86 and x86-64 systems. Starting
503 with GCC 4.7, there is also a 64-bit ‘<samp class="samp">amd64-*-solaris2*</samp>’ or
504 ‘<samp class="samp">x86_64-*-solaris2*</samp>’ configuration that corresponds to
505 ‘<samp class="samp">sparcv9-sun-solaris2*</samp>’.
506 </p>
507 <hr />
508 <a class="anchor" id="ia64-x-linux"></a><h3 class="heading" id="ia64-*-linux"><span>ia64-*-linux<a class="copiable-link" href="#ia64-*-linux"> ¶</a></span></h3>
509 <p>IA-64 processor (also known as IPF, or Itanium Processor Family)
510 running GNU/Linux.
511 </p>
512 <p>If you are using the installed system libunwind library with
513 <samp class="option">--with-system-libunwind</samp>, then you must use libunwind 0.98 or
514 later.
515 </p>
516 <hr />
517 <a class="anchor" id="ia64-x-hpux"></a><h3 class="heading" id="ia64-*-hpux*"><span>ia64-*-hpux*<a class="copiable-link" href="#ia64-*-hpux*"> ¶</a></span></h3>
518 <p>Building GCC on this target requires the GNU Assembler. The bundled HP
519 assembler will not work. To prevent GCC from using the wrong assembler,
520 the option <samp class="option">--with-gnu-as</samp> may be necessary.
521 </p>
522 <hr />
523 <!-- rs6000-ibm-aix*, powerpc-ibm-aix* -->
524 <a class="anchor" id="x-ibm-aix"></a><h3 class="heading" id="g_t*-ibm-aix*"><span>*-ibm-aix*<a class="copiable-link" href="#g_t*-ibm-aix*"> ¶</a></span></h3>
525 <p>Support for AIX version 3 and older was discontinued in GCC 3.4.
526 Support for AIX version 4.2 and older was discontinued in GCC 4.5.
527 </p>
528 <p>“out of memory” bootstrap failures may indicate a problem with
529 process resource limits (ulimit). Hard limits are configured in the
530 <samp class="file">/etc/security/limits</samp> system configuration file.
531 </p>
532 <p>GCC 4.9 and above require a C++ compiler for bootstrap. IBM VAC++ / xlC
533 cannot bootstrap GCC. xlc can bootstrap an older version of GCC and
534 G++ can bootstrap recent releases of GCC.
535 </p>
536 <p>GCC can bootstrap with recent versions of IBM XLC, but bootstrapping
537 with an earlier release of GCC is recommended. Bootstrapping with XLC
538 requires a larger data segment, which can be enabled through the
539 <var class="var">LDR_CNTRL</var> environment variable, e.g.,
540 </p>
541 <div class="example smallexample">
542 <pre class="example-preformatted">% LDR_CNTRL=MAXDATA=0x50000000
543 % export LDR_CNTRL
544 </pre></div>
545
546 <p>One can start with a pre-compiled version of GCC to build from
547 sources. One may delete GCC’s “fixed” header files when starting
548 with a version of GCC built for an earlier release of AIX.
549 </p>
550 <p>To speed up the configuration phases of bootstrapping and installing GCC,
551 one may use GNU Bash instead of AIX <code class="command">/bin/sh</code>, e.g.,
552 </p>
553 <div class="example smallexample">
554 <pre class="example-preformatted">% CONFIG_SHELL=/opt/freeware/bin/bash
555 % export CONFIG_SHELL
556 </pre></div>
557
558 <p>and then proceed as described in <a class="uref" href="build.html">the build
559 instructions</a>, where we strongly recommend specifying an absolute path
560 to invoke <var class="var">srcdir</var>/configure.
561 </p>
562 <p>Because GCC on AIX is built as a 32-bit executable by default,
563 (although it can generate 64-bit programs) the GMP and MPFR libraries
564 required by gfortran must be 32-bit libraries. Building GMP and MPFR
565 as static archive libraries works better than shared libraries.
566 </p>
567 <p>Errors involving <code class="code">alloca</code> when building GCC generally are due
568 to an incorrect definition of <code class="code">CC</code> in the Makefile or mixing files
569 compiled with the native C compiler and GCC. During the stage1 phase of
570 the build, the native AIX compiler <strong class="strong">must</strong> be invoked as <code class="command">cc</code>
571 (not <code class="command">xlc</code>). Once <code class="command">configure</code> has been informed of
572 <code class="command">xlc</code>, one needs to use ‘<samp class="samp">make distclean</samp>’ to remove the
573 configure cache files and ensure that <code class="env">CC</code> environment variable
574 does not provide a definition that will confuse <code class="command">configure</code>.
575 If this error occurs during stage2 or later, then the problem most likely
576 is the version of Make (see above).
577 </p>
578 <p>The native <code class="command">as</code> and <code class="command">ld</code> are recommended for
579 bootstrapping on AIX. The GNU Assembler, GNU Linker, and GNU
580 Binutils version 2.20 is the minimum level that supports bootstrap on
581 AIX 5. The GNU Assembler has not been updated to support AIX 6 or
582 AIX 7. The native AIX tools do interoperate with GCC.
583 </p>
584 <p>AIX 7.1 added partial support for DWARF debugging, but full support
585 requires AIX 7.1 TL03 SP7 that supports additional DWARF sections and
586 fixes a bug in the assembler. AIX 7.1 TL03 SP5 distributed a version
587 of libm.a missing important symbols; a fix for IV77796 will be
588 included in SP6.
589 </p>
590 <p>AIX 5.3 TL10, AIX 6.1 TL05 and AIX 7.1 TL00 introduced an AIX
591 assembler change that sometimes produces corrupt assembly files
592 causing AIX linker errors. The bug breaks GCC bootstrap on AIX and
593 can cause compilation failures with existing GCC installations. An
594 AIX iFix for AIX 5.3 is available (APAR IZ98385 for AIX 5.3 TL10, APAR
595 IZ98477 for AIX 5.3 TL11 and IZ98134 for AIX 5.3 TL12). AIX 5.3 TL11 SP8,
596 AIX 5.3 TL12 SP5, AIX 6.1 TL04 SP11, AIX 6.1 TL05 SP7, AIX 6.1 TL06 SP6,
597 AIX 6.1 TL07 and AIX 7.1 TL01 should include the fix.
598 </p>
599 <p>Building <samp class="file">libstdc++.a</samp> requires a fix for an AIX Assembler bug
600 APAR IY26685 (AIX 4.3) or APAR IY25528 (AIX 5.1). It also requires a
601 fix for another AIX Assembler bug and a co-dependent AIX Archiver fix
602 referenced as APAR IY53606 (AIX 5.2) or as APAR IY54774 (AIX 5.1)
603 </p>
604 <a class="anchor" id="TransferAixShobj"></a><p>‘<samp class="samp">libstdc++</samp>’ in GCC 3.4 increments the major version number of the
605 shared object and GCC installation places the <samp class="file">libstdc++.a</samp>
606 shared library in a common location which will overwrite the and GCC
607 3.3 version of the shared library. Applications either need to be
608 re-linked against the new shared library or the GCC 3.1 and GCC 3.3
609 versions of the ‘<samp class="samp">libstdc++</samp>’ shared object needs to be available
610 to the AIX runtime loader. The GCC 3.1 ‘<samp class="samp">libstdc++.so.4</samp>’, if
611 present, and GCC 3.3 ‘<samp class="samp">libstdc++.so.5</samp>’ shared objects can be
612 installed for runtime dynamic loading using the following steps to set
613 the ‘<samp class="samp">F_LOADONLY</samp>’ flag in the shared object for <em class="emph">each</em>
614 multilib <samp class="file">libstdc++.a</samp> installed:
615 </p>
616 <p>Extract the shared objects from the currently installed
617 <samp class="file">libstdc++.a</samp> archive:
618 </p><div class="example smallexample">
619 <pre class="example-preformatted">% ar -x libstdc++.a libstdc++.so.4 libstdc++.so.5
620 </pre></div>
621
622 <p>Enable the ‘<samp class="samp">F_LOADONLY</samp>’ flag so that the shared object will be
623 available for runtime dynamic loading, but not linking:
624 </p><div class="example smallexample">
625 <pre class="example-preformatted">% strip -e libstdc++.so.4 libstdc++.so.5
626 </pre></div>
627
628 <p>Archive the runtime-only shared object in the GCC 3.4
629 <samp class="file">libstdc++.a</samp> archive:
630 </p><div class="example smallexample">
631 <pre class="example-preformatted">% ar -q libstdc++.a libstdc++.so.4 libstdc++.so.5
632 </pre></div>
633
634 <p>Eventually, the
635 <a class="uref" href="./configure.html#WithAixSoname"><samp class="option">--with-aix-soname=svr4</samp></a>
636 configure option may drop the need for this procedure for libraries that
637 support it.
638 </p>
639 <p>Linking executables and shared libraries may produce warnings of
640 duplicate symbols. The assembly files generated by GCC for AIX always
641 have included multiple symbol definitions for certain global variable
642 and function declarations in the original program. The warnings should
643 not prevent the linker from producing a correct library or runnable
644 executable.
645 </p>
646 <p>AIX 4.3 utilizes a “large format” archive to support both 32-bit and
647 64-bit object modules. The routines provided in AIX 4.3.0 and AIX 4.3.1
648 to parse archive libraries did not handle the new format correctly.
649 These routines are used by GCC and result in error messages during
650 linking such as “not a COFF file”. The version of the routines shipped
651 with AIX 4.3.1 should work for a 32-bit environment. The <samp class="option">-g</samp>
652 option of the archive command may be used to create archives of 32-bit
653 objects using the original “small format”. A correct version of the
654 routines is shipped with AIX 4.3.2 and above.
655 </p>
656 <p>Some versions of the AIX binder (linker) can fail with a relocation
657 overflow severe error when the <samp class="option">-bbigtoc</samp> option is used to link
658 GCC-produced object files into an executable that overflows the TOC. A fix
659 for APAR IX75823 (OVERFLOW DURING LINK WHEN USING GCC AND -BBIGTOC) is
660 available from IBM Customer Support and from its
661 <a class="uref" href="https://techsupport.services.ibm.com/">techsupport.services.ibm.com</a>
662 website as PTF U455193.
663 </p>
664 <p>The AIX 4.3.2.1 linker (bos.rte.bind_cmds Level 4.3.2.1) will dump core
665 with a segmentation fault when invoked by any version of GCC. A fix for
666 APAR IX87327 is available from IBM Customer Support and from its
667 <a class="uref" href="https://techsupport.services.ibm.com/">techsupport.services.ibm.com</a>
668 website as PTF U461879. This fix is incorporated in AIX 4.3.3 and above.
669 </p>
670 <p>The initial assembler shipped with AIX 4.3.0 generates incorrect object
671 files. A fix for APAR IX74254 (64BIT DISASSEMBLED OUTPUT FROM COMPILER FAILS
672 TO ASSEMBLE/BIND) is available from IBM Customer Support and from its
673 <a class="uref" href="https://techsupport.services.ibm.com/">techsupport.services.ibm.com</a>
674 website as PTF U453956. This fix is incorporated in AIX 4.3.1 and above.
675 </p>
676 <p>AIX provides National Language Support (NLS). Compilers and assemblers
677 use NLS to support locale-specific representations of various data
678 formats including floating-point numbers (e.g., ‘<samp class="samp">.</samp>’ vs ‘<samp class="samp">,</samp>’ for
679 separating decimal fractions). There have been problems reported where
680 GCC does not produce the same floating-point formats that the assembler
681 expects. If one encounters this problem, set the <code class="env">LANG</code>
682 environment variable to ‘<samp class="samp">C</samp>’ or ‘<samp class="samp">En_US</samp>’.
683 </p>
684 <p>A default can be specified with the <samp class="option">-mcpu=<var class="var">cpu_type</var></samp>
685 switch and using the configure option <samp class="option">--with-cpu-<var class="var">cpu_type</var></samp>.
686 </p>
687 <hr />
688 <a class="anchor" id="iq2000-x-elf"></a><h3 class="heading" id="iq2000-*-elf"><span>iq2000-*-elf<a class="copiable-link" href="#iq2000-*-elf"> ¶</a></span></h3>
689 <p>Vitesse IQ2000 processors. These are used in embedded
690 applications. There are no standard Unix configurations.
691 </p>
692 <hr />
693 <a class="anchor" id="lm32-x-elf"></a><h3 class="heading" id="lm32-*-elf"><span>lm32-*-elf<a class="copiable-link" href="#lm32-*-elf"> ¶</a></span></h3>
694 <p>Lattice Mico32 processor.
695 This configuration is intended for embedded systems.
696 </p>
697 <hr />
698 <a class="anchor" id="lm32-x-uclinux"></a><h3 class="heading" id="lm32-*-uclinux"><span>lm32-*-uclinux<a class="copiable-link" href="#lm32-*-uclinux"> ¶</a></span></h3>
699 <p>Lattice Mico32 processor.
700 This configuration is intended for embedded systems running uClinux.
701 </p>
702 <hr />
703 <a class="anchor" id="loongarch"></a><h3 class="heading" id="LoongArch"><span>LoongArch<a class="copiable-link" href="#LoongArch"> ¶</a></span></h3>
704 <p>LoongArch processor.
705 The following LoongArch targets are available:
706 </p><dl class="table">
707 <dt><code class="code">loongarch64-linux-gnu*</code></dt>
708 <dd><p>LoongArch processor running GNU/Linux. This target triplet may be coupled
709 with a small set of possible suffixes to identify their default ABI type:
710 </p><dl class="table">
711 <dt><code class="code">f64</code></dt>
712 <dd><p>Uses <code class="code">lp64d/base</code> ABI by default.
713 </p></dd>
714 <dt><code class="code">f32</code></dt>
715 <dd><p>Uses <code class="code">lp64f/base</code> ABI by default.
716 </p></dd>
717 <dt><code class="code">sf</code></dt>
718 <dd><p>Uses <code class="code">lp64s/base</code> ABI by default.
719 </p></dd>
720 </dl>
721
722 </dd>
723 <dt><code class="code">loongarch64-linux-gnu</code></dt>
724 <dd><p>Same as <code class="code">loongarch64-linux-gnuf64</code> for legacy support.
725 </p></dd>
726 </dl>
727
728 <p>More information about LoongArch can be found at
729 <a class="uref" href="https://github.com/loongson/LoongArch-Documentation">https://github.com/loongson/LoongArch-Documentation</a>.
730 </p>
731 <hr />
732 <a class="anchor" id="m32c-x-elf"></a><h3 class="heading" id="m32c-*-elf"><span>m32c-*-elf<a class="copiable-link" href="#m32c-*-elf"> ¶</a></span></h3>
733 <p>Renesas M32C processor.
734 This configuration is intended for embedded systems.
735 </p>
736 <hr />
737 <a class="anchor" id="m32r-x-elf"></a><h3 class="heading" id="m32r-*-elf"><span>m32r-*-elf<a class="copiable-link" href="#m32r-*-elf"> ¶</a></span></h3>
738 <p>Renesas M32R processor.
739 This configuration is intended for embedded systems.
740 </p>
741 <hr />
742 <a class="anchor" id="m68k-x-x"></a><h3 class="heading" id="m68k-*-*"><span>m68k-*-*<a class="copiable-link" href="#m68k-*-*"> ¶</a></span></h3>
743 <p>By default,
744 ‘<samp class="samp">m68k-*-elf*</samp>’, ‘<samp class="samp">m68k-*-rtems</samp>’, ‘<samp class="samp">m68k-*-uclinux</samp>’ and
745 ‘<samp class="samp">m68k-*-linux</samp>’
746 build libraries for both M680x0 and ColdFire processors. If you only
747 need the M680x0 libraries, you can omit the ColdFire ones by passing
748 <samp class="option">--with-arch=m68k</samp> to <code class="command">configure</code>. Alternatively, you
749 can omit the M680x0 libraries by passing <samp class="option">--with-arch=cf</samp> to
750 <code class="command">configure</code>. These targets default to 5206 or 5475 code as
751 appropriate for the target system when
752 configured with <samp class="option">--with-arch=cf</samp> and 68020 code otherwise.
753 </p>
754 <p>The ‘<samp class="samp">m68k-*-netbsd</samp>’ and
755 ‘<samp class="samp">m68k-*-openbsd</samp>’ targets also support the <samp class="option">--with-arch</samp>
756 option. They will generate ColdFire CFV4e code when configured with
757 <samp class="option">--with-arch=cf</samp> and 68020 code otherwise.
758 </p>
759 <p>You can override the default processors listed above by configuring
760 with <samp class="option">--with-cpu=<var class="var">target</var></samp>. This <var class="var">target</var> can either
761 be a <samp class="option">-mcpu</samp> argument or one of the following values:
762 ‘<samp class="samp">m68000</samp>’, ‘<samp class="samp">m68010</samp>’, ‘<samp class="samp">m68020</samp>’, ‘<samp class="samp">m68030</samp>’,
763 ‘<samp class="samp">m68040</samp>’, ‘<samp class="samp">m68060</samp>’, ‘<samp class="samp">m68020-40</samp>’ and ‘<samp class="samp">m68020-60</samp>’.
764 </p>
765 <p>GCC requires at least binutils version 2.17 on these targets.
766 </p>
767 <hr />
768 <a class="anchor" id="m68k-x-uclinux"></a><h3 class="heading" id="m68k-*-uclinux"><span>m68k-*-uclinux<a class="copiable-link" href="#m68k-*-uclinux"> ¶</a></span></h3>
769 <p>GCC 4.3 changed the uClinux configuration so that it uses the
770 ‘<samp class="samp">m68k-linux-gnu</samp>’ ABI rather than the ‘<samp class="samp">m68k-elf</samp>’ ABI.
771 It also added improved support for C++ and flat shared libraries,
772 both of which were ABI changes.
773 </p>
774 <hr />
775 <a class="anchor" id="microblaze-x-elf"></a><h3 class="heading" id="microblaze-*-elf"><span>microblaze-*-elf<a class="copiable-link" href="#microblaze-*-elf"> ¶</a></span></h3>
776 <p>Xilinx MicroBlaze processor.
777 This configuration is intended for embedded systems.
778 </p>
779 <hr />
780 <a class="anchor" id="mips-x-x"></a><h3 class="heading" id="mips-*-*"><span>mips-*-*<a class="copiable-link" href="#mips-*-*"> ¶</a></span></h3>
781 <p>If on a MIPS system you get an error message saying “does not have gp
782 sections for all it’s [sic] sectons [sic]”, don’t worry about it. This
783 happens whenever you use GAS with the MIPS linker, but there is not
784 really anything wrong, and it is okay to use the output file. You can
785 stop such warnings by installing the GNU linker.
786 </p>
787 <p>It would be nice to extend GAS to produce the gp tables, but they are
788 optional, and there should not be a warning about their absence.
789 </p>
790 <p>The libstdc++ atomic locking routines for MIPS targets requires MIPS II
791 and later. A patch went in just after the GCC 3.3 release to
792 make ‘<samp class="samp">mips*-*-*</samp>’ use the generic implementation instead. You can also
793 configure for ‘<samp class="samp">mipsel-elf</samp>’ as a workaround. The
794 ‘<samp class="samp">mips*-*-linux*</samp>’ target continues to use the MIPS II routines. More
795 work on this is expected in future releases.
796 </p>
797
798 <p>The built-in <code class="code">__sync_*</code> functions are available on MIPS II and
799 later systems and others that support the ‘<samp class="samp">ll</samp>’, ‘<samp class="samp">sc</samp>’ and
800 ‘<samp class="samp">sync</samp>’ instructions. This can be overridden by passing
801 <samp class="option">--with-llsc</samp> or <samp class="option">--without-llsc</samp> when configuring GCC.
802 Since the Linux kernel emulates these instructions if they are
803 missing, the default for ‘<samp class="samp">mips*-*-linux*</samp>’ targets is
804 <samp class="option">--with-llsc</samp>. The <samp class="option">--with-llsc</samp> and
805 <samp class="option">--without-llsc</samp> configure options may be overridden at compile
806 time by passing the <samp class="option">-mllsc</samp> or <samp class="option">-mno-llsc</samp> options to
807 the compiler.
808 </p>
809 <p>MIPS systems check for division by zero (unless
810 <samp class="option">-mno-check-zero-division</samp> is passed to the compiler) by
811 generating either a conditional trap or a break instruction. Using
812 trap results in smaller code, but is only supported on MIPS II and
813 later. Also, some versions of the Linux kernel have a bug that
814 prevents trap from generating the proper signal (<code class="code">SIGFPE</code>). To enable
815 the use of break, use the <samp class="option">--with-divide=breaks</samp>
816 <code class="command">configure</code> option when configuring GCC. The default is to
817 use traps on systems that support them.
818 </p>
819 <hr />
820 <a class="anchor" id="moxie-x-elf"></a><h3 class="heading" id="moxie-*-elf"><span>moxie-*-elf<a class="copiable-link" href="#moxie-*-elf"> ¶</a></span></h3>
821 <p>The moxie processor.
822 </p>
823 <hr />
824 <a class="anchor" id="msp430-x-elf"></a><h3 class="heading" id="msp430-*-elf*"><span>msp430-*-elf*<a class="copiable-link" href="#msp430-*-elf*"> ¶</a></span></h3>
825 <p>TI MSP430 processor.
826 This configuration is intended for embedded systems.
827 </p>
828 <p>‘<samp class="samp">msp430-*-elf</samp>’ is the standard configuration with most GCC
829 features enabled by default.
830 </p>
831 <p>‘<samp class="samp">msp430-*-elfbare</samp>’ is tuned for a bare-metal environment, and disables
832 features related to shared libraries and other functionality not used for
833 this device. This reduces code and data usage of the GCC libraries, resulting
834 in a minimal run-time environment by default.
835 </p>
836 <p>Features disabled by default include:
837 </p><ul class="itemize mark-bullet">
838 <li>transactional memory
839 </li><li>__cxa_atexit
840 </li></ul>
841
842 <hr />
843 <a class="anchor" id="nds32le-x-elf"></a><h3 class="heading" id="nds32le-*-elf"><span>nds32le-*-elf<a class="copiable-link" href="#nds32le-*-elf"> ¶</a></span></h3>
844 <p>Andes NDS32 target in little endian mode.
845 </p>
846 <hr />
847 <a class="anchor" id="nds32be-x-elf"></a><h3 class="heading" id="nds32be-*-elf"><span>nds32be-*-elf<a class="copiable-link" href="#nds32be-*-elf"> ¶</a></span></h3>
848 <p>Andes NDS32 target in big endian mode.
849 </p>
850 <hr />
851 <a class="anchor" id="nvptx-x-none"></a><h3 class="heading" id="nvptx-*-none"><span>nvptx-*-none<a class="copiable-link" href="#nvptx-*-none"> ¶</a></span></h3>
852 <p>Nvidia PTX target.
853 </p>
854 <p>Instead of GNU binutils, you will need to install
855 <a class="uref" href="https://github.com/SourceryTools/nvptx-tools">nvptx-tools</a>.
856 Tell GCC where to find it:
857 <samp class="option">--with-build-time-tools=[install-nvptx-tools]/nvptx-none/bin</samp>.
858 </p>
859 <p>You will need newlib 4.3.0 or later. It can be
860 automatically built together with GCC. For this, add a symbolic link
861 to nvptx-newlib’s <samp class="file">newlib</samp> directory to the directory containing
862 the GCC sources.
863 </p>
864 <p>Use the <samp class="option">--disable-sjlj-exceptions</samp> and
865 <samp class="option">--enable-newlib-io-long-long</samp> options when configuring.
866 </p>
867 <p>The <samp class="option">--with-arch</samp> option may be specified to override the
868 default value for the <samp class="option">-march</samp> option, and to also build
869 corresponding target libraries.
870 The default is <samp class="option">--with-arch=sm_30</samp>.
871 </p>
872 <p>For example, if <samp class="option">--with-arch=sm_70</samp> is specified,
873 <samp class="option">-march=sm_30</samp> and <samp class="option">-march=sm_70</samp> target libraries are
874 built, and code generation defaults to <samp class="option">-march=sm_70</samp>.
875 </p>
876 <hr />
877 <a class="anchor" id="or1k-x-elf"></a><h3 class="heading" id="or1k-*-elf"><span>or1k-*-elf<a class="copiable-link" href="#or1k-*-elf"> ¶</a></span></h3>
878 <p>The OpenRISC 1000 32-bit processor with delay slots.
879 This configuration is intended for embedded systems.
880 </p>
881 <hr />
882 <a class="anchor" id="or1k-x-linux"></a><h3 class="heading" id="or1k-*-linux"><span>or1k-*-linux<a class="copiable-link" href="#or1k-*-linux"> ¶</a></span></h3>
883 <p>The OpenRISC 1000 32-bit processor with delay slots.
884 </p>
885 <hr />
886 <a class="anchor" id="powerpc-x-x"></a><h3 class="heading" id="powerpc-*-*"><span>powerpc-*-*<a class="copiable-link" href="#powerpc-*-*"> ¶</a></span></h3>
887 <p>You can specify a default version for the <samp class="option">-mcpu=<var class="var">cpu_type</var></samp>
888 switch by using the configure option <samp class="option">--with-cpu-<var class="var">cpu_type</var></samp>.
889 </p>
890 <p>You will need GNU binutils 2.20 or newer.
891 </p>
892 <hr />
893 <a class="anchor" id="powerpc-x-darwin"></a><h3 class="heading" id="powerpc-*-darwin*"><span>powerpc-*-darwin*<a class="copiable-link" href="#powerpc-*-darwin*"> ¶</a></span></h3>
894 <p>PowerPC running Darwin (Mac OS X kernel).
895 </p>
896 <p>Pre-installed versions of Mac OS X may not include any developer tools,
897 meaning that you will not be able to build GCC from source. Tool
898 binaries are available at
899 <a class="uref" href="https://opensource.apple.com">https://opensource.apple.com</a>.
900 </p>
901 <p>This version of GCC requires at least cctools-590.36. The
902 cctools-590.36 package referenced from
903 <a class="uref" href="https://gcc.gnu.org/ml/gcc/2006-03/msg00507.html">https://gcc.gnu.org/ml/gcc/2006-03/msg00507.html</a> will not work
904 on systems older than 10.3.9 (aka darwin7.9.0).
905 </p>
906 <hr />
907 <a class="anchor" id="powerpc-x-elf"></a><h3 class="heading" id="powerpc-*-elf"><span>powerpc-*-elf<a class="copiable-link" href="#powerpc-*-elf"> ¶</a></span></h3>
908 <p>PowerPC system in big endian mode, running System V.4.
909 </p>
910 <hr />
911 <a class="anchor" id="powerpc-x-linux-gnu"></a><h3 class="heading" id="powerpc*-*-linux-gnu*"><span>powerpc*-*-linux-gnu*<a class="copiable-link" href="#powerpc*-*-linux-gnu*"> ¶</a></span></h3>
912 <p>PowerPC system in big endian mode running Linux.
913 </p>
914 <hr />
915 <a class="anchor" id="powerpc-x-netbsd"></a><h3 class="heading" id="powerpc-*-netbsd*"><span>powerpc-*-netbsd*<a class="copiable-link" href="#powerpc-*-netbsd*"> ¶</a></span></h3>
916 <p>PowerPC system in big endian mode running NetBSD.
917 </p>
918 <hr />
919 <a class="anchor" id="powerpc-x-eabisim"></a><h3 class="heading" id="powerpc-*-eabisim"><span>powerpc-*-eabisim<a class="copiable-link" href="#powerpc-*-eabisim"> ¶</a></span></h3>
920 <p>Embedded PowerPC system in big endian mode for use in running under the
921 PSIM simulator.
922 </p>
923 <hr />
924 <a class="anchor" id="powerpc-x-eabi"></a><h3 class="heading" id="powerpc-*-eabi"><span>powerpc-*-eabi<a class="copiable-link" href="#powerpc-*-eabi"> ¶</a></span></h3>
925 <p>Embedded PowerPC system in big endian mode.
926 </p>
927 <hr />
928 <a class="anchor" id="powerpcle-x-elf"></a><h3 class="heading" id="powerpcle-*-elf"><span>powerpcle-*-elf<a class="copiable-link" href="#powerpcle-*-elf"> ¶</a></span></h3>
929 <p>PowerPC system in little endian mode, running System V.4.
930 </p>
931 <hr />
932 <a class="anchor" id="powerpcle-x-eabisim"></a><h3 class="heading" id="powerpcle-*-eabisim"><span>powerpcle-*-eabisim<a class="copiable-link" href="#powerpcle-*-eabisim"> ¶</a></span></h3>
933 <p>Embedded PowerPC system in little endian mode for use in running under
934 the PSIM simulator.
935 </p>
936 <hr />
937 <a class="anchor" id="powerpcle-x-eabi"></a><h3 class="heading" id="powerpcle-*-eabi"><span>powerpcle-*-eabi<a class="copiable-link" href="#powerpcle-*-eabi"> ¶</a></span></h3>
938 <p>Embedded PowerPC system in little endian mode.
939 </p>
940 <hr />
941 <a class="anchor" id="rl78-x-elf"></a><h3 class="heading" id="rl78-*-elf"><span>rl78-*-elf<a class="copiable-link" href="#rl78-*-elf"> ¶</a></span></h3>
942 <p>The Renesas RL78 processor.
943 This configuration is intended for embedded systems.
944 </p>
945 <hr />
946 <a class="anchor" id="riscv32-x-elf"></a><h3 class="heading" id="riscv32-*-elf"><span>riscv32-*-elf<a class="copiable-link" href="#riscv32-*-elf"> ¶</a></span></h3>
947 <p>The RISC-V RV32 instruction set.
948 This configuration is intended for embedded systems.
949 This (and all other RISC-V) targets require the binutils 2.30 release.
950 </p>
951 <hr />
952 <a class="anchor" id="riscv32-x-linux"></a><h3 class="heading" id="riscv32-*-linux"><span>riscv32-*-linux<a class="copiable-link" href="#riscv32-*-linux"> ¶</a></span></h3>
953 <p>The RISC-V RV32 instruction set running GNU/Linux.
954 This (and all other RISC-V) targets require the binutils 2.30 release.
955 </p>
956 <hr />
957 <a class="anchor" id="riscv64-x-elf"></a><h3 class="heading" id="riscv64-*-elf"><span>riscv64-*-elf<a class="copiable-link" href="#riscv64-*-elf"> ¶</a></span></h3>
958 <p>The RISC-V RV64 instruction set.
959 This configuration is intended for embedded systems.
960 This (and all other RISC-V) targets require the binutils 2.30 release.
961 </p>
962 <hr />
963 <a class="anchor" id="riscv64-x-linux"></a><h3 class="heading" id="riscv64-*-linux"><span>riscv64-*-linux<a class="copiable-link" href="#riscv64-*-linux"> ¶</a></span></h3>
964 <p>The RISC-V RV64 instruction set running GNU/Linux.
965 This (and all other RISC-V) targets require the binutils 2.30 release.
966 </p>
967 <hr />
968 <a class="anchor" id="rx-x-elf"></a><h3 class="heading" id="rx-*-elf"><span>rx-*-elf<a class="copiable-link" href="#rx-*-elf"> ¶</a></span></h3>
969 <p>The Renesas RX processor.
970 </p>
971 <hr />
972 <a class="anchor" id="s390-x-linux"></a><h3 class="heading" id="s390-*-linux*"><span>s390-*-linux*<a class="copiable-link" href="#s390-*-linux*"> ¶</a></span></h3>
973 <p>S/390 system running GNU/Linux for S/390.
974 </p>
975 <hr />
976 <a class="anchor" id="s390x-x-linux"></a><h3 class="heading" id="s390x-*-linux*"><span>s390x-*-linux*<a class="copiable-link" href="#s390x-*-linux*"> ¶</a></span></h3>
977 <p>zSeries system (64-bit) running GNU/Linux for zSeries.
978 </p>
979 <hr />
980 <a class="anchor" id="s390x-ibm-tpf"></a><h3 class="heading" id="s390x-ibm-tpf*"><span>s390x-ibm-tpf*<a class="copiable-link" href="#s390x-ibm-tpf*"> ¶</a></span></h3>
981 <p>zSeries system (64-bit) running TPF. This platform is
982 supported as cross-compilation target only.
983 </p>
984 <hr />
985 <a class="anchor" id="x-x-solaris2"></a><h3 class="heading" id="g_t*-*-solaris2*"><span>*-*-solaris2*<a class="copiable-link" href="#g_t*-*-solaris2*"> ¶</a></span></h3>
986 <p>Support for Solaris 11.3 and earlier has been obsoleted in GCC 13, but
987 can still be enabled by configuring with <samp class="option">--enable-obsolete</samp>.
988 Support for Solaris 10 has been removed in GCC 10. Support for Solaris
989 9 has been removed in GCC 5. Support for Solaris 8 has been removed in
990 GCC 4.8. Support for Solaris 7 has been removed in GCC 4.6.
991 </p>
992 <p>Solaris 11.3 provides GCC 4.5.2, 4.7.3, and 4.8.2 as
993 <code class="command">/usr/gcc/4.5/bin/gcc</code> or similar. Solaris 11.4
994 provides one or more of GCC 5, 7, 9, 10, 11, and 12.
995 </p>
996 <p>You need to install the <code class="code">system/header</code>, <code class="code">system/linker</code>, and
997 <code class="code">developer/assembler</code> packages.
998 </p>
999 <p>Trying to use the compatibility tools in <samp class="file">/usr/ucb</samp>, from the
1000 <code class="code">compatibility/ucb</code> package, to install GCC has been observed to
1001 cause trouble. The fix is to remove <samp class="file">/usr/ucb</samp> from your
1002 <code class="env">PATH</code>.
1003 </p>
1004 <p>The build process works more smoothly with the legacy Solaris tools so,
1005 if you have <samp class="file">/usr/xpg4/bin</samp> in your <code class="env">PATH</code>, we recommend that
1006 you place <samp class="file">/usr/bin</samp> before <samp class="file">/usr/xpg4/bin</samp> for the duration
1007 of the build.
1008 </p>
1009 <p>We recommend the use of the Solaris assembler or the GNU assembler, in
1010 conjunction with the Solaris linker.
1011 </p>
1012 <p>The GNU <code class="command">as</code> versions included in Solaris 11.3, from GNU
1013 binutils 2.23.1 or newer (in <samp class="file">/usr/bin/gas</samp> and
1014 <samp class="file">/usr/gnu/bin/as</samp>), are known to work. The version from GNU
1015 binutils 2.40 is known to work as well. Recent versions of the Solaris
1016 assembler in <samp class="file">/usr/bin/as</samp> work almost as well, though. To use GNU
1017 <code class="command">as</code>, configure with the options <samp class="option">--with-gnu-as
1018 --with-as=/usr/gnu/bin/as</samp>.
1019 </p>
1020 <p>For linking, the Solaris linker is preferred. If you want to use the
1021 GNU linker instead, the version in Solaris 11.3, from GNU binutils
1022 2.23.1 or newer (in <samp class="file">/usr/gnu/bin/ld</samp> and <samp class="file">/usr/bin/gld</samp>),
1023 works, as does the version from GNU binutils 2.40. However, it
1024 generally lacks platform specific features, so better stay with Solaris
1025 <code class="command">ld</code>. To use the LTO linker plugin
1026 (<samp class="option">-fuse-linker-plugin</samp>) with GNU <code class="command">ld</code>, GNU binutils
1027 <em class="emph">must</em> be configured with <samp class="option">--enable-largefile</samp>. To use
1028 Solaris <code class="command">ld</code>, we recommend to configure with
1029 <samp class="option">--without-gnu-ld --with-ld=/usr/bin/ld</samp> to guarantee the
1030 right linker is found irrespective of the user’s <code class="env">PATH</code>.
1031 </p>
1032 <p>Note that your mileage may vary if you use a combination of the GNU
1033 tools and the Solaris tools: while the combination GNU <code class="command">as</code> and
1034 Solaris <code class="command">ld</code> works well, the reverse combination Solaris
1035 <code class="command">as</code> with GNU <code class="command">ld</code> may fail to build or cause memory
1036 corruption at runtime in some cases for C++ programs.
1037 </p>
1038 <p>To enable symbol versioning in ‘<samp class="samp">libstdc++</samp>’ and other runtime
1039 libraries with the Solaris linker, you need to have any version of GNU
1040 <code class="command">c++filt</code>, which is part of GNU binutils. Symbol versioning
1041 will be disabled if no appropriate version is found. Solaris
1042 <code class="command">c++filt</code> from the Solaris Studio compilers does <em class="emph">not</em>
1043 work.
1044 </p>
1045 <p>In order to build the GNU Ada compiler, GNAT, a working GNAT is needed.
1046 Since Solaris 11.4 SRU 39, GNAT 11 or 12 is bundled in the
1047 <code class="code">developer/gcc/gcc-gnat</code> package.
1048 </p>
1049 <p>In order to build the GNU D compiler, GDC, a working ‘<samp class="samp">libphobos</samp>’ is
1050 needed. That library wasn’t built by default in GCC 9–11 on SPARC, or
1051 on x86 when the Solaris assembler is used, but can be enabled by
1052 configuring with <samp class="option">--enable-libphobos</samp>. Also, GDC 9.4.0 is
1053 required on x86, while GDC 9.3.0 is known to work on SPARC.
1054 </p>
1055 <p>The versions of the GNU Multiple Precision Library (GMP), the MPFR
1056 library and the MPC library bundled with Solaris 11.3 and later are
1057 usually recent enough to match GCC’s requirements. There are two
1058 caveats:
1059 </p>
1060 <ul class="itemize mark-bullet">
1061 <li>While the version of the GMP library in Solaris 11.3 works with GCC, you
1062 need to configure with <samp class="option">--with-gmp-include=/usr/include/gmp</samp>.
1063
1064 </li><li>The version of the MPFR libary included in Solaris 11.3 is too old; you
1065 need to provide a more recent one.
1066
1067 </li></ul>
1068
1069 <hr />
1070 <a class="anchor" id="sparc-x-x"></a><h3 class="heading" id="sparc*-*-*"><span>sparc*-*-*<a class="copiable-link" href="#sparc*-*-*"> ¶</a></span></h3>
1071 <p>This section contains general configuration information for all
1072 SPARC-based platforms. In addition to reading this section, please
1073 read all other sections that match your target.
1074 </p>
1075 <p>Newer versions of the GNU Multiple Precision Library (GMP), the MPFR
1076 library and the MPC library are known to be miscompiled by earlier
1077 versions of GCC on these platforms. We therefore recommend the use
1078 of the exact versions of these libraries listed as minimal versions
1079 in <a class="uref" href="prerequisites.html">the prerequisites</a>.
1080 </p>
1081 <hr />
1082 <a class="anchor" id="sparc-sun-solaris2"></a><h3 class="heading" id="sparc-sun-solaris2*"><span>sparc-sun-solaris2*<a class="copiable-link" href="#sparc-sun-solaris2*"> ¶</a></span></h3>
1083 <p>When GCC is configured to use GNU binutils 2.14 or later, the binaries
1084 produced are smaller than the ones produced using Solaris native tools;
1085 this difference is quite significant for binaries containing debugging
1086 information.
1087 </p>
1088 <p>Starting with Solaris 7, the operating system is capable of executing
1089 64-bit SPARC V9 binaries. GCC 3.1 and later properly supports
1090 this; the <samp class="option">-m64</samp> option enables 64-bit code generation.
1091 </p>
1092 <p>When configuring the GNU Multiple Precision Library (GMP), the MPFR
1093 library or the MPC library on Solaris, the canonical target triplet must
1094 be specified as the <code class="command">build</code> parameter on the <code class="command">configure</code>
1095 line. This target triplet can be obtained by invoking
1096 <code class="command">./config.guess</code> in the toplevel source directory of GCC (and
1097 not that of GMP or MPFR or MPC). For example:
1098 </p>
1099 <div class="example smallexample">
1100 <pre class="example-preformatted">% <var class="var">srcdir</var>/configure --build=sparc-sun-solaris2.11 --prefix=<var class="var">dirname</var>
1101 </pre></div>
1102
1103 <hr />
1104 <a class="anchor" id="sparc-x-linux"></a><h3 class="heading" id="sparc-*-linux*"><span>sparc-*-linux*<a class="copiable-link" href="#sparc-*-linux*"> ¶</a></span></h3>
1105
1106 <hr />
1107 <a class="anchor" id="sparc64-x-solaris2"></a><h3 class="heading" id="sparc64-*-solaris2*"><span>sparc64-*-solaris2*<a class="copiable-link" href="#sparc64-*-solaris2*"> ¶</a></span></h3>
1108 <p>This is a synonym for ‘<samp class="samp">sparcv9-*-solaris2*</samp>’.
1109 </p>
1110 <hr />
1111 <a class="anchor" id="sparcv9-x-solaris2"></a><h3 class="heading" id="sparcv9-*-solaris2*"><span>sparcv9-*-solaris2*<a class="copiable-link" href="#sparcv9-*-solaris2*"> ¶</a></span></h3>
1112
1113 <p>When configuring a 64-bit-default GCC on Solaris/SPARC, you must use a
1114 build compiler that generates 64-bit code, either by default or by
1115 specifying ‘<samp class="samp">CC='gcc -m64' CXX='g++ -m64' GDC='gdc -m64'</samp>’ to <code class="command">configure</code>.
1116 Additionally, you <em class="emph">must</em> pass <samp class="option">--build=sparcv9-sun-solaris2.11</samp>
1117 or <samp class="option">--build=sparc64-sun-solaris2.11</samp> because <samp class="file">config.guess</samp>
1118 misdetects this situation, which can cause build failures.
1119 </p>
1120 <p>When configuring the GNU Multiple Precision Library (GMP), the MPFR
1121 library or the MPC library, the canonical target triplet must be specified
1122 as the <code class="command">build</code> parameter on the <code class="command">configure</code> line. For example:
1123 </p>
1124 <div class="example smallexample">
1125 <pre class="example-preformatted">% <var class="var">srcdir</var>/configure --build=sparcv9-sun-solaris2.11 --prefix=<var class="var">dirname</var>
1126 </pre></div>
1127
1128 <hr />
1129 <a class="anchor" id="c6x-x-x"></a><h3 class="heading" id="c6x-*-*"><span>c6x-*-*<a class="copiable-link" href="#c6x-*-*"> ¶</a></span></h3>
1130 <p>The C6X family of processors. This port requires binutils-2.22 or newer.
1131 </p>
1132 <hr />
1133 <a class="anchor" id="visium-x-elf"></a><h3 class="heading" id="visium-*-elf"><span>visium-*-elf<a class="copiable-link" href="#visium-*-elf"> ¶</a></span></h3>
1134 <p>CDS VISIUMcore processor.
1135 This configuration is intended for embedded systems.
1136 </p>
1137 <hr />
1138 <a class="anchor" id="x-x-vxworks"></a><h3 class="heading" id="g_t*-*-vxworks*"><span>*-*-vxworks*<a class="copiable-link" href="#g_t*-*-vxworks*"> ¶</a></span></h3>
1139 <p>Support for VxWorks is in flux. At present GCC supports <em class="emph">only</em> the
1140 very recent VxWorks 5.5 (aka Tornado 2.2) release, and only on PowerPC.
1141 We welcome patches for other architectures supported by VxWorks 5.5.
1142 Support for VxWorks AE would also be welcome; we believe this is merely
1143 a matter of writing an appropriate “configlette” (see below). We are
1144 not interested in supporting older, a.out or COFF-based, versions of
1145 VxWorks in GCC 3.
1146 </p>
1147 <p>VxWorks comes with an older version of GCC installed in
1148 <samp class="file"><var class="var">$WIND_BASE</var>/host</samp>; we recommend you do not overwrite it.
1149 Choose an installation <var class="var">prefix</var> entirely outside <var class="var">$WIND_BASE</var>.
1150 Before running <code class="command">configure</code>, create the directories <samp class="file"><var class="var">prefix</var></samp>
1151 and <samp class="file"><var class="var">prefix</var>/bin</samp>. Link or copy the appropriate assembler,
1152 linker, etc. into <samp class="file"><var class="var">prefix</var>/bin</samp>, and set your <var class="var">PATH</var> to
1153 include that directory while running both <code class="command">configure</code> and
1154 <code class="command">make</code>.
1155 </p>
1156 <p>You must give <code class="command">configure</code> the
1157 <samp class="option">--with-headers=<var class="var">$WIND_BASE</var>/target/h</samp> switch so that it can
1158 find the VxWorks system headers. Since VxWorks is a cross compilation
1159 target only, you must also specify <samp class="option">--target=<var class="var">target</var></samp>.
1160 <code class="command">configure</code> will attempt to create the directory
1161 <samp class="file"><var class="var">prefix</var>/<var class="var">target</var>/sys-include</samp> and copy files into it;
1162 make sure the user running <code class="command">configure</code> has sufficient privilege
1163 to do so.
1164 </p>
1165 <p>GCC’s exception handling runtime requires a special “configlette”
1166 module, <samp class="file">contrib/gthr_supp_vxw_5x.c</samp>. Follow the instructions in
1167 that file to add the module to your kernel build. (Future versions of
1168 VxWorks will incorporate this module.)
1169 </p>
1170 <hr />
1171 <a class="anchor" id="x86-64-x-x"></a><h3 class="heading" id="x86_005f64-*-*_002c-amd64-*-*"><span>x86_64-*-*, amd64-*-*<a class="copiable-link" href="#x86_005f64-*-*_002c-amd64-*-*"> ¶</a></span></h3>
1172 <p>GCC supports the x86-64 architecture implemented by the AMD64 processor
1173 (amd64-*-* is an alias for x86_64-*-*) on GNU/Linux, FreeBSD and NetBSD.
1174 On GNU/Linux the default is a bi-arch compiler which is able to generate
1175 both 64-bit x86-64 and 32-bit x86 code (via the <samp class="option">-m32</samp> switch).
1176 </p>
1177 <hr />
1178 <a class="anchor" id="x86-64-x-solaris2"></a><h3 class="heading" id="x86_005f64-*-solaris2*"><span>x86_64-*-solaris2*<a class="copiable-link" href="#x86_005f64-*-solaris2*"> ¶</a></span></h3>
1179 <p>GCC also supports the x86-64 architecture implemented by the AMD64
1180 processor (‘<samp class="samp">amd64-*-*</samp>’ is an alias for ‘<samp class="samp">x86_64-*-*</samp>’).
1181 Unlike other systems, without special options a
1182 bi-arch compiler is built which generates 32-bit code by default, but
1183 can generate 64-bit x86-64 code with the <samp class="option">-m64</samp> switch. Since
1184 GCC 4.7, there is also a configuration that defaults to 64-bit code, but
1185 can generate 32-bit code with <samp class="option">-m32</samp>. To configure and build
1186 this way, you have to provide all support libraries like <samp class="file">libgmp</samp>
1187 as 64-bit code, configure with <samp class="option">--target=x86_64-pc-solaris2.11</samp>
1188 and ‘<samp class="samp">CC=gcc -m64</samp>’.
1189 </p>
1190 <hr />
1191 <a class="anchor" id="xtensa-x-elf"></a><h3 class="heading" id="xtensa*-*-elf"><span>xtensa*-*-elf<a class="copiable-link" href="#xtensa*-*-elf"> ¶</a></span></h3>
1192 <p>This target is intended for embedded Xtensa systems using the
1193 ‘<samp class="samp">newlib</samp>’ C library. It uses ELF but does not support shared
1194 objects. Designed-defined instructions specified via the
1195 Tensilica Instruction Extension (TIE) language are only supported
1196 through inline assembly.
1197 </p>
1198 <p>The Xtensa configuration information must be specified prior to
1199 building GCC. The <samp class="file">include/xtensa-config.h</samp> header
1200 file contains the configuration information. If you created your
1201 own Xtensa configuration with the Xtensa Processor Generator, the
1202 downloaded files include a customized copy of this header file,
1203 which you can use to replace the default header file.
1204 </p>
1205 <hr />
1206 <a class="anchor" id="xtensa-x-linux"></a><h3 class="heading" id="xtensa*-*-linux*"><span>xtensa*-*-linux*<a class="copiable-link" href="#xtensa*-*-linux*"> ¶</a></span></h3>
1207 <p>This target is for Xtensa systems running GNU/Linux. It supports ELF
1208 shared objects and the GNU C library (glibc). It also generates
1209 position-independent code (PIC) regardless of whether the
1210 <samp class="option">-fpic</samp> or <samp class="option">-fPIC</samp> options are used. In other
1211 respects, this target is the same as the
1212 <a class="uref" href="#xtensa*-*-elf">‘<samp class="samp">xtensa*-*-elf</samp>’</a> target.
1213 </p>
1214 <hr />
1215 <a class="anchor" id="windows"></a><h3 class="heading" id="Microsoft-Windows"><span>Microsoft Windows<a class="copiable-link" href="#Microsoft-Windows"> ¶</a></span></h3>
1216
1217 <h4 class="subheading" id="Intel-16-bit-versions"><span>Intel 16-bit versions<a class="copiable-link" href="#Intel-16-bit-versions"> ¶</a></span></h4>
1218 <p>The 16-bit versions of Microsoft Windows, such as Windows 3.1, are not
1219 supported.
1220 </p>
1221 <p>However, the 32-bit port has limited support for Microsoft
1222 Windows 3.11 in the Win32s environment, as a target only. See below.
1223 </p>
1224 <h4 class="subheading" id="Intel-32-bit-versions"><span>Intel 32-bit versions<a class="copiable-link" href="#Intel-32-bit-versions"> ¶</a></span></h4>
1225 <p>The 32-bit versions of Windows, including Windows 95, Windows NT, Windows
1226 XP, and Windows Vista, are supported by several different target
1227 platforms. These targets differ in which Windows subsystem they target
1228 and which C libraries are used.
1229 </p>
1230 <ul class="itemize mark-bullet">
1231 <li>Cygwin <a class="uref" href="#x-x-cygwin">*-*-cygwin</a>: Cygwin provides a user-space
1232 Linux API emulation layer in the Win32 subsystem.
1233 </li><li>MinGW <a class="uref" href="#x-x-mingw32">*-*-mingw32</a>: MinGW is a native GCC port for
1234 the Win32 subsystem that provides a subset of POSIX.
1235 </li><li>MKS i386-pc-mks: NuTCracker from MKS. See
1236 <a class="uref" href="https://www.mkssoftware.com">https://www.mkssoftware.com</a> for more information.
1237 </li></ul>
1238
1239 <h4 class="subheading" id="Intel-64-bit-versions"><span>Intel 64-bit versions<a class="copiable-link" href="#Intel-64-bit-versions"> ¶</a></span></h4>
1240 <p>GCC contains support for x86-64 using the mingw-w64
1241 runtime library, available from <a class="uref" href="https://www.mingw-w64.org/downloads/">https://www.mingw-w64.org/downloads/</a>.
1242 This library should be used with the target triple x86_64-pc-mingw32.
1243 </p>
1244 <h4 class="subheading" id="Windows-CE"><span>Windows CE<a class="copiable-link" href="#Windows-CE"> ¶</a></span></h4>
1245 <p>Windows CE is supported as a target only on Hitachi
1246 SuperH (sh-wince-pe), and MIPS (mips-wince-pe).
1247 </p>
1248 <h4 class="subheading" id="Other-Windows-Platforms"><span>Other Windows Platforms<a class="copiable-link" href="#Other-Windows-Platforms"> ¶</a></span></h4>
1249 <p>GCC no longer supports Windows NT on the Alpha or PowerPC.
1250 </p>
1251 <p>GCC no longer supports the Windows POSIX subsystem. However, it does
1252 support the Interix subsystem. See above.
1253 </p>
1254 <p>Old target names including *-*-winnt and *-*-windowsnt are no longer used.
1255 </p>
1256 <p>UWIN support has been removed due to a lack of maintenance.
1257 </p>
1258 <hr />
1259 <a class="anchor" id="x-x-cygwin"></a><h3 class="heading" id="g_t*-*-cygwin"><span>*-*-cygwin<a class="copiable-link" href="#g_t*-*-cygwin"> ¶</a></span></h3>
1260 <p>Ports of GCC are included with the
1261 <a class="uref" href="http://www.cygwin.com/">Cygwin environment</a>.
1262 </p>
1263 <p>GCC will build under Cygwin without modification; it does not build
1264 with Microsoft’s C++ compiler and there are no plans to make it do so.
1265 </p>
1266 <p>The Cygwin native compiler can be configured to target any 32-bit x86
1267 cpu architecture desired; the default is i686-pc-cygwin. It should be
1268 used with as up-to-date a version of binutils as possible; use either
1269 the latest official GNU binutils release in the Cygwin distribution,
1270 or version 2.20 or above if building your own.
1271 </p>
1272 <hr />
1273 <a class="anchor" id="x-x-mingw32"></a><h3 class="heading" id="g_t*-*-mingw32"><span>*-*-mingw32<a class="copiable-link" href="#g_t*-*-mingw32"> ¶</a></span></h3>
1274 <p>GCC will build with and support only MinGW runtime 3.12 and later.
1275 Earlier versions of headers are incompatible with the new default semantics
1276 of <code class="code">extern inline</code> in <code class="code">-std=c99</code> and <code class="code">-std=gnu99</code> modes.
1277 </p>
1278 <p>To support emitting DWARF debugging info you need to use GNU binutils
1279 version 2.16 or above containing support for the <code class="code">.secrel32</code>
1280 assembler pseudo-op.
1281 </p>
1282 <hr />
1283 <a class="anchor" id="older"></a><h3 class="heading" id="Older-systems"><span>Older systems<a class="copiable-link" href="#Older-systems"> ¶</a></span></h3>
1284 <p>GCC contains support files for many older (1980s and early
1285 1990s) Unix variants. For the most part, support for these systems
1286 has not been deliberately removed, but it has not been maintained for
1287 several years and may suffer from bitrot.
1288 </p>
1289 <p>Starting with GCC 3.1, each release has a list of “obsoleted” systems.
1290 Support for these systems is still present in that release, but
1291 <code class="command">configure</code> will fail unless the <samp class="option">--enable-obsolete</samp>
1292 option is given. Unless a maintainer steps forward, support for these
1293 systems will be removed from the next release of GCC.
1294 </p>
1295 <p>Support for old systems as hosts for GCC can cause problems if the
1296 workarounds for compiler, library and operating system bugs affect the
1297 cleanliness or maintainability of the rest of GCC. In some cases, to
1298 bring GCC up on such a system, if still possible with current GCC, may
1299 require first installing an old version of GCC which did work on that
1300 system, and using it to compile a more recent GCC, to avoid bugs in the
1301 vendor compiler. Old releases of GCC 1 and GCC 2 are available in the
1302 <samp class="file">old-releases</samp> directory on the <a class="uref" href="../mirrors.html">GCC mirror
1303 sites</a>. Header bugs may generally be avoided using
1304 <code class="command">fixincludes</code>, but bugs or deficiencies in libraries and the
1305 operating system may still cause problems.
1306 </p>
1307 <p>Support for older systems as targets for cross-compilation is less
1308 problematic than support for them as hosts for GCC; if an enthusiast
1309 wishes to make such a target work again (including resurrecting any of
1310 the targets that never worked with GCC 2, starting from the last
1311 version before they were removed), patches
1312 <a class="uref" href="../contribute.html">following the usual requirements</a> would be
1313 likely to be accepted, since they should not affect the support for more
1314 modern targets.
1315 </p>
1316 <p>For some systems, old versions of GNU binutils may also be useful,
1317 and are available from <samp class="file">pub/binutils/old-releases</samp> on
1318 <a class="uref" href="https://sourceware.org/mirrors.html">sourceware.org mirror sites</a>.
1319 </p>
1320 <p>Some of the information on specific systems above relates to
1321 such older systems, but much of the information
1322 about GCC on such systems (which may no longer be applicable to
1323 current GCC) is to be found in the GCC texinfo manual.
1324 </p>
1325 <hr />
1326 <a class="anchor" id="elf"></a><h3 class="heading" id="all-ELF-targets-_0028SVR4_002c-Solaris_002c-etc_002e_0029"><span>all ELF targets (SVR4, Solaris, etc.)<a class="copiable-link" href="#all-ELF-targets-_0028SVR4_002c-Solaris_002c-etc_002e_0029"> ¶</a></span></h3>
1327 <p>C++ support is significantly better on ELF targets if you use the
1328 <a class="uref" href="./configure.html#with-gnu-ld">GNU linker</a>; duplicate copies of
1329 inlines, vtables and template instantiations will be discarded
1330 automatically.
1331 </p>
1332
1333 <hr />
1334 <p>
1335 <p><a class="uref" href="./index.html">Return to the GCC Installation page</a>
1336 </p>
1337
1338
1339
1340
1341
1342 </body>
1343 </html>
1344