source: part3intro/toolchaintechnotes.xml@ a665a20

multilib trunk
Last change on this file since a665a20 was a665a20, checked in by Pierre Labastie <pierre.labastie@…>, 5 months ago

Remove spaces at end o lines - part3intro

  • Property mode set to 100644
File size: 17.0 KB
1<?xml version="1.0" encoding="ISO-8859-1"?>
2<!DOCTYPE sect1 PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
3 "" [
4 <!ENTITY % general-entities SYSTEM "../general.ent">
5 %general-entities;
8<sect1 id="ch-tools-toolchaintechnotes" xreflabel="Toolchain Technical Notes">
9 <?dbhtml filename="toolchaintechnotes.html"?>
11 <title>Toolchain Technical Notes</title>
13 <para>This section explains some of the rationale and technical details
14 behind the overall build method. It is not essential to immediately
15 understand everything in this section. Most of this information will be
16 clearer after performing an actual build. This section can be referred
17 to at any time during the process.</para>
19 <para>The overall goal of <xref linkend="chapter-cross-tools"/> and <xref
20 linkend="chapter-temporary-tools"/> is to produce a temporary area that
21 contains a known-good set of tools that can be isolated from the host system.
22 By using <command>chroot</command>, the commands in the remaining chapters
23 will be contained within that environment, ensuring a clean, trouble-free
24 build of the target LFS system. The build process has been designed to
25 minimize the risks for new readers and to provide the most educational value
26 at the same time.</para>
28 <para>The build process is based on the process of
29 <emphasis>cross-compilation</emphasis>. Cross-compilation is normally used
30 for building a compiler and its toolchain for a machine different from
31 the one that is used for the build. This is not strictly needed for LFS,
32 since the machine where the new system will run is the same as the one
33 used for the build. But cross-compilation has the great advantage that
34 anything that is cross-compiled cannot depend on the host environment.</para>
36 <sect2 id="cross-compile" xreflabel="About Cross-Compilation">
38 <title>About Cross-Compilation</title>
40 <para>Cross-compilation involves some concepts that deserve a section on
41 their own. Although this section may be omitted in a first reading, it
42 is strongly suggested to come back to it later in order to get a full
43 grasp of the build process.</para>
45 <para>Let us first define some terms used in this context:</para>
47 <variablelist>
48 <varlistentry><term>build</term><listitem>
49 <para>is the machine where we build programs. Note that this machine
50 is referred to as the <quote>host</quote> in other
51 sections.</para></listitem>
52 </varlistentry>
54 <varlistentry><term>host</term><listitem>
55 <para>is the machine/system where the built programs will run. Note
56 that this use of <quote>host</quote> is not the same as in other
57 sections.</para></listitem>
58 </varlistentry>
60 <varlistentry><term>target</term><listitem>
61 <para>is only used for compilers. It is the machine the compiler
62 produces code for. It may be different from both build and
63 host.</para></listitem>
64 </varlistentry>
66 </variablelist>
68 <para>As an example, let us imagine the following scenario (sometimes
69 referred to as <quote>Canadian Cross</quote>): we may have a
70 compiler on a slow machine only, let's call the machine A, and the compiler
71 ccA. We may have also a fast machine (B), but with no compiler, and we may
72 want to produce code for another slow machine (C). To build a
73 compiler for machine C, we would have three stages:</para>
75 <informaltable align="center">
76 <tgroup cols="5">
77 <colspec colnum="1" align="center"/>
78 <colspec colnum="2" align="center"/>
79 <colspec colnum="3" align="center"/>
80 <colspec colnum="4" align="center"/>
81 <colspec colnum="5" align="left"/>
82 <thead>
83 <row><entry>Stage</entry><entry>Build</entry><entry>Host</entry>
84 <entry>Target</entry><entry>Action</entry></row>
85 </thead>
86 <tbody>
87 <row>
88 <entry>1</entry><entry>A</entry><entry>A</entry><entry>B</entry>
89 <entry>build cross-compiler cc1 using ccA on machine A</entry>
90 </row>
91 <row>
92 <entry>2</entry><entry>A</entry><entry>B</entry><entry>C</entry>
93 <entry>build cross-compiler cc2 using cc1 on machine A</entry>
94 </row>
95 <row>
96 <entry>3</entry><entry>B</entry><entry>C</entry><entry>C</entry>
97 <entry>build compiler ccC using cc2 on machine B</entry>
98 </row>
99 </tbody>
100 </tgroup>
101 </informaltable>
103 <para>Then, all the other programs needed by machine C can be compiled
104 using cc2 on the fast machine B. Note that unless B can run programs
105 produced for C, there is no way to test the built programs until machine
106 C itself is running. For example, for testing ccC, we may want to add a
107 fourth stage:</para>
109 <informaltable align="center">
110 <tgroup cols="5">
111 <colspec colnum="1" align="center"/>
112 <colspec colnum="2" align="center"/>
113 <colspec colnum="3" align="center"/>
114 <colspec colnum="4" align="center"/>
115 <colspec colnum="5" align="left"/>
116 <thead>
117 <row><entry>Stage</entry><entry>Build</entry><entry>Host</entry>
118 <entry>Target</entry><entry>Action</entry></row>
119 </thead>
120 <tbody>
121 <row>
122 <entry>4</entry><entry>C</entry><entry>C</entry><entry>C</entry>
123 <entry>rebuild and test ccC using itself on machine C</entry>
124 </row>
125 </tbody>
126 </tgroup>
127 </informaltable>
129 <para>In the example above, only cc1 and cc2 are cross-compilers, that is,
130 they produce code for a machine different from the one they are run on.
131 The other compilers ccA and ccC produce code for the machine they are run
132 on. Such compilers are called <emphasis>native</emphasis> compilers.</para>
134 </sect2>
136 <sect2 id="lfs-cross">
137 <title>Implementation of Cross-Compilation for LFS</title>
139 <note>
140 <para>Almost all the build systems use names of the form
141 cpu-vendor-kernel-os referred to as the machine triplet. An astute
142 reader may wonder why a <quote>triplet</quote> refers to a four component
143 name. The reason is history: initially, three component names were enough
144 to designate unambiguously a machine, but with new machines and systems
145 appearing, that proved insufficient. The word <quote>triplet</quote>
146 remained. A simple way to determine your machine triplet is to run
147 the <command>config.guess</command>
148 script that comes with the source for many packages. Unpack the binutils
149 sources and run the script: <userinput>./config.guess</userinput> and note
150 the output. For example, for a 32-bit Intel processor the
151 output will be <emphasis>i686-pc-linux-gnu</emphasis>. On a 64-bit
152 system it will be <emphasis>x86_64-pc-linux-gnu</emphasis>.</para>
154 <para>Also be aware of the name of the platform's dynamic linker, often
155 referred to as the dynamic loader (not to be confused with the standard
156 linker <command>ld</command> that is part of binutils). The dynamic linker
157 provided by Glibc finds and loads the shared libraries needed by a
158 program, prepares the program to run, and then runs it. The name of the
159 dynamic linker for a 32-bit Intel machine will be <filename
160 class="libraryfile"></filename> (<filename
161 class="libraryfile"></filename> for 64-bit systems). A
162 sure-fire way to determine the name of the dynamic linker is to inspect a
163 random binary from the host system by running: <userinput>readelf -l
164 &lt;name of binary&gt; | grep interpreter</userinput> and noting the
165 output. The authoritative reference covering all platforms is in the
166 <filename>shlib-versions</filename> file in the root of the Glibc source
167 tree.</para>
168 </note>
170 <para>In order to fake a cross compilation, the name of the host triplet
171 is slightly adjusted by changing the &quot;vendor&quot; field in the
172 <envar>LFS_TGT</envar> variable. We also use the
173 <parameter>--with-sysroot</parameter> option when building the cross linker and
174 cross compiler to tell them where to find the needed host files. This
175 ensures that none of the other programs built in <xref
176 linkend="chapter-temporary-tools"/> can link to libraries on the build
177 machine. Only two stages are mandatory, and one more for tests:</para>
179 <informaltable align="center">
180 <tgroup cols="5">
181 <colspec colnum="1" align="center"/>
182 <colspec colnum="2" align="center"/>
183 <colspec colnum="3" align="center"/>
184 <colspec colnum="4" align="center"/>
185 <colspec colnum="5" align="left"/>
186 <thead>
187 <row><entry>Stage</entry><entry>Build</entry><entry>Host</entry>
188 <entry>Target</entry><entry>Action</entry></row>
189 </thead>
190 <tbody>
191 <row>
192 <entry>1</entry><entry>pc</entry><entry>pc</entry><entry>lfs</entry>
193 <entry>build cross-compiler cc1 using cc-pc on pc</entry>
194 </row>
195 <row>
196 <entry>2</entry><entry>pc</entry><entry>lfs</entry><entry>lfs</entry>
197 <entry>build compiler cc-lfs using cc1 on pc</entry>
198 </row>
199 <row>
200 <entry>3</entry><entry>lfs</entry><entry>lfs</entry><entry>lfs</entry>
201 <entry>rebuild and test cc-lfs using itself on lfs</entry>
202 </row>
203 </tbody>
204 </tgroup>
205 </informaltable>
207 <para>In the above table, <quote>on pc</quote> means the commands are run
208 on a machine using the already installed distribution. <quote>On
209 lfs</quote> means the commands are run in a chrooted environment.</para>
211 <para>Now, there is more about cross-compiling: the C language is not
212 just a compiler, but also defines a standard library. In this book, the
213 GNU C library, named glibc, is used. This library must
214 be compiled for the lfs machine, that is, using the cross compiler cc1.
215 But the compiler itself uses an internal library implementing complex
216 instructions not available in the assembler instruction set. This
217 internal library is named libgcc, and must be linked to the glibc
218 library to be fully functional! Furthermore, the standard library for
219 C++ (libstdc++) also needs being linked to glibc. The solution to this
220 chicken and egg problem is to first build a degraded cc1 based libgcc,
221 lacking some functionalities such as threads and exception handling, then
222 build glibc using this degraded compiler (glibc itself is not
223 degraded), then build libstdc++. But this last library will lack the
224 same functionalities as libgcc.</para>
226 <para>This is not the end of the story: the conclusion of the preceding
227 paragraph is that cc1 is unable to build a fully functional libstdc++, but
228 this is the only compiler available for building the C/C++ libraries
229 during stage 2! Of course, the compiler built during stage 2, cc-lfs,
230 would be able to build those libraries, but (1) the build system of
231 GCC does not know that it is usable on pc, and (2) using it on pc
232 would be at risk of linking to the pc libraries, since cc-lfs is a native
233 compiler. So we have to build libstdc++ later, in chroot.</para>
235 </sect2>
237 <sect2 id="other-details">
239 <title>Other procedural details</title>
241 <para>The cross-compiler will be installed in a separate <filename
242 class="directory">$LFS/tools</filename> directory, since it will not
243 be part of the final system.</para>
245 <para>Binutils is installed first because the <command>configure</command>
246 runs of both GCC and Glibc perform various feature tests on the assembler
247 and linker to determine which software features to enable or disable. This
248 is more important than one might first realize. An incorrectly configured
249 GCC or Glibc can result in a subtly broken toolchain, where the impact of
250 such breakage might not show up until near the end of the build of an
251 entire distribution. A test suite failure will usually highlight this error
252 before too much additional work is performed.</para>
254 <para>Binutils installs its assembler and linker in two locations,
255 <filename class="directory">$LFS/tools/bin</filename> and <filename
256 class="directory">$LFS/tools/$LFS_TGT/bin</filename>. The tools in one
257 location are hard linked to the other. An important facet of the linker is
258 its library search order. Detailed information can be obtained from
259 <command>ld</command> by passing it the <parameter>--verbose</parameter>
260 flag. For example, <command>$LFS_TGT-ld --verbose | grep SEARCH</command>
261 will illustrate the current search paths and their order. It shows which
262 files are linked by <command>ld</command> by compiling a dummy program and
263 passing the <parameter>--verbose</parameter> switch to the linker. For
264 example,
265 <command>$LFS_TGT-gcc dummy.c -Wl,--verbose 2&gt;&amp;1 | grep succeeded</command>
266 will show all the files successfully opened during the linking.</para>
268 <para>The next package installed is GCC. An example of what can be
269 seen during its run of <command>configure</command> is:</para>
271<screen><computeroutput>checking what assembler to use... /mnt/lfs/tools/i686-lfs-linux-gnu/bin/as
272checking what linker to use... /mnt/lfs/tools/i686-lfs-linux-gnu/bin/ld</computeroutput></screen>
274 <para>This is important for the reasons mentioned above. It also
275 demonstrates that GCC's configure script does not search the PATH
276 directories to find which tools to use. However, during the actual
277 operation of <command>gcc</command> itself, the same search paths are not
278 necessarily used. To find out which standard linker <command>gcc</command>
279 will use, run: <command>$LFS_TGT-gcc -print-prog-name=ld</command>.</para>
281 <para>Detailed information can be obtained from <command>gcc</command> by
282 passing it the <parameter>-v</parameter> command line option while compiling
283 a dummy program. For example, <command>gcc -v dummy.c</command> will show
284 detailed information about the preprocessor, compilation, and assembly
285 stages, including <command>gcc</command>'s included search paths and their
286 order.</para>
288 <para>Next installed are sanitized Linux API headers. These allow the
289 standard C library (Glibc) to interface with features that the Linux
290 kernel will provide.</para>
292 <para>The next package installed is Glibc. The most important
293 considerations for building Glibc are the compiler, binary tools, and
294 kernel headers. The compiler is generally not an issue since Glibc will
295 always use the compiler relating to the <parameter>--host</parameter>
296 parameter passed to its configure script; e.g. in our case, the compiler
297 will be <command>$LFS_TGT-gcc</command>. The binary tools and kernel
298 headers can be a bit more complicated. Therefore, take no risks and use
299 the available configure switches to enforce the correct selections. After
300 the run of <command>configure</command>, check the contents of the
301 <filename>config.make</filename> file in the <filename
302 class="directory">build</filename> directory for all important details.
303 Note the use of <parameter>CC="$LFS_TGT-gcc"</parameter> (with
304 <envar>$LFS_TGT</envar> expanded) to control which binary tools are used
305 and the use of the <parameter>-nostdinc</parameter> and
306 <parameter>-isystem</parameter> flags to control the compiler's include
307 search path. These items highlight an important aspect of the Glibc
308 package&mdash;it is very self-sufficient in terms of its build machinery
309 and generally does not rely on toolchain defaults.</para>
311 <para>As said above, the standard C++ library is compiled next, followed in
312 <xref linkend="chapter-temporary-tools"/> by all the programs that need
313 themselves to be built. The install step of all those packages uses the
314 <envar>DESTDIR</envar> variable to have the
315 programs land into the LFS filesystem.</para>
317 <para>At the end of <xref linkend="chapter-temporary-tools"/> the native
318 lfs compiler is installed. First binutils-pass2 is built,
319 with the same <envar>DESTDIR</envar> install as the other programs,
320 then the second pass of GCC is constructed, omitting libstdc++
321 and other non-important libraries. Due to some weird logic in GCC's
322 configure script, <envar>CC_FOR_TARGET</envar> ends up as
323 <command>cc</command> when the host is the same as the target, but is
324 different from the build system. This is why
325 <parameter>CC_FOR_TARGET=$LFS_TGT-gcc</parameter> is put explicitly into
326 the configure options.</para>
328 <para>Upon entering the chroot environment in <xref
329 linkend="chapter-chroot-temporary-tools"/>, the first task is to install
330 libstdc++. Then temporary installations of programs needed for the proper
331 operation of the toolchain are performed. From this point onwards, the
332 core toolchain is self-contained and self-hosted. In
333 <xref linkend="chapter-building-system"/>, final versions of all the
334 packages needed for a fully functional system are built, tested and
335 installed.</para>
337 </sect2>
Note: See TracBrowser for help on using the repository browser.