[673b0d8] | 1 | <?xml version="1.0" encoding="ISO-8859-1"?>
|
---|
[b06ca36] | 2 | <!DOCTYPE sect1 PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
|
---|
| 3 | "http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd" [
|
---|
[673b0d8] | 4 | <!ENTITY % general-entities SYSTEM "../general.ent">
|
---|
| 5 | %general-entities;
|
---|
| 6 | ]>
|
---|
[b28fd35] | 7 |
|
---|
[673b0d8] | 8 | <sect1 id="ch-tools-toolchaintechnotes">
|
---|
[b28fd35] | 9 | <?dbhtml filename="toolchaintechnotes.html"?>
|
---|
| 10 |
|
---|
| 11 | <title>Toolchain Technical Notes</title>
|
---|
| 12 |
|
---|
| 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 | back to at any time during the process.</para>
|
---|
| 18 |
|
---|
| 19 | <para>The overall goal of <xref linkend="chapter-temporary-tools"/> is to
|
---|
| 20 | provide a temporary environment that can be chrooted into and from which can be
|
---|
| 21 | produced a clean, trouble-free build of the target LFS system in <xref
|
---|
| 22 | linkend="chapter-building-system"/>. Along the way, we separate the new system
|
---|
| 23 | from the host system as much as possible, and in doing so, build a
|
---|
| 24 | self-contained and self-hosted toolchain. It should be noted that the build
|
---|
| 25 | process has been designed to minimize the risks for new readers and provide
|
---|
| 26 | maximum educational value at the same time.</para>
|
---|
| 27 |
|
---|
| 28 | <important>
|
---|
| 29 | <para>Before continuing, be aware of the name of the working platform,
|
---|
| 30 | often referred to as the target triplet. Many times, the target
|
---|
| 31 | triplet will probably be <emphasis>i686-pc-linux-gnu</emphasis>. A
|
---|
| 32 | simple way to determine the name of the target triplet is to run the
|
---|
| 33 | <command>config.guess</command> script that comes with the source for
|
---|
| 34 | many packages. Unpack the Binutils sources and run the script:
|
---|
| 35 | <userinput>./config.guess</userinput> and note the output.</para>
|
---|
| 36 |
|
---|
| 37 | <para>Also be aware of the name of the platform's dynamic linker, often
|
---|
| 38 | referred to as the dynamic loader (not to be confused with the standard
|
---|
| 39 | linker <command>ld</command> that is part of Binutils). The dynamic linker
|
---|
| 40 | provided by Glibc finds and loads the shared libraries needed by a program,
|
---|
| 41 | prepares the program to run, and then runs it. The name of the dynamic
|
---|
| 42 | linker will usually be <filename class="libraryfile">ld-linux.so.2</filename>.
|
---|
| 43 | On platforms that are less prevalent, the name might be <filename
|
---|
| 44 | class="libraryfile">ld.so.1</filename>, and newer 64 bit platforms might
|
---|
| 45 | be named something else entirely. The name of the platform's dynamic linker
|
---|
| 46 | can be determined by looking in the <filename class="directory">/lib</filename>
|
---|
| 47 | directory on the host system. A sure-fire way to determine the name is to
|
---|
| 48 | inspect a random binary from the host system by running:
|
---|
| 49 | <userinput>readelf -l <name of binary> | grep interpreter</userinput>
|
---|
| 50 | and noting the output. The authoritative reference covering all platforms
|
---|
| 51 | is in the <filename>shlib-versions</filename> file in the root of the Glibc
|
---|
| 52 | source tree.</para>
|
---|
| 53 | </important>
|
---|
| 54 |
|
---|
| 55 | <para>Some key technical points of how the <xref
|
---|
| 56 | linkend="chapter-temporary-tools"/> build method works:</para>
|
---|
| 57 |
|
---|
| 58 | <itemizedlist>
|
---|
| 59 | <listitem>
|
---|
| 60 | <para>The process is similar in principle to cross-compiling, whereby
|
---|
| 61 | tools installed in the same prefix work in cooperation, and thus utilize
|
---|
| 62 | a little GNU <quote>magic</quote></para>
|
---|
| 63 | </listitem>
|
---|
| 64 | <listitem>
|
---|
| 65 | <para>Careful manipulation of the standard linker's library search path
|
---|
| 66 | ensures programs are linked only against chosen libraries</para>
|
---|
| 67 | </listitem>
|
---|
| 68 | <listitem>
|
---|
| 69 | <para>Careful manipulation of <command>gcc</command>'s
|
---|
| 70 | <filename>specs</filename> file tells the compiler which target dynamic
|
---|
| 71 | linker will be used</para>
|
---|
| 72 | </listitem>
|
---|
| 73 | </itemizedlist>
|
---|
| 74 |
|
---|
| 75 | <para>Binutils is installed first because the <command>configure</command>
|
---|
| 76 | runs of both GCC and Glibc perform various feature tests on the assembler
|
---|
| 77 | and linker to determine which software features to enable or disable. This
|
---|
| 78 | is more important than one might first realize. An incorrectly configured
|
---|
| 79 | GCC or Glibc can result in a subtly broken toolchain, where the impact of
|
---|
| 80 | such breakage might not show up until near the end of the build of an
|
---|
| 81 | entire distribution. A test suite failure will usually highlight this error
|
---|
| 82 | before too much additional work is performed.</para>
|
---|
| 83 |
|
---|
| 84 | <para>Binutils installs its assembler and linker in two locations,
|
---|
| 85 | <filename class="directory">/tools/bin</filename> and <filename
|
---|
| 86 | class="directory">/tools/$TARGET_TRIPLET/bin</filename>. The tools in one
|
---|
| 87 | location are hard linked to the other. An important facet of the linker is
|
---|
| 88 | its library search order. Detailed information can be obtained from
|
---|
| 89 | <command>ld</command> by passing it the <parameter>--verbose</parameter>
|
---|
| 90 | flag. For example, an <userinput>ld --verbose | grep SEARCH</userinput>
|
---|
| 91 | will illustrate the current search paths and their order. It shows which
|
---|
| 92 | files are linked by <command>ld</command> by compiling a dummy program and
|
---|
| 93 | passing the <parameter>--verbose</parameter> switch to the linker. For example,
|
---|
| 94 | <userinput>gcc dummy.c -Wl,--verbose 2>&1 | grep succeeded</userinput>
|
---|
| 95 | will show all the files successfully opened during the linking.</para>
|
---|
| 96 |
|
---|
| 97 | <para>The next package installed is GCC. An example of what can be
|
---|
| 98 | seen during its run of <command>configure</command> is:</para>
|
---|
| 99 |
|
---|
| 100 | <screen><computeroutput>checking what assembler to use...
|
---|
[81fd230] | 101 | /tools/i686-pc-linux-gnu/bin/as
|
---|
| 102 | checking what linker to use... /tools/i686-pc-linux-gnu/bin/ld</computeroutput></screen>
|
---|
| 103 |
|
---|
[b28fd35] | 104 | <para>This is important for the reasons mentioned above. It also demonstrates
|
---|
| 105 | that GCC's configure script does not search the PATH directories to find which
|
---|
| 106 | tools to use. However, during the actual operation of <command>gcc</command>
|
---|
| 107 | itself, the same search paths are not necessarily used. To find out which
|
---|
| 108 | standard linker <command>gcc</command> will use, run:
|
---|
| 109 | <userinput>gcc -print-prog-name=ld</userinput>.</para>
|
---|
| 110 |
|
---|
| 111 | <para>Detailed information can be obtained from <command>gcc</command> by
|
---|
| 112 | passing it the <parameter>-v</parameter> command line option while compiling
|
---|
| 113 | a dummy program. For example, <userinput>gcc -v dummy.c</userinput> will show
|
---|
| 114 | detailed information about the preprocessor, compilation, and assembly stages,
|
---|
| 115 | including <command>gcc</command>'s included search paths and their order.</para>
|
---|
| 116 |
|
---|
| 117 | <para>The next package installed is Glibc. The most important considerations
|
---|
| 118 | for building Glibc are the compiler, binary tools, and kernel headers. The
|
---|
| 119 | compiler is generally not an issue since Glibc will always use the
|
---|
| 120 | <command>gcc</command> found in a <envar>PATH</envar> directory. The binary
|
---|
| 121 | tools and kernel headers can be a bit more complicated. Therefore, take no
|
---|
| 122 | risks and use the available configure switches to enforce the correct
|
---|
| 123 | selections. After the run of <command>configure</command>, check the contents
|
---|
| 124 | of the <filename>config.make</filename> file in the <filename
|
---|
| 125 | class="directory">glibc-build</filename> directory for all important details.
|
---|
| 126 | Note the use of <parameter>CC="gcc -B/tools/bin/"</parameter> to control which
|
---|
| 127 | binary tools are used and the use of the <parameter>-nostdinc</parameter>
|
---|
| 128 | and <parameter>-isystem</parameter> flags to control the compiler's include
|
---|
| 129 | search path. These items highlight an important aspect of the Glibc
|
---|
| 130 | package—it is very self-sufficient in terms of its build machinery and
|
---|
| 131 | generally does not rely on toolchain defaults.</para>
|
---|
| 132 |
|
---|
| 133 | <para>After the Glibc installation, make some adjustments to ensure that
|
---|
| 134 | searching and linking take place only within the <filename
|
---|
| 135 | class="directory">/tools</filename> prefix. Install an adjusted
|
---|
| 136 | <command>ld</command>, which has a hard-wired search path limited to
|
---|
| 137 | <filename class="directory">/tools/lib</filename>. Then amend
|
---|
| 138 | <command>gcc</command>'s specs file to point to the new dynamic linker in
|
---|
| 139 | <filename class="directory">/tools/lib</filename>. This last step is vital
|
---|
| 140 | to the whole process. As mentioned above, a hard-wired path to a dynamic
|
---|
| 141 | linker is embedded into every Executable and Link Format (ELF)-shared
|
---|
| 142 | executable. This can be inspected by running:
|
---|
| 143 | <userinput>readelf -l <name of binary> | grep interpreter</userinput>.
|
---|
| 144 | Amending gcc's specs file ensures that every program compiled from here
|
---|
| 145 | through the end of this chapter will use the new dynamic linker in
|
---|
| 146 | <filename class="directory">/tools/lib</filename>.</para>
|
---|
| 147 |
|
---|
| 148 | <para>The need to use the new dynamic linker is also the reason why
|
---|
| 149 | the Specs patch is applied for the second pass of GCC. Failure to do
|
---|
| 150 | so will result in the GCC programs themselves having the name of the
|
---|
| 151 | dynamic linker from the host system's <filename
|
---|
| 152 | class="directory">/lib</filename> directory embedded into them, which
|
---|
| 153 | would defeat the goal of getting away from the host.</para>
|
---|
| 154 |
|
---|
| 155 | <para>During the second pass of Binutils, we are able to utilize the
|
---|
| 156 | <parameter>--with-lib-path</parameter> configure switch to control
|
---|
| 157 | <command>ld</command>'s library search path. From this point onwards,
|
---|
| 158 | the core toolchain is self-contained and self-hosted. The remainder of
|
---|
| 159 | the <xref linkend="chapter-temporary-tools"/> packages all build against
|
---|
| 160 | the new Glibc in <filename class="directory">/tools</filename>.</para>
|
---|
| 161 |
|
---|
| 162 | <para>Upon entering the chroot environment in <xref
|
---|
| 163 | linkend="chapter-building-system"/>, the first major package to be
|
---|
| 164 | installed is Glibc, due to its self-sufficient nature mentioned above.
|
---|
| 165 | Once this Glibc is installed into <filename
|
---|
| 166 | class="directory">/usr</filename>, perform a quick changeover of the
|
---|
| 167 | toolchain defaults, then proceed in building the rest of the target
|
---|
| 168 | LFS system.</para>
|
---|
| 169 |
|
---|
[004616a] | 170 | <!-- FIXME: Removed as part of the fix for bug 1061 - we no longer build pass1
|
---|
| 171 | packages statically, therefore this explanation isn't required
|
---|
[b28fd35] | 172 |
|
---|
[004616a] | 173 | <sect2>
|
---|
[b28fd35] | 174 | <title>Notes on Static Linking</title>
|
---|
| 175 |
|
---|
| 176 | <para>Besides their specific task, most programs have to perform many
|
---|
| 177 | common and sometimes trivial operations. These include allocating
|
---|
| 178 | memory, searching directories, reading and writing files, string
|
---|
| 179 | handling, pattern matching, arithmetic, and other tasks. Instead of
|
---|
| 180 | obliging each program to reinvent the wheel, the GNU system provides
|
---|
| 181 | all these basic functions in ready-made libraries. The major library
|
---|
| 182 | on any Linux system is Glibc.</para>
|
---|
| 183 |
|
---|
| 184 | <para>There are two primary ways of linking the functions from a
|
---|
| 185 | library to a program that uses them—statically or dynamically. When
|
---|
| 186 | a program is linked statically, the code of the used functions is
|
---|
| 187 | included in the executable, resulting in a rather bulky program. When
|
---|
| 188 | a program is dynamically linked, it includes a reference to the
|
---|
| 189 | dynamic linker, the name of the library, and the name of the function,
|
---|
| 190 | resulting in a much smaller executable. A third option is to use the
|
---|
| 191 | programming interface of the dynamic linker (see <filename>dlopen(3)</filename>
|
---|
| 192 | for more information).</para>
|
---|
| 193 |
|
---|
| 194 | <para>Dynamic linking is the default on Linux and has three major
|
---|
| 195 | advantages over static linking. First, only one copy of the executable
|
---|
| 196 | library code is needed on the hard disk, instead of having multiple
|
---|
| 197 | copies of the same code included in several programs, thus saving
|
---|
| 198 | disk space. Second, when several programs use the same library
|
---|
| 199 | function at the same time, only one copy of the function's code is
|
---|
| 200 | required in core, thus saving memory space. Third, when a library
|
---|
| 201 | function gets a bug fixed or is otherwise improved, only the one
|
---|
| 202 | library needs to be recompiled instead of recompiling all programs
|
---|
| 203 | that make use of the improved function.</para>
|
---|
| 204 |
|
---|
| 205 | <para>If dynamic linking has several advantages, why then do we
|
---|
| 206 | statically link the first two packages in this chapter? The reasons
|
---|
| 207 | are threefold—historical, educational, and technical. The
|
---|
| 208 | historical reason is that earlier versions of LFS statically linked
|
---|
| 209 | every program in this chapter. Educationally, knowing the difference
|
---|
| 210 | between static and dynamic linking is useful. The technical benefit is
|
---|
| 211 | a gained element of independence from the host, meaning that those
|
---|
| 212 | programs can be used independently of the host system. However, it is
|
---|
| 213 | worth noting that an overall successful LFS build can still be
|
---|
| 214 | achieved when the first two packages are built dynamically.</para>
|
---|
| 215 |
|
---|
| 216 | </sect2>-->
|
---|
[673b0d8] | 217 |
|
---|
| 218 | </sect1>
|
---|