[98c95f5] | 1 | <sect1 id="ch05-toolchaintechnotes">
|
---|
| 2 | <title>Toolchain technical notes</title>
|
---|
| 3 | <?dbhtml filename="toolchaintechnotes.html" dir="chapter05"?>
|
---|
| 4 |
|
---|
| 5 | <para>This section attempts to explain some of the rationale and technical
|
---|
| 6 | details behind the overall build method. It's not essential that you understand
|
---|
| 7 | everything here immediately. Most of it will make sense once you have performed
|
---|
| 8 | an actual build. Feel free to refer back here at any time.</para>
|
---|
| 9 |
|
---|
[7f1fcd8] | 10 | <para>The overall goal of <xref linkend="chapter05"/> is to provide a sane,
|
---|
| 11 | temporary environment that we can chroot into, and from which we can produce a
|
---|
| 12 | clean, trouble-free build of the target LFS system in
|
---|
| 13 | <xref linkend="chapter06"/>. Along the way, we attempt to divorce ourselves
|
---|
| 14 | from the host system as much as possible, and in so doing build a
|
---|
| 15 | self-contained and self-hosted toolchain. It should be noted that the
|
---|
[98c95f5] | 16 | build process has been designed in such a way so as to minimize the risks for
|
---|
[1668d8e] | 17 | new readers and provide maximum educational value at the same time. In other
|
---|
| 18 | words, more advanced techniques could be used to build the system.</para>
|
---|
[98c95f5] | 19 |
|
---|
| 20 | <important>
|
---|
| 21 | <para>Before continuing, you really should be aware of the name of your working
|
---|
| 22 | platform, often also referred to as the <emphasis>target triplet</emphasis>. For
|
---|
| 23 | many folks the target triplet will be, for example:
|
---|
| 24 | <emphasis>i686-pc-linux-gnu</emphasis>. A simple way to determine your target
|
---|
| 25 | triplet is to run the <filename>config.guess</filename> script that comes with
|
---|
| 26 | the source for many packages. Unpack the Binutils sources and run the script:
|
---|
| 27 | <userinput>./config.guess</userinput> and note the output.</para>
|
---|
| 28 |
|
---|
| 29 | <para>You'll also need to be aware of the name of your platform's
|
---|
| 30 | <emphasis>dynamic linker</emphasis>, often also referred to as the
|
---|
| 31 | <emphasis>dynamic loader</emphasis>, not to be confused with the standard linker
|
---|
| 32 | <emphasis>ld</emphasis> that is part of Binutils. The dynamic linker is provided
|
---|
| 33 | by Glibc and has the job of finding and loading the shared libraries needed by a
|
---|
| 34 | program, preparing the program to run and then running it. For most folks, the
|
---|
| 35 | name of the dynamic linker will be <emphasis>ld-linux.so.2</emphasis>. On
|
---|
| 36 | platforms that are less prevalent, the name might be
|
---|
| 37 | <emphasis>ld.so.1</emphasis> and newer 64 bit platforms might even have
|
---|
| 38 | something completely different. You should be able to determine the name
|
---|
| 39 | of your platform's dynamic linker by looking in the
|
---|
| 40 | <filename class="directory">/lib</filename> directory on your host system. A
|
---|
| 41 | surefire way is to inspect a random binary from your host system by running:
|
---|
[a3f6e124] | 42 | <userinput>'readelf -l <name of binary> | grep interpreter'</userinput>
|
---|
[98c95f5] | 43 | and noting the output. The authoritative reference covering all platforms is in
|
---|
| 44 | the <filename>shlib-versions</filename> file in the root of the Glibc source
|
---|
| 45 | tree.</para>
|
---|
| 46 | </important>
|
---|
| 47 |
|
---|
[7f1fcd8] | 48 | <para>Some key technical points of how the <xref linkend="chapter05"/> build
|
---|
| 49 | method works:</para>
|
---|
[98c95f5] | 50 |
|
---|
| 51 | <itemizedlist>
|
---|
| 52 | <listitem><para>Similar in principle to cross compiling whereby tools installed
|
---|
| 53 | into the same prefix work in cooperation and thus utilize a little GNU
|
---|
| 54 | "magic".</para></listitem>
|
---|
| 55 |
|
---|
| 56 | <listitem><para>Careful manipulation of the standard linker's library search
|
---|
| 57 | path to ensure programs are linked only against libraries we
|
---|
| 58 | choose.</para></listitem>
|
---|
| 59 |
|
---|
[1668d8e] | 60 | <listitem><para>Careful manipulation of <userinput>gcc</userinput>'s
|
---|
| 61 | <emphasis>specs</emphasis> file to tell the compiler which target dynamic
|
---|
| 62 | linker will be used.</para></listitem>
|
---|
[98c95f5] | 63 | </itemizedlist>
|
---|
| 64 |
|
---|
| 65 | <para>Binutils is installed first because both GCC and Glibc perform various
|
---|
| 66 | feature tests on the assembler and linker during their respective runs of
|
---|
[1668d8e] | 67 | <userinput>./configure</userinput> to determine which software features to enable
|
---|
[98c95f5] | 68 | or disable. This is more important than one might first realize. An incorrectly
|
---|
| 69 | configured GCC or Glibc can result in a subtly broken toolchain where the impact
|
---|
[1668d8e] | 70 | of such breakage might not show up until near the end of the build of a whole
|
---|
[98c95f5] | 71 | distribution. Thankfully, a test suite failure will usually alert us before too
|
---|
[1668d8e] | 72 | much time is wasted.</para>
|
---|
[98c95f5] | 73 |
|
---|
| 74 | <para>Binutils installs its assembler and linker into two locations,
|
---|
| 75 | <filename class="directory">/tools/bin</filename> and
|
---|
| 76 | <filename class="directory">/tools/$TARGET_TRIPLET/bin</filename>. In reality,
|
---|
[1668d8e] | 77 | the tools in one location are hard linked to the other. An important facet of
|
---|
| 78 | the linker is its library search order. Detailed information can be obtained
|
---|
| 79 | from <userinput>ld</userinput> by passing it the <emphasis>--verbose</emphasis>
|
---|
[a3f6e124] | 80 | flag. For example: <userinput>'ld --verbose | grep SEARCH'</userinput> will
|
---|
[1668d8e] | 81 | show you the current search paths and their order. You can see what files are
|
---|
| 82 | actually linked by <userinput>ld</userinput> by compiling a dummy program and
|
---|
| 83 | passing the <emphasis>--verbose</emphasis> switch. For example:
|
---|
[a3f6e124] | 84 | <userinput>'gcc dummy.c -Wl,--verbose 2>&1 | grep succeeded'</userinput>
|
---|
[98c95f5] | 85 | will show you all the files successfully opened during the link.</para>
|
---|
| 86 |
|
---|
| 87 | <para>The next package installed is GCC and during its run of
|
---|
[1668d8e] | 88 | <userinput>./configure</userinput> you'll see, for example:</para>
|
---|
[98c95f5] | 89 |
|
---|
| 90 | <blockquote><screen>checking what assembler to use... /tools/i686-pc-linux-gnu/bin/as
|
---|
| 91 | checking what linker to use... /tools/i686-pc-linux-gnu/bin/ld</screen></blockquote>
|
---|
| 92 |
|
---|
| 93 | <para>This is important for the reasons mentioned above. It also demonstrates
|
---|
| 94 | that GCC's configure script does not search the $PATH directories to find which
|
---|
[1668d8e] | 95 | tools to use. However, during the actual operation of <userinput>gcc</userinput>
|
---|
| 96 | itself, the same search paths are not necessarily used. You can find out which
|
---|
| 97 | standard linker <userinput>gcc</userinput> will use by running:
|
---|
[a3f6e124] | 98 | <userinput>'gcc -print-prog-name=ld'</userinput>.
|
---|
[1668d8e] | 99 | Detailed information can be obtained from <userinput>gcc</userinput> by passing
|
---|
| 100 | it the <emphasis>-v</emphasis> flag while compiling a dummy program. For
|
---|
[a3f6e124] | 101 | example: <userinput>'gcc -v dummy.c'</userinput> will show you detailed
|
---|
[1668d8e] | 102 | information about the preprocessor, compilation and assembly stages, including
|
---|
| 103 | <userinput>gcc</userinput>'s include search paths and their order.</para>
|
---|
[98c95f5] | 104 |
|
---|
| 105 | <para>The next package installed is Glibc. The most important considerations for
|
---|
| 106 | building Glibc are the compiler, binary tools and kernel headers. The compiler
|
---|
[1668d8e] | 107 | is generally no problem as Glibc will always use the <userinput>gcc</userinput>
|
---|
| 108 | found in a $PATH directory. The binary tools and kernel headers can be a little
|
---|
| 109 | more troublesome. Therefore we take no risks and use the available configure
|
---|
| 110 | switches to enforce the correct selections. After the run of
|
---|
| 111 | <userinput>./configure</userinput> you can check the contents of the
|
---|
[98c95f5] | 112 | <filename>config.make</filename> file in the
|
---|
| 113 | <filename class="directory">glibc-build</filename> directory for all the
|
---|
| 114 | important details. You'll note some interesting items like the use of
|
---|
| 115 | <userinput>CC="gcc -B/tools/bin/"</userinput> to control which binary tools are
|
---|
[1668d8e] | 116 | used, and also the use of the <emphasis>-nostdinc</emphasis> and
|
---|
| 117 | <emphasis>-isystem</emphasis> flags to control the compiler's include search
|
---|
| 118 | path. These items help to highlight an important aspect of the Glibc package:
|
---|
| 119 | it is very self-sufficient in terms of its build machinery and generally does
|
---|
| 120 | not rely on toolchain defaults.</para>
|
---|
[98c95f5] | 121 |
|
---|
| 122 | <para>After the Glibc installation, we make some adjustments to ensure that
|
---|
[1668d8e] | 123 | searching and linking take place only within our <filename>/tools</filename>
|
---|
| 124 | prefix. We install an adjusted <userinput>ld</userinput>, which has a hard-wired
|
---|
| 125 | search path limited to <filename class="directory">/tools/lib</filename>. Then
|
---|
| 126 | we amend <userinput>gcc</userinput>'s specs file to point to our new dynamic
|
---|
| 127 | linker in <filename class="directory">/tools/lib</filename>. This last step is
|
---|
[98c95f5] | 128 | <emphasis>vital</emphasis> to the whole process. As mentioned above, a
|
---|
[f57e3d1] | 129 | hard-wired path to a dynamic linker is embedded into every ELF shared
|
---|
| 130 | executable. You can inspect this by running:
|
---|
[a3f6e124] | 131 | <userinput>'readelf -l <name of binary> | grep interpreter'</userinput>.
|
---|
[1668d8e] | 132 | By amending <userinput>gcc</userinput>'s specs file, we are ensuring that every
|
---|
[7f1fcd8] | 133 | program compiled from here through the end of <xref linkend="chapter05"/> will
|
---|
| 134 | use our new dynamic linker in
|
---|
| 135 | <filename class="directory">/tools/lib</filename>.</para>
|
---|
[98c95f5] | 136 |
|
---|
| 137 | <para>The need to use the new dynamic linker is also the reason why we apply the
|
---|
[1668d8e] | 138 | Specs patch for the second pass of GCC. Failure to do so will result in the GCC
|
---|
| 139 | programs themselves having the name of the dynamic linker from the host system's
|
---|
[98c95f5] | 140 | <filename class="directory">/lib</filename> directory embedded into them, which
|
---|
[1668d8e] | 141 | would defeat our goal of getting away from the host.</para>
|
---|
[98c95f5] | 142 |
|
---|
| 143 | <para>During the second pass of Binutils, we are able to utilize the
|
---|
[1668d8e] | 144 | <emphasis>--with-lib-path</emphasis> configure switch to control
|
---|
| 145 | <userinput>ld</userinput>'s library search path. From this point onwards, the
|
---|
| 146 | core toolchain is self-contained and self-hosted. The remainder of the
|
---|
[7f1fcd8] | 147 | <xref linkend="chapter05"/> packages all build against the new Glibc in
|
---|
[1668d8e] | 148 | <filename class="directory">/tools</filename> and all is well.</para>
|
---|
[98c95f5] | 149 |
|
---|
[7f1fcd8] | 150 | <para>Upon entering the chroot environment in <xref linkend="chapter06"/>, the
|
---|
| 151 | first major package we install is Glibc, due to its self-sufficient nature that
|
---|
| 152 | we mentioned above. Once this Glibc is installed into
|
---|
| 153 | <filename class="directory">/usr</filename>, we perform a quick changeover of
|
---|
| 154 | the toolchain defaults, then proceed for real in building the rest of the
|
---|
| 155 | target <xref linkend="chapter06"/> LFS system.</para>
|
---|
[98c95f5] | 156 |
|
---|
| 157 | <sect2>
|
---|
| 158 | <title>Notes on static linking</title>
|
---|
| 159 |
|
---|
| 160 | <para>Most programs have to perform, beside their specific task, many rather
|
---|
[8b437193] | 161 | common and sometimes trivial operations. These include allocating memory,
|
---|
[98c95f5] | 162 | searching directories, reading and writing files, string handling, pattern
|
---|
[8b437193] | 163 | matching, arithmetic and many other tasks. Instead of obliging each program to
|
---|
[98c95f5] | 164 | reinvent the wheel, the GNU system provides all these basic functions in
|
---|
| 165 | ready-made libraries. The major library on any Linux system is
|
---|
| 166 | <emphasis>Glibc</emphasis>.</para>
|
---|
| 167 |
|
---|
| 168 | <para>There are two primary ways of linking the functions from a library to a
|
---|
| 169 | program that uses them: statically or dynamically. When a program is linked
|
---|
| 170 | statically, the code of the used functions is included in the executable,
|
---|
| 171 | resulting in a rather bulky program. When a program is dynamically linked, what
|
---|
| 172 | is included is a reference to the dynamic linker, the name of the library, and
|
---|
[1668d8e] | 173 | the name of the function, resulting in a much smaller executable. (A third way
|
---|
| 174 | is to use the programming interface of the dynamic linker. See the
|
---|
| 175 | <emphasis>dlopen</emphasis> man page for more information.)</para>
|
---|
[98c95f5] | 176 |
|
---|
| 177 | <para>Dynamic linking is the default on Linux and has three major advantages
|
---|
| 178 | over static linking. First, you need only one copy of the executable library
|
---|
| 179 | code on your hard disk, instead of having many copies of the same code included
|
---|
| 180 | into a whole bunch of programs -- thus saving disk space. Second, when several
|
---|
| 181 | programs use the same library function at the same time, only one copy of the
|
---|
| 182 | function's code is required in core -- thus saving memory space. Third, when a
|
---|
| 183 | library function gets a bug fixed or is otherwise improved, you only need to
|
---|
| 184 | recompile this one library, instead of having to recompile all the programs that
|
---|
| 185 | make use of the improved function.</para>
|
---|
| 186 |
|
---|
[7f1fcd8] | 187 | <para>If dynamic linking has several advantages, why then do we statically link
|
---|
[b4f6bac] | 188 | the first two packages in this chapter? The reasons are threefold: historical,
|
---|
[7f1fcd8] | 189 | educational, and technical. Historical, because earlier versions of LFS
|
---|
| 190 | statically linked every program in this chapter. Educational, because knowing
|
---|
| 191 | the difference is useful. Technical, because we gain an element of independence
|
---|
| 192 | from the host in doing so, meaning that those programs can be used
|
---|
[98c95f5] | 193 | independently of the host system. However, it's worth noting that an overall
|
---|
[7f1fcd8] | 194 | successful LFS build can still be achieved when the first two packages are
|
---|
| 195 | built dynamically.</para>
|
---|
[98c95f5] | 196 |
|
---|
| 197 | </sect2>
|
---|
| 198 |
|
---|
| 199 | </sect1>
|
---|
| 200 |
|
---|