source: chapter05/gcc-pass2-inst.xml@ 2723b39

10.0 10.0-rc1 10.1 10.1-rc1 11.0 11.0-rc1 11.0-rc2 11.0-rc3 11.1 11.1-rc1 11.2 11.2-rc1 11.3 11.3-rc1 12.0 12.0-rc1 12.1 12.1-rc1 6.0 6.1 6.1.1 6.3 6.4 6.5 6.6 6.7 6.8 7.0 7.1 7.2 7.3 7.4 7.5 7.5-systemd 7.6 7.6-systemd 7.7 7.7-systemd 7.8 7.8-systemd 7.9 7.9-systemd 8.0 8.1 8.2 8.3 8.4 9.0 9.1 arm bdubbs/gcc13 ml-11.0 multilib renodr/libudev-from-systemd s6-init trunk v5_0 v5_1 v5_1_1 xry111/arm64 xry111/arm64-12.0 xry111/clfs-ng xry111/lfs-next xry111/loongarch xry111/loongarch-12.0 xry111/loongarch-12.1 xry111/mips64el xry111/pip3 xry111/rust-wip-20221008 xry111/update-glibc
Last change on this file since 2723b39 was 7acb4c3, checked in by Alex Gronenwoud <alex@…>, 21 years ago

Bringing order to the presentation of patches.

git-svn-id: http://svn.linuxfromscratch.org/LFS/trunk/BOOK@2938 4aa44e1e-78dd-0310-a6d2-fbcd4c07a689

  • Property mode set to 100644
File size: 7.8 KB
Line 
1<sect2><title>&nbsp;</title><para>&nbsp;</para></sect2>
2
3<sect2>
4<title>Re-installation of GCC</title>
5
6<para>The tools required to test GCC and Binutils are installed now (Tcl, Expect
7and DejaGnu). We can continue on rebuilding GCC and Binutils, link them against
8the new Glibc, and test them properly. One thing to note, however, is that these
9test suites are highly dependent on properly functioning pseudo terminals (PTYs)
10which are provided by your host distribution. These days, PTYs are most commonly
11implemented via the <emphasis>devpts</emphasis> file system. You can quickly
12check if your host system is set up correctly in this regard by performing a
13simple test:</para>
14
15<para><screen><userinput>expect -c "spawn ls"</userinput></screen></para>
16
17<para>If you receive the message:</para>
18
19<blockquote><screen>The system has no more ptys. Ask your system administrator to create more.</screen></blockquote>
20
21<para>Your host distribution is not set up for proper PTY operation. In this
22case there is no point in running the test suites for GCC and Binutils until you
23are able to resolve the issue. You can consult the LFS Wiki at
24<ulink url="http://wiki.linuxfromscratch.org/"/> for more information on how to
25get PTYs working.</para>
26
27<note><para>It's worth pointing out that the GCC test suite we run in this
28section is considered not as important as the one we run in Chapter 6.</para></note>
29
30<para>Unpack all three GCC tarballs (-core, -g++, and -testsuite) in one and the
31same working directory. They will all unfold into a single
32<filename>gcc-&gcc-version;/</filename> subdirectory.</para>
33
34<para>First correct one problem and make an essential adjustment:</para>
35
36<para><screen><userinput>patch -Np1 -i ../&gcc-nofixincludes-patch;
37patch -Np1 -i ../&gcc-specs-patch;</userinput></screen></para>
38
39<para>The first patch disables the GCC "fixincludes" script. We mentioned this
40briefly earlier, but a slightly more in-depth explanation of the fixincludes
41process is warranted here. Under normal circumstances, the GCC fixincludes
42script scans your system for header files that need to be fixed. It might find
43that some Glibc header files on your host system need to be fixed, fix them and
44put them in the GCC private include directory. Then, later on in Chapter 6,
45after we've installed the newer Glibc, this private include directory would be
46searched before the system include directory, resulting in GCC finding the
47fixed headers from the host system, which would most likely not match the Glibc
48version actually used for the LFS system.</para>
49
50<para>The last patch changes GCC's default location of the dynamic linker
51(typically <filename>ld-linux.so.2</filename>). It also removes
52<filename class="directory">/usr/include</filename> from GCC's include search
53path. Patching now rather than adjusting the specs file after installation
54ensures that our new dynamic linker gets used during the actual build of GCC.
55That is, all the final (and temporary) binaries created during the build will
56link against the new Glibc.</para>
57
58<important><para>These patches are <emphasis>critical</emphasis> in ensuring a
59successful overall build. Do not forget to apply them.</para></important>
60
61<para>Create a separate build directory again:</para>
62
63<para><screen><userinput>mkdir ../gcc-build
64cd ../gcc-build</userinput></screen></para>
65
66<para>Before starting to build GCC, remember to unset any environment
67variables that override the default optimization flags.</para>
68
69<para>Now prepare GCC to be compiled:</para>
70
71<para><screen><userinput>../gcc-&gcc-version;/configure --prefix=/tools \
72&nbsp;&nbsp;&nbsp;&nbsp;--with-local-prefix=/tools \
73&nbsp;&nbsp;&nbsp;&nbsp;--enable-clocale=gnu --enable-shared \
74&nbsp;&nbsp;&nbsp;&nbsp;--enable-threads=posix --enable-__cxa_atexit \
75&nbsp;&nbsp;&nbsp;&nbsp;--enable-languages=c,c++</userinput></screen></para>
76
77<para>The meaning of the new configure options:</para>
78
79<itemizedlist>
80<listitem><para><userinput>--enable-threads=posix</userinput>: This enables
81C++ exception handling for multi-threaded code.</para></listitem>
82
83<listitem><para><userinput>--enable-__cxa_atexit</userinput>: This option
84allows use of __cxa_atexit, rather than atexit, to register C++ destructors for
85local statics and global objects and is essential for fully standards-compliant
86handling of destructors. It also affects the C++ ABI and therefore results in
87C++ shared libraries and C++ programs that are interoperable with other Linux
88distributions.</para></listitem>
89
90<listitem><para><userinput>--enable-clocale=gnu</userinput>: This option ensures
91the correct locale model is selected for the C++ libraries under all
92circumstances. If the configure script finds the <emphasis>de_DE</emphasis>
93locale installed, it will select the correct model of <emphasis>gnu</emphasis>.
94However, people who don't install the <emphasis>de_DE</emphasis> locale, run the
95risk of building ABI incompatible C++ libraries due to the wrong locale model of
96<emphasis>generic</emphasis> being selected.</para></listitem>
97
98<listitem><para><userinput>--enable-languages=c,c++</userinput>: This option is
99needed to ensure that both C and C++ compilers are built.</para></listitem>
100</itemizedlist>
101
102<para>Compile the package:</para>
103
104<para><screen><userinput>make</userinput></screen></para>
105
106<para>There is no need to use the <userinput>bootstrap</userinput> target now,
107as the compiler we're using to compile this GCC was built from the exact same
108version of the GCC sources we used earlier.</para>
109
110<para>Test the results:</para>
111
112<para><screen><userinput>make -k check</userinput></screen></para>
113
114<para>The <userinput>-k</userinput> flag is used to make the test suite run
115through to completion and not stop at the first failure. The GCC test suite is
116very comprehensive and is almost guaranteed to generate a few failures. To get
117a summary of the test suite results, run this:</para>
118
119<para><screen><userinput>../gcc-&gcc-version;/contrib/test_summary | less</userinput></screen></para>
120
121<para>You can compare your results to those posted to the gcc-testresults
122mailing list for similar configurations to your own. For an example of how
123current GCC-3.3.1 should look on i686-pc-linux-gnu, see
124<ulink url="http://gcc.gnu.org/ml/gcc-testresults/2003-08/msg01612.html"/>.</para>
125
126<para>Note that the results contain:</para>
127
128<screen>* 1 XPASS (unexpected pass) for g++
129* 1 FAIL (unexpected failure) for g++
130* 2 FAIL for gcc
131* 26 XPASS's for libstdc++</screen>
132
133<para>The unexpected pass for g++ is due to the use of
134<userinput>--enable-__cxa_atexit</userinput>. Apparently not all platforms
135supported by GCC have support for "__cxa_atexit" in their C libraries, so this
136test is not always expected to pass.</para>
137
138<para>The 26 unexpected passes for libstdc++ are due to the use of
139<userinput>--enable-clocale=gnu</userinput>, which is the correct choice on
140Glibc-based systems of versions 2.2.5 and above. The underlying locale support
141in the GNU C library is superior to that of the otherwise selected "generic"
142model (which may be applicable if for instance you were using Newlibc, Sun-libc
143or whatever libc). The libstdc++ test suite is apparently expecting the
144"generic" model, hence those tests are not always expected to pass.</para>
145
146<para>Unexpected failures often cannot be avoided. The GCC developers are
147usually aware of them but haven't yet gotten around to fixing them. In short,
148unless your results are vastly different from those at the above URL, it is safe
149to continue on.</para>
150
151<para>And finally install the package:</para>
152
153<para><screen><userinput>make install</userinput></screen></para>
154
155<note><para>At this point it is strongly recommended to repeat the sanity check
156we performed earlier in the chapter. Refer back to
157<xref linkend="ch05-locking-glibc"/> and repeat the check. If the results are
158wrong then most likely, you forgot to apply the above mentioned GCC Specs
159patch.</para></note>
160
161</sect2>
162
Note: See TracBrowser for help on using the repository browser.