1 | <sect2><title> </title><para> </para></sect2>
|
---|
2 |
|
---|
3 | <sect2>
|
---|
4 | <title>Re-installation of GCC</title>
|
---|
5 |
|
---|
6 | <para>Unpack all three GCC tarballs in one and the same working directory.
|
---|
7 | They will all unfold into a single <filename>gcc-&gcc-version;/</filename>
|
---|
8 | subdir.</para>
|
---|
9 |
|
---|
10 | <para>First correct two problems and make an essential adjustment:</para>
|
---|
11 |
|
---|
12 | <para><screen><userinput>patch -Np1 -i ../gcc-&gcc-version;-mmap_test.patch
|
---|
13 | patch -Np1 -i ../gcc-&gcc-version;-no_fixincludes.patch
|
---|
14 | patch -Np1 -i ../gcc-&gcc-specs-version;.patch</userinput></screen></para>
|
---|
15 |
|
---|
16 | <para>The last patch changes GCC's default location of the dynamic linker,
|
---|
17 | a simple substitution of "/lib/ld-linux.so.2" with "/stage1/lib/ld-linux.so.2"
|
---|
18 | in <filename>config/i386/linux.h</filename>. Patching now rather than adjusting
|
---|
19 | the specs file after installation ensures that our new dynamic linker gets
|
---|
20 | used during the actual build of GCC. That is, all the final (and temporary)
|
---|
21 | binaries created during the build will link against the new Glibc.</para>
|
---|
22 |
|
---|
23 | <para>Create a separate build directory again:</para>
|
---|
24 |
|
---|
25 | <para><screen><userinput>mkdir ../gcc-build
|
---|
26 | cd ../gcc-build</userinput></screen></para>
|
---|
27 |
|
---|
28 | <para>Before starting to build GCC, remember to unset any environment
|
---|
29 | variables that override the default optimization flags.</para>
|
---|
30 |
|
---|
31 | <para>Now prepare GCC to be compiled:</para>
|
---|
32 |
|
---|
33 | <para><screen><userinput>../gcc-&gcc-version;/configure --prefix=/stage1 \
|
---|
34 | --with-local-prefix=/stage1 \
|
---|
35 | --enable-clocale=gnu --enable-shared \
|
---|
36 | --enable-threads=posix --enable-__cxa_atexit \
|
---|
37 | --enable-languages=c,c++</userinput></screen></para>
|
---|
38 |
|
---|
39 | <para>Compile the package:</para>
|
---|
40 |
|
---|
41 | <para><screen><userinput>make</userinput></screen></para>
|
---|
42 | <para>There is no need to use the <userinput>bootstrap</userinput> target now,
|
---|
43 | as the compiler we're using to compile this GCC has been built from the exact
|
---|
44 | same sources.</para>
|
---|
45 |
|
---|
46 | <para>Test the results:</para>
|
---|
47 |
|
---|
48 | <para><screen><userinput>make -k check</userinput></screen></para>
|
---|
49 |
|
---|
50 | <para>The <userinput>-k</userinput> flag is used to make the test suite run
|
---|
51 | through to completion and not stop at the first failure. The GCC test suite is
|
---|
52 | very comprehensive and is almost guaranteed to generate a few failures. To get
|
---|
53 | a summary of the test suite results, run this:</para>
|
---|
54 |
|
---|
55 | <para><screen><userinput>../gcc-3*/contrib/test_summary | less</userinput></screen></para>
|
---|
56 |
|
---|
57 | <para>You can compare your results to those posted to the gcc-testresults
|
---|
58 | mailing list for similar configurations to your own. For an example of how
|
---|
59 | current GCC-3.2.x should look on i686-pc-linux-gnu, see
|
---|
60 | <ulink url="http://gcc.gnu.org/ml/gcc-testresults/2003-02/msg00204.html"/>.</para>
|
---|
61 |
|
---|
62 | <para>Note that the results contain:</para>
|
---|
63 |
|
---|
64 | <screen>* 1 XPASS (unexpected pass) for g++
|
---|
65 | * 26 XPASS's for libstdc++
|
---|
66 | * 1 FAIL for libstdc++</screen>
|
---|
67 |
|
---|
68 | <para>The unexpected pass for g++ is due to the use of
|
---|
69 | <userinput>--enable-__cxa_atexit</userinput>. Apparently not all platforms
|
---|
70 | supported by GCC have support for "__cxa_atexit" in their C libraries, so this
|
---|
71 | test is not always expected to pass.</para>
|
---|
72 |
|
---|
73 | <para>The 26 unexpected passes for libstdc++ are due to the use of
|
---|
74 | <userinput>--enable-clocale=gnu</userinput>, which is the correct choice on
|
---|
75 | Glibc-based systems of versions 2.2.5 and above. The underlying locale support
|
---|
76 | in the GNU C library is superior to that of the otherwise selected "generic"
|
---|
77 | model (which may be applicable if for instance you were using Newlibc, Sun-libc
|
---|
78 | or whatever libc). The libstdc++ test suite is apparently expecting the
|
---|
79 | "generic" model, hence those tests are not always expected to pass.</para>
|
---|
80 |
|
---|
81 | <para>The failure for libstdc++ is in
|
---|
82 | <filename>26_numerics/c99_classification_macros_c.cc</filename> and is a
|
---|
83 | long-standing known failure (since at least January 2002) that the developers
|
---|
84 | are apparently unable to easily fix.</para>
|
---|
85 |
|
---|
86 | <para>And finally install the package:</para>
|
---|
87 |
|
---|
88 | <para><screen><userinput>make install</userinput></screen></para>
|
---|
89 |
|
---|
90 | </sect2>
|
---|
91 |
|
---|