gcc-pass2-inst.xml 7.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161
  1. <sect2><title>&nbsp;</title><para>&nbsp;</para></sect2>
  2. <sect2>
  3. <title>Re-installation of GCC</title>
  4. <para>The tools required to test GCC and Binutils are installed now (Tcl, Expect
  5. and DejaGnu). We can continue on rebuilding GCC and Binutils, link them against
  6. the new Glibc, and test them properly. One thing to note, however, is that these
  7. test suites are highly dependent on properly functioning pseudo terminals (PTYs)
  8. which are provided by your host distribution. These days, PTYs are most commonly
  9. implemented via the <emphasis>devpts</emphasis> file system. You can quickly
  10. check if your host system is set up correctly in this regard by performing a
  11. simple test:</para>
  12. <para><screen><userinput>expect -c "spawn ls"</userinput></screen></para>
  13. <para>If you receive the message:</para>
  14. <blockquote><screen>The system has no more ptys. Ask your system administrator to create more.</screen></blockquote>
  15. <para>Your host distribution is not set up for proper PTY operation. In this
  16. case there is no point in running the test suites for GCC and Binutils until you
  17. are able to resolve the issue. You can consult the LFS Wiki at
  18. <ulink url="http://wiki.linuxfromscratch.org/"/> for more information on how to
  19. get PTYs working.</para>
  20. <note><para>It's worth pointing out that the GCC test suite we run in this
  21. section is considered not as important as the one we run in Chapter 6.</para></note>
  22. <para>Unpack all three GCC tarballs (-core, -g++, and -testsuite) in one and the
  23. same working directory. They will all unfold into a single
  24. <filename>gcc-&gcc-version;/</filename> subdirectory.</para>
  25. <para>First correct one problem and make an essential adjustment:</para>
  26. <para><screen><userinput>patch -Np1 -i ../gcc-&gcc-nofixincludes-patch-version;.patch
  27. patch -Np1 -i ../gcc-&gcc-specs-patch-version;.patch</userinput></screen></para>
  28. <para>The first patch disables the GCC "fixincludes" script. We mentioned this
  29. briefly earlier, but a slightly more in-depth explanation of the fixincludes
  30. process is warranted here. Under normal circumstances, the GCC fixincludes
  31. script scans your system for header files that need to be fixed. It might find
  32. that some Glibc header files on your host system need to be fixed, fix them and
  33. put them in the GCC private include directory. Then, later on in Chapter 6,
  34. after we've installed the newer Glibc, this private include directory would be
  35. searched before the system include directory, resulting in GCC finding the
  36. fixed headers from the host system, which would most likely not match the Glibc
  37. version actually used for the LFS system.</para>
  38. <para>The last patch changes GCC's default location of the dynamic linker
  39. (typically <filename>ld-linux.so.2</filename>). It also removes
  40. <filename class="directory">/usr/include</filename> from GCC's include search
  41. path. Patching now rather than adjusting the specs file after installation
  42. ensures that our new dynamic linker gets used during the actual build of GCC.
  43. That is, all the final (and temporary) binaries created during the build will
  44. link against the new Glibc.</para>
  45. <important><para>These patches are <emphasis>critical</emphasis> in ensuring a
  46. successful overall build. Do not forget to apply them.</para></important>
  47. <para>Create a separate build directory again:</para>
  48. <para><screen><userinput>mkdir ../gcc-build
  49. cd ../gcc-build</userinput></screen></para>
  50. <para>Before starting to build GCC, remember to unset any environment
  51. variables that override the default optimization flags.</para>
  52. <para>Now prepare GCC to be compiled:</para>
  53. <para><screen><userinput>../gcc-&gcc-version;/configure --prefix=/tools \
  54. &nbsp;&nbsp;&nbsp;&nbsp;--with-local-prefix=/tools \
  55. &nbsp;&nbsp;&nbsp;&nbsp;--enable-clocale=gnu --enable-shared \
  56. &nbsp;&nbsp;&nbsp;&nbsp;--enable-threads=posix --enable-__cxa_atexit \
  57. &nbsp;&nbsp;&nbsp;&nbsp;--enable-languages=c,c++</userinput></screen></para>
  58. <para>The meaning of the new configure options:</para>
  59. <itemizedlist>
  60. <listitem><para><userinput>--enable-threads=posix</userinput>: This enables
  61. C++ exception handling for multi-threaded code.</para></listitem>
  62. <listitem><para><userinput>--enable-__cxa_atexit</userinput>: This option
  63. allows use of __cxa_atexit, rather than atexit, to register C++ destructors for
  64. local statics and global objects and is essential for fully standards-compliant
  65. handling of destructors. It also affects the C++ ABI and therefore results in
  66. C++ shared libraries and C++ programs that are interoperable with other Linux
  67. distributions.</para></listitem>
  68. <listitem><para><userinput>--enable-clocale=gnu</userinput>: This option ensures
  69. the correct locale model is selected for the C++ libraries under all
  70. circumstances. If the configure script finds the <emphasis>de_DE</emphasis>
  71. locale installed, it will select the correct model of <emphasis>gnu</emphasis>.
  72. However, people who don't install the <emphasis>de_DE</emphasis> locale, run the
  73. risk of building ABI incompatible C++ libraries due to the wrong locale model of
  74. <emphasis>generic</emphasis> being selected.</para></listitem>
  75. <listitem><para><userinput>--enable-languages=c,c++</userinput>: This option is
  76. needed to ensure that both C and C++ compilers are built.</para></listitem>
  77. </itemizedlist>
  78. <para>Compile the package:</para>
  79. <para><screen><userinput>make</userinput></screen></para>
  80. <para>There is no need to use the <userinput>bootstrap</userinput> target now,
  81. as the compiler we're using to compile this GCC was built from the exact same
  82. version of the GCC sources we used earlier.</para>
  83. <note><para>At this point it is strongly recommended to repeat the sanity check
  84. we performed earlier in the chapter. Refer back to the "Locking in" Glibc section
  85. and repeat the check. If the results are wrong then most likely, you forgot to
  86. apply the abovementioned GCC Specs patch.</para></note>
  87. <para>Test the results:</para>
  88. <para><screen><userinput>make -k check</userinput></screen></para>
  89. <para>The <userinput>-k</userinput> flag is used to make the test suite run
  90. through to completion and not stop at the first failure. The GCC test suite is
  91. very comprehensive and is almost guaranteed to generate a few failures. To get
  92. a summary of the test suite results, run this:</para>
  93. <para><screen><userinput>../gcc-&gcc-version;/contrib/test_summary | less</userinput></screen></para>
  94. <para>You can compare your results to those posted to the gcc-testresults
  95. mailing list for similar configurations to your own. For an example of how
  96. current GCC-3.3.1 should look on i686-pc-linux-gnu, see
  97. <ulink url="http://gcc.gnu.org/ml/gcc-testresults/2003-08/msg01612.html"/>.</para>
  98. <para>Note that the results contain:</para>
  99. <screen>* 1 XPASS (unexpected pass) for g++
  100. * 1 FAIL (unexpected failure) for g++
  101. * 2 FAIL for gcc
  102. * 26 XPASS's for libstdc++</screen>
  103. <para>The unexpected pass for g++ is due to the use of
  104. <userinput>--enable-__cxa_atexit</userinput>. Apparently not all platforms
  105. supported by GCC have support for "__cxa_atexit" in their C libraries, so this
  106. test is not always expected to pass.</para>
  107. <para>The 26 unexpected passes for libstdc++ are due to the use of
  108. <userinput>--enable-clocale=gnu</userinput>, which is the correct choice on
  109. Glibc-based systems of versions 2.2.5 and above. The underlying locale support
  110. in the GNU C library is superior to that of the otherwise selected "generic"
  111. model (which may be applicable if for instance you were using Newlibc, Sun-libc
  112. or whatever libc). The libstdc++ test suite is apparently expecting the
  113. "generic" model, hence those tests are not always expected to pass.</para>
  114. <para>Unexpected failures often cannot be avoided. The GCC developers are
  115. usually aware of them but haven't yet gotten around to fixing them. In short,
  116. unless your results are vastly different from those at the above URL, it is safe
  117. to continue on.</para>
  118. <para>And finally install the package:</para>
  119. <para><screen><userinput>make install</userinput></screen></para>
  120. </sect2>