内核编译模式:built-in,将相应功能编译到内核中;module,将相应功能编译成模块,生成.0文件,动态调用
为什么要编译内核:主要是定制内核,根据需要将相应的功能加到内核中;
 
主要任务
1).获得内核源文件:www.kernel.org;
2).配置内核(获得内核配置文件.config):make config|menuconfig|xconfig|oldconfig
3).备份相关文件
4)建立内核,相关命令:
A.make dep #建立依赖关系表;
B.make clean or make mrpropper #删除安装过程中产生的大量临时文件;(可选)
C.make bzImage #建立内核,通常在/usr/src/arch/i386/boot目录下;
D.make modules #编译内核模块;(可选)
E.mkinitrd /boot/initrd-2.6.31.5 #
F.make module install(把内核模块存放到相应的位置) (可选)
5). 安装内核
cp /usr/src/arch/i386/bzImage /boot/vmlinuz-2.6.31
cp /usr/src/linux/System.map /boot/system.map
make install ,
6).编辑/boot/grub/grub.conf文件;
 
步骤:
 
S1.了解系统相关信息 uname -a
 
S2.下载内核源文件(linux-2.6.31.5.tar.bz2)及补丁(patch-2.6.31.5)
cd /usr/src
bzip2 -dc linux-2.6.31.5.tar.bz2 | tar vxf - #解包
bzip2 -dc patch-2.6.31.5.bz2 | patch -p1  #解包,给内核源代码打补丁(未执行成功)
(或使用patch命令给内核源码打补丁patch -p0 original <patch name)
 
S3.配置内核
cd ./linux-2.6.31.5
make menuconfig
 
S4.编译内核
?A.make dep #建立依赖关系表;
    make clean or make mrpropper #删除安装过程中产生的大量临时文件
B.make bzImage #建立内核,通常在/usr/src/arch/i386/boot目录下;
   make modules #编译内核模块;(可选)
 
C.mkinitrd /boot/initrd-2.6.31.5 #
  make module install(把内核模块存放到相应的位置) (可选)
 
make[1]: *** No rule to make target `|', needed by `firmware/keyspan/usa19qi.fw.gen.S'.  Stop.
make: *** [firmware] Error 2

 
S5. 安装内核
cp /usr/src/arch/i386/bzImage /boot/vmlinuz-2.6.31
cp /usr/src/linux/System.map /boot/system.map
make install
 
S6.vi  /boot/grub/grub.conf
 
S7.重启机器测试结果
 
 
 
以下内容为 make help内容:
Cleaning targets:
  clean    - Remove most generated files but keep the config and
                    enough build support to build external modules
  mrproper   - Remove all generated files + config + various backup files
  distclean   - mrproper + remove editor backup and patch files
Configuration targets:
  config   - Update current config utilising a line-oriented program
  menuconfig   - Update current config utilising a menu based program
  xconfig   - Update current config utilising a QT based front-end
  gconfig   - Update current config utilising a GTK based front-end
  oldconfig   - Update current config utilising a provided .config as base
  silentoldconfig - Same as oldconfig, but quietly, additionally update deps
  randconfig   - New config with random answer to all options
  defconfig   - New config with default answer to all options
  allmodconfig   - New config selecting modules when possible
  allyesconfig   - New config where all options are accepted with yes
  allnoconfig   - New config where all options are answered with no
Other generic targets:
  all    - Build all targets marked with [*]
* vmlinux   - Build the bare kernel
* modules   - Build all modules
  modules_install - Install all modules to INSTALL_MOD_PATH (default: /)
  firmware_install- Install all firmware to INSTALL_FW_PATH
                    (default: $(INSTALL_MOD_PATH)/lib/firmware)
  dir/            - Build all files in dir and below
  dir/file.[ois]  - Build specified target only
  dir/file.ko     - Build module including final link
  modules_prepare - Set up for building external modules
  tags/TAGS   - Generate tags file for editors
  cscope   - Generate cscope index
  kernelrelease   - Output the release version string
  kernelversion   - Output the version stored in Makefile
  headers_install - Install sanitised kernel headers to INSTALL_HDR_PATH
                    (default: /usr/src/linux-2.6.31.5/usr)
Static analysers
  checkstack      - Generate a list of stack hogs
  namespacecheck  - Name space analysis on compiled kernel
  versioncheck    - Sanity check on version.h usage
  includecheck    - Check for duplicate included header files
  export_report   - List the usages of all exported symbols
  headers_check   - Sanity check on exported headers
  headerdep       - Detect inclusion cycles in headers
Kernel packaging:
  rpm-pkg         - Build both source and binary RPM kernel packages
  binrpm-pkg      - Build only the binary kernel package
  deb-pkg         - Build the kernel as an deb package
  tar-pkg         - Build the kernel as an uncompressed tarball
  targz-pkg       - Build the kernel as a gzip compressed tarball
  tarbz2-pkg      - Build the kernel as a bzip2 compressed tarball
Documentation targets:
 Linux kernel internal documentation in different formats:
  htmldocs        - HTML
  pdfdocs         - PDF
  psdocs          - Postscript
  xmldocs         - XML DocBook
  mandocs         - man pages
  installmandocs  - install man pages generated by mandocs
  cleandocs       - clean all generated DocBook files
Architecture specific targets (x86):
* bzImage      - Compressed kernel p_w_picpath (arch/x86/boot/bzImage)
  install      - Install kernel using (your) ~/bin/installkernel or  (distribution) /sbin/installkernel or
                  install to $(INSTALL_PATH) and run lilo
  fdp_w_picpath      - Create 1.4MB boot floppy p_w_picpath (arch/x86/boot/fdp_w_picpath)
  fdp_w_picpath144   - Create 1.4MB boot floppy p_w_picpath (arch/x86/boot/fdp_w_picpath)
  fdp_w_picpath288   - Create 2.8MB boot floppy p_w_picpath (arch/x86/boot/fdp_w_picpath)
  isop_w_picpath     - Create a boot CD-ROM p_w_picpath (arch/x86/boot/p_w_picpath.iso)
                  bzdisk/fdp_w_picpath*/isop_w_picpath also accept:
                  FDARGS="..."  arguments for the booted kernel
                  FDINITRD=file initrd for the booted kernel
  i386_defconfig           - Build for i386
  x86_64_defconfig         - Build for x86_64
  make V=0|1 [targets] 0 => quiet build (default), 1 => verbose build
  make V=2   [targets] 2 => give reason for rebuild of target
  make O=dir [targets] Locate all output files in "dir", including .config
  make C=1   [targets] Check all c source with $CHECK (sparse by default)
  make C=2   [targets] Force check of all c source with $CHECK
 
Execute "make" or "make all" to build all targets marked with [*]
For further info see the ./README file

以下内容为下载的内核源文件包中README部分内容:
DOCUMENTATION:
 - There is a lot of documentation available both in electronic form on   the Internet and in books, both Linux-specific and pertaining to   general UNIX questions.  I'd recommend looking into the documentation   subdirectories on any Linux FTP site for the LDP (Linux Documentation   Project) books.  This README is not meant to be documentation on the   system: there are much better sources available.
 - There are various README files in the Documentation/ subdirectory:   these typically contain kernel-specific installation notes for some    drivers for example. See ocumentation/00-INDEX for a list of what   is contained in each file.  Please read the Changes file, as it   contains information about the problems, which may result by upgrading   your kernel.
 - The Documentation/DocBook/ subdirectory contains several guides for   kernel developers and users.  These guides can be rendered in a   number of formats:  PostScript (.ps), PDF, HTML, & man-pages, among others.   After installation, "make psdocs", "make pdfdocs", "make htmldocs",   or "make mandocs" will render the documentation in the requested format.

INSTALLING the kernel source:
 - If you install the full sources, put the kernel tarball in a   directory  (eg. your home directory) and   unpack it:
  gzip -cd linux-2.6.XX.tar.gz | tar xvf -
   or
  bzip2 -dc linux-2.6.XX.tar.bz2 | tar xvf -

     Do NOT use the /usr/src/linux area! This area has a (usually
   incomplete) set of kernel headers that are used by the library header
   files.  They should match the library, and not get messed up by
   whatever the kernel-du-jour happens to be.
 - You can also upgrade between 2.6.xx releases by patching.  Patches are
   distributed in the traditional gzip and the newer bzip2 format.  To
   install by patching, get all the newer patch files, enter the
   top level directory of the kernel source (linux-2.6.xx) and execute:
  gzip -cd ../patch-2.6.xx.gz | patch -p1
   or
  bzip2 -dc ../patch-2.6.xx.bz2 | patch -p1
   (repeat xx for all versions bigger than the version of your current
   source tree, _in_order_) and you should be ok.  You may want to remove
   the backup files (xxx~ or xxx.orig), and make sure that there are no
   failed patches (xxx# or xxx.rej). If there are, either you or me has
   made a mistake.
   Unlike patches for the 2.6.x kernels, patches for the 2.6.x.y kernels
   (also known as the -stable kernels) are not incremental but instead apply
   directly to the base 2.6.x kernel.  Please read
   Documentation/applying-patches.txt for more information.
   Alternatively, the script patch-kernel can be used to automate this
   process.  It determines the current kernel version and applies any
   patches found
.
  linux/scripts/patch-kernel linux
   The first argument in the command above is the location of the
   kernel source.  Patches are applied from the current directory, but
   an alternative directory can be specified as the second argument.
 - If you are upgrading between releases using the stable series patches
   (for example, patch-2.6.xx.y), note that these "dot-releases" are
   not incremental and must be applied to the 2.6.xx base tree. For
   example, if your base kernel is 2.6.12 and you want to apply the
   2.6.12.3 patch, you do not and indeed must not first apply the
   2.6.12.1 and 2.6.12.2 patches. Similarly, if you are running kernel
   version 2.6.12.2 and want to jump to 2.6.12.3, you must first
   reverse the 2.6.12.2 patch (that is, patch -R) _before_ applying
   the 2.6.12.3 patch.
   You can read more on this in Documentation/applying-patches.txt
 - Make sure you have no stale .o files and dependencies lying around:
  cd linux
  make mrproper
   You should now have the sources correctly installed.
SOFTWARE REQUIREMENTS
   Compiling and running the 2.6.xx kernels requires up-to-date
   versions of various software packages.  Consult
   Documentation/Changes for the minimum version numbers required
   and how to get updates for these packages.  Beware that using
   excessively old versions of these packages can cause indirect
   errors that are very difficult to track down, so don't assume that
   you can just update packages when obvious problems arise during
   build or operation.
BUILD directory for the kernel:
   When compiling the kernel all output files will per default be
   stored together with the kernel source code.
   Using the option "make O=output/dir" allow you to specify an alternate
   place for the output files (including .config).
   Example:
     kernel source code: /usr/src/linux-2.6.N
     build directory:  /home/name/build/kernel
   To configure and build the kernel use:
   cd /usr/src/linux-2.6.N
   make O=/home/name/build/kernel menuconfig
   make O=/home/name/build/kernel
   sudo make O=/home/name/build/kernel modules_install install
   Please note: If the 'O=output/dir' option is used then it must be
   used for all invocations of make.
 
CONFIGURING the kernel:
   Do not skip this step even if you are only upgrading one minor
   version.  New configuration options are added in each release, and
   odd problems will turn up if the configuration files are not set up
   as expected.  If you want to carry your existing configuration to a
   new version with minimal work, use "make oldconfig", which will
   only ask you for the answers to new questions.
 - Alternate configuration commands are:
 "make config"      Plain text interface.
 "make menuconfig"  Text based color menus, radiolists & dialogs.
 "make xconfig"     X windows (Qt) based configuration tool.
 "make gconfig"     X windows (Gtk) based configuration tool.
 "make oldconfig"   Default all questions based on the contents of
      your existing ./.config file and asking about
      new config symbols.
 "make silentoldconfig"
      Like above, but avoids cluttering the screen
      with questions already answered.
      Additionally updates the dependencies.
 "make defconfig"   Create a ./.config file by using the default
      symbol values from either arch/$ARCH/defconfig
      or arch/$ARCH/configs/${PLATFORM}_defconfig,
      depending on the architecture.
 "make ${PLATFORM}_defconfig"
     Create a ./.config file by using the default
     symbol values from
     arch/$ARCH/configs/${PLATFORM}_defconfig.
     Use "make help" to get a list of all available
     platforms of your architecture.
 "make allyesconfig"
      Create a ./.config file by setting symbol
      values to 'y' as much as possible.
 "make allmodconfig"
      Create a ./.config file by setting symbol
      values to 'm' as much as possible.
 "make allnoconfig" Create a ./.config file by setting symbol
      values to 'n' as much as possible.
 "make randconfig"  Create a ./.config file by setting symbol
      values to random values.
   You can find more information on using the Linux kernel config tools
   in Documentation/kbuild/kconfig.txt.
 NOTES on "make config":
 - having unnecessary drivers will make the kernel bigger, and can
   under some circumstances lead to problems: probing for a
   nonexistent controller card may confuse your other controllers
 - compiling the kernel with "Processor type" set higher than 386
   will result in a kernel that does NOT work on a 386.  The
   kernel will detect this on bootup, and give up.
 - A kernel with math-emulation compiled in will still use the
   coprocessor if one is present: the math emulation will just
   never get used in that case.  The kernel will be slightly larger,
   but will work on different machines regardless of whether they
   have a math coprocessor or not.
 - the "kernel hacking" configuration details usually result in a
   bigger or slower kernel (or both), and can even make the kernel
   less stable by configuring some routines to actively try to
   break bad code to find kernel problems (kmalloc()).  Thus you
   should probably answer 'n' to the questions for
          "development", "experimental", or "debugging" features.
COMPILING the kernel:
 - Make sure you have at least gcc 3.2 available.   For more information, refer to Documentation/Changes.   Please note that you can still run a.out user programs with this kernel.
 - Do a "make" to create a compressed kernel p_w_picpath. It is also   possible to do "make install" if you have lilo installed to suit the   kernel makefiles, but you may want to check your particular lilo setup first.   To do the actual install you have to be root, but none of the normal   build should require that. Don't take the name of root in vain.
 - If you configured any of the parts of the kernel as `modules', you   will also have to do "make modules_install".
 - Verbose kernel compile/build output:
   Normally the kernel build system runs in a fairly quiet mode (but not   totally silent).  However, sometimes you or other kernel developers need   to see compile, link, or other commands exactly as they are executed.
   For this, use "verbose" build mode.  This is done by inserting   "V=1" in the "make" command.  E.g.: make V=1 all
   To have the build system also tell the reason for the rebuild of each   target, use "V=2".  The default is "V=0".
 - Keep a backup kernel handy in case something goes wrong.  This is    especially true for the development releases, since each new release   contains new code which has not been debugged.  Make sure you keep a  backup of the modules corresponding to that kernel, as well.  If you   are installing a new kernel with the same version number as your
   working kernel, make a backup of your modules directory before you   do a "make modules_install".
   Alternatively, before compiling, use the kernel config option
   "LOCALVERSION" to append a unique suffix to the regular kernel version.
   LOCALVERSION can be set in the "General Setup" menu.
 - In order to boot your new kernel, you'll need to copy the kernel   p_w_picpath (e.g. .../linux/arch/i386/boot/bzImage after compilation)   to the place where your regular bootable kernel is found.
 - Booting a kernel directly from a floppy without the assistance of a   bootloader such as LILO, is no longer supported.
   If you boot Linux from the hard drive, chances are you use LILO which   uses the kernel p_w_picpath as specified in the file /etc/lilo.conf.  The   kernel p_w_picpath file is usually /vmlinuz, /boot/vmlinuz, /bzImage or   /boot/bzImage.  To use the new kernel, save a copy of the old p_w_picpath   and copy the new p_w_picpath over the old one.  Then, you MUST RERUN LILO   to update the loading map!! If you don't, you won't be able to boot
   the new kernel p_w_picpath.   Reinstalling LILO is usually a matter of running /sbin/lilo.
   You may wish to edit /etc/lilo.conf to specify an entry for your   old kernel p_w_picpath (say, /vmlinux.old) in case the new one does not   work.  See the LILO docs for more information.    After reinstalling LILO, you should be all set.  Shutdown the system,
   reboot, and enjoy!   If you ever need to change the default root device, video mode,
   ramdisk size, etc.  in the kernel p_w_picpath, use the 'rdev' program (or   alternatively the LILO boot options when appropriate).  No need to   recompile the kernel to change these parameters.
 - Reboot with the new kernel and enjoy.
IF SOMETHING GOES WRONG:
 - If you have problems that seem to be due to kernel bugs, please check
   the file MAINTAINERS to see if there is a particular person associated
   with the part of the kernel that you are having trouble with. If there
   isn't anyone listed there, then the second best thing is to mail
   them to me (torvalds@linux-foundation.org), and possibly to any other
   relevant mailing-list or to the newsgroup.
 - In all bug-reports, *please* tell what kernel you are talking about,
   how to duplicate the problem, and what your setup is (use your common
   sense).  If the problem is new, tell me so, and if the problem is
   old, please try to tell me when you first noticed it.
 - If the bug results in a message like
 unable to handle kernel paging request at address C0000010
 Oops: 0002
 EIP:   0010:XXXXXXXX
 eax: xxxxxxxx   ebx: xxxxxxxx   ecx: xxxxxxxx   edx: xxxxxxxx
 esi: xxxxxxxx   edi: xxxxxxxx   ebp: xxxxxxxx
 ds: xxxx  es: xxxx  fs: xxxx  gs: xxxx
 Pid: xx, process nr: xx
 xx xx xx xx xx xx xx xx xx xx
   or similar kernel debugging information on your screen or in your
   system log, please duplicate it *exactly*.  The dump may look
   incomprehensible to you, but it does contain information that may
   help debugging the problem.  The text above the dump is also
   important: it tells something about why the kernel dumped code (in
   the above example it's due to a bad kernel pointer). More information
   on making sense of the dump is in Documentation/oops-tracing.txt
 - If you compiled the kernel with CONFIG_KALLSYMS you can send the dump
   as is, otherwise you will have to use the "ksymoops" program to make
   sense of the dump (but compiling with CONFIG_KALLSYMS is usually preferred).
   This utility can be downloaded from
   .kernel.org/pub/linux/utils/kernel/ksymoops/">ftp://ftp.<country>.kernel.org/pub/linux/utils/kernel/ksymoops/ .
   Alternately you can do the dump lookup by hand:
 - In debugging dumps like the above, it helps enormously if you can
   look up what the EIP value means.  The hex value as such doesn't help
   me or anybody else very much: it will depend on your particular
   kernel setup.  What you should do is take the hex value from the EIP
   line (ignore the "0010:"), and look it up in the kernel namelist to
   see which kernel function contains the offending address.
   To find out the kernel function name, you'll need to find the system
   binary associated with the kernel that exhibited the symptom.  This is
   the file 'linux/vmlinux'.  To extract the namelist and match it against
   the EIP from the kernel crash, do:
  nm vmlinux | sort | less
   This will give you a list of kernel addresses sorted in ascending
   order, from which it is simple to find the function that contains the
   offending address.  Note that the address given by the kernel
   debugging messages will not necessarily match exactly with the
   function addresses (in fact, that is very unlikely), so you can't
   just 'grep' the list: the list will, however, give you the starting
   point of each kernel function, so by looking for the function that
   has a starting address lower than the one you are searching for but
   is followed by a function with a higher address you will find the one
   you want.  In fact, it may be a good idea to include a bit of
   "context" in your problem report, giving a few lines around the
   interesting one.
   If you for some reason cannot do the above (you have a pre-compiled
   kernel p_w_picpath or similar), telling me as much about your setup as
   possible will help.  Please read the REPORTING-BUGS document for details.
 - Alternately, you can use gdb on a running kernel. (read-only; i.e. you
   cannot change values or set break points.) To do this, first compile the
   kernel with -g; edit arch/i386/Makefile appropriately, then do a "make
   clean". You'll also need to enable CONFIG_PROC_FS (via "make config").
   After you've rebooted with the new kernel, do "gdb vmlinux /proc/kcore".
   You can now use all the usual gdb commands. The command to look up the
   point where your system crashed is "l *0xXXXXXXXX". (Replace the XXXes
   with the EIP value.)
   gdb'ing a non-running kernel currently fails because gdb (wrongly)
   disregards the starting offset for which the kernel is compiled.