TITLE: Crosscompiling on IA-32 (x86) platforms LFS VERSION: All (fixes for 4.1) AUTHOR: Daniel Baumann SYNOPSIS: How to compile a LFS system on a fast machine and use it on your old box HINT: CHANGELOG ========= * 2003-07-23 Initial revision DISCLAIMER ========== You use this LFS-Hint at your own risk. Neither the author, nor the Linux From Scratch project accepts any reponsibility for anything that happens when using this document or associated files. This hint is originally written by Nicholas Dille , now updated and maintained by Daniel Baumann . I would like to thank Nicholas for allowing to continue his good work. INTRODUCTION ============ Alright, we have gathered here to mourn our failing in compiling our favourite LFS system for our beloved outdated hardware and have either dived right into it being stopped by unforseen difficulties or been hinted at this document to circumvent some known problems. Lets say you call an old i486 your own but moved on to some i686 based box some time ago. Now you decided that the i486 might serve wonderfully as your personal file/printer server, internet gateway, Ogg player, a combination or some entirely different purpose. Without even trying you can imagine that it will probably take quite some time to compile a LFS system caused by the lack of memory, harddisk space and mere performance. Wouldn't it be nice to have your, by comparison, blindingly fast i686 compile LFS and then use the resulting system on your i486? I can assure you that several people had the same idea and tried solving the difficulties encountered. I hope to compile most of those experiences here to help you find the least time consuming path from a bare i486 to a lovely lfs'd linux box. To give you a rough outline how I organized the hint let me elaborate shortly: We will learn what problems you may encounter, under what conditions, why they exist AFAIK and then at last come to the part where you can actively profit from this by giving it a try. In any case if you feel the hint missing something important, doesn't explain something to your convenience, completely missed some point or if you'd just like to give me a big hug, I'd be honestly happy to hear you out. Nicholas like to thank: * Gerard Beekmans, for this wonderful project * Tommy Wareing, for the uname hack * Arthur H. Johnson II, for bullying me into rewriting this hint and helping collecting new material * Christophe Devine, for the kernel module uname hack * Yann Guidon, for extending the kernel module uname hack PREPARATION =========== Why the naive approach fails ---------------------------- In case you only paste the commands provided by the LFS book to your console, wait some time and come back for the next package you might want to read through this to understand what really happens. Let us devide the package which you are about to compile into two destinct categories: the friendly and the painful packages. Friendly packages are configured using GNU autoconf (./configure [...]) which usually check for a sane environment and also try to determine which system it will be compiled on. This information is then passed on to gcc which then knows which architecture specific features it is allows to use. These packages are easily persuaded to believe they are being compiled on a weaker platform. In contrast, the painful packages do not provide any automatic check whether their prerequisites are satisfied and trust the user to know what he is doing. In this case the decision which platform is present is left to gcc which uses uname and CFLAGS/CXXFLAGS to guess it. Unfortunately these packages are harder to force to compile for a weaker platform. So, if you leave any of those packages alone they will compile for the platform they detect either by the configure script or by means of gcc. We need to make autoconf and gcc think that the underlying architecture is of a different type. $CHOST ------ Autoconf normally tries to guess the type of machine we are compiling on. Because we use our new computer instead of the old, we have to fake this value manually. You can get the correct value with # uname -m from your target computer. If you haven't already a running Linux system on the target machine, use a corresponding value from this list instead: Intel 386: i386 Intel 486: i486 Intel Pentium 1 and MMX: i586 Intel Pentium Pro, 2, 3 and 4: i686 AMD K6-1, K6-2 and K6-3: i586 AMD Athlon (all versions): i686 Export this to the build machine with: # export CHOST=$VALUE-pc-linux-gnu or # export CHOST=i586-pc-linux-gnu as an example for an Intel Pentium 1. $CFLAGS/$CXXFLAGS ----------------- CFLAGS is an environment variable which is used by any honourable package to pass user flags on to gcc. It is often used to set the optimization level (-On where n is a number). For details please refer to the optimization hint. In our case we additionally try to force gcc to compile for the desired architecture by setting CFLAGS. Choose from this list: Intel 386: i386 Intel 486: i486 Intel Pentium 1: pentium Intel Pentium MMX: pentium-mmx Intel Pentium Pro: pentiumpro Intel Pentium 2: pentium2 Intel Pentium 3: pentium3 Intel Pentium 4: pentium4 AMD K6-1: k6 AMD K6-2: k6-2 AMD K6-3: k6-3 AMD Athlon 1 (Classic): athlon AMD Athlon 2 (Thunderbird): athlon-tbird AMD Athlon 3 (XP): athlon-xp AMD Athlon 4 (Palomino): athlon-4 AMD Athlon 5 (MP): athlon-mp Set it to the system with # export CFLAGS=-march=$VALUE # export CXXFLAGS=-march=$VALUE or # export CFLAGS=-march=pentium # export CXXFLAGS=-march=pentium as an example for Intel Pentium 1. Consult the gcc online-manual on http://gcc.gnu.org/ for other values. If you know what you do, you can also set some optimization flags, see the optimization hint for that. Installing friendly packages ============================ As I have stated before configure scripts are specially easy to persuade into believing to be running on a different architecture. You will have to add "--host=$CHOST --target=$CHOST" to the configure scripts' arguments. As an example, for Bash in Chapter 5, you have to use # ./configure --host=$CHOST --target=$CHOST \ --enable-static-link --prefix=$LFS/static --with-curses Installing painful packages =========================== This chapter will provide you with a list of packages which do not come with a configure script or which come with one that does not honour '--host' and '--target' options. Use the following fixes to proper install the packages. Bin86 0.16.3 ------------ Use the following commands to write your CFLAGS variable into bin86's Makefile. # make CFLAGS="-D_POSIX_SOURCE $CFLAGS" Bzip2 1.0.2 ----------- Use the following commands to patch bzip2: # cp Makefile Makefile.backup # sed -e 's%$(BIGFILES)%$(BIGFILES) $(OPT)%' \ Makefile.backup > Makefile # cp Makefile-libbz2_so Makefile-libbz2_so.backup # sed -e 's%$(BIGFILES)%$(BIGFILES) $(OPT)%' \ Makefile-libbz2_so.backup > Makefile-libbz2_so Add 'OPT="$CFLAGS"' to the make commands: For chapter 5: # make CC="gcc -static" OPT="$CFLAGS" For chapter 6: # make -f Makefile-libbz2_so OPT="$CFLAGS" # make OPT="$CFLAGS" Gcc 3.1.2 --------- Gcc compilation is devided into three parts. Stage1 is the first run, where CFLAGS/CXXFLAGS are respected well. In stage2, gcc compiles itselfs. To make use of our flags, we use BOOT_CFLAGS to pass through our settings. Stage3 is just a check-run to verify a working gcc. For chapter 5: # make BOOT_CFLAGS=$CFLAGS BOOT_LDFLAGS="-static" bootstrap For chapter 6: # make BOOT_CFLAGS=$CFLAGS bootstrap Kbd 1.08 -------- Use the following commands to write your CFLAGS variable into kbd's Makefile. # make CFLAGS="$CFLAGS" Lilo 22.2 --------- Use the following make commands: # make OPT="$CFLAGS" Man 1.5k -------- Use the following commands to add your CFLAGS variable to man2html's Makefile: # cp man2html/Makefile.in man2html/Makefile.in.backup # sed -e "s/CFLAGS = /CFLAGS = $CFLAGS /" \ man2html/Makefile.in.backup > man2html/Makefile.in Net-tools 1.60 -------------- Use the following make command: # make COPTS="-D_GNU_SOURCE -Wall $CFLAGS" Netkit-base 0.17 ---------------- This package does not need any changes to the book's commands. It will pick the value of the CFLAGS variable and apply it correctly. Perl 5.8.0 ---------- The configure scripts shipped with perl are not autoconf-scripts. They use diffrent flags for the same meaning: # ./configure.gnu -Dhost=$CHOST -Dtarget=$CHOST -Darchname=$CHOST \ --prefix=/usr Procinfo 18 ----------- Add 'CFLAGS="$CFLAGS"' to the make command: # make LDLIBS=-lncurses CFLAGS="$CFLAGS" Procps 3.1.5 ------------ Add 'OPT="$CFLAGS"' to the make command: # make OPT="$CFLAGS" Please note that the build process will fail with top.c if your CFLAGS variable did not contain the -On switch. If that is the case please use the following line: # make OPT="$CFLAGS -O2" This switch will not add any additional kind of optimization to your build pro- cess because -O2 is usually assumed by gcc in case it is not told otherwise. I will contact the author about this behaviour of procps and update the hint accordingly. Psmisc 21.2 ----------- Add 'AM_CFLAGS="$CFLAGS"' to the make command: # make AM_CFLAGS="$CFLAGS" Sysklogd 1.4.1 -------------- Add 'RPM_OPT_FLAGS="$CFLAGS"' to the make command: # make RPM_OPT_FLAGS="$CFLAGS" Sysvinit 2.84 ------------- Add 'CFLAGS="-Wall -D_GNU_SOURCE $CFLAGS"' to the make command: # make -C src CFLAGS="-Wall -D_GNU_SOURCE $CFLAGS" APPENDIX ======== The uname hack -------------- Generally spoken the uname hack will fool the system. It pretends that linux is running on different hardware. For example, you are using an i686 based system to compile a lfs for an old i486. Then the uname hack would force uname to lie about the underlying hardware - not reporting i686 but i486. . Traditional Save the original uname program with: # mv /bin/uname /bin/uname.backup Create the new uname with: cat > /bin/uname < "EOF" #!/bin/sh /bin/uname.backup "$@" | sed "s/i[456]86/$ARCH/" EOF And give it the correct permissions: # chmod 755 /bin/uname The traditional uname hack will have to be installed three times: once before you start your lfs compilation to replace the uname binary of your host system, then when you enter chroot to force your statically linked environment into assuming a different architecture and last after installing sh-utils in chapter 6 because they will replace the uname binary. . Kernel module Please check out the niftiness of this version of the uname hack! Put the code below into the file uname_i586.c and compile with the following command: # gcc -I/usr/src/linux/include -c -DMODULE \ -DUNAME_DUMB_STEPPING=\'4\' uname_i586.c By adjusting the '4' in the compilation command you will be able to specify the desired platform at compile time without changing the code :) Listing of uname_i586.c: /* This simple piece of code simply turns your ix86 into a i586 - useful if you're cross-compiling for a weaker platform. Compile with: gcc -I/usr/src/linux/include -c -DMODULE uname_i586.c and then: insmod ./uname_i586.o Revision : whygee@f-cpu.org, Wed Apr 10 07:45:11 CEST 2002 : added the UNAME_DUMB_STEPPING parameter/ifdef so the user can modify it at compile time. it's just a quick hack. your command line for simulating a 486 will look like : gcc -I/usr/src/linux/include -c -DMODULE \ -DUNAME_DUMB_STEPPING=\'4\' uname_i586.c */ #include #include #ifndef UNAME_DUMB_STEPPING #define UNAME_DUMB_STEPPING '5'; /* #error "no stepping specified." */ #endif char save; int init_module( void ) { save = system_utsname.machine[1]; system_utsname.machine[1] = UNAME_DUMB_STEPPING; return( 0 ); } void cleanup_module( void ) { system_utsname.machine[1] = save; } Notes ----- . Compiling Glibc without optimization fails (need at leas -O1) END OF CROSSCOMPILING-X86-HINT