Saving Space mini-HOWTO By Guido Gonzato, mailto:guido@ibogeo.df.unibo.it v1.0.1, 7 April 1999 ------------------------------------------------------------------------------- This mini-HOWTO gives you directions for squeezing your Linux installation into the least possible space. It's particularly aimed at notebook users. ------------------------------------------------------------------------------- 1. Introduction I've got a notebook I installed Linux on, beside Windows 95 that was pre-- installed. I squeezed the Windows partition to 500 Mb, making space for a 240 Mb Linux one. Small though the latter may seem, I've managed to install a fairly complete Linux system on it, based on Red Hat 4.1 and some magic to save as much space as possible. If you're in need for space, the indications you'll find in the following sections will free up a considerable chunk of hard disk. The only caveat I'm giving you is: don't blame me if something goes wrong! We'll be using a couple of programs that worked fine for me, but are inherently dangerous. You've been warned. 2. Software requirements You need: * a fully working Linux system (any version should be OK); * the common gzip compression utility, or alternatively bzip2 that you find on ftp://sunsite.unc.edu:/pub/Linux/utils/compress; this tool compresses better than gzip, but it's also a wee bit slower and consumes a lot of memory; * the upx executable compressor, whose home page is at http://cdata.tvnet.hu/ ~ml/upx.html and http://wildsau.idv.uni-linz.ac.at/mfx/upx.html; * the zlibc package, available on ftp://sunsite.unc.edu:/pub/Linux/libs/ compression. It's called zlibc-X.X.tar.gz, where X.X is the latest version. There are other executable compressors. One is gzexe (forget it), while tcx was probably the best before upx became available; yet another is tzx, in theory better than tcx. The thing is, it screwed everything up when I tried it on a spare PC, though I'm not sure if I did something wrong. You had better stick with upx: it's reliable, efficient, and very handy. There are kernel patches that provide transparent file system compression a la Stacker, but as of this writing none of them has yet a reputation for stability and reliability. For safety's sake, steer clear. 3. The procedure 3.1 Removing the Kernel Sources The kernel sources take up more than 20 Mb, and you may want to remove them. If so, I suggest that you compile a new kernel that is tailored to your machine once and for all, then get rid of the sources. Take care, though. I don't recommend that you remove the kernel sources unless you are sure that your machine is properly configured. Besides, you need the kernel #includes to compile C programs. Think twice! If you do decide so, do not remove the include/linux tree unless you know you'll never compile applications on your machine. 3.2 Pruning Out Applications Now, decide which applications you really need. Some may prove redundant: for instance, are you sure you can't live without emacs? You could use jed instead. It's up to you to decide what you want to keep. Some general points: * gcc is a fairly large package. It's needed to recompile the kernel and all the applications for which there are no pre--built Linux binaries. It's also needed, of course, if you write your own C or Fortran (with f2c or g77) programs; consider your needs before removing it. Other compilers like lcc are fine, but not up to gcc level. I suggest that you keep it; * X11 is awfully large, but it's a nice thing to have. If you decide you can't give it up, at least try to make do with as little as necessary: only the right X server, one simple window manager, only one xterm, no 100 dpi fonts, and so on; * TeX and accompanying packages are very large indeed. Giving up LaTeX and sticking to plain TeX saves a lot of space; giving up X11 previewers like xdvi and ghostview is possible if you use dvitty, dvivga and some such. Eliminating the need for X11 dviware could even make X11 redundant; * games are never ``necessary''. 3.3 Stripping Binaries Let's start reducing the size of binaries. Move to /usr/bin and issue the command machine:/usr/bin# strip * which will do away with symbols embedded in binaries. Repeat this step in /usr/ X11R6/bin/ and other directories containing executables you may have (don't forget to locate TeX's and gcc's binaries), but do not run it under /sbin, / bin, or /usr/sbin/ if you value your installation! 3.4 Compressing with upx First, install upx and read its documentation. Then go to /usr/bin and run it with the command upx *; it will compress all executables, including suid ones (tcx wouldn't). Repeat this step in other directories as seen above. Remember to compress the executables whenever you install a new package! 3.5 Gzipping around There are lots of other files that can be compressed once and for all. Let's start with /usr/doc/; move to this directory and issue the command machine:/usr/doc# find . -type f -exec gzip -9 {} \; 2> /dev/null Remember to compress the docs whenever you install a new package! Repeat this step in the directory containing the documentation for TeX (on my system, /usr/lib/texmf/texmf/doc/. If you're really sure, remove these directories altogether. Now, install zlibc and compile it. If your system is like mine, the compilation process will abort complaining about a missing (static) libc. Never mind; you'll find a file called uncompress.o that is what's needed. Move it to /usr/ local/lib/ and add this line to your /etc/profile: export LD_ELF_PRELOAD=/usr/local/lib/uncompress.o Now, you can compress with gzip not only documentation, but also data files: the applications that use them will be able to use them nonetheless. In theory, the trick should work with most applications, but in practice your degree of success may vary. Mine was rather unimpressive. 4. A Real Life Example This is what I got applying the above procedure to one of my machines. Before the treatment, df reported I used 398,798 1024--blocks: * I didn't remove the kernel sources and the kernel headers; * I uninstalled several applications and all of the games, but I left X11, X11 development, C and Fortran development, Tcl/Tk, networking tools, and a few other standard applications. df reported 244,668 used blocks; * I ran upx on /usr/bin, /usr/X11R6/bin, /usr/lib/texmf/bin/i586-linux, and / usr/lib/gcc-lib/i386-linux/2.7.2.1. 226,270 used blocks; * I compressed the documentation under /usr/doc and /usr/lib/texmf/texmf/doc: 198,745 used blocks. To sum up, I started with 398,798 blocks and finished with 198,745. Think of the stuff you can shove in those 200,000 spared blocks! I would have saved even more if I had used bzip2 instead of gzip. On average, if you are careful from the beginning and install only the necessary applications, compressing executables and documents will save you some 20 Mb. On a notebook, this can be a lifesaver. 5. The End 5.1 Copyright Unless otherwise stated, Linux HOWTO documents are copyrighted by their respective authors. Linux HOWTO documents may be reproduced and distributed in whole or in part, in any medium physical or electronic, as long as this copyright notice is retained on all copies. Commercial redistribution is allowed and encouraged; however, the author would like to be notified of any such distributions. All translations, derivative works, or aggregate works incorporating any Linux HOWTO documents must be covered under this copyright notice. That is, you may not produce a derivative work from a HOWTO and impose additional restrictions on its distribution. Exceptions to these rules may be granted under certain conditions; please contact the Linux HOWTO coordinator at the address given below. In short, we wish to promote dissemination of this information through as many channels as possible. However, we do wish to retain copyright on the HOWTO documents, and would like to be notified of any plans to redistribute the HOWTOs. If you have questions, please contact Tim Bynum, the Linux HOWTO coordinator, at tjbinum@sunsite.unc.edu via email. 5.2 Disclaimer ``Saving Space mini-HOWTO''was written by Guido Gonzato, REMOVE_MEguido@ibogeo.df.unibo.it. This document is provided ``as is''. I put great effort into writing it as accurately as I could, but you use the information contained in it at your own risk. In no event shall I be liable for any damages resulting from the use of this work. Feedback is welcome. For any requests, suggestions, flames, etc., feel free to contact me. Enjoy Linux and life, Guido =8-)