Category Archives: shell

How to extract from tgz, rpm, and deb files

Posted on by
Reply

The most common way to move a bundle of files around in Linux is a combination of tar (tape archive), which appends all of the files together into a single large file,  and gzip, a compression utility.  These are often referred to as tarball. They will have an extension like tar.gz or tgz.  Sometimes bzip2 is used as the comprseeion utility, and they file will end with tar.bz2.

To see the list of files in the  tarball for sage, run:

tar -ztf  sage-2.10.2-linux-ubuntu64-opteron-x86_64-Linux.tar.gz

To extract the files, create a target directory and change to it, thenextract the file from it’s original location

mkdir /tmp/sage

cd /tmp/sage

tar -zxf  ~/Desktop/sage-2.10.2-linux-ubuntu64-opteron-x86_64-Linux.tar.gz

All of the files will be put into the current directory.  There is no rule that says that all the files in a tarball are under on subdirectory, so it really behooves you to do this in a empty directory.  That way you know all of the files you see post extraction are files from the tarball.

Debian uses a package management system called dpkg based on this technology.  The packages will end with .deb, but you can see what Linux thinks the  file type is by using the file command.  Here it is run on automake_1.10+nogfdl-1_all.deb:

 adyoung@adyoung-devd$ file ~/Desktop/automake_1.10+nogfdl-1_all.deb
/home/adyoung/Desktop/automake_1.10+nogfdl-1_all.deb: Debian binary package (format 2.0)

To See the list of included files, run:

dpkg –contents automake_1.10+nogfdl-1_all.deb

And to extract use the –extract command line parameter.  Note that you have to supply the target directory as well.

dpkg –extract automake_1.10+nogfdl-1_all.deb /tmp/deb/

Again, make sure the target directory is empty to avoid intermingling your own files and files from the package.

The letters RPM stand for Redhat Package Manager.  The file extension rpm is used for packages of software designed to be installed on a some distributions of GNU/Linux.   RPM is used on Redhat, SuSE, and distributions based off of these two major distributions, such as Fedora, CentOS, and  OpenSuSE.  RPMs are shipped in a format called cpio. This format has the advantage of allowing longer file names, and providing stroage and compression all in one utility and format.  However, RPMs are not exactly cpio format, and you have to run a converter first, before you can extract the files.  This converter is called rpm2cpio.  It reads the filename in as the first command line parameter, and outputs the cpio file to standard output.  So if you run it without redirecting output, you are going to spew binary data all over your terminal.  Better to redirect it into the cpio utility, with the command line switches -di.  These switches mean extract the files, and build any subdirectories required.  Again, run this in a clean directory:

rpm2cpio ../testware-e.x.p-00000.i386.rpm | cpio -di

Reading C++ symbols in a binary

Posted on by
Reply

objdump -t will pull the raw symbols out of an elf binary, but it is mangled format like this

0000000000000000       F *UND*  0000000000000006              _ZNSt13basic_ostreamIwSt11char_traitsIwEElsEPFRS2_S3_E@@GLIBCXX_3.4

c++ file will translate this to a human readly string that maps to the rigianlfunction definition:

 adyoung@adyoung-devd$ echo _ZNSt13basic_ostreamIwSt11char_traitsIwEElsEPFRS2_S3_E@@GLIBCXX_3.4 | c++filt
std::basic_ostream<wchar_t, std::char_traits<wchar_t> >::operator<<(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& (*)(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >&))@@GLIBCXX_3.4

To do this for an entire file, you want to print only the last column, a task custom made for awk:

 objdump -t   casttest | awk ‘{print $(NF)}’ | c++filt

dpkg-others

Posted on by
Reply

dpkg-query -S `which $1` | cut -d: -f1 | xargs dpkg-query -L

This command finds other files in the same debian package as a given executable.  I put it into a file called dpkg-others.  Example run:

/home/adyoung/bin/dpkg-others ODBCConfig
/.
/usr
/usr/bin
/usr/bin/DataManager
/usr/bin/DataManagerII
/usr/bin/ODBCConfig
/usr/bin/odbctest
/usr/share
/usr/share/doc
/usr/share/doc/unixodbc-bin
/usr/share/doc/unixodbc-bin/copyright
/usr/share/doc/unixodbc-bin/changelog.gz
/usr/share/doc/unixodbc-bin/changelog.Debian.gz
/usr/share/menu
/usr/share/menu/unixodbc-bin
/usr/share/man
/usr/share/man/man1
/usr/share/man/man1/ODBCConfig.1.gz

How to reset the root password on a GRUB based Linux boot

Posted on by
Reply

If you forget or somehow manage to change the root password on a machine running various flavors of glibc based security, and you are running a Linux kernel here are the steps to reset it.

  1.  Reboot the  machine.  This assume physical access, but reset the root password requires that anyway.
  2. At the GRUB prompt select the kernel you want and hit ‘e’ for edit.
  3. At the end of the kernel boot parameters add the word ‘single’.  This means boot into single user mode, and should bypass the need to type in a password.
  4. hit ‘b’ to boot.
  5. Once a Command prompt appears, use the passwd utility to reset the machine.
  6. Reboot.  Or, you can type ‘init 3’ or ‘init 5’ to complete the bot process.  Use 3 for servers, 5 for machines with graphical displays.

rpm and deb commands: files to packages and back

Posted on by
1

These two commands tell you which package owns a particular file for an RPM or DEB based system respectively.

rpmquery –whatprovides <path-to-file>

dpkg-query –search <path-to-file>

(The short form for dpkg is -S)

To list the files owned by a package use:

rpmquery –list <package>

dpkg-query -L <package>

(This long form for debian is –listfiles )

A useful utility to find all the other files in a package for a given binary (in this case lsof) is:

rpmquery –list $(rpmquery –whatprovides $(which lsof ) )

Or

dpkg-query -L $(dpkg-query –search $( which lsof ) | sed ‘s!\:.*!!’ )

extracting files from an rpm

Posted on by
Reply

rpms are saved in a file format called cpio, with a few minor changes.  To convert an rpm to this format use rpm2cpio.  The cpio binary is responsible for manipulating these files.  So if you want to treat that rpm as just another archive, and extract the files in it, you can link a couple of tools.  Here’s the line:

rpm2cpio your.rpm | cpio -di

If you don’t have arrays…but you have awk

Posted on by
Reply

A small project I have in the embedded side of things requires keeping track of information per index, what one would normally do in an array. However, the ash shell from busy box does not support arrays. Busybox does support awk. The following function is used to increment, decrement, and reset and array of counters of single digits stored as characters in a string.

#mod_field takes three parameters: OP INDEX DATA

#OP is the opperation: inc, dec, reset

#INDEX is the offset into the string from 1 to 9

#DATA is the state stored as a string

#OUTPUT is DATA with DATA[INDEX] modified

mod_field(){
OP=$1
INDEX=$2
DATA=$3

echo $OP $INDEX $DATA | \
awk ‘
{
VAL=substr($3,$2,1);
FIRST=substr($3,0,$2-1);
TAIL=substr($3,$2+1) ;
}
$1 ~ /inc/ { VAL++; }
$1 ~ /dec/ { VAL–; }
$1 ~ /reset/{ VAL=0; }
END{ print FIRST VAL TAIL;}
‘
}

#Here is how to call it:
DATA=0000000000
DATA=`mod_field inc 3 $DATA `
echo $DATA

#This should produce 0010000000

DATA=`mod_field dec 3 $DATA `
echo $DATA

#And set DATA back to 0000000000
for i in 1 2 2 3 4 4 4 4 5 6 6 7 8 8 8 8 8 8 9
do
DATA=`mod_field inc $i $DATA `
done
echo $DATA
#Should produce 1214121610

for i in 1 2 2 3 4 4 4 4 5 6 6 7 8 8 8 8 8 8 9
do
DATA=`mod_field dec $i $DATA `
done
echo $DATA

#Should produce 0000000000

DATA=1111111111111
DATA=`mod_field reset 3 $DATA`
echo $DATA

#Should produce 0000000000