Using the 'patch' and 'diff' utilities

Windows Research Kernel @ HPI

In some upcoming postings, we will describe simple demo modifications of the Windows Research Kernel. The examples will be downloadable from this site as kernel patches and small demo applications.

The required tools can be downloaded from the Win32 GNU tools sourceforge page. You need the DiffUtils and the Patch tool. Download and install the tools on your workstation.

The diff tool can be used to write the differences between two source trees (normally a modified and an unmodified version) into a single (text) file. With patch a created patch file can be applied to an unmodified source tree. In this way, kernel modifications can be distributed without having to provide the whole source code and without a common repository.

Creating a patch

It is assumed that you have a directory %temp% with two subdirectories %temp%\original and %temp%\modified. Each of this subdirectories should only contain the source code without object files or other intermediately created files. In a cmd-shell opened in the %temp% directory a patch can be created by typing

diff -Naur original modified --strip-trailing-cr > test.patch.

The -Naur option tells the diff command that (1) missing files in the either directory are considered as empty files, (2) all files are considered as text files, (3) three lines of "context" information are put into the patch file, and (3) the source code directories are traversed and compared recursively. Afterwards, the file test.patch contains the patch information which can be distributed by EMail or put on a website for download.

A simple example: the following boxes show the original text file and the modified version. Both files are saved in a directory structure as described above.


abcdefg !!!

The diff utility creates the following patch file:

diff -Naur original/test.txt modified/test.txt
- original/test.txt 2007-05-22 14:50:51.841199200 +0200
+++ modified/test.txt 2007-05-22 14:56:16.670944400 +0200
@@ -1,5 +1,4 @@
+abcdefg !!!
\ Kein Zeilenumbruch am Dateiende.

The '+' and '-' signs at the beginning of the lines in the patch file indicate whether a line was added/modified or removed.

Applying a patch

It is assumed that you have an unmodified base source code tree of the original source code version saved in %original%.

If you get a patch file created as described above, you can copy it to the %original% directory and (if the patch file is named %name%.patch) type

patch -p1 < %name%.patch.

This will apply the patch to the unmodified version and transform it into the patched version.

The p option tells the patch tool to strip a specific number of leading components from file names used in the patch file. An example: the patch file shown above contains the directory names ("original" and "modified") used by the patch creator. If you have a directory "foo" which contains the orginal source code, the patch utility needs information about how to resolve the file names. Therefore, the first component (the patch creation directory) can be stripped from the filenames by using -p1.


With diff and patch patches which contain only the differences between two source code trees can be created and re-applied to unmodified versions. In this way modified source code versions can be easily distributed or made available for download.


Comments are closed.