doc: document building and usage of iptables with bpfilter

Add doc/iptables.md to document how to build iptables to be used with
bpfilter.

Signed-off-by: Quentin Deslandes <[email protected]>

README.md

@@ -26,6 +26,8 @@ make -C build test
 
 `bpfilter` daemon will be in `$BUILD/src/bpfilter`, and `libbpfilter.so` will be in `$BUILD/lib/libbpfilter.so`.
 
+For more details about build and using `iptables` with `bpfilter`, see [doc/iptables.md](doc/iptables.md).
+
 ## License
 
 bpfilter is GPLv2 licensed, as found in the COPYING file.

doc/iptables.md

@@ -0,0 +1,77 @@
+# `bpfilter` and `iptables`
+
+The purpose of this document is to guide you through the process of using `iptables` as a front-end for `bpfilter`. Although `iptables` does not officially support `bpfilter`, it can still route its requests directly to the `bpfilter` daemon instead of the Linux kernel.
+
+To facilitate the linking of `iptables` to `bpfilter`, we recommend defining a directory for installing both. You can do this by using the following command:
+
+```shell
+export INSTALL_DIRECTORY=$HOME/bpfilter_install
+```
+
+## Build `bpfilter`
+
+There is no special flag to build `bpfilter` to be used with `iptables`, however you need to override `CMAKE_INSTALL_PREFIX` with your custom install directory. To do this, navigate to the bpfilter source directory and run the following commands:
+
+```shell
+# Configure CMake
+cmake -S . -B build -DCMAKE_INSTALL_PREFIX=$INSTALL_DIRECTORY
+
+# Build and install bpfilter
+make -C build install
+```
+
+If you wish to build bpfilter in debug mode, simply add `-DCMAKE_BUILD_TYPE=debug` to the CMake command.
+
+Upon successful installation, `bpfilter` will be available at `$INSTALL_DIRECTORY/bin/bpfilter`. The `libbpfilter.so` file can be found in the `lib64` directory, which is adjacent to the `bin` directory.
+
+## Build `iptables`
+
+To use `iptables` with `bpfilter`, you need to clone [this fork](https://github.com/qdeslandes/iptables.git) and switch to the `bpfilter` branch. This version of `iptables` is identical to the one on your system, with the exception of the added `--bpf` option. This option allows `iptables` to communicate with `bpfilter` instead of the kernel.
+
+To configure `iptables`, navigate to its source directory and run the following commands:
+
+```shell
+./autogen.sh
+
+# Instruct the configure script where to find bpfilter and enable it.
+PKG_CONFIG_PATH=$INSTALL_DIRECTORY/share/pkgconfig ./configure \
+    --prefix=$INSTALL_DIRECTORY \
+    --disable-nftables \
+    --enable-libipq \
+    --enable-bpfilter
+```
+
+Upon completion of the configuration step, a summary of `iptables`'s options will be displayed, including "bpfilter support".
+
+You can then proceed to build and install your custom `iptables` with the following commands:
+
+```shell
+make
+make install
+```
+
+Please note that running `make install` alone is not sufficient to properly build and install `iptables`. You must first build it using `make`, and then install it with `make install`.
+
+## Usage
+
+With everything set up, you can now use `iptables` with `bpfilter`. To initiate the daemon, use the following command:
+
+```shell
+sudo $INSTALL_DIRECTORY/bin/bpfilter --transient --verbose
+```
+
+Given that `bpfilter` is a project under active development, it's recommended to run it in transient mode (`--transient`). This ensures that no cache or BPF programs remain on your system once the daemon is stopped. The `--verbose` option, while not mandatory, can be helpful in understanding the daemon's operations.
+
+To use `iptables` and ensure that requests are routed to `bpfilter`, use the `--bpf` switch:
+
+```shell
+# List existing rules and counters
+sudo $INSTALL_DIRECTORY/sbin/iptables-legacy -L -v --bpf
+
+# Filter incoming ICMP packets
+sudo $INSTALL_DIRECTORY/sbin/iptables-legacy -I INPUT -p icmp -j DROP --bpf
+```
+
+The above example only filters incoming packets based on the protocol field. However, you're free to use the `FORWARD` or `OUTPUT` chains, and filter based on source or destination addresses, or ports.
+
+If you encounter any issues or have any questions, don't hesitate to open an issue!