[Buildroot] [PATCH 1/3] Documentation update : add tips to build manual, add information about buildroot toolchain not being relocable and put some hints to use it, move the Beyond Buildroot section before FAQs and add content
lambert.willy at gmail.com
Tue Feb 5 20:51:29 UTC 2013
From: "A.R.D" <contact at team-ard.com>
Signed-off-by: A.R.D <contact at team-ard.com>
docs/manual/beyond-buildroot.txt | 83 ++++++++++++++++++++++++++++++-----
docs/manual/customize-toolchain.txt | 45 ++++++++++++++++---
docs/manual/make-tips.txt | 12 +++++
docs/manual/manual.txt | 4 +-
4 files changed, 125 insertions(+), 19 deletions(-)
diff --git a/docs/manual/beyond-buildroot.txt b/docs/manual/beyond-buildroot.txt
index a87b584..d9bdd32 100644
@@ -3,17 +3,77 @@
-Boot the generated images
+After having run Buildroot, you will have a brand new filesystem for your target exported in buildroot/output/images. The content of this folder depends of the option you choosed during buildroot configuration in the "Filesystem images" section.
+So what's next ? You will probably want to :
-To achieve NFS-boot, enable _tar root filesystem_ in the _Filesystem
+* generate a virtual disk to dump to real system or to use in virtualisation systems (http://wiki.qemu.org/Main_Page[Qemu], https://www.virtualbox.org/[VirtualBox],...)
+* test your kernel and your rootfs in virtualisation systems
+* install your rootfs on the target to test it
+This stuff is really depending on each project and hardware, so we cannot describe every solution here, and this is where Buildroot's work ends. The following section aims at guiding new user on what to do next to avoid staying in the dark. It is *not* an exact guide on how to precisely do what is described. Please take the time to have a look to refered projets to get those details.
+Generate a bootable raw disk file
+If you plan to use virtual machines, or to copy a binary bootable image to your target, you will need to create a "virtual disk". To create a bootable raw disk file you will need to :
+* create an empty file with "dd"
+* partition the file as a disk with "fdisk"
+* set the MBR
+* create virtual devices in /dev pointing to your virtual disk partitions (as you will have with /dev/sda, /dev/sda1, /dev/sda2,...) with http://robert.penz.name/73/kpartx-a-tool-for-mounting-partitions-within-an-image-file/[kpartx]
+* mount one (or several) partition of your virtual disk with "mount"
+* extract your rootfs, boot folder,home directory, ... into them.
+Some device doesn't have external memory and need to be booted to be able to install the rootfs. A nice way of doing this is to boot from the network. If your device allows you to do that you will be able to :
+* test the in-progress-rootfs without installing it on your system (it prevents ruining flash memories)
+* create a second buildroot project with a minimalist installer that will install your production rootfs on the target.
+Network bootloaders (PXE,iPXE)
+To fully boot on the network you need a network bootloader. This is optionnal and you could use your classic bootloader to mount a NFS rootfs.
+http://download.intel.com/design/archives/wfm/downloads/pxespec.pdf[PXE] is a specification that has been implemented at least by http://www.syslinux.org/wiki/index.php/PXELINUX[PXELINUX] and http://ipxe.org/[iPXE].
+The main idea is to have a DHCP server that provide a link to a generic boot ROM that is accessible from a simple FTP server (TFTP). Then your target boots with it and come back to the TFTP server to get the specific stuff (for instance it's boot menu).
+Here is some hints on how to setup this :
+NFS rootfs mount on "/"
+The idea is to mount "/" with a network shared folder from a http://tldp.org/HOWTO/NFS-HOWTO/index.html[NFS] server (in general your dev host). To enable the NFS-boot, you shall activate the _tar root filesystem_ option in the _Filesystem images_ menu.
+Your embedded kernel need at least the following option :
+* NFS filesystem support (CONFIG_NFS_FS).
+* Root file system on NFS (CONFIG_ROOT_NFS).
+* Ethernet (CONFIG_NET_ETHERNET).
+* The ethernet driver for the embedded network card.
+* IP: kernel level autoconfiguration (CONFIG_IP_PNP, CONFIG_IP_PNP_BOOTTP, CONFIG_IP_PNP_DHCP).
After a complete build, just run the following commands to setup the
+NFS-root directory of the server :
sudo tar -xavf /path/to/output_dir/rootfs.tar -C /path/to/nfs_root_dir
@@ -21,10 +81,13 @@ sudo tar -xavf /path/to/output_dir/rootfs.tar -C /path/to/nfs_root_dir
Remember to add this path to +/etc/exports+.
-Then, you can execute a NFS-boot from your target.
+Then, you can execute a NFS-boot from your target. Here is a documentation hints :
+Chroot into generated image
If you want to chroot in a generated image, then there are few thing
you should be aware of:
diff --git a/docs/manual/customize-toolchain.txt b/docs/manual/customize-toolchain.txt
index 2b24412..1f146ad 100644
@@ -16,6 +16,18 @@ generate it.
It also requires to set the Buildroot settings according to the toolchain ones
+Using the Crosstool-NG backend
+The http://crosstool-ng.org[crosstool-NG] toolchain backend enables a rather
+limited set of settings under the Buildroot +Toolchain+ menu:
+* The http://crosstool-ng.org[crosstool-NG] configuration file
+* Gdb and some toolchain options
+Then, the toolchain can be fine-tuned by invoking +make ctng-menuconfig+.
Using the internal Buildroot toolchain backend
@@ -33,14 +45,33 @@ However, it allows to tune major settings, such as:
These settings are available after selecting the +Buildroot toolchain+ type in
the menu +Toolchain+.
-Using the Crosstool-NG backend
-The http://crosstool-ng.org[crosstool-NG] toolchain backend enables a rather
-limited set of settings under the Buildroot +Toolchain+ menu:
+The common use case is to have a buildroot setup where you disable everything from the following menus:
-* The http://crosstool-ng.org[crosstool-NG] configuration file
+* System configuration
-* Gdb and some toolchain options
+* Package Selection for the target
+* Filesystem images
+And build this out of sources with :
+ $ make menuconfig O=/path/to/somewhere
+Then configure your working configuration with :
+* Toolchain type (External toolchain)
+* Toolchain (Custom toolchain)
+* Toolchain origin (Pre-installed toolchain)
+* (/path/to/somewhere/host/usr) Toolchain path
+Note that builroot toolchains are *not* relocable, so if you plan to share it with other machines, you will *have to* install it in the same position ("/path/to/somewhere")
-Then, the toolchain can be fine-tuned by invoking +make ctng-menuconfig+.
diff --git a/docs/manual/make-tips.txt b/docs/manual/make-tips.txt
index 8cd77c0..80574bd 100644
@@ -53,6 +53,18 @@ and target trees, the images and the toolchain):
$ make clean
+The present manual sources are in the buildroot/docs/manual folder. You may compile them with :
+ $ make manual-clean
+ $ make manual
+The output documentation is provided in the buildroot/output/docs/manual folder.
+*Note* : asciidoc is required to build it. There is a known issue that you can't build it under Dedian Squeeze.
To delete all build products as well as the configuration:
diff --git a/docs/manual/manual.txt b/docs/manual/manual.txt
index c34e0ca..2f49674 100644
@@ -17,6 +17,8 @@ include::starting-up.txt
@@ -25,8 +27,6 @@ include::developer-guide.txt
More information about the buildroot