RPi Advanced Setup

From eLinux.org
Jump to: navigation, search

RaspPi.png Back to the Hub

Getting Started:

Buying Guide - for advice on buying the Raspberry Pi.

SD Card Setup - for information on how to prepare the SD Card used to boot your Raspberry Pi.

Basic Setup - for help with buying / selecting other hardware and setting it up.

Beginners Guide - you are up and running, now what can you do?

Latest RPi 4 Topics - Recent topics on Raspberry Pi 4 and the Pi OS

Advanced Setup - for more extensive information on setting up.

Trouble Shooting - some things to check if things don't work as expected.


WARNING: This page is not suitable for the average user. Information in this page is for experienced hackers only.

This page is a community work in progress in preparation for the first users. If something doesn't work or isn't covered in these guides, please feel free to ask on the Forum. But before you ask there, make sure you read the FAQs.

This page is based on BeagleBoardBeginners so the serial port info is applicable only under explained circumstances. Also, many Raspberry Pi users will buy pre-programmed SD cards and can skip to reading RPi Hardware Basic Setup. We expect that once Raspberry Pi boards become generally available helpful volunteers will update this page to match Raspberry Pi completely or point to better information elsewhere.

This page in a major work in progress!

Finding hardware and setting up

Main article: see RPi Hardware Basic Setup

You'll need to copy an image to a suitable SD card (or make your own image). You'll also need a USB keyboard, TV/Monitor (with HDMI/DVI/Composite/SCART input), and power supply (USB charger or a USB port from a powered USB Hub or another computer).

You'll likely also want a USB mouse, a case, and a USB Hub (a necessity for Model A). A powered USB Hub will reduce the demand on the RPi. To connect to the Internet, you'll need either an Ethernet/LAN cable (Model B) or a USB WiFi adaptor (either model). See RPi VerifiedPeripherals for more information on supported peripherals.

All this basic verified hardware is available from both Pi Hut's Raspberry Pi Store and ModMyPi's Raspberry Pi Shop

When setting up, it is advisable to connect the power after everything else is ready. See RPi_Hardware_Basic_Setup#Connecting_Together.

Serial connection

For help setting up a serial connection with the Raspberry Pi, see RPi_Serial_Connection.

Advanced SD card setup

Now we want to use an SD card to install some GNU/Linux distro in it and get more space for our stuff. You can use either an SD or SDHC card. In the latter case of course take care that your PC card reader also supports SDHC. Be aware that you are not dealing with an x86 processor, but instead a completely different architecture called ARM, so don't forget to install the ARM port for the distro you are planning to use.

Our first step will be the formatting of the SD card.

Formatting the SD card via the mkcard.txt script

  1. Download mkcard.txt from http://downloads.angstrom-distribution.org/demo/beaglebone/mkcard.txt
  2. $ chmod +x mkcard.txt
  3. $ ./mkcard.txt /dev/sdX, Where X is the letter of the card. You can find this by inserting your card and then running dmesg | tail. You should see the messages about the device being mounted in the log. Mine mounts as sdc.

Once run, your card should be formatted.

Formatting the SD card via fdisk "Expert mode"

First, lets clear the partition table:

$ sudo fdisk /dev/sdb

Command (m for help): o
Building a new DOS disklabel. Changes will remain in memory only,
until you decide to write them. After that, of course, the previous
content won't be recoverable.

Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite) 

Print card info:

Command (m for help): p

Disk /dev/sdb: 128 MB, 128450560 bytes

Note card size in bytes. Needed later below.

Then go into "Expert mode":

Command (m for help): x

Now we want to set the geometry to 255 heads, 63 sectors and calculate the number of cylinders required for the particular SD/MMC card:

Expert command (m for help): h
Number of heads (1-256, default 4): 255

Expert command (m for help): s
Number of sectors (1-63, default 62): 63
Warning: setting sector offset for DOS compatiblity

NOTE: Be especially careful in the next step. First calculate the number of cylinders as follows:

  • B = Card size in bytes (you got it before, in the second step when you printed the info out)
  • C = Number of cylinders

When you get the number, you round it DOWN. Thus, if you got 108.8 you'll be using 108 cylinders.

Expert command (m for help): c
Number of cylinders (1-1048576, default 1011): 15

In this case 128MB card is used (reported as 128450560 bytes by fdisk above), thus 128450560 / 255 / 63 / 512 = 15.6 rounded down to 15 cylinders. Numbers there are 255 heads, 63 sectors, 512 bytes per sector.

So far so good, now we want to create two partitions. One for the boot image, one for our distro.

Create the FAT32 partition for booting and transferring files from Windows. Mark it as bootable.

Expert command (m for help): r
Command (m for help): n
Command action
  e   extended
  p   primary partition (1-4)
Partition number (1-4): 1
First cylinder (1-245, default 1): (press Enter)
Using default value 1
Last cylinder or +size or +sizeM or +sizeK (1-245, default 245): +50

Command (m for help): t
Selected partition 1
Hex code (type L to list codes): c
Changed system type of partition 1 to c (W95 FAT32 (LBA))

Command (m for help): a
Partition number (1-4): 1

Create the Linux partition for the root file system.

Command (m for help): n
Command action
  e   extended
  p   primary partition (1-4)
Partition number (1-4): 2
First cylinder (52-245, default 52): (press Enter)
Using default value 52
Last cylinder or +size or +sizeM or +sizeK (52-245, default 245):(press Enter)
Using default value 245

Print and save the new partition records.

Command (m for help): p

Disk /dev/sdc: 2021 MB, 2021654528 bytes
255 heads, 63 sectors/track, 245 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

  Device Boot      Start         End      Blocks   Id  System
/dev/sdc1   *           1          51      409626    c  W95 FAT32 (LBA)
/dev/sdc2              52         245     1558305   83  Linux

Command (m for help): w
The partition table has been altered!

Calling ioctl() to re-read partition table.

WARNING: Re-reading the partition table failed with error 16: Device or resource busy.
The kernel still uses the old table. The new table will be used at the next reboot.

WARNING: If you have created or modified any DOS 6.x partitions, please see the fdisk 
manual page for additional information.
Syncing disks.

Now we've got both partitions, next step is formatting them.

NOTE: If the partitions (/dev/sdc1 and /dev/sdc2) does not exist, you should unplug the card and plug it back in. Linux will now be able to detect the new partitions.

$ sudo mkfs.msdos -F 32 /dev/sdc1 -n LABEL
mkfs.msdos 2.11 (12 Mar 2005)

$ sudo mkfs.ext3 /dev/sdc2
mke2fs 1.40-WIP (14-Nov-2006)
Filesystem label=
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
195072 inodes, 389576 blocks
19478 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=402653184
12 block groups
32768 blocks per group, 32768 fragments per group
16256 inodes per group
Superblock backups stored on blocks: 
       32768, 98304, 163840, 229376, 294912

Writing inode tables: done                            
Creating journal (8192 blocks): done
Writing superblocks and filesystem accounting information: 

All done!

NOTE: For convenience, you can add the -L option to the mkfs.ext3 command to assign a volume label to the new ext3 filesystem. If you do that, the new (automatic) mount point under /media when you insert that SD card into some Linux hosts will be based on that label. If there's no label, the new mount point will most likely be a long hex string, so assigning a label makes manual mounting on the host more convenient.

Setting up the boot partition

The boot partition must contain the following files, get them from one of the official images:(bootable/fat32 partition)

  • bootcode.bin : 2nd stage bootloader, starts with SDRAM disabled
  • loader.bin : 3rd stage bootloader, starts with SDRAM enabled
  • start.elf: The GPU binary firmware image, provided by the foundation.
  • kernel.img: The OS kernel to load on the ARM processor. Normally this is Linux - see instructions for compiling a kernel.
  • cmdline.txt: Parameters passed to the kernel on boot.

Optional files:

  • config.txt: A configuration file read by the GPU. Use this to override set the video mode, alter system clock speeds, voltages, etc.
  • vlls directory: Additional GPU code, e.g. extra codecs. Not present in the initial release.

Additional files supplied by the foundation

These files are also present on the SD card images supplied by the foundation.

Additional kernels. Rename over kernel.img to use them (ensure you have a backup of the original kernel.img first!):

  • kernel_emergency.img : kernel with busybox rootfs. You can use this to repair the main linux partition using e2fsck if the linux partition gets corrupted.

Before the introduction of the 512MB Raspberry PI additional GPU firmware images (start.elf files) were delivered, to be copied over start.elf to use them:

the file called start.elf actually determines how much of the available 256MB of memory is assigned to the GPU, but previously each start.elf file gave a fixed split size. The other splits were simply very similar files with a different filename, which were copied over the one called start.elf that was actually to be used, the others had names like arm192_start.elf and such.

  • arm128_start.elf : 128M ARM, 128M GPU split (use this for heavy 3D work, possibly also required for some video decoding)
  • arm192_start.elf : 192M ARM, 64M GPU split (this is the default)
  • arm224_start.elf : 224M ARM, 32M GPU split (use this for Linux only with no 3D or video processing. Its enough for the 1080p framebuffer, but not much else)
  • arm240_start.elf : 240M ARM, 16M GPU split (use this for headless mode only. It is enough to boot the system, but you can not use video at all.)

Note that actually there is no "default" split, the nature of the software determines what is the most suitable split. So a "distro" that is very heavy multimedia oriented will normally use the 128/128MB split as the GPU needs a lot of RAM, but a generic desktop distro will probably use the 64/192 MB split, and a game that doesn't use the GPU will probably use the 32/224MB split.

However for a situation where there are two memory sizes (256 and 512) this isn't a very smart system. For example the maximum GPU memory spit of 256MB /2 = 128 MB would in the past give the CPU 128 MB, and the GPU also 128 MB, but when you use that memory split on a 512 MB PI it would stil give 128 MB to the CPU, and the rest, a whopping 384 MB to the GPU, which couldn't do anything with it.

So the RPF changed the firmware so that a single start.elf you now can give the GPU exactly the amount you want, in chunks of 16MB, with 16MB as minimum, and 128MB as maximum.

The new syntax is to use:

gpu_mem=(number of megabytes for the GPU)

So for example putting


will give the GPU 64 MB and whatever the rest is (either 192 or 448 MB) to the ARM CPU.

Finally booting GNU/Linux

Important steps

<To be completed>

Setting up for remote access / headless operation

If you're anything like me (lazy, with a limited number of monitors), you'll want to get your Raspberry Pi set up for remote access as soon as possible. Luckily, this is easy. These instructions assume you're using the official Debian distribution for the Raspberry Pi. Steps 0 & 1 are based on information from Steve Smith.

  • Step 0. Before you set up SSH, you might want to change the default password on the Raspberry Pi, especially if it'll end up Internet-facing. Do this on the Raspberry Pi's console with the following command:
  • Step 1: Enable SSH with the following command:
    sudo mv /boot/boot_enable_ssh.rc /boot/boot.rc
    This will enable sshd on the next boot. Restart the Raspberry Pi. On reboot, you should see a line like the following:
    Starting OpenBSD Secure Shell server: sshd
    near the end of the boot sequence. This indicates that sshd is enabled, and you should be able to ssh into the Raspberry Pi. You'll need the Raspberry Pi's IP address to do that; get that at the Raspberry Pi's console with:
    ip addr
    You may also find it useful to copy an SSH key to the Raspberry Pi so you don't need to enter a password each time you connect. To do that, first check if you've already got a public ssh keyfile:
    ls ~/.ssh/id_rsa.pub
    If you haven't, you can generate one with:
    ssh-keygen -t rsa -C "your_email@youremail.com"
    Finally, copy the keyfile to ~/.ssh/authorized_keys on the Raspberry Pi (there's a few different ways to do this, I used Transmit [1] to copy it over SFTP, since I'm a Mac user. Windows users have WinSCP [2], and Linux users probably already know how to do it ;) ). This file contains all of the keys authorised to connect to the Raspberry Pi, so it will probably be blank or non-existent on a new Raspberry Pi. If so, just copy id_rsa.pub to this location. If it already exists, add the key from id_rsa.pub to the end of the file.
  • Step 2: IP address config. If your Raspberry Pi is going to be always-on, or your network is set up in a such a way that devices always get the same IP address, you can skip the step. However, if your Raspberry Pi's IP address is likely to change frequently (say, for instance, you're just powering it up every so often to play, and your network assigns IP addresses first-come first-served {like most home routers), it's a good idea to set up a consistent network address for your Raspberry Pi. There are two ways to do this: The quick (but brittle) way and the more flexible way.
    • The quick way: Assign a static IP address to your Raspberry Pi. This is simple, but runs the risk of clashing IP addresses with other devices on your network since your Raspberry Pi's address will no longer be managed by DHCP. I haven't tried this myself, but here's some instructions from Andrew Munsell. He's using for his Raspberry Pi, since that's outside the range assigned by his router. Change this address to whatever suits.
      You can do this in Debian Squeeze on the Raspberry Pi by modifying the /etc/network/interfaces file.
      I removed the original iface eth0 line and replaced it with the following:
      iface eth0 inet static
      On reboot, your Raspberry Pi should now be using the static address specified in /etc/network/interfaces.
    • The flexible way: Set up avahi / zeroconf. Zeroconf is 'a set of techniques that automatically creates a usable Internet Protocol (IP) network without manual operator intervention or special configuration servers.'[3]. Avahi is an implementation of zeroconf which 'ships with most Linux and *BSD distributions' [4], but not the Raspberry Pi's Debian distribution. Zeroconf will be familiar to Apple users as Bonjour and is pretty clever tech which means that things Just Work when sharing stuff across computers on a network. In this context, it means that once we've set it up on the Raspberry Pi, we'll be able to address it as:
      regardless of what IP address it's been assigned on your local network. This is handy if its IP address is likely to change regularly, and even means we'll continue to be able to address it if we're on a different network (say, shuffling between home and work networks).
      Information in this section largely gathered from 4dc5.
  1. Install avahi with the following commands on the Raspberry Pi:
    sudo apt-get install avahi-daemon
    and then on older Debian installs:
    sudo update-rc.d avahi-daemon defaults
    or on newer Raspbian installs:
    sudo insserv avahi-daemon
    (if in doubt, you're probably on the newer one).
  2. Create a configfile for Avahi at /etc/avahi/services/multiple.service. I did this with the following command:
    sudo pico /etc/avahi/services/multiple.service
    The contents of this file should be something like the following, courtesy of aXon on the Rasperry Pi forums:
    <?xml version="1.0" standalone='no'?>
    <!DOCTYPE service-group SYSTEM "avahi-service.dtd">
            <name replace-wildcards="yes">%h</name>
  3. Apply the new configuration with:
    sudo /etc/init.d/avahi-daemon restart
    The Raspberry Pi should now be addressable from other machines as raspberrypi.local, for example:
    ssh pi@raspberrypi.local
  4. Get Windows to play nice with avahi
    If you've done the first steps correctly and you open up PuTTY and you try to address your Raspberry Pi as raspberrypi.local; it will tell you:
    This happens for a very good reason: your Windows PC can't interpret the UDP-datagrams avahi sends and most firewalls don't even allow them to get read. So you'll have to do a couple of things extra to get it working.

  1. Get Bonjour for Windows
    http://support.apple.com/kb/DL999 Just install it, the quick next next next next procedure will suffice. Now your computer is able to interpret the UDP datagrams which are multicasted by the Raspberry Pi. But we're not out of the woods just yet; if you try to ping to your Raspberry Pi:
    C:\Windows\System32>ping raspberrypi.local
    Ping-request cannot find host raspberrypi.local.
    Check the name and try again.
  2. Tell your firewall: trust me, I'm an engineer.
    Stereotypically, the firewall forbids us to have some fun. First of all, the 5353 UDP port is blocked on most firewalls, so you have to add an exception for it. Also, you'll have to grant Internet access to the mDNSresponder.exe. This way, whenever your computer tries to connect with a host *.local, mDNS sends a multicast over the local subnet to ask whether anyone calls himself *.local. If mDNS isn't granted network access, nothing gets multicasted and nobody answers the phone.
    Note: If you have the McAfee firewall, you'll also have to enable UDP control. If it isn't enabled, all UDP datagrams are ignored.
  3. Enjoy the pleasure of typing raspberrypi.local in PuTTY
    After these steps, you should be able to ping to raspberrypi.local and even address it that way in PuTTY. YMMV, if you're still having troubles at this point, try to ping to raspberrypi.local with the firewall turned off. If it works: hey presto, you've got your culprit, and you can start an educated Google search.

--Live long and geeky 17:14, 17 November 2012 (UTC)

Software development/proving

A supported platform for the Raspberry is Qt, which is already being worked on. C/C++ is supported through a gcc cross-compiling toolchain. On Debian/Ubuntu systems, the packages gcc-4.6-arm-linux-gnueabi and g++-4.6-arm-linux-gnueabi provide suitable compilers. For other build platforms, Chris has good instructions for building a cross-compiler - this should also work in a Cygwin environment on Windows. MinGW may also be supported.

Python is being pushed forward by the foundation. (Status ??)

After compiling, using QEMU and a Linux VM would be one way of testing your apps (this also works on Windows). Search the forum for the ready-made ARM images.

The choice of programming languages, IDEs and other tools on the R-Pi is only determined by:

1) The operating system compatibility (at the moment the specific Linux distro used)

2) The status of the respective ARM package repositories and their binary compatibility

3) The possibilty to build other software + its dependencies for the R-Pi from sources (depends on C cross-compiler ???)

What kind of software development and testing loop has been proven effective please (from someone who's been there and done it)?

For me (and others, hopefully) that would be very useful.

Sample setup scripts to run after a fresh Raspbian install


Further reading

The main Raspberry Pi resources are:

An alternative startup guide for beginners can be found on h2g2: Introducing the Raspberry Pi

For more guides and projects involving the Raspberry Pi, see RPi Projects.

Thanks to

  • Nabax, _vlad, jkridner, ds2 and the other BeagleBoard wiki contributors on elinux.org for an excellent BeagleBoardBeginners resource, which we used as the template for this page.