DM365 Leopard SDK Legacy Getting Started Guide

From RidgeRun Developer Connection
Revision as of 08:40, 15 October 2010 by Tfischer (Talk | contribs)

Jump to: navigation, search

This guide includes instructions and tips users should follow to get their DM365 Leopard Board to work with the RR Evaluation SDK. For this guide different commands need to be executed on the host machine and the target. Commands to be run on the Ubuntu host have a yellow background. Commands to be run on the Ubuntu target have an aqua background.


The e-mail received with the evaluation SDK installation links should have:

  • The SVN repository link.
  • The toolchain download link.
  • An username and password in order to access the download link for the toolchain.

GNU Toolchain Installation

The LeopardBoard 365 Evaluation SDK uses the GNU toolchain for cross compilation. You need to download and install the toolchain before you can build the SDK.

Toolchain Download

The email you received when you requested the LeopardBoard 365 Evaluation SDK contained a link to the toolchain tarball, along with the user name and password to access the files available for download.


I saved the toolchain to the default $HOME/Downloads directory.

Toolchain Installation

Assuming the toolchain tarball is in the $HOME/Downloads directory, you can follow these steps to install the toolchain into the location expected by the SDK.

sudo tar -C / -xjf $HOME/Downloads/toolchain.tar.bz2 

Verify your GNU toolchain is installed properly.

ls /opt/arm-linux-gnueabi/bin

and you should see around 35 tools in the directory.

SDK Installation

With this SVN repository Linux from the email, check out a development directory from the repository:

mkdir -p ~/work
cd ~/work
svn co <repositoryLink>

I put all my development directories in the work directory in my home folder. You can have more than one development directory installed, just remember to have the DEVDIR shell variable set properly.

Build SDK

Several packages are needed in order to compile the SDK. The SDK build process will let you know what packages are needed. While compiling you will be prompted to install them using sudo apt-get <packages>.

Configure the features you want to have in your board using:

make config

Set the PATH to include the toolchain and aslo set the DEVDIR shell variable

$(make env) # this is the same as using grave accents to execute the output from the command - `make env`

Compile the SDK:


Install bootloader, kernel, and target file system on LeopardBoard 365

Verify bootloader is active

First a minicom (or any other terminal emulator used) session must be opened in order to stop the booting process at the bootloader prompt:


Reboot the board and stop the autoboot when

Hit any key to stop autoboot: ...

is displayed. Then close your minicom session.

Save bootloader to target hardware

The next step is to install the bootloader on the target. Typically you only install the SDK bootloader once.

make installbootloader

Update target hardware with new images

Power cycle your target hardware, and install uboot environment / kernel / FS on the target (per your SDK configuration choices):

make install

Similar to the bootloader installation, to install the devdir on the target the booting process should be stopped at the bootloader prompt.

SDK Hints

SD boot recover

If you brick your board, this can be recovered from an SD card. In order to make a bootable SD card you first need to have a successfully compiled SDK.

  • SD card preparation:

1. Insert the SDcard to the host port

2. Find out the device mounting point (e.g. /dev/sdd)

3. On your devdir root directory type:

make sd_boot_recover device=/dev/sdX

where sdX is your device mounting point.

  • Boot recover

1. Insert the SD card (if you already inserted it, disconnect and connect again).

2. Type

make install sd_boot_recover

3. Set switch 2 of DIPSW1 to ON, 1 and 3 to OFF. This is SD card boot mode.

4. Power up the board, it will automatically erase the NAND flash, and burn UBL, u-boot, kernel and filesystem

5. Set all pos of DIPSW1 to OFF for NAND boot mode.

6. Connect serial cable to PC and power cycle the board (it should boot up). Insert the SD card to the board.

Having this done the bootloader and devdir must be installed (check Installation for instructions)

Setting up video components

The output and mode used for video playback/capture can be set during the SDK configuration process. The first step is to enter the configuration script:

make config

Inside Architecture configurations select the options you desire on Video Output and Video Default Resolution.

Finally compile your changes and install

make install

When you use the SDK to set your video output and resolution, the SDK build process creates $DEVDIR/images/cmdline with different kernel command line parameter settings. The table below shows the various video output settings for three of the common settings

Video Output DVI Component Composite
Resolution 720P-60 720P-60 NTSC
davinci_enc_mngr.ch0_output DVI COMPOSITE COMPONENT
davinci_enc_mngr.ch0_mode 720P-60 720P-60 NTSC
davinci_display.cont2_bufsize 6291456 6291456 3145728
vpfe_capture.cont_bufoffset 6291456 6291456 3145728
vpfe_capture.cont_bufsize 6291456 6291456 6291456

Activation of video capture modules

While configuring the SDK go to Kernel configuration->Device drivers->Multimedia support->Video capture adapters and activate the devices you desire.

Video initialization

When you boot your board the RidgeRun logo will be displayed, in order to display video the following commands must be executed:
fbset -disable

Component output initialization

There's a known bug with the component output initialization. Before you play or capture video using the component output this commands must be executed on the target:

echo "COMPOSITE" > /sys/class/davinci_display/ch0/output
echo "COMPONENT" > /sys/class/davinci_display/ch0/output

CMEM pools modification

Depending on the video displayed, the CMEM pools need to be modified to meet the requirements. This is needed when you execute a pipeline and the following error is displayed:

CMEMK Error: Failed to find a pool which fits <poolsize>

First you need the start and end addresses of the CMEM pools, these can be obtained executing and checking the output: a line like the next one will be printed

Loading CMEM from 0x<startaddr> to 0x<endaddr>

then type:

rmmod cmemk
modprobe cmemk phys_start=<startaddr> phys_end=<endaddr> pools=<somepools>, <mynewpools>

where <mynewpools> must be in format: <numberofpools>x<poolsize> (e.g 2x128). You may also need to resize/delete <somepools>

Output volume modification

To increase/decrease the volume for audio playback (or other audio options) type


and modify as desired.