Digi Home Health Hub (HHH)

• Download Debian AMD64 Net installation disk¹
• Start VMWare Fusion and install Debian from the ISO
• Adjust the Processor/Memory to 1024 MiB of memory
• Adjust disk size to 20 GiB
• Under Network, tell this virtual machine to make its own connection to the network (so TFTP will work)
• If Debian can't find the network during the install, switch to hardwired Ethernet, reboot the MAC, then try again
• Have VMWare Connect Directly To the MAC's network using autodetect. If you don't do this, the kit won't be able to find our Linux image.

These instructions will install the minimum number of packages necessary to make a working development system.

• Select base-system and ssh-server for the Debian install
• Install
• Login to the system as the root-user
• Edit /etc/apt/apt.conf to not install recommends and suggests

  echo 'APT::Install-Recommends "0";' >>  /etc/apt/apt.conf
  echo 'APT::Install-Suggests "0";'   >>  /etc/apt/apt.conf
  

• Update the system: apt-get update && apt-get upgrade
• Install vim: apt-get install vim
• (optional) Install emacs-nox: apt-get install emacs23-nox
• (optional) Set emacs as the default editor: update-alternatives --config editor
• Install sudo: apt-get install sudo
• Use visudo to edit the file /etc/sudoers/ as below

  XXX ALL=(ALL) PASSWD: ALL , NOPASSWD: /usr/bin/rpm, /opt/freescale/ltib/usr/bin/rpm
  

  Replace XXX with the name of your user account. The above line should be added as the last line of the file. This will let you properly run Freescale's Linux Target Image Builder (LTIB) tool without constantly entering your password, thus facilitating unattended builds.

Make sure you are not the root user, then run the following commands:

sudo apt-get install build-essential
sudo apt-get install nfs-kernel-server
sudo apt-get install nfs-common
sudo apt-get install portmap
sudo apt-get install tftpd-hpa
sudo apt-get install tftp
sudo apt-get install rpm
sudo apt-get install wget
sudo apt-get install bison
sudo apt-get install flex
sudo apt-get install ncurses
sudo apt-get install ncurses-dev
sudo apt-get install libncurses5
sudo apt-get install libncurses5-dev
sudo apt-get install tcl
sudo apt-get install tclreadline
sudo apt-get install zlib1g
sudo apt-get install zlib1g-dev
sudo apt-get install liblzo2-2
sudo apt-get install liblzo2-dev
sudo apt-get install uuid
sudo apt-get install uuid-dev
sudo apt-get install libuuid1
sudo apt-get install gettext
sudo apt-get install libgtk2.0-dev
sudo apt-get install libdbus-glib-1-dev
sudo apt-get install liborbit2-dev
sudo apt-get install intltool
sudo apt-get install ccache
sudo apt-get install libtool
sudo apt-get install tcl

# Because we are on a 64-bit machine
sudo apt-get install ia32-libs
sudo apt-get install libc6-dev-i386
sudo apt-get install lib32z1

su # become root
perl -MCPAN -eshell
install CPAN
reload CPAN
install LWP::UserAgent
exit # exit Perl

exit # exit su
whoami # should be you, not root

Follow the instructions in the i.MX28 Linux BSP User Guide. Once everything is unpacked do:

./ltib  # this will take awhile
./ltib --selectype
	# choose platform type  -->  imx28
	# choose packages profile  -->  mfg firmware profile
	# save and exit

Assuming ltib completes, you will now have two new files updater.sb and updater_ivt.sb in the ltib directory.

You will need the development machine to run some network services so that you can easily move files back and forth between the host and kit.

The firmware on the kit reports itself as Freescale USB-to-Serial. Check /dev/tty* to figure out where it is on your machine.

We need the following items:

imx28_ivt_uboot_v3.sb

This is the bootlet that gets passed to the iMX28.

When the iMX28 is reset, it executes its ROM. There is no alternative, no other code is permitted to handle the reset exception. The ROM code reads boot mode pins to detect the boot source (USB, SD/MMC, NAND, etc.) and negotiates with that source to retrieve a boot stream. A boot stream is an executable collection of bytes in Safe Boot (SB) format. This file is the boot stream for loading uboot.

Note, the ivt in the filename indicates that the High Assurance Boot (HAB) is operating. Specifically, that the HAB_DISABLE bit (HW_OCOTP_ROM7:0x8002C210:bit11) on the iMX28 is zero (0).

h3_flash_init

This is a uboot initialization script. See the section below on modifying this to your needs.

uImage_heg

This is the Linux kernel image.

rootfs_XXX.ubi

In my case, this was rootfs_3893.ubi. This is a root file system image in UBIFS (Unsorted Block Image File System) format.

demo.tgz

The demonstration software.

The overall process is to tell the kit to boot from USB. This is done by setting SW1-2 to the ON position. When the iMX28 powers up, its ROM code will detect that setting and attempt to retrieve a boot stream via USB. On the other end of the USB, is a program named sb_loader.exe. This program is part of Freescale's iMX Manufacturing Tools bundle, which can be found by starting at the i.MX28 Product Page and then following the link to the i.MX Manufacturing Toolkit. There are a few versions available. These instructions correspond to Mfgtools-Rel-1.6.2.032.zip.

Once the iMX28 has established contact with the sb_loader program over USB, we need to send the processor a boot stream. We will send it a version of the u-boot bootloader in safe boot format contained in file imx28_ivt_uboot_v3.sb.

Once the iMX28 is running u-boot, we want to prepare the board for hosting Linux. This is the job of the h3_flash_init script. This script will setup a proper u-boot environment, load the Linux kernel, and flash the root file system.

Once the Linux kernel is running, program the boot stream (imx28_ivt_uboot_v3.sb) into NAND flash. That way, when the kit is booted with SW1-2 in the OFF position, the iMX28 will find a boot stream.

Here we want to modify the environment that will be programmed into u-boot. The script just automates the creation of the environment, we could enter its contents by hand if we needed to.

• Install necessary tools: sudo apt-get install uboot-mkimage
• Unpack the u-boot source code

  tar xzf u-boot-2011.09.tgz
  cd u-boot-2011.09/
  

• Modify h3_flash_init.txt. I changed bootargs_net to derive NFS options from environment variables as in:

  nfsroot=${serverip}:${nfsroot}
  

• Rebuild the boot script (this just puts a header on the contents so u-boot knows how to parse it).

  mkimage -T script -C none -n 'Uboot Init Script' \
  -d u-boot-2011.09/h3_flash_init.txt /tftboot/h3_flash_init
  

This is the detailed procedure for booting the kit.

1. Ensure that your TFTP server is properly setup. Ours is at /tftpboot/.
2. Copy the u-boot script, kernel, root filesystem, and demonstration software into the TFTP server root.

   cp  h3_flash_init    /tftpboot
   cp  uImage_heg       /tftpboot
   cp  rootfs_3893.ubi  /tftpboot
   cp  demo.tgz         /tftpboot
   

3. On a Windows machine, install the Freescale iMX Manufacturing Tools.
4. On the Windows machine, in the same directory where the program sb_loader.exe, was installed, copy the u-boot safe boot-formatted boot stream file (imx28_ivt_uboot_v3.sb).
5. On the Windows machine, open a command prompt and change to the directory where sb_loader.exe was installed.
6. On the hardware, set SW1-1 to OFF and SW1-2 to ON. This will tell the iMX28 to boot from USB.
7. Connect two USB cables to the kit. One to J5 for debug message and one to J7 for the iMX28 boot stream.
8. Connect the USB cables to the Windows box. Using the Device Manager, determine which COM port the debug messages will be connected to. Open a terminal emulator on that COM port.
9. Power on the kit.
10. Execute the u-boot boot stream by entering the command below into the Windows command prompt.

    sb_loader.exe /f imx28_ivt_uboot_v3.sb
    

11. Verify that debug messages are appearing in the terminal emulator.
12. Interrupt u-boot by pressing any key before the 3-second timeout.
13. Adjust the u-boot environment to reflect your network settings:

    printenv  # note current settings
    
    setenv  ethaddr   00:04:9F:01:03:29 # adjust to your network
    setenv  serverip  10.160.42.128     # adjust to your network
    setenv  bootfile  h3_flash_init
    setenv  tftp_kernel  uImage_heg
    setenv  tftp_rootfs_ubifs  rootfs_3893.ubi
    setenv  tftp_prefix  # you may or may not need this
    saveenv
    

    Note the tftp_prefix environment variable. Depending on your setup, you may or may not need this. We did not.

14. Execute the boot script

    dhcp
    source
    

    You will see a lot of messages come by the debug terminal. If all goes well, you will see the kernel and the root file system being downloaded and burned into flash.

15. Login as root with password root
16. Flash the bootloader

    cd  /boot
    kobs-ng  init  imx28_ivt_uboot_v3.sb
    sync
    

17. Flash the demonstration software

    cd  /heg
    dhclient  eth0
    tftp  -g  -r demo.tgz  <your dev server ip>
    
    # At this point you might want to check the file integrity
    # using the md5sum command on the TFTP server and the device
    
    tar xzf demo.tgz
    rm demo.tgz
    sync
    

18. Properly shutdown the board: poweroff
19. Power off the board
20. Set SW1-2 to OFF so that the next time the iMX28 boots, it will look for a boot stream in NAND flash rather than on USB.

After this point, you should be able to leave the Windows box and connect to the kit's debug serial port using the cu program:

cu -l /dev/ttyACM0 -s 115200
~. # to exit

Assuming the kit is setup to configure a network interface and launch dropbear, you can also just SSH into it.

curl -o idigi-connector.tgz \
http://ftp1.digi.com/support/sampleapplications/40003007_E.tgz

md5sum idigi-connector.tgz
07916d03dfa8a8647e1475e3a189d002  idigi-connector.tgz

tar xzf idigi-connector.tgz

You should do the steps above for the following project in idigi/public/run/samples/. Note, you should test these in the order listed, consulting the iDigi connector documentation as necessary.

• compile_and_link
• connect_to_idigi
• send_data

If the above sample projects work, then your development environment should be properly configured.

The number of active panic buttons is computed by the active_buttons() function inside pb_status.c. By going through the PB Manager code in zorglub you can determine that performing the IOCTL IOCTL_ISM_GET_STAT on the file /dev/ism will give you some idea of the number of panic buttons currently associated with the HHH and the state that they are in.

icarm_fd = open("/dev/ism", O_RDONLY)) < 0 );
ioctl(icarm_fd, IOCTL_ISM_GET_STAT, stat_table);

/* For items in stat_table, if bit-1 is set, the button is active. */

The program's main loop occurs in the function application_run() inside application.c. This repeatedly calls app_send_put_request() and then sleeps.

The function app_send_put_request() calls active_buttons() and includes its return value in the data streamed to iDigi. This function also sends some sample, made-up data such as; pulse, breaths, systolic, and diastolic to make the demonstration more interesting.