I used only two wires connection where the DS’s VDD is connected to the ground. DS’s datasheet requires an external pull up 4.7 kohm resistor on the data line but I successfully used the MSP’s internal pull up resistor.

Download

Here.

Example of communication with DS18B20

Lets connect the thermometer e.g. to pin P1.7.

#include <msp430.h>
#include "onewire.h"
#include "delay.h"
 
int main()
{
  onewire_t ow;
  int i;
  uint8_t scratchpad[9];
 
  WDTCTL = WDTPW + WDTHOLD; //Stop watchdog timer
  BCSCTL1 = CALBC1_8MHZ;
  DCOCTL = CALDCO_8MHZ;
 
  ow.port_out = &P1OUT;
  ow.port_in = &P1IN;
  ow.port_ren = &P1REN;
  ow.port_dir = &P1DIR;
  ow.pin = BIT7;
 
  onewire_reset(&ow);
  onewire_write_byte(&ow, 0xcc); // skip ROM command
  onewire_write_byte(&ow, 0x44); // convert T command
  onewire_line_high(&ow);
  DELAY_MS(800); // at least 750 ms for the default 12-bit resolution
  onewire_reset(&ow);
  onewire_write_byte(&ow, 0xcc); // skip ROM command
  onewire_write_byte(&ow, 0xbe); // read scratchpad command
  for (i = 0; i < 9; i++) scratchpad[i] = onewire_read_byte(&ow);
 
  _BIS_SR(LPM0_bits + GIE);
  return 0;
}

Search ROM command

Materials for inspiration:

• DS18B20 datasheet contains a flowchart but it doesn’t show very well the recursive nature of the algorithm
• 1-Wire Search Algorithm from Maxim

Following example shows usage of search ROM command (0xF0). I used printf() and UART for debugging.

#include <msp430.h>
#include <stdio.h>
#include <stdint.h>
 
#include "onewire.h"
#include "delay.h"
 
/***************************************************************/
 
int putchar(int c)
{
  if (c == '\n') putchar('\r');
  while (!(IFG2&UCA0TXIFG));
  UCA0TXBUF = c;
  return 0;
}
 
/***************************************************************/
 
void uart_setup()
{
  P1SEL = BIT1 + BIT2;
  P1SEL2 = BIT1 + BIT2;
  P1DIR &= ~ BIT1;
  P1DIR |= BIT2;
  UCA0CTL1 |= UCSSEL_2;                     // SMCLK
  UCA0BR0 = 0x41;                            // 8MHz 9600
  UCA0BR1 = 0x03;                              // 8MHz 9600
  UCA0MCTL = UCBRS0;                        // Modulation UCBRSx = 1
  UCA0CTL1 &= ~UCSWRST;
}
 
/***************************************************************/
 
void search(onewire_t *ow, uint8_t *id, int depth, int reset)
{
  int i, b1, b2;
 
  if (depth == 64)
  {
    // we have all 64 bit in this search branch
    printf("found: ");
    for (i = 0; i < 8; i++) printf("%02x", id[i]);
    printf("\n");
    return;
  }
 
  if (reset)
  {
    if (onewire_reset(ow) != 0) { printf("reset failed\n"); return; }
    onewire_write_byte(ow, 0xF0); // search ROM command
 
    // send currently recognized bits
    for (i = 0; i < depth; i++)
    {
      b1 = onewire_read_bit(ow);
      b2 = onewire_read_bit(ow);
      onewire_write_bit(ow, id[i / 8] & (1 << (i % 8)));
    }
  }
 
  // check another bit
  b1 = onewire_read_bit(ow);
  b2 = onewire_read_bit(ow);
  if (b1 && b2) return; // no response to search
  if (!b1 && !b2) // two devices with different bits on this position
  {
    // check devices with this bit = 0
    onewire_write_bit(ow, 0);
    id[depth / 8] &= ~(1 << (depth % 8));
    search(ow, id, depth + 1, 0);
    // check devices with this bit = 1
    id[depth / 8] |= 1 << (depth % 8);
    search(ow, id, depth + 1, 1); // different branch, reset must be issued
  } else if (b1) {
    // devices have 1 on this position
    onewire_write_bit(ow, 1);
    id[depth / 8] |= 1 << (depth % 8);
    search(ow, id, depth + 1, 0);
  } else if (b2) {
    // devices have 0 on this position
    onewire_write_bit(ow, 0);
    id[depth / 8] &= ~(1 << (depth % 8));
    search(ow, id, depth + 1, 0);
  }
}
 
/***************************************************************/
 
int main()
{
  onewire_t ow;
  uint8_t id[8]; // 64 bits
 
  WDTCTL = WDTPW + WDTHOLD; //Stop watchdog timer
  BCSCTL1 = CALBC1_8MHZ;
  DCOCTL = CALDCO_8MHZ;
  uart_setup();
 
  ow.port_out = &P1OUT;
  ow.port_in = &P1IN;
  ow.port_ren = &P1REN;
  ow.port_dir = &P1DIR;
  ow.pin = BIT7;
 
  printf("start\n");
  search(&ow, id, 0, 1);
  printf("done\n");
 
  _BIS_SR(LPM0_bits + GIE);
  return 0;
}

First of all you need the disk image and appropriate linux kernel. Then download QEMU for Windows. Unzip the QEMU somewhere. Copy the the disk image debian6-17-02-2012.img and the kernel zImage_3.1.9 into the directory with unpacked QEMU. Then open shell/cmd and run:

bin\qemu-system-arm.exe -M versatilepb -cpu arm1136-r2 -hda debian6-17-02-2012.img
                       -kernel zImage_3.1.9 -m 192 -append "root=/dev/sda2"

Be patient because it could take few minutes to get even to the login phase. Login/password is pi/suse. After you log in run “startx” to enter the graphical environment. That’s it!

Similar procedure is possible for Linux as well. In Ubuntu you need qemu-kvm-extras package which contains qemu-system-arm executable.

There are 2 options: Visual C++ Express or Windows SDK. SDK is better because:

• contains 64-bit compiler,
• contains debugging and profiling tools.

I had two little problems with running this SDK toolchain under Windows 7. There is a broken vcvars32.bat. Running it gives an error:

ERROR: Cannot determine the location of the VS Common Tools folder.

Solution is simple. Patch mentioned in the forum answer has a little flaw. Paths must end with a backslash. Then the compiler gave me another error:

fatal error C1083: Cannot open include file: ‘windows.h’: No such file or directory

Insert following line at the beginning of vcvars32.bat:

@SET WindowsSdkDir=c:\Program Files\Microsoft SDKs\Windows\v7.1\

First of all I was confused at the beginning by no ability to load any program at all. It failed on erasing the chip. Solution: do not use programming command

$ mspdebug rf2500 "prog test.elf"

but use

$ mspdebug rf2500 "load test.elf"

because FRAM chips don’t need erasing.

After few loading rounds the big trouble appeared. It took me some googling to find this discussion. This clever guy Donald created a simple program for LaunchPad to use BSL to mass erase the experimenter board. I tried to compile and load the program on my own but without success. The red LED always stayed on. I tried it with and without soldered crystal, both failed. Then I used Donald’s prebuilt binary with soldered crystal and it worked.

Now I will wait how long it will take to break again.

Update 2012-02-09

It didn’t take very long to jam again. Actually it took about 5 rounds of programming. My frustration was huge. After lots of another trials and failures I stumbled on a little code in mspdebug/drivers/fet.c:

/* This packet seems to be necessary in order to program on the
 * MSP430FR5739 development board.
 */
if (xfer(dev, 0x30, NULL, 0, 0) < 0)
  printc_dbg("fet: warning: message 0x30 failed\n");

At least in my case it is not neccessary. I removed the code, experimenter board programming works fine and it does not jam any more. What a lucky shot

Update 2012-02-12

Well, card is blocked again. But this time it takes much more programming cycles then before. Better than nothing. Stay tuned.

Install package python-stdeb and then execute

python setup.py --command-packages=stdeb.command bdist_deb

And your DEB is ready!

First I booted into the Live CD to make the disk preparations. I splitted the disk into 3 partitions:

• 250 GB for root
• 60 GB encrypted partition
• 10 GB swap

For the encryption I decided to utilize aes-ebc-plain with key size 128 bits for its speed (I don’t need military grade safety). Partitions formatting:

mkfs.ext4 /dev/sda1
cryptsetup luksFormat -c aes-ecb-null -s 128 /dev/sda2
cryptsetup luksOpen /dev/sda2 encpart
mkfs.ext4 /dev/mapper/encpart
cryptsetup luksClose encpart
mkswap /dev/sda3

First idea was to have the encrypted partition mounted directly to /home/bobalice in my final system but I wasn’t able to make the mount point owned by bobalice. So I created a directory home/bobalice on the encrypted partition and used symbolic link to “attach it”. See below. After the disk preparation I ran ordinary Ubuntu installation using the root and swap partition.

Login mount

Next step was to setup automatic mount of the encrypted partition. It was a kind of easy. I installed package libpam-mount and added lines to /etc/security/pam_mount.conf.xml:

<mkmountpoint enable="1" remove="true" />
<volume user="bobalice" fstype="crypt"
    path="/dev/disk/by-uuid/55f028b1-3306-4de6-b420-6478ea649604"
    mountpoint="/mnt/localcrypt" />

It says that only on bobalice login the mounting should be performed. I advice to use by-uuid partition reference because sd* labels are not constant. Now if I login as bobalice the partition will be mounted and ready. Next:

mv /home/bobalice/.??* /enc/home/bobalice
rm -Rf /home/bobalice
usermod -d /enc/home/bobalice bobalice

Now I have my home directory on the encrypted partition.

GDM and initial chdir

My first idea of home directory mapping was to use a symbolic link (ln -sf /enc/home/bobalice /home/bobalice) but it caused a little trouble with GDM. When I logged in with GDM it changed directory to the path on the encrypted mount, it followed the symlink. It is not a big deal but current working directory was different from $PATH.

MONTH=`python -c "import time; print int(time.time()-3600*24*30)*1000000"`
echo "DELETE FROM moz_historyvisits WHERE visit_date < $MONTH;" | sqlite3 places.sqlite
for z in *.sqlite; do echo "VACUUM;" | sqlite3 "$z"; done

This will remove items older than 30 days and maintain all the sqlite databases. Speedup is often very striking.

Compilation of source code for MSP430 is the same like in the previous article. Lets focus on debugging. I had troubles to make it working with the original MSPGCC project. In other words I didn’t make it working. But GDB 7.0.1 works better.

Mspdebug supports breakpoints only from newest versions, use git to get it (git clone git://mspdebug.git.sourceforge.net/gitroot/mspdebug). It currently supports only 1 breakpoint at a time.

First run the mspdebug in a stub mode:

$ mspdebug -R gdb
MSPDebug version 0.7 - debugging tool for MSP430 MCUs
Copyright (C) 2009, 2010 Daniel Beer <daniel@tortek.co.nz>
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Trying to open interface 1 on 002
rf2500: warning: can't detach kernel driver: No data available
FET protocol version is 30000002
Configured for Spy-Bi-Wire
Set Vcc: 3000 mV
Device: MSP430F22xx
Sending magic messages for >= 30000000
Bound to port 2000. Now waiting for connection...

Then run debugger. I like DDD:

$ ddd --debugger msp430-gdb test.elf

In the debugger console run a command to connect to the stub:

(gdb) target remote :2000
0x0000e000 in _reset_vector__ ()

Now you can debug at will.

GCC

The easiest way for users of Debian and derived distributions is to add following repository to /etc/apt/sources.list:

deb http://merlin.fit.vutbr.cz/FITkit/download/debian debian fitkit

Then run as root:

# wget http://merlin.fit.vutbr.cz/FITkit/download/debian/debkey.asc -O- | apt-key add -
# apt-get update
# apt-get install msp430-devtools

After that you will have full toolchain. Lets try to compile the following C source:

#include <io.h>
 
void wait(void)          //delay function
{
  volatile int i;        //declare i as volatile int
  for(i=0;i<32000;i++);  //repeat 32000 times
}
 
int main(void)
{
  WDTCTL = WDTPW | WDTHOLD;
 
  P1DIR=0xFF;            //port 1 = output
  P1OUT=0x01;            //set bit 0 in port 1
 
  for(;;) {              //infinite loop
    P1OUT=~P1OUT;        //invert port 1
    wait();              //call delay function
  }
 
}

Compile the source and generate output suitable for programming the MCU:

$ msp430-gcc -Os -mmcu=msp430x2234 -o test.elf test.c
$ msp430-objdump -DS test.elf > test.lst      # assembler output
$ msp430-objcopy -O ihex test.elf test.hex
$ ls
test.c  test.elf  test.hex  test.lst

It’s done!

Programming

For this purpose there is all-in-one tool called mspdebug. It was originaly created for EZ430-RF2500 but now it supports all the MSP430 programmers. I’ve tested it with the RF2500 programmer. The easiest way to program the MCU’s flash is to run the following command:

$ mspdebug -R "prog test.hex"
MSPDebug version 0.6 - debugging tool for MSP430 MCUs
Copyright (C) 2009, 2010 Daniel Beer <daniel@tortek.co.nz>
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Trying to open interface 1 on 002
rf2500: warning: can't detach kernel driver: No data available
FET protocol version is 30000002
Configured for Spy-Bi-Wire
Set Vcc: 3000 mV
Device: MSP430F22xx
Erasing...
Writing 120 bytes to e000...
Writing  32 bytes to ffe0...

It’s done!

Debugging

Next time…

Besides the PNG transparency Internet Explorer has many other holes. Missing CSS selectors, wrong box model, … I found IE7 script which patches most of those holes. So great! Now I can use selectors like DIV > DIV or DIV + DIV everywhere. It also fixes the box model (total width = margin + border + padding + width) and min-height/min-width. It tries to fix the PNG transparency too but it is not as well tuned as the iepngfix. You should remove this path of the script.