FTDI support under Puppy linux

[volhout]

Dear Support team,

I have been following this thread with great interest, since I am a great puppy lover (since 2.17), but have never taken the step to do anything else than assembly language, a bit of bash, and (in the early days) basic.

Your joined effort to get BaCon working in Wary made me decide to drop Puppy 5 and install Wary 5.1.1 with success. Installed the devpack, and downloaded hug.bac. Using NicoEdit.
And I have been programming since, taking Barry's proxy program as a reference. I am impressed, and would like to use this configuration for my hobby (model railway automation). In this I need digital IO. One of the options for this is USB IO chips like the FTDI range (I own a few FT245BM modules).

But I have no clue on how to add the drivers for these USB chips to the BaCon environment. There is a Linux library (1.0.4) for Mandrake 9, but that would (probably) need re-compilation for Wary ?

www.ftdichip.com/Drivers/D2XX.htm

And even then, how can I access this library from BaCon ?? Do I use the EXEC command ?

Can someone help me ??

Volhout

P.S. I originally posted this at the puppy linux forum but was directed here. I hope this is the right place...

[Pjot]

Hi volhout,

Can you show me a piece of code (preferably C) which uses these drivers? From there we can try to convert it to BaCon.

Thanks,
Peter

[volhout]

Hi Peter,

There are several code examples in the xxx.tar file that you download when you click in the matrix at 1.0.4 for the Linux version of the FTDI driver.

Is that sufficient to get an idea how difficult it is to port it ??
Agian, I am not looking for a "seamles" integration. Anything you can do to make it work is fine. When you have trouble working with the official FTDI driver, I may have found an alternative, but I am not sure how good the support and compatibility is.


www.intra2net.com/en/developer/libftdi/index.php

C code example

www.intra2net.com/en/developer/libftdi/documentation/

Not sure how it must be tailored for Wary 5.1.1 (Puppy) Linux.

Many of the USB-RS232 adapters use FTDI chips, so it may be possible for you to find a adapter locally (something to check of the library connects via USB at all). Alternatively I could send you one of mine if needed.

Volhout.

[Pjot]

Hi Volhout,

Basically it comes down to IMPORTing the necessary C calls into BaCon and then use them in your prgram. You can find many examples of this on the BaCon website under the section "Interfaces".

A very useful source of information is the FTDI header file on basis of which I have defined the IMPORTS and CONSTANTS below.

If we look at the FTDI example program, we can port it as follows:

TRAP LOCAL

' Import some calls first
CONST library$ = "libftdi.so.1"

' Get the functions from the library
IMPORT ftdi_init(long) FROM library$ TYPE int
IMPORT ftdi_usb_open(long,int,int) FROM library$ TYPE int
IMPORT ftdi_get_error_string(long) FROM library$ TYPE char* ALIAS ftdi_get_error_string$
IMPORT ftdi_read_chipid(long,unsigned int) FROM library$ TYPE int
IMPORT ftdi_usb_close(long) FROM library$ TYPE int
IMPORT ftdi_deinit(long) FROM library$ TYPE void

CONST TYPE_AM = 0
CONST TYPE_BM = 1
CONST TYPE_2232C = 2
CONST TYPE_R = 3
CONST TYPE_2232H = 4
CONST TYPE_4232H = 5

' BaCon does not know types so we define
'   a RECORD with the needed members
RECORD ftdic
' USB specific
LOCAL usb_dev TYPE long
LOCAL usb_read_timeout TYPE int
LOCAL usb_write_timeout TYPE int
' FTDI specific
LOCAL type TYPE int
LOCAL baudrate TYPE int
LOCAL bitbang_enabled TYPE unsigned char
LOCAL *readbuffer TYPE unsigned char
LOCAL readbuffer_offset TYPE unsigned int
LOCAL readbuffer_remaining TYPE unsigned int
LOCAL readbuffer_chunksize TYPE unsigned int
LOCAL writebuffer_chunksize TYPE unsigned int
LOCAL max_packet_size TYPE unsigned int
' FTDI FT2232C requirecments
LOCAL interface TYPE int
LOCAL index TYPE int
' Endpoints
LOCAL in_ep TYPE int
LOCAL out_ep TYPE int
LOCAL bitbang_mode TYPE unsigned char
LOCAL eeprom_size TYPE int
LOCAL error_str TYPE char*
LOCAL async_usb_buffer TYPE char*
LOCAL async_usb_buffer_size TYPE unsigned int
END RECORD

LOCAL chipid

IF ftdi_init(ADDRESS(ftdic)) < 0 THEN
PRINT "ftdi_init failed"
END 1
END IF

ret = ftdi_usb_open(ADDRESS(ftdic), 0x0403, 0x6001)

IF ret < 0 THEN
PRINT "unable to open ftdi device: ", ret, " (", ftdi_get_error_string$(ADDRESS(ftdic)), ")"
END 1
END IF

' Read out FTDIChip-ID of R type chips
IF ftdic.type = TYPE_R THEN
PRINT "ftdi_read_chipid: ", ftdi_read_chipid(ADDRESS(ftdic), ADDRESS(chipid))
PRINT "FTDI chipid: ", chipid
END IF

ret = ftdi_usb_close(ADDRESS(ftdic))

IF ret < 0 THEN
PRINT "unable to close ftdi device: ", ret, " (", ftdi_get_error_string$(ADDRESS(ftdic)), ")"
END 1
END IF

ftdi_deinit(ADDRESS(ftdic))

END

Copy and paste this code into a file, for example "ftdi.bac", then compile. On my Ubuntu 10.10 64bit I see the following:

peter@Solarstriker:~/programming/bacon$ bacon fdti
Starting conversion... done.
Starting compilation... done.
Program 'fdti' ready.
peter@Solarstriker:~/programming/bacon$ ./fdti
unable to open ftdi device: -3 (device not found)
peter@Solarstriker:~/programming/bacon$

This makes sense of course as I do not have such a chip attached to my system :-)

Best regards
Peter

[2lss]

I was wondering the same thing about ftdi devices.

However I was thinking of a different approach.

I have a Arduino (which is a little micro controller, coupled with an ftdi chip for usb support) and I used a bash script to read the output from /dev/ttyUSB0.

I don't have the script at the moment, but it used the 'tail' command like so:

tail -f /dev/ttyUSB0

This just ran in a loop and executed different programs based on the output of the Arduino.

I'm curious if something similar could be done with bacon. Like:

OPEN "/dev/ttyUSB0" FOR READING AS ftdi_device
READLN txt$ FROM ftdi_device
do stuff
CLOSE FILE ftdi_device

[volhout]

@peter
I saw your replay and started doing everything that was described in the FTDI readme file. When I was ready, I compiled your program, and got an error message. Only then, I realized your example was based on the open source (the second option) library.

I tried to find a list of actions to perform to build the open source library, but could not find one. I am stuck. There is nop compiled library in the tar file, and I have not the faintest clue where to start to compile it (have never done this before). It must be very straight forward, but how

2lss
From what you write, you are using the FTDI VSP library (virtual serial port). That is the default library for most systems since the primary market for these chips is USB serial port extension. When you plug a FTDI device in (almost any) linux or windows PC it will show up as a serial port.
But there is a second library The D2xx (direct access) that allows this so called "bit-bang mode. I will have to blacklist the VSP driver to allow this to work.

Volhout

[Pjot]

I tried to find a list of actions to perform to build the open source library, but could not find one. I am stuck.

It's not a big deal, really. Just download the FTDI source package, untar and gunzip, then go into the newly created source directory and run './configure' (this tests for packages needed). When finished just run 'make' to compile and then 'make install' to install the library on your system. There are a lot of sites about Linux which provide additional information, for example:

www.tuxfiles.org/linuxhelp/softinstall.html

Best regards
Peter

[volhout]

Peter,

Thanks a lot. I have managed to configure, make, and install the driver.
After this I get the same error message as you did....and when I plug in the FT245B, the error message is gone. So I have a good starting point. I will respond to this thread when I make progress with the project.

Thanks again for the prompt reply and help.

Volhout.....

[Pjot]

volhout: thanks for feedback!

2lss: this should work:

OPEN "/dev/ttyUSB0" FOR READING AS ftdi_device
WHILE TRUE
READLN txt$ FROM ftdi_device
IF LEN(txt$) > 0 THEN PRINT txt$
SLEEP 50
WEND


Regards
Peter

[2lss]

I gave the 'bitbang' program (post four) with my arduino and here are the results:
ftdi_read_chipid: 0
FTDI chipid: 1919903481

Not sure what this means, but I'm assuming its good since it recognized the device.

Then I tried the example above. Unfortunately this seg faults at the READ line and causes some problems at the kernel level. Here is my demsg:

[35269.857134] usb 5-2: new full speed USB device using uhci_hcd and address 4
[35270.054147] usb 5-2: New USB device found, idVendor=0403, idProduct=6001
[35270.054156] usb 5-2: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[35270.054163] usb 5-2: Product: FT232R USB UART
[35270.054167] usb 5-2: Manufacturer: FTDI
[35270.054172] usb 5-2: SerialNumber: A900cexD
[35270.054390] usb 5-2: configuration #1 chosen from 1 choice
[35270.062227] ftdi_sio 5-2:1.0: FTDI USB Serial Device converter detected
[35270.062289] usb 5-2: Detected FT232RL
[35270.062294] usb 5-2: Number of endpoints 2
[35270.062299] usb 5-2: Endpoint 1 MaxPacketSize 64
[35270.062304] usb 5-2: Endpoint 2 MaxPacketSize 64
[35270.062308] usb 5-2: Setting MaxPacketSize 64
[35270.063254] usb 5-2: FTDI USB Serial Device converter now attached to ttyUSB0
[36112.486782] ftdi_kernel[28834]: segfault at 0 ip 00007f9440f6e13d sp 00007fffc71bf070 error 4 in libc-2.11.2.so[7f9440f0b000+158000]

The segfault happened after I ran the second example program.

I also found my bash script that I used to read ttyUSB0. Turns out I was wrong. I used the 'head' command instead of the 'tail' command.

sudo sh -c 'head -c 32 < /dev/ttyUSB0'

EDIT: This was all done on Debian Stable

[Pjot]


sudo sh -c 'head -c 32 < /dev/ttyUSB0'

Then the BaCon code should be different also - 'head -c 32' means the first 32K bytes?

Maybe first you can try to do this:

info$ = EXEC$("head -c 32 < /dev/ttyUSB0")
PRINT info$

Regards
Peter

[2lss]

Mar 21, 2011 10:40:19 GMT 1 Pjot said:

Then the BaCon code should be different also - 'head -c 32' means the first 32K bytes?

I checked the manual for 'head' and yes -c 32 means the first 32 kb. I'm not sure what the significance of 32 kbits means as its been a while since I experimented with the ardiuno.

Mar 21, 2011 10:40:19 GMT 1 Pjot said:

info$ = EXEC$("head -c 32 < /dev/ttyUSB0")
PRINT info$

I that a try and it works as long as its run as root.

I am slightly confused as to why READLN doesn't work. The bacon documentation says:

Reads a line of ASCII data from a file identified by <handle> into variable <var>

and the arduino documentation says (for the serial.print function):

Prints data to the serial port as human-readable ASCII text

www.arduino.cc/en/Serial/Print

So it seems like it should work, unless I am missing something.

[Pjot]

The BaCon READLN is based on the standard C fgets call, with an additional wrapper around it to handle large memory blocks. Note that 'fgets' always expects a CR (0x10) or an EOF on the file descriptor.

The 'print' command where Arduino is pointing at, is a method of their "Serial" object. It is propriety software created to read from a serial interface, so it probably has a different implementation structure. Though the result of this command may look the same as the result of the READLN.

We would need some glance at the way their 'print' is working to see if we can mimic this in BaCon also.

For the 'head' command, we could try to use a binary read limited by 32K instead. This can be done using the GETBYTE statement. The 'head -c 32' could be done as follows:

TRAP LOCAL

' How much data we read in total
CONST amount = 32 * 1024

' Open file
OPEN "/dev/ttyUSB0" FOR READING AS ftdi_device

' Cliam memory to store result
mem = MEMORY(amount)

' Read from file descriptor
GETBYTE mem FROM ftdi_device SIZE amount

' Close descriptor
CLOSE FILE ftdi_device

' Now print the binary data in HRF (Human Readable Format)
FOR x = 0 TO amount - 1
PRINT CHR$(PEEK(mem+x));
NEXT

FREE mem


Regards
Peter

[2lss]

I gave the code above a try and it works just like 'head' Thank you for posting that. Would writing to the serial port be similar? ie using PUTBYTE?

Mar 22, 2011 13:03:27 GMT 1 Pjot said:

The 'print' command where Arduino is pointing at, is a method of their "Serial" object. It is propriety software created to read from a serial interface, so it probably has a different implementation structure. Though the result of this command may look the same as the result of the READLN.

We would need some glance at the way their 'print' is working to see if we can mimic this in BaCon also.

I did a little digging and found the git repo for the arduino project. Here is the source for the print function.
github.com/arduino/Arduino/blob/master/hardware/arduino/cores/arduino/Print.cpp

In addition to that, from what I read, the arduino language is just a set of functions that work with the avr-libc (part of the avr32-toolchain). So I'm assuming that the serial functions would be included in that?

But anyway, I'm getting off topic. For my purposes I think the above example will will be fine. Thanks again for taking the time to help with this.

[Pjot]

Would writing to the serial port be similar? ie using PUTBYTE?

Yes, indeed. I'll leave this as an exercise to you

The 'print' method code you show is C++ based. It shows so-called "overloaded" methods which do not demonstrate the Serial object implementation... anyway, this thread in the forum now has demonstrated 3 ways of accessing the ftdi chip, I hope one of them is useful

Best regards
Peter