Forth Lesson 13

From OLPC
Revision as of 13:36, 13 August 2012 by FGrose (talk | contribs) (Saving and loading)
Jump to: navigation, search
Mitch Bradley's Forth and
Open Firmware Lessons:

See also the first section of this page for editing files on an external device like a USB stick or SD card.

Review

In the previous lesson we learned that:

  • Console I/O operates directly on a framebuffer (text is rendered by Forth code)
  • OFW can boot from ELF images, Linux bzImage files, or Forth source files
  • Boot images can be explicitly loaded from either local storage devices or the network (TFTP, NFS, HTTP)

Overview

In this lesson we'll cover how to write Forth programs and save them in persistent storage on the XO.

Storage options

The OFW and EC code both reside in SPI FLASH. There is extra room there that, in principle, could be used to store user-entered stuff, but there are some practical problems. There is the risk of bricking the machine, from which recovery is possible, but extremely tedious, requiring a special dongle and an on-board connector that is often unpopulated. There is also the problem that, in order to write to the SPI FLASH, it is necessary to hold the EC's 8051 in reset (so it doesn't interfere with the programming by doing microcode instruction fetches from the same serial bus that the programming commands go over). While the 8051 is in reset, the keyboard doesn't work, and when you have finished programming, you have to restart the 8051, which resets the whole system.

OFW can write to files on FAT-16 and FAT-32 filesystems, so you can use USB keys and SD memory cards to store your own files. An SD memory card is a good choice for adding personal storage, so the examples will use filenames on the SD memory filesystem.

Text editing

First, if all you need to do is view a file without changing it, you can use:

  ok more u:\boot\olpc.fth

As of release Q2E15, OFW contains a microEMACS text editor. To run it, type

  ok emacs

You can also specify the name of a file on the command line, for example:

  ok emacs u:\boot\olpc.fth

microEMACS has built-in help; just look at the red window to get started. You can get more help by typing the F1 key then scrolling the pop-up help window. You can do simple editing by typing and using the arrow keys to move around. For more complex tasks you can use many of the common ctrl and esc commands that are present in full EMACS. (The one feature I really miss from full EMACS is filename completion.)

When you specify filenames to microEMACS, you have to use the OFW syntax, because microEMACS relies of OFW for all of its I/O and system functions. microEMACS can only write to filesystems that OFW knows how to write, so while you can use it to view files on NAND FLASH, you can't write changes back to NAND FLASH. You could, however, open a file from NAND, edit it, and write it back to a USB key.

If you have an open wireless access point, try this:

  ok wifi <your_essid>
  ok emacs http:\\dev.laptop.org\pub\firmware\scripts\batman.fth

The information below is mostly of historical interest if you have a recent OFW with microEMACS.

OFW has a simple editor; it is the same one that is used for editing command lines, but it also works for multi-line memory buffers.

  ok nvedit
  <Edit using arrow keys or basic emacs>
  ^C
  Store script to NVRAM [y/n]? y
  ok

It doesn't really store the data to NVRAM, because the OLPC machine doesn't have any (CMOS RAM is too small, and SPI FLASH has the problems listed above). What it does is save the data in memory, where you can get it back with "nvramrc ( -- adr len )".

  ok  nvramrc .s
  ff9f4baa 3d
  ok nvramrc list
  <whatever you put in there>
  ok nvramrc evaluate
  <executes the contents of nvramrc as Forth code>
  ok nvramrc eval              \ Same as above but shorter
  ok

Saving and loading

To save it to disk, you could write this definition:

  : nvwrite  \ filename  ( -- )
    safe-parse-word  $create-file >r    nvramrc " write" r@ $call-method
  drop  r> close-dev
  ;

then

  ok nvwrite sd:\myfile.fth

then one of these lines to evaluate the Forth code from disk:

  ok including sd:\myfile.fth       \ For standards-aware persons
  ok fload sd:\myfile.fth           \ Another way to say the same thing
  ok fl sd:\myfile.fth              \ For impatient people like me

$create-file will not overwrite an existing file. You can delete it with:

  ok  rm sd:\myfile.fth

Or you could modify nvwrite as follows:

  : nvwrite  \ filename  ( -- )
    safe-parse-word  2dup ?delete-file  if
      $create-file >r    nvramrc " write" r@ $call-method drop  r> close-dev
    then
  ;

Here is a definition that will read a file into nvramrc so you can edit it

  : nvread  \ filename  ( -- )
    reading
    load-base   dup h# 10000  ifd @ fgets   ( adr len )
    to nvramrc
    ifd @ fclose
  ;

The primitive editing routine that nvedit uses is "edit-file ( adr len maxlen -- newlen )" . If you want to do your own buffer management, you could use that to build a custom set of editing words.

Thus endeth the lesson.

Next Lesson: Some Cute Hacks