gemini - kennedy.gemi.dev

💾 Archived View for mirrors.apple2.org.za › archive › apple.cabi.net › System › GSOS.P8.Anatomy › FI… captured on 2024-02-05 at 14:49:08.

-=-=-=-=-=-=-

Mana

In this chapter, we familiarize you with the concept of a file and explain how the
I'roDOS file system organizes files on the disk drive medium. You need to know the
details of the ProDOS file system if you want to better comprehend the internal
GS/OS and ProDOS 8 file-handling commands described in Chapter 4. (GS/OS works
with non-ProDOS file systems as well, but most users will be using it with disks
formatted for the ProDOS file system.)

The concept of a file is fundamental to all disk operating systems. A file is just a
collection of data that can define an executable program, a letter to the editor, a
spreadsheet template, or any other document a program can deal with. The general
structure of a file is defined by the operating system itseli; the operating system also
provides the various commands for accessing the file in different ways: create, open,
read, write, close, destroy, rename, and so on.

NAMING FILES

When you first save a file to disk, you must assign it a unique filename that a program
can use to identifi' it thereafter. A ProDOS filename can be up to 15 characters long.
It must begin with an alphabetic letter (A to Z), but the other characters may be any
combination of letters, digits (0 to 9), and periods (.). You can use lowercase letters,
too, but ProDOS 8 and GS/OS automatically convert them to uppercase when dealing
with the ProDOS file system. Here are some examples of valid ProDOS filenames:
FORM.LE'1TEB
CONTBACT.3
CHAPTER. FOUR

Here are some examples of invalid filenames and the reasons they are invalid:

5.EASY.PIECES starts with a number
EXPLORING MARS contains an illegal space
THIS&THAT contains an illegal &
THIRD.AND.TWELVE too long

A common mistake that arises in naming files is the use of the space as a word separator
(as in the second example). This is permitted with DOS 3.3 but not ProDOS. Periods, not
spaces, must be used to separate words in a filename to improve readability. Some pro-
grams, like AppleWorks, allow users to enter spaces in filenames, but they internally con-
vert the spaces to periods before using the filenames with operating system commands.

GS/OS, of course, can work with disk volumes that have been formatted for foreign
operating systems (such as Macintosh HFS, MS-DOS, and High Sierra) if the appro-
priate file system translator files are on the boot disk. The naming rules for these file
systems are different from those for the ProDOS file system. Macintosh HFS, for
example, allows names up to 31 characters long; these names can contain any print-
able ASCII character except the colon. Refer to the appropriate operating system
reference manuals for the naming rules for other operating systems.

DIRECTORIES AND SUBDIRECTORIES

When you save a ProDOS file to disk, you can store it in any one of several directories
that may have been created on the disk. These directories are analogous to file folders
in that they are often used to hold groups of related files. (In fact, they are often
referred to as folders instead of directories.) For example, you may create one
directory to hold word processing documents, and another to hold Applesofr programs.
The ability to create separate directories on the same disk makes it much easier to
efficiently organize large numbers of files.

When you first format a disk, only one directory, the volume directory or root
directory, exists; you name it as part of the formatting procedure. (The rules for
naming directories are the same as for naming standard files.) The volume directory
for a ProDOS-formatted disk can hold the names of up to 51 files (whereas a DOS 3.3
directory can hold 105 files).

You can create additional directories (called subdirectories) within the volume
directory using the GS/OS or ProDOS 8 Create command. Indeed, you can even
create subdirectories within subdirectories. A subdirectory can hold the names of as
many files as you wish to store in it, although at some point the disk will become full.
This system of nested directories is called a hierarchical directory structure. Most
modern file systems, including Macintosh HFS, MS-DOS (version 2.x and higher),
and CD-ROM's High Sierra, use similar hierarchical directory structures.

14 Disk Volumes and File Management

To specify the directory a file is to be saved in, you normally add a special prefix to
the filename to create a unique identifier called a pathname. A pathname comprises
the names of a series of directories, beginning with the name of the volume directory
and continuing with the names of all the directories you must pass through to reach
the target directory, followed by the filename itself. Each directory name is separated
from the next by a special separator character, and a separator must precede the name
of the volume directory.
Under GS/OS, the separator character can be either a slash (I) or a colon (:). Under
ProDOS 8, it must be a slash. We use the slash as the separator character in the
following discussion.

The directory names in a pathname chain must define a continuous path - that is,
each directory specified must be contained within the preceding directory. For exam-
ple, suppose a disk has a volume directory called BASEBALL and two subdirectories
within BASEBALL called AMERICAN and NATIONAL. (Figure 2-1 shows such a
directory arrangement.) If you want to save a file called NY.YANKEES in the AMER-
ICAN subdirectory, you would specif' the following pathname:

/BASEBALL/AMERIcAN/NY YANKEES

If you had specified the name NY.YANKEES itself the file would have been saved in
the current directory, which is usually the volume directory (unless it has been
changed using the SetPrefix command described next).

Under GS/OS, you can specifi' a device name, instead of a volume directory name,
when forming a pathname. Device names begin with a period (.) and can be between
2 and 31 characters long. Examples of device names are .SCSI1, .DEV4, and
.APPLEDISK3.5A. If the NY.YANKEES file in the above example is on the disk in
the drive whose device name is .SCS11, you could identifi' it with the following
pathname instead:

.SCSI1/AMERIcAN/NY.YANKEES

This technique cannot be used with ProDOS 8 because ProDOS 8 does not use
device names.

As we saw above, the separator for a GS/OS pathname can be a slash or a colon, but
you can't use both as separators in a single pathname. GS/OS determines what the
separator is by scanning the pathname from left to right until it finds a slash or colon;
the character it finds is the separator.
If the GS/OS separator is a colon, you can use slashes in GS/OS filenames, which is
important if you're accessing files on a non-ProDOS disk volume through a GS/OS file
system translator. (Macintosh files, for example, can include slashes.) The reverse is
not true, however: If the separator is a slash, you cannot use a colon in a filename.
Thus it's best to always use the colon as a pathname separator in GS/OS applications.

Directories and Subdirectories 15

Figure 2-1 The ProDOS hierarchical directory structure

/BASEBALL/NATI0NAL/CHAMPS

AMERICAN/ PIRATES. 1960 TXT
NY.YANKEES TXT GIANTS. 1954 TXT

/BASEBALI,/AMERIcAN/CHAMPS

Prefixes

If most of the files you arc using are in the same subdirectory, it becomes annoying to
have to specify the same chain of directory names every time you want to access a file.

To abate this annoyance, GS/OS and ProDOS 8 have a SetPrefix command you can
use to set the chain of directory names to which any filename specified in a command
will be automatically appended. The chain is the default prefix and cannot be more
than 64 characters long under ProDOS 8 or 8K characters long under GS/OS.

For example, if you set the default prefix to ,BA5EBALL/AMERICAN/, you can refer

to any file ii' the directory at the end of this path (such as NY.YANKEES) by filename only.
A name that is a continuation of the default prefix could also be specified to access
files in lower-level subdirectories; such a nanle is called a paJ~al pathnamc. If the
default prefix has the value just described, and if AMERICAN contains a subdirectory
called CHAMPS that contains a file called TWINS.1987, you could access the file by
specifying a partial pathname of CHAMPS/TW1NS.1987. Here the pathname is not
preceded by a slash.

16 Disk Volumes and File Management

Under GS/OS (but not ProDOS 8), the default prefix also goes by the shorthand
name of 01. This means 01 is equivalent to 1BASEBALL1AMER1CAN1 if you've used
SetPrefix to assign 1BASEBALL1AMER1CAN/ to the 01 prefix. As Table 2-1 shows,
GS1O5 supports 32 different prefixes you can refer to by a number followed by a slash
(01 through 311) and a boot prefix called ~/. GS1OS sets */ to the name of the disk you
booted from; you cannot change */z 11 and 91 identify the directory in which the
current application resides, and 21 identifies the directory containing system library
files. You can change 11, 21, and 91 with the GS1O5 SetPrefix command, but it's
probably best to leave them alone. Use the user-definable prefixes if your application
needs to identifi' a particular directory using the convenient GS1O5 shorthand notation.

ProDOS 8 prefixes can be up to 64 characters long, including the preceding slash.
Partial pathnames can be up to 64 characters long as well. GS1OS has both short and
long prefixes. Short prefixes (*1 and 01 through 71) can be up to 64 characters long and
long prefixes (81 through 311) can be up to about 8192 characters long.

A good feature of GS1OS and ProDOS 8 is that whenever a command must locate
a file described by a pathname, it searches every disk available to the system. Contrast
this with the DOS 3.3 environment where you must explicitly specifi' the drive and
slot number for the file before you can access it (using the ,S# and ,D# parameters).
BASIC.SYSTEM, for reasons of compatibility, also permits the use of the ,S# and
,D# parameters. If you specifi' a filename or partial pathname in a command line, and
no default prefix has yet been defined, or if either the slot or drive parameter is used,
BASIC.SYSTEM automatically uses the name of the volume directory for the disk in
the specified slot and drive (or their defaults) to create the full pathname.

The advantages of using subdirectories are often not readily apparent to users of
floppy disks but are obvious to hard disk users. Hard disks have enough room for
hundreds of files. If all the files were held in one directory, you might have to wait a
long time to spot your file when the disk was cataloged, and even then you could well
miss it among the other files. Fortunately, the hierarchical directory structure ProDOS
uses allows related files to be grouped within the same subdirectory for easy access.

FUNDAMENTAL FILE-HANDLING CONCEPTS

As we see in Chapter 4, G51OS and ProDOS 8 both include a command interpreter
that understands a variety of file-handling commands. The most common commands
used with existing files are

Open open a file for 1/0 operations
Read read data from the file
Write write data to the file
Close close the file to 1/0 operations

(Four similar commands are also available from Applesoft when you are using the
BASIC.SYSTEM interpreter in a ProDOS 8 environment.) Let's review each of these
fundamental file-handling operations.

Fundamental File-Handling Concepts 17

Table 2-I Standard prefix numbers for GS/OS

The boot prefix. This is the name of the volume GS/OS was
booted from. This prefix cannot be changed by the user.

01 The default prefix. GS/OS automatically attaches it to any
filename or partial (rather than full) pathname you specify.

11 The application prefix. The pathname of the directory containing
the current application program.

21 The system library prefix. The pathname of the directory
containing library modules used by the current application. For a
standard GS/OS boot disk, this is 1MYDISK/SYSTEM1LIBS.

31 to 81
91

101 to 311

Opening a File

User-definable.
Same as for 11.
User-definable.

You must open a file before you can access it. Do this by using the Open command
and specifying the name of the file you wish to open. The operating system opens a file
by first locating it on the disk and then setting up a special buffer area for it in memory.
Part of the file buffer holds information that tells the operating system where the
file data is located on disk; another part holds the most recently accessed portion of
the file. Whenever you request a file 110 operation, the operating system determines
whether the portion of the file to be accessed is already sitting in the file buffer. If it
is, the operating system does not need, nor does it bother, to access that portion of the
file from the disk. Instead, it simply stores the data in the buffer (a write operation) or
reads the data from the buffer (a read operation). As a result, file operations occur
much more quickly than if unbuffered disk 110 techniques were used.

ProDOS 8 can open a file at one of sixteen different system file levels (numbered
from 0 to 15); GS1OS supports 256 different system file levels (0 to 255). Under
ProDOS 8, an application can specify the system file level by storing the level number
at a particular memory location ($BF94) just before opening the file. Under GS1OS, the
application must use the SetLevel command instead. The default system file level is 0.

The main advantage of having different file levels available is to make it easier to
write supervisory or executive programs. These types of programs typically open their
own work files, pass control to user programs, and regain control when the user
programs end. If a supervisory program bumps the file level by one before a user
program takes over, its work files can't be inadvertently closed by the user program,

18 Disk Volumes and File Management

even if the program tries to close all open files (unless the user program breaks a rule
and decrements the file level).

Reading and Writing a File

When the operating system opens a file, it initializes two important internal pointers it
uses for keeping track of the size of the file and the last position in the file that an
application accessed. These are called the EOF and Mark pointers. See Figure 2-2.

EOF is the end-of-file pointer, and it always points to the byte after the last byte in

the file. If you try to read data from the file past this position, an error occurs (the "end
of data" error). EOF normally changes only if an application writes information to the
end of a file; when this happens, EOF automatically increases by the appropriate
number of bytes, and if necessary, the operating system allocates more blocks on the
disk. But as we see in Chapter 4, GS/OS and ProDOS 8 also have a SetEOF command
you can use to set EOF to any specific value.

Mark is the position-in-the-file pointer, and it always contains the position at which

the next read or write operation will take place. It is set to 0 (the beginning of the file)
when you first open a file, but it automatically increases as information is read from or
written to the file. For example, if Mark is currently 10 (that is, it is pointing to the
11th byte in the file), and you read or write 14 more bytes of information, Mark
advances to 24.

It is also possible to explicitly set Mark to any position in the file so that you can
access the file randomly. This means a program can retrieve a record from a file
containing fixed-length records very quickly because it is not necessary to read
through all preceding records first.

Closing a File

You must close a file when you're finished dealing with it. This ensures that any data
written to the file buffer, but not yet stored on the disk itself is actually stored on the
disk. It also updates file information, such as size, in the directory.

Although it is not necessary to close a file immediately after you're finished with it
(you could wait until the program is about to end), it makes good sense to do so to
reduce the risk of data loss in the event of an unexpected power loss or a system reset.
Moreover, ProDOS 8 allows only so many files to be open simultaneously; if you have
a lot of inactive, but open, files lingering around, you could be faced with a surprising
error message the next time you open a file. Another compelling reason to close
unused files is to free up memory space; each open file reserves a buffer area that is
made available to the system when you close the file.

GS/OS DISK CACH1NG

To speed up disk operations like the ones described above, GS/OS supports the
caching of disk blocks. The cache is an area of memory where GS/OS saves copies of

GS/OS Disk Caching 19

Figure 2-2 The ProDOS 8 and GS/OS EOF and Mark pointers

82 83

Mark E0F

(b) EOF and Mark after 10 bytes of the file have been read:

E0F
and
Mark

NOTE: EOF is automatically extended.

disk blocks when it first reads them from disk. GS/OS also puts in the cache copies of
blocks it writes to disk. Once a block is in the cache, GS/OS can quickly get it from
memory whenever it needs to read the block again; GS/OS doesn't have to access the
relatively slow disk drive to get it.

The user usually sets the size of the disk cache with the Disk Cache desk accessory.
Like any desk accessory, Disk Cache appears in the Apple menu of most applications
which use the Apple IIcs Menu Manager, including the Finder. An application can
also set the cache size by calling the GS/OS ResetCache command after saving the
new cache size to Battery RAM with the WriteBParam function (see Chapter 4).
Generally speaking, the larger the cache, the better GS/OS will perform, but less
memory will be available to applications.

In most cases, the block cache is not large enough to hold all the blocks which
GS/OS may want to cache. When the cache is full, GS/OS throws out the least
recently used block to make room for the next block.

The GS/OS Read and Write commands (see Chapter 4) let you specify whether
specific disk blocks are to be cached or not. Applications should try to cache blocks
they expect to frequently access.

20 Disk Volumes and File Management

PRODOS FILE MANAGEMENT

Disk operating systems use different methods to organize files on disk and keep track of
what parts of the disk are being used for data storage so that files can be easily and effi-
ciently created, deleted, and accessed. In this section, we investigate the following topics:
z The structure of a ProDOS-formatted disk
z The structure of the ProDOS volume bit map
z The structure of ProDOS directories and subdirectories
z The structure of a ProDOS directory entry
z The indexing schemes ProDOS uses to locate files

ProDOS uses the same general method to organize files on every block-structured,
mass-storage device it works with (such as an Apple 5.25 Drive, an Apple 3.5 Drive,
an HD20SC, and the /RAM volume). Specific differences arise because the storage
capacities of these different devices vary. Furthermore, the sizes of two important data
structures stored on the media, the volume directory and the volume bit map, might
be different. We generally focus on the Apple 5.25 Drive (and its 5.25-inch floppy
disks) in this section; any specific differences for other devices that are not obvious
will be mentioned.

FORMATTING THE DISK MEDIUM

Before you can use a floppy disk (or any other disk medium) with GS/OS or ProDOS
8, it must be formatted into a state that GS/OS or ProDOS 8 recognizes. You can
format a disk with the Filer or System Utilities program on Apple's ProDOS 8 master
disk or the Apple IIcs Finder. GS/OS also has a Format command that applications
can use to format a disk.

The method used to format a disk depends on the nature of the disk device. When
you format a 5.25-inch floppy disk, for example, templates for 35 tracks on the disk are
created (numbered from 0 to 34), each of which can hold 4096 bytes of information.
These tracks are arranged in concentric rings around the central hub of the disk, with
track 0 at the outside edge and track 34 at the inside edge. The operating system can
access any track by causing a read/write head (located inside the disk drive) to move to
the desired track. This is done using 1/0 locations that activate a stepping motor that
controls the motion of a metal arm the read/write head is connected to. This arm
moves along a radial path beginning at the outside edge of the disk (track 0) and
ending at the inside edge (track 34).

Each of the 35 tracks formatted on a disk is subdivided into 16 smaller units, or
sectors. A sector is the smallest unit of data that can be written to or read from the disk
at one time. The sectors that make up a track are numbered from 0 to 15, and each can

Formatting the Disk Medium 21

hold 256 bytes of information. If you do the mathematics, you will quickly determine
that a disk can hold 560 sectors (140K) of information.

This is the last you'll hear about sectors, however, since ProDOS uses the 512-byte
block as the basic unit of file storage; each block is made up of two disk sectors. An
initialized disk is made up of 280 such blocks (numbered from 0 to 279). Fortunately,
it is rarely necessary to know where these blocks are actually located on the disk since
the operating system disk driver subroutine automatically maps block numbers to
actual physical locations on the disk.

DISK VOLUMES AND DISK DRIVES

A formatted floppy disk that is on line (placed in a system disk drive and ready to be
accessed) is often called a disk volume. ProDOS-formatted volumes have names that
follow the same naming rules as files, but they are often preceded with a slash (/) to
make them more recognizable as volume names.

Disk drives themselves also have unique identifiers. ProDOS 8 assigns a unit
number to each disk device it finds in the system. The value of the unit number is
formed from the slot number of the disk drive controller card and the drive number.
Figure 2-3 shows the format of the unit number byte.

In Figure 2-3, SLOT may actually be the number of a phantom, or logical, slot if

the system contains nonstandard disk devices like BAMdisks. The unit number for the
/RAM volume on a lIe, lIe, or IIGs is $B0, for example; in other words, /RAM is the
logical slot 3, drive 2 device.

DR indicates the drive number: It is 0 for drive 1 and 1 for drive 2. More than two
drives may be connected to the port 5 SmartPort. In this case, ProDOS 8 logically
assigns the next two drives to slot 2, drive 1 and slot 2, drive 2. ProDOS 8 ignores all
SmartPort drives after the first four.

GS/OS assigns unique device reference numbers to the disk devices (and character
devices) it finds - these numbers are consecutive integers beginning with 1. It also
assigns device names to each device; examples are .APPLEDISK3.5A, .SCSI1, and
.DEV3. (These names can be from 2 to 31 characters long.) GS/OS does not use the
unit number scheme that ProDOS 8 uses.

(See Chapter 7 for more detailed information on disk devices and naming conventions.)

DISK VOLUME BLOCK USAGE

We are now ready to examine the method ProDOS uses to manage files on a disk. Our
discussion includes an analysis of the structures of the directories that hold informa-
tion about files, of the volume bit map that keeps track of block usage on the disk, and
of the index blocks that contain the locations of the data blocks each file uses.
But before we continue, keep in mind that the following descriptions relate only to
the ProDOS file system and not to its predecessor, DOS 3.3, the Apple Pascal file
system, or any other foreign operating system.

22 Disk Volumes and File Management

Figure 2-3 The format of a ProDOS 8 unit number byte

7 6 $

4 3 2 1 0

OR SLOT [Unused]

As we have seen, a total of 280 blocks, holding 140K of data, are available on a
ProDOS-formatted 5.25-inch disk. If a standard disk-formatting program is used,
however, seven of these blocks (0-6) are not available for use by files becanse ProDOS
reserves them for special purposes. Figure 2-4 shows the usage of blocks on freshly
formatted 5.25- and 3.5-inch disks.

Blocks 0 and 1 contain a short assembly-language program that the firmware on the
drive controller card loads into memory and executes whenever it boots a disk. This
program is called the boot record, and it locates, loads, and executes a special system
file called PRODOS if it finds it on the disk. (A system file has a file type code of $FF
and a CATALOG mnemonic of SYS. We discuss file type codes later in this chapter.)
PRODOS is the program ultimately responsible for installing and activating the
operating system. (See Chapter 3.)

Blocks 2 through 5 are the blocks containing the volume directory for the disk. We
describe the structure of this directory later in this chapter.

Block 6 is the first volume bit map block for the disk. Each bit in the map indicates
whether the block it corresponds to is free or in use. ProDOS reserves one bit map
block for each 2Mb (4096 blocks) of storage space.

The blocks past the end of the bit map block (or blocks), a total of 273 for a
5.25-inch disk or 1593 for a 3.5-inch disk, are free for use by files stored on the disk.

THE VOLUME BIT MAP

The operating system accesses the volume bit map to determine the status of each
block on the disk. It reads the bit map whenever it allocates new space to a file so that
it can quickly locate free blocks on the disk. It writes to the bit map to reserve new file
blocks (this occurs when an existing file grows or a new one is saved) or to free up
blocks (this occurs when a file shrinks or is deleted).

Standard formatting routines use block 6 as the first block for a disk's volume bit map.
But block 6 is only the conventional location for the bit map; it is permissible to store the
map in any free block on the disk. For example, the volume bit map for the /RAM volume
is in block 3. As we see in the next section, the block number for the first bit map block
appears in the directory header that describes the characteristics of the disk volume.

For a 5.25-inch disk, only the first 35 bytes (280 bits) in the volume bit map block

are used, and each bit in each byte corresponds to a unique block number. A one-block bit
map such as this can handle volumes of up to 4096 blocks. For larger volumes, like a hard
disk, a continuation of the bit map can be found in the blocks on the disk immediately
following the first one used. For example, the old 9728-block Apple ProFile hard disk

The Volume Bit Map 23

Figure 2-4 Map of block usage on a 5.25-inch disk and a 3.5-inch disk

Each block holds 512 bytes.

Continuation of tne
volume bit map
(one block for each
2Mb of storage)

Start of the
volume bit map

Volume directory

Boot record

Total storage capacity is 280 blocks (140K) for a 5.25-inch disk
Total storage capacity is 1600 blocks (80OK) for a 3.5-inch disk

requires three blocks for its bit map; the standard formatting program stores the first part
of the map in block 6 and the continuation in blocks 7 and 8. (The operating system
determines the size of the volume bit map by examining 2 bytes in the volume directory
header that hold the size of the disk; the program used to format the disk places them
there. We look at volume directory headers later in this chapter.)

Figure 2-5 shows the structure of the volume bit map for 5.25-inch disks. As you
can see, the bits in each byte in the bit map block reflect the states of eight contiguous
blocks; bit 0 corresponds to the highest-numbered block in the octet and bit 7 to the

24 Disk Volumes and File Management