Files
Digital-Research-Source-Code/CPM OPERATING SYSTEMS/CPM 68K/1.0X SOURCES/v101/doc/clib.rno
Sepp J Morris 31738079c4 Upload
Digital Research
2020-11-06 18:50:37 +01:00

2517 lines
56 KiB
Plaintext
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

.tab stops 8,16,24,32,40,48,56,64,72,80
.rm 65
.nofill
.title ^^C Language Under CP/M
.da Revision 01.07
.dat Revision 01.07
.swap toc
.bl 5
.c
C Language Under CP/M
.c
---------------------
.bl 2
.c
Revision 01.07
.c
--------------
.bl 2
.c
September 23, 1982
.c
------------------
.bl 15
.c
Digital Research, Inc.
.c
P.O. Box 579
.c
Pacific Grove, CA 93950
.c
(408) 649-3896
.c
TWX 910 360 5001
.bl 4
.c
************************
.c
* Company Confidential *
.c
************************
.bl 5
.fill
Note: This document contains Digital Research proprietary information, and
must not be copied or distributed for any purpose whatsoever, without
prior written permission of Digital Research, Inc.
.bl 1
Address Comments to: Steve Williams, x6082
.break
.swap doc
.title ^^C Language Under CP/M
.chapter Introduction
.title ^^C Language Under CP/M
.subtitle Introduction
.page
.hl 1 Audience
.i 5
This document describes the C language implementation available with
CP/M-68K.
The intent is to provide both an implementation guide, and a preliminary
manual for early users. The reader should be familiar with the C language
as described by Kernighan and Ritchie in "The C Programming Language".
.bl 1
.subtitle Philosophy
.hl 1 Philosophy
.i 5
The implementation philosophy is to provide a system which can run
applications written in C on other operating systems. This usually
means programs written to run under UNIX. Programs which use the UNIX
fork/exec multi-tasking primitives or which rely upon reading UNIX file
structures directly (e.g., directories) obviously cannot be made to run
under CP/M. The vast majority of UNIX software, however, does not use
fork/exec except for chaining, and does not read file structure data. It is
this large body of software with which CP/M C seeks compatibility.
.subtitle Differences from PCC
.bl 1
.hl 1 Differences from PCC
.i 5
The most commonly accepted standard for the C language is the Portable C
Compiler (pcc) written by Stephen C. Johnson, as implemented on UNIX operating
system (Version 7). This compiler is also used in the Zilog, ONYX, XENIX,
Berkeley UNIX, and UNIQ systems. The language accepted by pcc is thus
a defacto standard among UNIX systems.
.bl 1
.i 5
The CP/M C compiler differs from pcc in the following areas:
.bl 1
.ls
.leb;The C "int" (default) data type is 16 bits long. Pointers are 32 bits
long. Care is required to ensure that all function definitions and function
calls which use long (32 bit ints) and pointer parameters use the proper
declarations.
.leb;"long", "int", and "char" register variables are assigned to "D"
registers. Five such registers are available in each procedure.
.leb;Any register variable used as a pointer is assigned to an "A" register.
There are 3 such registers available per procedure.
.leb;Floating point is not implemented.
.leb;Structure assignment, structure arguments, and structures returned from
procedures are not implemented.
.leb;All local declarations (within a function body) must precede the first
executable statement of the function.
.leb;Initialization of automatic variables is not supported.
.leb;Enumeration types are not supported.
.leb;Structure initialization is handled as if the structure were an array of
short integers, as in UNIX version 6.
.leb;Variable and function names must be unique in 8 characters. External
names must be unique in 7 characters.
.els
.subtitle Operating Instructions
.bl 1
.hl 1 Operating Instructions
.i 5
To compile a C program under CP/M, one uses the "c" and "clink" command files.
Separate compilations are handled easily using this method.
.bl 1
The compiler command line is:
.bl 1
.i 5
A>submit c file
.bl 1
where "file.c" is the filename. The compiler produces "file.o" as the
object. The Linker is invoked as follows:
.bl 1
.i 5
A>submit clink file
.bl 1
All libraries and header files are included automatically by the command
file. Multiple files may be specified if desired. For example, to compile
and link files "a.c", "b.c", and "c.c", the following commands would be
used:
.bl 1
.i 5
A>submit c a
.i 5
A>submit c b
.i 5
A>submit c c
.i 5
A>submit clink a b c
.bl 1
The resulting output would be placed in file "a.68k".
.subtitle Memory Layout
.test page 30
.hl 1 Memory Layout
.i 5
C programs under CP/M have a memory layout similar to UNIX C programs. The
program is divided up into three segments: "text" (instructions), "data"
(initialized data), and "bss" (uninitialized data). Automatic variables are
allocated on the stack. These sections are arranged in memory as follows:
.bl 1
.nofill
.lm 8
+----------------------------------+ TPA Low
! Text Segment !
+----------------------------------+ "etext"
! Data Segment !
+----------------------------------+ "edata"
! Bss Segment !
+----------------------------------+ "end"
! Heap (grows to higher addresses) !
+----------------------------------+ "break"
.
.
.
+----------------------------------+
! Stack (grows to lower addresses) !
+----------------------------------+ TPA High
.bl 1
.lm 0
.fill
There are two dynamic memory areas: the stack and the heap. The stack is
used for procedure calls and automatic variables. The heap is managed by the
"brk", "sbrk", "malloc", and "free" procedures described later. This area
is used for dynamically growing data structures, such as symbol tables.
.bl 1
.i 5
The locations "etext", "edata", and "end" are defined by the linkage
editor, and may be used to determine the ending addresses of the ".text",
".data", and ".bss" segments. The "break" location is the first free location
following the heap area.
.subtitle Calling Conventions
.hl 1 Calling Conventions
.i 5
Procedures are called in C via the jsr (Jump to Subroutine) instruction. A6
is used as the frame pointer (for referencing local storage). Arguments are
pushed (in reverse order) onto the A7 stack. Word and character arguments
to functions occupy 16 bits on the stack. Long and pointer arguments occupy
32 bits. D0 is used as the function value return register in all cases.
Functions which declare no return value will return garbage.
.bl 1
.test page 25
.i 5
The sequence:
.nofill
.bl 1
xyz() _{
long a;
int b;
char x;
register y;
.
.
.
b = blivot(x,a);
_}
.bl 1
Generates:
.bl 1
__xyz:
link a6,_#-8 * space for a,b,x
movem.l d6-d7,-(a7) * d7 used for "y"
. * d6 reserves space
.
.
move.l -4(a6),(a7) * Load Parameter "a"
move.b -8(a6),d0 * Load Parameter "x"
ext.w d0 * Extend to word size
move.w d0,-(a7) * Push it
jsr __blivot * Call subroutine
add.l _#2,a7 * Pop argument list
move.w d0,-6(a6) * Store return parameter
tst.l (a7)+ * Purge longword
movem.l (a7)+,d7 * Unsave register(s)
unlk a6 * Restore frame pointer
rts * Return to caller
.bl 1
.fill
Note that due to the varying length of arguments on the stack, older C code
which relied on the fact that all arguments were the same length may not work
without modification.
.bl 1
The compiler prepends an underscore ('__') character to the beginning of
each external variable or function name. This means that all external names
in C must be unique in 7 characters.
.bl 1
The compiler generated code maintains a long word at the top of the stack
for use in subroutine calls. This shortens the stack popping code
required on return from a procedure call. The "movem.l" instruction used
to save the registers contains an extra register to allocate this space.
.bl 1
Registers D3-D7 and A3-A5 are used by the compiler for register variables.
These registers must be saved and restored (if used) by a procedure called
from a C program. The compiler generated code saves only those registers
used, as illustrated above.
Registers D0-D2 and A0-A2 are scratch registers and may
be modified by the called procedure.
.subtitle Stack Frame
.test page 25
.hl 1 Stack Frame
The standard C stack frame has the following appearance:
.bl 1
.nofill
.lm 8
A7 -> +---------------------------------+
! Longword for procedure calls !
+---------------------------------+
! Saved Registers !
+---------------------------------+
! Local variable area !
+---------------------------------+
A6 -> ! Previous value of A6 !
+---------------------------------+
! Return address !
+---------------------------------+
! Argument 1 !
+---------------------------------+
! Argument 2 !
+---------------------------------+
.
.
.
.bl 1
.fill
.lm 0
Arguments are either 2 or 4 bytes, depending on argument type.
A6 is used by the compiler-generated code to reference all variables on the
stack.
.subtitle Command Line Interface
.hl 1 Command Line Interface
.i 5
The standard C "argc/argv" interface for arguments typed on the command line
also works under CP/M. For instance, the command:
.bl 1
.lm 8
command arg1 arg2 arg3 ... argn
.bl 1
.lm 0
causes the following setup:
.bl 1
.lm 8
.nofill
argc = n+1
argv[0] -> "C Runtime"
argv[1] -> "arg1"
argv[2] -> "arg2
.
.
.
argv[n] -> "argn"
.bl 1
.lm 0
.fill
Note that the command name cannot be obtained under CP/M. The argv[0]
argument will therefore always be incorrect.
.subtitle I/O Conventions
.hl 1 I/O Conventions
.i 5
UNIX C programs use two types of file and device I/O: regular files and
stream files. Regular files are identified by a unique number called a
"file descriptor". In CP/M, this number is in the range (0-15). Stream
files are identified by the address of a mysterious control block hidden
deep in the bowels of the run-time system. (This is a USER control block,
the file system does not rely on it).
.bl 1
.i 5
Stream files differ from regular files in that some form of intermediate
buffering is used, making single-byte I/O much more efficient.
.bl 1
.i 5
Under UNIX, printers, terminals, etc. may be referenced as "files", using
the special names "/dev/tty", "/dev/lp", etc. Under CP/M, these names will
be "CON:" for the console device, "LST:" for the listing device.
.bl 1
.i 5
ASCII files under CP/M are stored with a carriage return / line feed pair
following each line. End of file is indicated by a control-Z (0x1a)
character. C programs typically terminate lines with only a line
feed. This means that CP/M C library read and write operations to ASCII
files must insert and delete carriage return characters where appropriate.
The control-Z must be deleted on read and inserted on close for such files.
These operations are not desirable for binary files. An extra
entry point has been added to all file open and creation calls to distinguish
between ASCII and binary files.
.subtitle Standard Files
.hl 1 Standard Files
.i 5
C programs begin execution with three files already open. These files are
known as the standard input, standard output and standard error files, and can
be accessed as either stream or regular files within the C program. The
following definitions are available from the <stdio.h> file:
.bl 1
.nofill
.lm 8
File File Descriptor Stream Name
---- --------------- -----------
standard input STDIN stdin
standard output STDOUT stdout
standard error STDERR stderr
.bl 1
.fill
.lm 0
Standard files may be closed and reopened using the usual library routines.
In addition, I/O redirection is available from the command line.
.subtitle I/O Redirection
.hl 1 I/O Redirection
.i 5
The C program standard input and standard output may be redirected by using
the "<" and ">" characters. No space is allowed between the "<" or ">" and
the filename. The standard error file may not be redirected.
.bl 1
.i 5
For example, the command:
.bl 1
.i 8
test <a >lst: c d e f
.bl 1
executes the file "test.68k" with the standard input coming from disk file
"a" and the standard output going to the listing device. The argument list
is "c d e f".
.chapter Library Routines
.page
.c
Library Routines
.c
----------------
.bl 1
.i 5
The CP/M C library is a collection of routines dealing with I/O,
dynamic memory allocation, system traps, and data conversion.
Specific routines are provided for running on the 68000 or 8086/88
processors, as well as for accessing CP/M in a portable fashion.
.bl 1
.i 5
The C Style Guide (Chapter 3) provides important information on writing
C code in a portable manner. The reader is urged to pay careful attention
to the material found there.
.subtitle Differences from UNIX V7
.hl 1 Differences from UNIX V7
.i 5
Much emphasis has been placed on keeping the C library compatible with UNIX
V7, to allow programs to be easily moved from UNIX to CP/M.
Many UNIX operating system calls and features have been simulated.
Certain C functions available under UNIX simply cannot be implemented under
CP/M:
.ls
.leb;The UNIX fork / exec, kill, lock, nice, pause, ptrace, sync, and wait
primitives require a multi-tasking system, and
cannot be implemented under CP/M without unreasonable effort and poor
performance.
.leb;The UNIX acct system call is not implemented under CP/M.
.leb;The UNIX alarm function (timer), "stime" (set time), "time", "ftime" and
"times" system calls
are meaningless in CP/M, since CP/M
does not require a clock.
.leb;The UNIX "chdir" (Change Directory) system call is not implemented.
.leb;The UNIX "dup" and "dup2" (duplicate file descriptor) functions are
not implemented.
.leb;The UNIX "getuid", "getgid", "geteuid", "getegid", "setuid" and "setgid"
functions for
obtaining and setting the group and user IDs of a process are not implemented.
.leb;The UNIX "indir" (indirect system call) is not available from C under
UNIX, and is not implemented.
.leb;The UNIX "ioctl", "stty", and "gtty" system calls for manipulating
I/O devices are not implemented.
.leb;The UNIX "link" call to connect one file to another is
not implemented.
.leb;The UNIX "chdir", "chroot",
"mknod", "mount", "umount", "mpx", "pipe", "pkon", "pkoff",
"profil", "sync",
"stat", "fstat", "umask" and "utime" system calls
are UNIX file system specific, and
are not implemented.
.leb;The UNIX "phys" system call is not implemented.
.leb;The following UNIX library functions are not available under CP/M:
.ls
.leb;Assert -- program verification routines
.leb;Crypt -- DES Encryption routines
.leb;dbm -- Data base management routines
.leb;Getenv -- environment variable support
.leb;Getgrent, getlogin, getpw, and getpwent functions for reading the
login file
.leb;l3tol, ltol3 -- 3-byte to long conversion routines
.leb;monitor -- Execution profiling routines
.leb;itom, madd, msub, mult, mdiv, min, mout, pow, gcd, and rpow multi-word
integer math routines
.leb;nlist -- fetch name list values from a UNIX load module
.leb;pkopen, pkclose, pkread, pkwrite, and pkfail -- UNIX communications
driver simulation subroutines.
.leb;plot -- graphics interface
.leb;popen, pclose -- piping routines
.leb;sleep
.leb;system -- Execute UNIX command
.leb;ttyslot -- Return slot number in UNIX "ttys" file of controlling
terminal.
.els
.els
The present version of the library does not contain the floating
point routines available under UNIX.
.bl 1
.i 5
In addition to the above omissions, it was necessary to add entry points
to file OPEN and CREATE calls to distinguish between ASCII and binary files.
Byte level End of File is also not available for binary files. ASCII files,
however, are compatible with UNIX, and with the CP/M text editors and
other CP/M utilities which use ASCII files.
.bl 1
.i 5
Another incompatibility is that this documentation does not split up the
UNIX system calls and library functions, since all calls are library
functions under CP/M.
.subtitle abort Function
.page
.hl 1 abort Function
Description:
.bl 1
The abort function causes the current program to terminate with a severe
error. The exact error is system dependent. For the 68000, an Illegal
Instruction Trap is used. This invokes DDT-68K if the debugger is loaded
with the object program.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD code;
.bl 1
abort(code);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
code Is loaded into register D0 before abort
.bl 1
.lm 0
Returns:
.bl 1
.lm 8
The abort function never returns.
.lm 0
Bugs:
.bl 1
.lm 8
.lm 0
.fill
.subtitle abs Function
.page
.hl 1 abs Function
Description:
The abs function takes the absolute value of a single argument. This function
is implemented as a macro (in <stdio.h>), so arguments with side effects
will not work as expected. In particular, the call:
.bl 1
.i 8
a = abs(*x++);
.bl 1
will increment "x" twice.
.bl 1
.nofill
Calling Sequence:
.bl 1
.lm 8
WORD val;
WORD ret;
.bl 1
ret = abs(val);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
val is the input value
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret is the absolute value of "val"
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle access Function
.page
.hl 1 access Function
Description:
.bl 1
The access function checks to see whether a file may be legally accessed
by the invoking program. Under CP/M, the file may be accessed if it
exists.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *name;
WORD mode;
WORD ret;
.bl 1
ret = access(name,mode);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
name points to the null-terminate filename
.bl 1
mode is 4 to check read access
2 to check write access
1 to check exec access
0 to check directory path access
(this argument is ignored by CP/M)
.bl 1
.lm 0
Returns:
.bl 1
.lm 8
0 If access is allowed
-1 If access is not permitted
.lm 0
.fill
.bl 1
Bugs:
.bl 1
The only real check made here is to see if the file exists.
.subtitle atoi, atol Functions
.page
.hl 1 atoi, atol Functions
Description:
.bl 1
The "atoi" and "atol" functions convert an ASCII string to a
integer, or long (respectively) binary number. Strings converted
by atoi and atol have the following format:
.i 8
.bl 1
[ ...][-][+]dddddd ....
.bl 1
Leading spaces are ignored, and one leading sign is permitted.
.bl 1
.nofill
Calling Sequence:
.bl 1
.lm 8
BYTE *string;
WORD ival;
LONG lval,atol();
.bl 1
ival = atoi(string);
lval = atol(string);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
.fill
String is a null-terminated string containing the number to be converted.
Conversion proceeds until digits are exhausted.
Zero is returned if no digits are present.
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
Converted value as "ival" (int), or "lval" (long).
.lm 0
.bl 1
Bugs:
.bl 1
There is no provision for overflow detection or reporting.
It is not possible to limit the number of contiguous digits processed, or to
determine the number of digits actually processed.
.subtitle brk, sbrk Functions
.page
.hl 1 brk, sbrk Functions
Description:
.bl 1
The "brk" and "sbrk" functions
are used to extend the "heap" portion of the user program.
The "brk" function sets the upper bound of the program (the "break" in
UNIX terminology) to an absolute address. The "sbrk" function extends the
program by an incremental amount.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD brk();
BYTE *addr,*sbrk();
WORD incr;
WORD ret;
BYTE *start;
.bl 1
ret = brk(addr);
start = sbrk(incr)
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
addr Is the desired new break address
incr Is the incremental number of bytes desired
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
0 Is returned for success (brk)
-1 Is returned for failure (brk)
.bl 1
start Is the beginning of the allocated area (sbrk)
0 is returned for failure (sbrk)
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle calloc, malloc, realloc, free Functions
.page
.hl 1 calloc, malloc, realloc, free Functions
Description:
.bl 1
The "calloc", "malloc", "realloc" and "free" library functions manage the
dynamic area between the ".bss" region and the stack.
.bl 1
"malloc" allocates an area of contiguous bytes (aligned on a word boundary),
and returns the address of this area. "sbrk" is called to allocate additional
heap space if needed.
.bl 1
"calloc" allocates space for an array of elements, the size of which are
given in bytes.
.bl 1
"realloc" changes the size of a block. The address of the (possibly moved)
block is returned.
.bl 1
"free" is used to release a block previously allocated by "malloc".
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD size,number;
BYTE *addr,*malloc(),*calloc(),*realloc();
.bl 1
addr = malloc(size);
addr = calloc(number,size);
addr = realloc(addr,size);
free(addr);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
size Is the number of bytes desired
number Is the number of elements desired
"addr" Points to the allocated region
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
address of the allocated region if successful
0 if unsuccessful
.lm 0
.fill
.bl 1
Bugs:
.bl 1
Freeing a bogus address can have disastrous consequences.
.subtitle chmod, chown Functions
.page
.hl 1 chmod, chown Functions
Description:
.bl 1
The "chmod" and "chown" UNIX system calls allow a user under UNIX to change
the protection and owner ID of an existing file. These calls are treated
as NO-OPs by CP/M (if the file exists).
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *name;
WORD mode,owner,group,ret;
.bl 1
ret = chmod(name,mode);
ret = chown(name,owner,group);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
name Is the affected filename (null-terminated)
mode Is the new mode for the file
owner Is the new owner of the file
group Is the new group number
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret 0 (success) if file exists
-1 if file does not exist
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle close Function
.page
.hl 1 close Function
Description:
.bl 1
The close function is used to terminate access to a file or device. This
routine is used on files opened via the "open" or "creat" functions.
A file descriptor (not a stream) is specified for the operation. Stream files
are closed via the "fclose" function.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD fd,ret;
.bl 1
ret = close(fd);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
fd Is the file descriptor to be closed
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
0 For a successful close
-1 For an unknown file descriptor
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle creat, creata, creatb Functions
.page
.hl 1 creat, creata, creatb Functions
Description:
.bl 1
The creat function adds a new file to a disk directory. The file is then
opened for reference by file descriptor (not a stream file).
The creat and creata functions create an ASCII file. "Creatb" creates a
binary file.
.bl 1
.nofill
Calling Sequence:
.bl 1
.lm 8
BYTE *name;
WORD mode,fd;
.bl 1
fd = creat(name,mode);
fd = creata(name,mode);
fd = creatb(name,mode);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
name Is the filename string (null-terminated)
mode Is the UNIX file mode (ignored by CP/M)
.bl 1
.lm 0
Returns:
.lm 16
.fill
.bl 1
.i -8
fd Is the file descriptor for the opened file. A file descriptor is
an "int" quantity used to denote an open file in a read, write, or lseek
call.
.bl 1
.i -8
-1 Is returned if there are any errors.
.lm 0
.fill
.bl 1
Bugs:
.bl 1
UNIX programs which use binary files will compile successfully, but will not
execute properly.
.subtitle ctype Functions
.page
.hl 1 ctype Functions
Description:
.bl 1
The file "<ctype.h>" defines a number of functions used to classify
ASCII characters. These functions indicate whether a character belongs to
a certain character class, returning non-zero for true and zero for false.
.bl 1
The functions are:
.bl 1
.nofill
isalpha(c) c is a letter
isupper(c) c is upper case
islower(c) c is lower case
isdigit(c) c is a digit
isalnum(c) c is alphanumeric
isspace(c) c is a whitespace character
ispunct(c) c is a punctuation character
isprint(c) c is a printable character
iscntrl(c) c is a control character
isascii(c) c is an ascii character (< 0x80)
.bl 1
.fill
Whitespace characters are defined as the space (0x20), tab (0x09), carriage
return (0x0d), line feed (0x0a) and form feed (0x0c) characters. Punctuation
characters are those characters which are not control characters and
not alphanumeric. Printing characters are defined as space (0x20) through
tilde (0x7e). A control character is less than space (0x20).
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
_#include <ctype.h>
.bl 1
WORD ret;
BYTE c; /* or WORD c; */
.bl 1
ret = isalpha(c);
ret = isupper(c);
ret = islower(c);
ret = isdigit(c);
ret = isalnum(c);
ret = isspace(c);
ret = ispunct(c);
ret = isprint(c);
ret = iscntrl(c);
ret = isascii(c);
.lm 0
.bl 1
Arguments:
.bl 1
.lm 8
c Is the character to be classified
.bl 1
.lm 0
.test page 5
Returns:
.lm 8
.bl 1
ret = 0 for False
ret <>0 for True
.lm 0
.fill
.bl 1
Bugs:
.bl 1
These functions are implemented as macros, so arguments with side effects
(e.g., "*p++") will not function correctly in all cases.
.bl 1
Bogus values are returned if arguments are not ASCII characters (i.e.,
> 0x7f).
.subtitle end, etext, edata Locations
.page
.hl 1 end, etext, edata Locations
Description:
.bl 1
The labels "end", "etext", and "edata" are defined by the linkage editor
to be the first location past the ".bss", ".text", and ".data" regions
respectively. The program break location (last used location) is initially
set to "end". This location is altered by many library functions, however.
The "break" may be retrieved by 'sbrk(0)'.
.subtitle exit, __exit Functions
.page
.hl 1 exit, __exit Functions
Description:
.bl 1
The "exit" function is used to pass control back to CP/M. An optional
completion code may be returned. (This is presently ignored by CP/M).
.bl 1
All memory is deallocated, and any open files are closed. Buffer flushing
is also performed for stream output files.
.bl 1
"__exit" is used to return control immediately to CP/M without flushing or
closing open files.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD code;
.bl 1
exit(code);
__exit(code);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
code is the optional return code
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
(none)
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle fclose, fflush Functions
.page
.hl 1 fclose, fflush Functions
Description:
.bl 1
fclose and fflush are used to close and flush a stream file, respectively.
The stream address is used to identify which stream is to be operated on.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
FILE *stream;
.bl 1
ret = fclose(stream);
ret = fflush(stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
stream Is the stream address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
0 If successful
-1 If bad stream address or write failure
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle feof, ferror, clearerr, fileno Functions
.page
.hl 1 feof, ferror, clearerr, fileno Functions
Description:
.bl 1
These functions deal with manipulating file streams in a system-independent
manner.
.bl 1
The "feof" function returns non-zero if a specified stream is at end of
file, and zero otherwise.
.bl 1
"ferror" returns non-zero when an error has occurred on a specified stream.
"clearerr" clears this error condition. This is useful for functions such
as "putw" where no error indication is returned for output failures.
.bl 1
"fileno" returns the file descriptor associated with an open stream.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
FILE *stream;
WORD fd;
.bl 1
ret = feof(stream);
ret = ferror(stream);
clearerr(stream);
fd = fileno(stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
stream Is the stream address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Is a zero or non-zero indicator
fd Is the returned file descriptor
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle fopen, freopen, fdopen Functions
.page
.hl 1 fopen, freopen, fdopen Functions
Description:
.bl 1
The "fopen", "freopen", and "fdopen" functions are used to associate an I/O
stream with a file or device.
.bl 1
"fopen" and "fopena" open an existing ASCII file for I/O as a stream.
"fopenb" opens an existing binary file for I/O as a stream.
.bl 1
"freopen" and "freopa" substitute a new ASCII file in place of an open stream.
"freopb" substitutes a new binary file in place of an open stream.
.bl 1
"fdopen" associates a file which was opened by file descriptor (using
"open" or "creat") with a stream.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
FILE *fopen(),*fopena(),*fopenb();
FILE *freopen(),*freopa(),*freopb();
FILE *fdopen;
FILE *stream;
BYTE *name,*access;
WORD fd;
.bl 1
stream = fopen(name,access);
stream = fopena(name,access);
stream = fopenb(name,access);
stream = freopen(name,access,stream);
stream = freopa(name,access,stream);
stream = freopb(name,access,stream);
stream = fdopen(fd,access);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
name Is the null-terminated filename string
.bl 1
access Is the access string:
"r" For reading the file
"w" For writing the file
"a" For appending to a file
.bl 1
stream Is the stream address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
stream Stream address if open was successful
0 If unsuccessful
.lm 0
.fill
.bl 1
.test page 10
Bugs:
.bl 1
UNIX programs which use "fopen" on binary files will compile and link
correctly, but fail to execute properly.
.subtitle fread, fwrite Functions
.page
.hl 1 fread, fwrite Functions
Description:
.bl 1
"fread" and "fwrite" are used to transfer a stream of bytes between a
stream file and primary memory.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD nitems;
BYTE *buff;
WORD size;
FILE *stream;
.bl 1
nitems = fread(buff,size,nitems,stream);
nitems = fwrite(buff,size,nitems,stream);
.lm 0
.bl 1
Arguments:
.bl 1
.lm 8
buff Is the primary memory buffer address
size Is the number of bytes in each item
nitems Is the number of items to transfer
stream Is an open stream file
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
nitems The number of items actually read / written
0 If an error occurred (including EOF)
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle fseek, ftell, rewind Functions
.page
.hl 1 fseek, ftell, rewind Functions
Description:
.bl 1
The "fseek", "ftell", and "rewind" functions are used in positioning a stream
file.
.bl 1
"fseek" sets the read / write pointer to an arbitrary offset in the stream.
"rewind" sets the read / write pointer to the beginning of the stream.
These calls have no effect on the console device or the listing device.
.bl 1
The "ftell" function returns the present value of the read / write pointer
within the stream. This call returns a meaningless value for non-file
devices.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
FILE *stream;
LONG offset,ftell();
WORD ptrname;
.bl 1
ret = fseek(stream,offset,ptrname);
ret = rewind(stream);
offset = ftell(stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
stream Is the stream address
offset Is a signed offset measured in bytes
ptrname Is the sense of offset:
.bl 1
0 => from beginning of file
1 => from current position
2 => from end of file
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret 0 for success; -1 for failure
offset present offset in stream
.lm 0
.fill
.bl 1
Bugs:
.bl 1
ASCII file seek and tell operations do not account for carriage returns
which will eventually be deleted. Control-Z characters at the end of the
file are handled correctly.
.subtitle getc, getchar, fgetc, getw, getl Functions
.page
.hl 1 getc, getchar, fgetc, getw, getl Functions
Description:
.bl 1
The "getc", "getchar", "fgetc", "getw", and "getl"
functions are used to perform
input from a stream.
.bl 1
"getc" reads a single character from an input stream. This function is
implemented as a macro in <stdio.h>, and arguments should not have
side effects.
.bl 1
"getchar" reads a single character from the standard input. It is identical
to "getc(stdin)" in all respects.
.bl 1
"fgetc" is a function implementation of "getc", used to reduce object code
size.
.bl 1
"getw" reads a word (16 bits)
from the stream, high byte first (compatible with "read"
function call). No special alignment is required.
.bl 1
"getl" reads a long (32 bits) from the stream, in 68000 byte order. No
special alignment is required.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ichar;
FILE *stream;
WORD iword;
LONG ilong,getl();
.bl 1
ichar = getc(stream);
ichar = getchar();
ichar = fgetc(stream);
iword = getw(stream);
ilong = getl(stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
stream Is the stream address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ichar Character read from stream
iword Word read from stream
ilong Long word read from stream
-1 On read failures
.lm 0
.fill
.test page 6
.bl 1
Bugs:
.bl 1
Error return from getchar is incompatible with UNIX version 6 and previous.
.bl 1
Error return from getl or
getw is a valid value which might occur in the
file normally. Use feof or ferror to detect end of file or read error.
.subtitle getpass Function
.page
.hl 1 getpass Function
Description:
.bl 1
The "getpass" function reads a password from the console device. A prompt
is output, and the input read without echoing. A pointer is returned to
a 0-8 character null-terminated string.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *prompt;
BYTE *getpass;
BYTE *pass;
.bl 1
pass = getpass(prompt);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
prompt Is a null-terminated prompt string
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
pass points to the password read
.lm 0
.fill
.bl 1
Bugs:
.bl 1
The return value points to static data whose content is overwritten by
each call.
.subtitle getpid Function
.page
.hl 1 getpid Function
Description:
.bl 1
The getpid function is a bogus routine which returns a phony "process id".
This routine is strictly for UNIX compatibility, and serves no useful purpose
under CP/M. The return value is unpredictable in some implementations.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD pid;
.bl 1
pid = getpid();
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
(none)
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
pid Phony "process id"
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle gets, fgets Functions
.page
.hl 1 gets, fgets Functions
Description:
.bl 1
"gets" and "fgets" are used to read strings from stream files. "fgets" reads
a string including a newline (line feed) character. "gets" deletes the
newline, and reads only from the standard input. Both terminate the strings
with a null character.
.bl 1
A maximum count is specified with fgets, but not with gets. This maximum
count includes the terminating null character.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *addr;
BYTE *s;
BYTE *gets(),*fgets();
WORD n;
FILE *stream;
.bl 1
addr = gets(s);
addr = fgets(s,n,stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
s Is the string buffer area address
n Is the maximum character count
stream Is the input stream
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
addr Is the string buffer address
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle index, rindex Functions
.page
.hl 1 index, rindex Functions
Description:
.bl 1
The index and rindex functions are used to locate a given character in a
string. "index" returns a pointer to the first occurrence of the character,
while "rindex" returns a pointer to the last occurrence.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE c;
BYTE *s;
BYTE *ptr;
BYTE *index(),*rindex();
.bl 1
ptr = index(s,c);
ptr = rindex(s,c);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
s Is a null-terminated string pointer
c Is the character to look for
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ptr The desired character address
0 If character not in the string
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle isatty Function
.page
.hl 1 isatty Function
Description:
.bl 1
The "isatty" function can be used by a CP/M C program to determine if a
file descriptor is attached to the CP/M console device ("CON:").
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD fd;
WORD ret;
.bl 1
ret = isatty(fd);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
fd Is an open file descriptor
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
1 If fd attached to CON:
0 Otherwise
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle lseek, tell Functions
.page
.hl 1 lseek, tell Functions
Description:
.bl 1
"lseek" is used to position a file
referenced by file descriptor to an
arbitrary offset. This function should not be used with stream files, due
to the possibility of stale data in the stream buffer.
The "fseek" function should be used instead.
.bl 1
"tell" is used to determine the file offset of an open file descriptor.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD fd;
WORD ptrname;
LONG offset,lseek(),tell(),ret;
.bl 1
ret = lseek(fd,offset,ptrname);
ret = tell (fd);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
fd Is the open file descriptor
offset Is a signed byte offset in the file
ptrname Is the interpretation of "offset":
.bl 1
0 => from the beginning of the file
1 => from the current file position
2 => from the end of the file
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret resulting absolute file offset
-1 If an error occurs
.lm 0
.fill
.bl 1
Bugs:
.bl 1
Incompatible with versions 1-6 of UNIX
.subtitle mktemp Function
.page
.hl 1 mktemp Function
Description:
.bl 1
The "mktemp" function is used to create a temporary filename. The calling
argument is a character string ending in 6 "X" characters. These characters
are overwritten with the temporary filename.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *string;
BYTE *mktemp();
.bl 1
string = mktemp(string)
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
string Is the address of the template string
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
string Is the original address argument
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle open, opena, openb Functions
.page
.hl 1 open, opena, openb Functions
Description:
.bl 1
The open and opena functions open an existing ASCII
file by file descriptor. The openb function opens an existing binary file.
The file may
be opened for reading, writing, or updating.
.bl 1
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *name;
WORD mode;
WORD fd;
.bl 1
fd = open(name,mode);
fd = opena(name,mode);
fd = openb(name,mode);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
name Is the null-terminated filename string
.bl 1
mode Is the access desired:
0 => read only
1 => write only
2 => read/write (update)
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
fd Is the file descriptor for accessing the file
-1 Indicates open failure
.lm 0
.fill
.bl 1
Bugs:
.bl 1
UNIX programs which use binary files will compile correctly, but will not
execute properly.
.subtitle perror Function
.page
.hl 1 perror Function
Description:
.bl 1
The "perror" function produces a short message on the standard error file
which gives a description of the last system error encountered. An argument
string is printed first, then a colon, followed by the message.
.bl 1
The UNIX notion of an external variable, "errno", which contains the last
error returned from the operating system is simulated.
A list of the possible values of "errno" and of the messages printed by
"perror" is given in Appendix A.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *s;
.bl 1
perror(s);
.bl 1
.lm 0
Arguments:
.bl 1
.i 8
s Is the prefix string to be printed
.bl 1
Returns:
.bl 1
.i 8
(none)
.bl 1
Bugs:
.bl 1
.lm 8
.fill
Many messages are "Error undefined on CP/M".
.lm 0
.subtitle printf, fprintf, sprintf Functions
.page
.hl 1 printf, fprintf, sprintf Functions
Description:
.bl 1
The printf family of functions is used to format data for output. "printf"
outputs to the standard output stream. "fprintf" outputs to an arbitrary
stream file. "sprintf" outputs to a string (memory).
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
BYTE *fmt;
FILE *stream;
BYTE *string;
BYTE *sprintf(),rs;
/* Args can be any type */
.bl 1
ret = printf (fmt,arg1,arg2 ...);
ret = fprintf(stream,fmt,arg1,arg2 ...);
rs = sprintf(string,fmt,arg1,arg2 ...);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
fmt Is a format string with conversion specifiers
argn Are data arguments to be converted
stream Is an output stream file
string Is a buffer address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Number of characters output
-1 if error
rs Buffer string address
Null (zero) if error
.lm 0
.fill
.bl 1
Conversion Operators
.bl 1
The occurrence of a "%" character in the format string indicates the start of
a conversion operator. Values to be converted are obtained from the argument
list in order. Valid Conversion operators are:
.bl 1
.ls
.lec 'd' Convert a binary number to decimal ASCII and insert in output stream.
.lec 'o' Convert a binary number to octal ASCII and insert in output stream.
.lec 'x' Convert a binary number to hexadecimal ASCII and insert in output
stream.
.lec 'c' Use the argument as a single ASCII character.
.lec 's' Use the argument as a pointer to a null-terminated ASCII string, and
insert the string into the output stream.
.lec 'u' Convert a binary number to decimal ASCII (unsigned) and insert in
output stream.
.lec '%' Print a '%' character.
.els
Between the '%' character and the conversion operator, the following optional
information may be present:
.ls
.leb;An optional minus sign which specifies that the converted output be
left-adjusted rather than the default right adjustment.
.leb;An optional digit string specifying a field width. This value gives the
minimum width of the field. If the digit string begins with a '0' character,
then zero padding will be done instead of blank padding.
If an asterisk ("*") is used, the value of the width field will be taken as
the next argument in the argument list.
.leb;A '.' character which separates the field width from the precision
string.
.leb;A digit string which specifies the precision for floating point
conversion. (i.e., the number of digits following the decimal point).
If an asterisk ("*") is used, the value of the precision field will be taken
from the next argument in the argument list.
.leb;The character "l" or "L" which specifies that a long value (32 bits)
is to be converted. A capitalized conversion code will accomplish the
same thing.
.els
Bugs:
.bl 1
.subtitle putc, putchar, fputc, putw, putl Functions
.page
.hl 1 putc, putchar, fputc, putw, putl Functions
Description:
.bl 1
The "putc", "putchar", "fputc", and "putw" functions are used to output
characters and words (16-bits) to stream files.
.bl 1
The "putc" function outputs a single 8-bit character to a stream file. This
function is implemented as a macro in <stdio.h>, so arguments with side
effects should not be used. "fputc" provides the equivalent function as
a real function.
.bl 1
The putchar function outputs a character to the standard output stream file.
This function is also implemented as a macro in <stdio.h>. Side effects
should be avoided with putchar.
.bl 1
The "putw" function outputs a 16-bit word to the specified stream file. The
word is output high byte first, compatible with the write function call.
.bl 1
The "putl" function outputs a 32-bit longword to the stream file. The bytes
are output in 68000 order, as with the "write" function call.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE c;
FILE *stream;
WORD w,ret;
LONG lret,putl(),l;
.bl 1
ret = putc(c,stream);
ret = fputc(c,stream);
ret = putchar(c);
ret = putw(w,stream);
lret = putl(l,stream);
.bl 1
Arguments:
.bl 1
.lm 8
c Is the character to be output
stream Is the output stream address
w Is the word to be output
l Is the long to be output
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Is the word or character output
lret Is the long output with putl
-1 Indicates an output error
.lm 0
.fill
.bl 1
Bugs:
.bl 1
A -1 return from "putw" or "putl" is a perfectly valid integer or long
value. "ferror" should
be used to detect write errors.
.subtitle puts, fputs Functions
.page
.hl 1 puts, fputs Functions
Description:
.bl 1
"puts" and "fputs" are used to output a null-terminated string to an output
stream.
.bl 1
"puts" outputs the string to the standard output, and appends a newline
character.
.bl 1
"fputs" outputs the string to a named output stream. No newline character
is appended.
.bl 1
Neither routine copies the trailing null to the output stream.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
BYTE *s;
FILE *stream;
.bl 1
ret = puts(s);
ret = fputs(s,stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
s Is the string to be output
stream Is the output stream
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Is the last character output
-1 Indicates an error
.lm 0
.fill
.bl 1
Bugs:
.bl 1
The newline incompatibility is not very elegant, but is required to be
compatible with UNIX.
.subtitle qsort Function
.page
.hl 1 qsort Function
Description:
.bl 1
"qsort" is a quicksort routine. The user supplies a vector of elements
and a function to compare two elements. The vector is returned in sorted
order.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
BYTE *base;
WORD number;
WORD size;
WORD compare();
.bl 1
ret = qsort(base,number,size,compare);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
base Is the base address of the element vector
number Is the number of elements to sort
size Is the size of each element in bytes
compare Is the address of the comparison function
.bl 1
This function is called by:
.bl 1
ret = compare(a,b);
.bl 1
The return is:
.bl 1
< 0 if a < b
= 0 if a = b
> 0 if a > b
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
0 always
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle rand, srand Functions
.page
.hl 1 rand, srand Functions
Description:
.bl 1
"rand" and "srand" constitute the C language random number generator. The
generator is initialized by calling "srand" with the seed, and random numbers
are retrieved by calls to "rand". The random numbers are C "int" quantities.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD seed;
WORD rnum;
.bl 1
rnum = srand(seed);
rnum = rand();
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
seed Is an "int" random number seed
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
rnum Desired random number
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle read Function
.page
.hl 1 read Function
Description:
.bl 1
The "read" function is used to read data from a file opened by file descriptor
(using "open" or "creat"). Any arbitrary number of bytes may be read,
starting at the current file pointer.
.bl 1
Under CP/M, the most efficient reads begin and end on 128-byte boundaries.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
WORD fd;
BYTE *buffer;
WORD bytes;
.bl 1
ret = read(fd,buffer,bytes);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
fd Is a file descriptor open for read
buffer Is the buffer address
bytes Is the number of bytes to be read
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Number of bytes actually read
-1 Indicates error
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle scanf, sscanf, fscanf Functions
.page
.hl 1 scanf, sscanf, fscanf Functions
Description:
.bl 1
The scanf family of functions is used to do input format conversions. "scanf"
reads from the standard input, "fscanf" reads from an open stream file, and
"sscanf" reads from a null-terminated string.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *format,*string;
WORD nitems;
FILE *stream;
/* Args can be any type */
.bl 1
nitems = scanf(format,arg1,arg2 ...);
nitems = fscanf(stream,format,arg1,arg2 ...);
nitems = sscanf(string,format,arg1,arg2 ...);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
format Is the control string
argn Are pointers to converted data locations
stream Is an open input stream file
string Is a null-terminated input string
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
nitems The number of items converted
-1 If an I/O error occurred
.lm 0
.fill
.bl 1
Control String Format
.bl 1
The control string consists of:
.ls
.leb;Blanks, tabs, or newlines (line feeds) which match optional white
space in the input.
.leb;An ASCII character (not '%') which must match the next character of the
input stream.
.leb;Conversion specifications, which consist of a leading '%', an optional
'*' (which suppresses assignment), and a conversion character. The next
input field is converted and assigned to the next argument, up to the
next inappropriate character in the input, or until the field width is
exhausted.
.els
Conversion characters indicate the interpretation of the next input field.
Valid conversion characters are:
.ls
.lec '%###'A single '%' is matched in the input at this point; no conversion
is performed.
.lec 'd###'A decimal ASCII integer is converted and stored where the next
argument points.
.lec 'o###'An octal ASCII integer is converted.
.lec 'x###'A hexadecimal ASCII integer is converted.
.lec 's###'A space-terminated character string is input. The argument pointer
is assumed to point to a character array big enough to contain the string
and a trailing null character (which will be added).
.lec 'c###'A single ASCII character will be stored. Spaces are not skipped.
To find the next non-blank character, "%1s" should be used.
.lec '_[##'#A string not terminated by spaces is stored. The left bracket is
followed by a string of characters and a right bracket. If the first
character following the bracket is not "_^", the input will be read up
until the first character not between the brackets. If the first character
following the left bracket is "_^", then the input will be read until the
first character which is between the brackets.
.els
.bl 1
Bugs:
.bl 1
The success of literal matches and suppressed assignments cannot be
determined.
.subtitle setjmp, longjmp Functions
.page
.hl 1 setjmp, longjmp Functions
Description:
.bl 1
Setjmp and longjmp are used to execute a "non-local goto". The setjmp
function is called initially to specify a return location. Longjmp may then
be called from the procedure which invoked setjmp or any procedure which is
called subsequently. Longjmp causes a simulated return from setjmp in the
procedure which originally invoked setjmp. A setjmp
return value is passed from the
longjmp call.
The procedure invoking setjmp must not return before
longjmp is called.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
_#include <setjmp.h>
WORD xret,ret;
jmp__buf env;
.
.
.
xret = setjmp(env);
.
.
.
longjmp(env,ret);
.lm 0
.bl 1
Arguments:
.bl 1
.lm 8
env Contains the saved environment
ret Is the desired return value from setjmp
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
xret 0 when setjmp invoked initially
Copied from "ret" when longjmp called
.lm 0
.fill
.bl 1
Bugs:
.bl 1
Definitely kludgy.
.subtitle signal Function
.page
.hl 1 signal Function
Description:
.bl 1
The signal function connects a C function with a 68000 exception condition.
Each possible exception condition is indicated by a number. The following
exceptions are recognized:
.ls
.lec '4' Illegal Instruction Trap (Includes illegal instructions, privilege
violation, and Line A and Line F traps).
.lec '5' Trace Trap
.lec '6' Trap Instruction other than 2 or 3 (used by BDOS and BIOS).
.lec '8' Arithmetic traps (Zero Divide, CHK instruction, and TRAPV
instruction).
.lec '10' BUSERR (non-existant memory) or addressing (boundary) error trap.
.els
All other values are ignored, to be compatible with UNIX.
.bl 1
Return from the procedure activated by the signal causes
resumption of normal processing. Registers and condition codes are
preserved by the library routines.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret,sig;
WORD func();
.bl 1
ret = signal(sig,func);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
sig Is the signal number given above
func Is the address of a C function
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret 0 if no error, -1 if sig out of range
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle strcat, strncat Functions
.page
.hl 1 strcat, strncat Functions
Description:
.bl 1
Functions "strcat" and "strncat" are used for string concatenation. "strcat"
concatenates two null-terminated strings. "strncat" copies at most a
specified number of characters.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *s1,*s2,*ret;
BYTE *strcat(),*strncat();
WORD n;
.bl 1
ret = strcat(s1,s2);
ret = strncat(s1,s2,n);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
s1 Is the first string.
s2 Is the second string, appended to s1.
n Is the maximum number of characters in s1.
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Is a pointer to s1
.lm 0
.fill
.bl 1
Bugs:
.bl 1
"strcat(s1,s1)" never terminates, and usually destroys the operating system
in the process.
.subtitle strcmp, strncmp Functions
.page
.hl 1 strcmp, strncmp Functions
Description:
.bl 1
The "strcmp" and "strncmp" functions are used for string comparison. "strcmp"
uses null termination, and "strncmp" limits the comparison to a specified
number of characters.
.bl 1
.nofill
Calling Sequence:
.bl 1
.lm 8
BYTE *s1,*s2;
WORD val,n;
.bl 1
val = strcmp(s1,s2);
val = strncmp(s1,s2,n);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
s1 Is a null-terminated string address
s2 Is a null-terminated string address
n Is the maximum number of characters to compare
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
val Is the comparison result:
.bl 1
< 0 => s1 < s2
= 0 => s1 = s2
> 0 => s1 > s2
.lm 0
.fill
.bl 1
Bugs:
.bl 1
Different machines (and compilers)
interpret the characters as signed or unsigned.
.subtitle strcpy, strncpy Functions
.page
.hl 1 strcpy, strncpy Functions
Description:
.bl 1
"strcpy" and "strncpy" are used to copy one null-terminated string to
another. "strcpy" uses null-termination, while "strncpy" imposes a maximum
count on the copied string.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *s1,*s2,*ret;
BYTE *strcpy(),*strncpy();
WORD n;
.bl 1
ret = strcpy(s1,s2);
ret = strncpy(s1,s2,n);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
s1 Is the destination string
s2 Is the source string
n Is the maximum character count
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Is the address of s1
.lm 0
.fill
.bl 1
Bugs:
.bl 1
If the count is exceeded in strncpy, the destination string is not
null-terminated.
.subtitle strlen Function
.page
.hl 1 strlen Function
Description:
.bl 1
"strlen" returns the length of a null-terminated string.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *s;
WORD len;
.bl 1
len = strlen(s);
.lm 0
.bl 1
Arguments:
.bl 1
.lm 8
s Is the string address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
len Is the string length
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle swab Function
.page
.hl 1 swab Function
Description:
.bl 1
The "swab" function is used to copy an area of memory to another area
of memory. The high and low bytes in the destination copy are reversed.
This function may be used to copy binary data from a PDP-11 or VAX to the
68000. The number of bytes to swap must be even.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
BYTE *from,*to;
WORD nbytes;
.bl 1
ret = swab(from,to,nbytes);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
from Is the address of the source buffer
to Is the address of the destination
nbytes Is the number of bytes to copy
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Always 0
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle ttyname Function
.page
.hl 1 ttyname Function
Description:
.bl 1
"ttyname" returns a pointer to the null-terminated filename of the terminal
device associated with an open file descriptor.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE *name,*ttyname();
WORD fd;
.bl 1
name = ttyname(fd);
.lm 0
.bl 1
Arguments:
.bl 1
.lm 8
fd Is an open file descriptor
.bl 1
.lm 0
Returns:
.lm 8
.fill
.bl 1
A pointer to the null-terminated string "CON:" if the file descriptor is
open and attached to the CP/M console device. Zero (NULL) is returned
otherwise.
.lm 0
.bl 1
Bugs:
.bl 1
.subtitle ungetc Function
.page
.hl 1 ungetc Function
Description:
.bl 1
"ungetc" pushes a character back onto an input stream. The next "getc",
"getw", or "getchar" operation will incorporate the character. One character
of buffering is guaranteed, provided something has been read from the stream.
The "fseek" function erases any pushed back
characters.
Attempts to "ungetc" EOF (-1) are rejected.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
BYTE c;
FILE *stream;
WORD ret;
.bl 1
ret = ungetc(c,stream);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
c Is the character to push back
stream Is the stream address
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret c if the character successfully pushed back
-1 for errors
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle unlink Function
.page
.hl 1 unlink Function
Description:
.bl 1
The unlink function deletes a named file from the file system. The removal
operation fails if the file is presently open, or if the file does not
exist.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD ret;
BYTE *name;
.bl 1
ret = unlink(name);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
name Is the null-terminated filename
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
0 If unlink operation successful
-1 If unlink operation failed
.lm 0
.fill
.bl 1
Bugs:
.bl 1
.subtitle write Function
.page
.hl 1 write Function
Description:
.bl 1
The "write" function transfers data to a file open by file descriptor.
Transfer begins at the present file pointer, as set by previous transfers or
by the "lseek" function. Any arbitrary number of bytes may be written to the
file. The number of bytes actually written is returned. It is an error
condition when the number of bytes written does not match the number
requested.
.bl 1
Under CP/M, the most efficient writes begin and end on 128-byte boundaries.
.nofill
.bl 1
Calling Sequence:
.bl 1
.lm 8
WORD fd;
BYTE *buffer;
WORD bytes;
WORD ret;
.bl 1
ret = write(fd,buffer,bytes);
.bl 1
.lm 0
Arguments:
.bl 1
.lm 8
fd Is the open file descriptor
buffer Is the starting buffer address
bytes Is the number of bytes to write
.bl 1
.lm 0
Returns:
.lm 8
.bl 1
ret Is the number of bytes actually written
-1 For errors
.lm 0
.fill
.bl 1
Bugs:
.bl 1
Due to the buffering scheme used, all data is not written to the file until
the file is closed.