ack/modules/src/read_em/read_em.3

326 lines
9.8 KiB
Groff

.TH READ_EM 3ACK "March 17, 1986"
.ad
.SH NAME
EM_open, EM_getinstr, EM_close,
EM_mkcalls\ \-\ a module to read EM assembly code
.SH SYNOPSIS
.B #include <em_spec.h>
.br
.B #include <em_mnem.h>
.br
.B #include <em_pseu.h>
.br
.B #include <em_flag.h>
.br
.B #include <em_ptyp.h>
.br
.B #include <em.h>
.br
.B #include <em_comp.h>
.PP
.B int EM_open(filename)
.br
.B EM_close()
.br
.B char *filename;
.PP
.B int EM_getinstr(instr)
.B struct e_instr *instr;
.PP
.B int EM_mkcalls(instr)
.br
.B struct e_instr *instr;
.PP
.B char *EM_error;
.PP
.B unsigned int EM_lineno;
.PP
.B char *EM_filename;
.PP
.B int EM_wordsize, EM_pointersize;
.SH DESCRIPTION
This package provides routines to read EM assembly code.
The object is to simplify the program
writer's task of reading EM assembly code,
either in compact or human-readable form.
.PP
\fIEM_open\fR must be called as initializer of the package.
If \fIfilename\fR is a null pointer, reading is done from standard input,
otherwise it is done from the file \fIfilename\fR.
\fIEM_open\fR returns 1 on success and 0 on failure
with an error message in \fIEM_error\fR.
\fIEM_close\fR must be called after all other calls to this package.
.PP
\fIEM_getinstr\fR reads an EM instruction, and
returns it in the structure pointed to by \fIinstr\fR.
This structure has the following layout:
.br
.PP
.ta \w'struct\ \ \ 'u +\w'struct e_instr *\ \ \ \ \ \ 'u +\w'em_opcode\ \ \ 'u +\w'*emu_string\ \ \ 'u
.nf
/*
* (c) copyright 1987 by the Vrije Universiteit, Amsterdam, The Netherlands.
* See the copyright notice in the ACK home directory, in the file "Copyright".
*/
/* $Header$ */
struct e_arg {
int ema_argtype; /* type of this argument */
union e_simple_arg {
arith emu_cst; /* a cst */
label emu_dlb; /* a numeric data label */
label emu_ilb; /* an instruction label */
char *emu_dnam; /* a data label */
char *emu_pnam; /* a procedure name */
char *emu_string; /* a string (fcon,icon,ucon,scon) */
} ema_arg;
arith ema_szoroff;
};
#define ema_cst ema_arg.emu_cst
#define ema_dlb ema_arg.emu_dlb
#define ema_ilb ema_arg.emu_ilb
#define ema_dnam ema_arg.emu_dnam
#define ema_pnam ema_arg.emu_pnam
#define ema_nlocals ema_szoroff
#define ema_string ema_arg.emu_string
struct e_instr {
int em_type; /* Type of this instr */
#define EM_MNEM 256 /* A machine instruction */
#define EM_PSEU 257 /* A pseudo */
#define EM_STARTMES 258 /* Start of a MES pseudo */
#define EM_MESARG 259 /* A member in a MES list */
#define EM_ENDMES 260 /* End of a MES pseudo */
#define EM_DEFILB 261 /* An instruction label definition */
#define EM_DEFDLB 262 /* A numeric data label definition */
#define EM_DEFDNAM 263 /* A non-numeric data label def */
#define EM_ERROR 264 /* Recoverable error */
#define EM_FATAL 265 /* Unrecoverable error */
#define EM_EOF 266 /* End of file */
int em_opcode;
struct e_arg em_arg;
};
extern arith
EM_holsize, EM_bsssize;
extern int
EM_holinit, EM_bssinit;
#define em_ilb em_arg.ema_ilb
#define em_dlb em_arg.ema_dlb
#define em_dnam em_arg.ema_dnam
#define em_argtype em_arg.ema_argtype
#define em_cst em_arg.ema_cst
#define em_pnam em_arg.ema_pnam
#define em_nlocals em_arg.ema_nlocals
#define em_string em_arg.ema_string
#define em_off em_arg.ema_szoroff
#define em_size em_arg.ema_szoroff
/* arguments for EXC pseudo: */
#define em_exc1 em_arg.ema_cst
#define em_exc2 em_arg.ema_szoroff
extern char
*EM_error, *EM_filename;
extern unsigned int
EM_lineno;
extern int
EM_wordsize, EM_pointersize;
.fi
.PP
The named types \fBarith\fR and \fBlabel\fR refer to types on the local machine
that are suitable for doing arithmetic and storing EM numeric labels
respectively.
Common definitions are \fBlong\fR for \fBarith\fR and \fBunsigned int\fR for
\fBlabel\fR.
.PP
The \fIe_instr\fR structure consists of the fields
\fIem_type\fR, containing the type of this \fIe_instr\fR,
\fIem_opcode\fR, containing the opcode of an instruction,
and \fIem_arg\fR, containing a possible argument.
.PP
The possible values of
\fIem_type\fR, defined in <em_comp.h>, are summarized below:
.br
.ta \w'EM_STARTMES\ \ \ 'u +\w'em_defdnam\ \ \ 'u
.di xx
\ka
.br
.di
.IP "Value Selector" \nau
Meaning
.IP "EM_MNEM em_opcode" \nau
an EM machine instruction.
.br
.PD 0
.IP " em_arg" \nau
The \fIem_opcode\fR field
contains the opcode of the instruction, and \fIem_arg\fR may contain an
argument. If \fIem_argtype\fR indicates that there is no argument,
\fIem_cst\fR is set to 0.
.IP "EM_PSEU em_opcode" \nau
an EM pseudo instruction.
.IP " em_arg" \nau
The \fIem_opcode\fR field
contains the opcode, and \fIem_arg\fR may contain an argument.
As consecutive CON-pseudos are allocated consecutively, a CON delivered by
\fIEM_getinstr\fR has exactly one argument.
If the CON-pseudo read has more, they are delivered as separate CON's.
The same holds for ROM-pseudos.
Also, if the length of a string constant exceeds 256 characters, it will be
delivered as several CON's or ROM's.
There are two "special" pseudo's, that use other variables, HOL and BSS.
They use EM_holsize, EM_holinit, EM_bsssize, and EM_bssinit, because those
arguments do not fit in the \fIe_arg\fR structure, and we want to keep the
\fIe_arg\fR structure as small as possible.
The EXC pseudo has its arguments encoded as indicated in the #defines.
The PRO pseudo has its second argument in \fIem_nlocals\fR. If it is
-1, it was omitted.
.IP "EM_STARTMES em_arg" \nau
the start of a MES pseudo.
.br
There is one argument: the message number.
The other arguments, if any, are delivered as separate EM_MESARG's.
.IP "EM_MESARG em_arg" \nau
an argument of a MES pseudo.
.IP "EM_ENDMES none" \nau
the end of a MES pseudo.
.IP "EM_DEFILB em_ilb" \nau
an instruction label definition.
.br
The field \fIem_ilb\fR contains the label (instruction labels are always
numeric).
.IP "EM_DEFDLB em_dlb" \nau
a numeric data label definition.
.br
The field \fIem_dlb\fR contains the label.
.IP "EM_DEFDNAM em_dnam" \nau
a non-numeric data label definition.
.br
The field \fIem_dnam\fR contains the label.
.IP "EM_ERROR none" \nau
an error in the input that makes the rest of the data in the structure
meaningless.
.br
\fIEM_error\fR
contains an error message.
.IP "EM_FATAL none" \nau
a fatal error.
.br
\fIEM_error\fR contains an
error message.
.IP "EM_EOF none" \nau
end of file
.PD
.PP
The \fIe_arg\fR structure consists of the fields
the field \fIema_argtype\fR, containing the type of this argument or 0
if absent,
the field \fIema_arg\fR, containing the value of the argument,
and \fIema_szoroff\fR, containing an optional offset or size.
The possible values of \fIema_argtype\fR, defined in <em_ptyp.h>,
are summarized below:
.br
.ta \w'dlb_ptyp\ \ \ \ 'u +\w'em_opcode\ \ \ 'u
.di xx
\ka
.br
.di
.IP "Value Selector" \nau
Meaning
.IP "0 none" \nau
no argument.
.IP "ilb_ptyp em_ilb" \nau
an instruction label.
.PD 0
.IP "nof_ptyp em_dlb" \nau
an offset from a numeric data label.
.IP " em_off" \nau
The
\fIem_off\fR field contains the offset and the
\fIem_dlb\fR field contains the label.
.IP "sof_ptyp em_dnam" \nau
an offset from a non-numeric data label.
.IP " em_off" \nau
The \fIem_off\fR field contains the offset and the \fIem_dnam\fR field
contains the label, represented as a string.
.IP "cst_ptyp em_cst" \nau
a numeric constant.
.IP "pro_ptyp em_pnam" \nau
a procedure name, not including the '$',
represented as a string.
.IP "str_ptyp em_string" \nau
a string constant.
.IP " em_size" \nau
The string is found in \fIem_string\fR, represented as a row of bytes, of
length \fIem_size\fR.
.IP "ico_ptyp em_string" \nau
an integer constant.
.IP " em_size" \nau
A string representation of the constant is found in \fIem_string\fR.
It has size \fIem_size\fR bytes on the target machine.
.IP "uco_ptyp em_string" \nau
an unsigned constant.
.IP " em_size" \nau
A string representation of the constant is found in \fIem_string\fR.
It has size \fIem_size\fR bytes on the target machine.
.IP "fco_ptyp em_string" \nau
a floating constant.
.IP " em_size" \nau
A string representation of the constant is found in \fIem_string\fR.
It has size \fIem_size\fR bytes on the target machine.
.PD
.PP
When an error occurs, \fIEM_error\fR is set to indicate the reason.
\fIEM_getinstr\fR returns 1 if all goes well, 0 if it does not.
The EM_ERROR described above is only set when the error
is serious enough.
.PP
The routine \fIEM_mkcalls\fR "translates" the EM instruction indicated
by \fIinstr\fR
into calls of the procedural interface defined in \fIem_code\fR(3L).
It returns 1 if it succeeds, 0 if it fails for some reason. The
reason can then be found in \fIEM_error\fR.
.PP
\fIEM_lineno\fR contains the line number of the last line read by
\fIEM_getinstr\fR.
.PP
\fIEM_filename\fR contains a filename. It usually contains the value
given as parameter to \fIEM_open\fR, but may have a different value, when
the input was the result of some preprocessing.
.PP
.I EM_wordsize
and
.I EM_pointersize
contain the wordsize and pointersize, but only after the first
(pseudo-)instruction has successfully been read.
.SH FILES
.nf
~em/modules/h/em.h
~em/h/em_ptyp.h
~em/modules/h/em_comp.h
~em/modules/lib/libread_emk.a: non-checking library for reading compact EM code
~em/modules/lib/libread_emkV.a: checking library for reading compact EM code
~em/modules/lib/libread_emeV.a: checking library for reading human-readable EM code
.fi
.SH MODULES
em_code(3), string(3), system(3), ~em/lib/em_data.a
.SH "SEE ALSO"
em_code(3)
.br
A.S. Tanenbaum, H. v Staveren, E.G. Keizer, J.W. Stevenson, "\fBDescription
of a Machine Architecture for use with Block Structured Languages\fR",
Informatica Rapport IR-81, Vrije Universiteit, Amsterdam, 1983.
.SH REMARKS
All strings must be considered to be contained in a static area, so
must be copied to be saved.
.SH BUGS
As CON's and ROM's may be delivered in several parts, the count fields in
a static exchange may be wrong.
.PP
Please report bugs to the author.
.SH AUTHOR
Ceriel J.H. Jacobs <ceriel@cs.vu.nl>