This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Reapply some perldoc changes not in 3.09.
[perl5.git] / ext / B / t / assembler.t
CommitLineData
f4abc3e7
JH
1#!./perl -w
2
3=pod
4
5=head1 TEST FOR B::Assembler.pm AND B::Disassembler.pm
6
7=head2 Description
8
9The general idea is to test by assembling a choice set of assembler
10instructions, then disassemble them, and check that we've completed the
11round trip. Also, error checking of Assembler.pm is tested by feeding
12it assorted errors.
13
14Since Assembler.pm likes to assemble a file, we comply by writing a
15text file. This file contains three sections:
16
17 testing operand categories
18 use each opcode
19 erronous assembler instructions
20
21An "operand category" is identified by the suffix of the PUT_/GET_
22subroutines as shown in the C<%Asmdata::insn_data> initialization, e.g.
23opcode C<ldsv> has operand category C<svindex>:
24
25 insn_data{ldsv} = [1, \&PUT_svindex, "GET_svindex"];
26
27Because Disassembler.pm also assumes input from a file, we write the
28resulting object code to a file. And disassembled output is written to
29yet another text file which is then compared to the original input.
30(Erronous assembler instructions still generate code, but this is not
31written to the object file; therefore disassembly bails out at the first
32instruction in error.)
33
34All files are kept in memory by using TIEHASH.
35
36
37=head2 Caveats
38
39An error where Assembler.pm and Disassembler.pm agree but Assembler.pm
40generates invalid object code will not be detected.
41
42Due to the way this test has been set up, failure of a single test
43could cause all subsequent tests to fail as well: After an unexpected
44assembler error no output is written, and disassembled lines will be
45out of sync for all lines thereafter.
46
47Not all possibilities for writing a valid operand value can be tested
48because disassembly results in a uniform representation.
49
50
51=head2 Maintenance
52
53New opcodes are added automatically.
54
55A new operand category will cause this program to die ("no operand list
56for XXX"). The cure is to add suitable entries to C<%goodlist> and
57C<%badlist>. (Since the data in Asmdata.pm is autogenerated, it may also
58happen that the corresponding assembly or disassembly subroutine is
59missing.) Note that an empty array as a C<%goodlist> entry means that
60opcodes of the operand category do not take an operand (and therefore the
61corresponding entry in C<%badlist> should have one). An C<undef> entry
62in C<%badlist> means that any value is acceptable (and thus there is no
63way to cause an error).
64
65Set C<$dbg> to debug this test.
66
67=cut
68
69package VirtFile;
70use strict;
71
72# Note: This is NOT a general purpose package. It implements
73# sequential text and binary file i/o in a rather simple form.
74
75sub TIEHANDLE($;$){
76 my( $class, $data ) = @_;
77 my $obj = { data => defined( $data ) ? $data : '',
78 pos => 0 };
79 return bless( $obj, $class );
80}
81
82sub PRINT($@){
83 my( $self ) = shift;
84 $self->{data} .= join( '', @_ );
85}
86
87sub WRITE($$;$$){
88 my( $self, $buf, $len, $offset ) = @_;
89 unless( defined( $len ) ){
90 $len = length( $buf );
91 $offset = 0;
92 }
93 unless( defined( $offset ) ){
94 $offset = 0;
95 }
96 $self->{data} .= substr( $buf, $offset, $len );
97 return $len;
98}
99
100
101sub GETC($){
102 my( $self ) = @_;
103 return undef() if $self->{pos} >= length( $self->{data} );
104 return substr( $self->{data}, $self->{pos}++, 1 );
105}
106
107sub READLINE($){
108 my( $self ) = @_;
109 return undef() if $self->{pos} >= length( $self->{data} );
110 my $lfpos = index( $self->{data}, "\n", $self->{pos} );
111 if( $lfpos < 0 ){
112 $lfpos = length( $self->{data} );
113 }
114 my $pos = $self->{pos};
115 $self->{pos} = $lfpos + 1;
116 return substr( $self->{data}, $pos, $self->{pos} - $pos );
117}
118
119sub READ($@){
120 my $self = shift();
121 my $bufref = \$_[0];
122 my( undef, $len, $offset ) = @_;
123 if( $offset ){
124 die( "offset beyond end of buffer\n" )
125 if ! defined( $$bufref ) || $offset > length( $$bufref );
126 } else {
127 $$bufref = '';
128 $offset = 0;
129 }
130 my $remlen = length( $self->{data} ) - $self->{pos};
131 $len = $remlen if $remlen < $len;
132 return 0 unless $len;
133 substr( $$bufref, $offset, $len ) =
134 substr( $self->{data}, $self->{pos}, $len );
135 $self->{pos} += $len;
136 return $len;
137}
138
139sub TELL($){
140 my $self = shift();
141 return $self->{pos};
142}
143
144sub CLOSE($){
145 my( $self ) = @_;
146 $self->{pos} = 0;
147}
148
1491;
150
151package main;
152
153use strict;
154use Test::More;
155use Config qw(%Config);
156
ff924b36
CB
157BEGIN {
158 if (($Config{'extensions'} !~ /\bByteLoader\b/) ){
159 print "1..0 # Skip -- Perl configured without ByteLoader module\n";
160 exit 0;
161 }
dc9e9a31 162 print "1..0\n"; exit 0;
ff924b36
CB
163}
164
f4abc3e7
JH
165use B::Asmdata qw( %insn_data );
166use B::Assembler qw( &assemble_fh );
167use B::Disassembler qw( &disassemble_fh &get_header );
168
169my( %opsByType, @code2name );
170my( $lineno, $dbg, $firstbadline, @descr );
171$dbg = 0; # debug switch
172
173# $SIG{__WARN__} handler to catch Assembler error messages
174#
175my $warnmsg;
176sub catchwarn($){
177 $warnmsg = $_[0];
178 print "error: $warnmsg\n" if $dbg;
179}
180
181# Callback for writing assembled bytes. This is where we check
182# that we do get an error.
183#
184sub putobj($){
185 if( ++$lineno >= $firstbadline ){
186 ok( $warnmsg && $warnmsg =~ /^\d+:\s/, $descr[$lineno] );
187 undef( $warnmsg );
188 } else {
189 my $l = syswrite( OBJ, $_[0] );
190 }
191}
192
193# Callback for writing a disassembled statement.
194#
195sub putdis(@){
196 my $line = join( ' ', @_ );
197 ++$lineno;
198 print DIS "$line\n";
199 printf "%5d %s\n", $lineno, $line if $dbg;
200}
201
202# Generate assembler instructions from a hash of operand types: each
203# existing entry contains a list of good or bad operand values. The
204# corresponding opcodes can be found in %opsByType.
205#
206sub gen_type($$$){
207 my( $href, $descref, $text ) = @_;
e53790c1 208 for my $odt ( sort( keys( %opsByType ) ) ){
f4abc3e7
JH
209 my $opcode = $opsByType{$odt}->[0];
210 my $sel = $odt;
211 $sel =~ s/^GET_//;
212 die( "no operand list for $sel\n" ) unless exists( $href->{$sel} );
213 if( defined( $href->{$sel} ) ){
214 if( @{$href->{$sel}} ){
215 for my $od ( @{$href->{$sel}} ){
216 ++$lineno;
217 $descref->[$lineno] = "$text: $code2name[$opcode] $od";
218 print ASM "$code2name[$opcode] $od\n";
219 printf "%5d %s %s\n", $lineno, $code2name[$opcode], $od if $dbg;
220 }
221 } else {
222 ++$lineno;
223 $descref->[$lineno] = "$text: $code2name[$opcode]";
224 print ASM "$code2name[$opcode]\n";
225 printf "%5d %s\n", $lineno, $code2name[$opcode] if $dbg;
226 }
227 }
228 }
229}
230
231# Interesting operand values
232#
233my %goodlist = (
234comment_t => [ '"a comment"' ], # no \n
235none => [],
236svindex => [ 0x7fffffff, 0 ],
237opindex => [ 0x7fffffff, 0 ],
238pvindex => [ 0x7fffffff, 0 ],
239U32 => [ 0xffffffff, 0 ],
240U8 => [ 0xff, 0 ],
241PV => [ '""', '"a string"', ],
242I32 => [ -0x80000000, 0x7fffffff ],
243IV64 => [ '0x000000000', '0x0ffffffff', '0x000000001' ], # disass formats 0x%09x
244IV => $Config{ivsize} == 4 ?
245 [ -0x80000000, 0x7fffffff ] :
246 [ '0x000000000', '0x0ffffffff', '0x000000001' ],
247NV => [ 1.23456789E3 ],
248U16 => [ 0xffff, 0 ],
249pvcontents => [],
250strconst => [ '""', '"another string"' ], # no NUL
251op_tr_array => [ join( ',', 0..255 ) ],
252 );
253
254# Erronous operand values
255#
256my %badlist = (
257comment_t => [ '"multi-line\ncomment"' ], # no \n
258none => [ '"spurious arg"' ],
259svindex => [ 0xffffffff * 2, -1 ],
260opindex => [ 0xffffffff * 2, -2 ],
261pvindex => [ 0xffffffff * 2, -3 ],
262U32 => [ 0xffffffff * 2, -4 ],
263U16 => [ 0x5ffff, -5 ],
264U8 => [ 0x6ff, -6 ],
265PV => [ 'no quote"' ],
266I32 => [ -0x80000001, 0x80000000 ],
267IV64 => undef, # PUT_IV64 doesn't check - no integrity there
268IV => $Config{ivsize} == 4 ?
269 [ -0x80000001, 0x80000000 ] : undef,
270NV => undef, # PUT_NV accepts anything - it shouldn't, real-ly
271pvcontents => [ '"spurious arg"' ],
272strconst => [ 'no quote"', '"with NUL '."\0".' char"' ], # no NUL
273op_tr_array => [ join( ',', 1..42 ) ],
274 );
275
276
277# Determine all operand types from %Asmdata::insn_data
278#
279for my $opname ( keys( %insn_data ) ){
280 my ( $opcode, $put, $getname ) = @{$insn_data{$opname}};
281 push( @{$opsByType{$getname}}, $opcode );
282 $code2name[$opcode] = $opname;
283}
284
285
286# Write instruction(s) for correct operand values each operand type class
287#
288$lineno = 0;
289tie( *ASM, 'VirtFile' );
290gen_type( \%goodlist, \@descr, 'round trip' );
291
292# Write one instruction for each opcode.
293#
294for my $opcode ( 0..$#code2name ){
295 next unless defined( $code2name[$opcode] );
296 my $sel = $insn_data{$code2name[$opcode]}->[2];
297 $sel =~ s/^GET_//;
298 die( "no operand list for $sel\n" ) unless exists( $goodlist{$sel} );
299 if( defined( $goodlist{$sel} ) ){
300 ++$lineno;
301 if( @{$goodlist{$sel}} ){
302 my $od = $goodlist{$sel}[0];
303 $descr[$lineno] = "round trip: $code2name[$opcode] $od";
304 print ASM "$code2name[$opcode] $od\n";
305 printf "%5d %s %s\n", $lineno, $code2name[$opcode], $od if $dbg;
306 } else {
307 $descr[$lineno] = "round trip: $code2name[$opcode]";
308 print ASM "$code2name[$opcode]\n";
309 printf "%5d %s\n", $lineno, $code2name[$opcode] if $dbg;
310 }
311 }
312}
313
314# Write instruction(s) for incorrect operand values each operand type class
315#
316$firstbadline = $lineno + 1;
317gen_type( \%badlist, \@descr, 'asm error' );
318
319# invalid opcode is an odd-man-out ;-)
320#
321++$lineno;
322$descr[$lineno] = "asm error: Gollum";
323print ASM "Gollum\n";
324printf "%5d %s\n", $lineno, 'Gollum' if $dbg;
325
326close( ASM );
327
328# Now that we have defined all of our tests: plan
329#
330plan( tests => $lineno );
331print "firstbadline=$firstbadline\n" if $dbg;
332
333# assemble (guard against warnings and death from assembly errors)
334#
335$SIG{'__WARN__'} = \&catchwarn;
336
337$lineno = -1; # account for the assembly header
338tie( *OBJ, 'VirtFile' );
339eval { assemble_fh( \*ASM, \&putobj ); };
340print "eval: $@" if $dbg;
341close( ASM );
342close( OBJ );
343$SIG{'__WARN__'} = 'DEFAULT';
344
345# disassemble
346#
347print "--- disassembling ---\n" if $dbg;
348$lineno = 0;
349tie( *DIS, 'VirtFile' );
350disassemble_fh( \*OBJ, \&putdis );
351close( OBJ );
352close( DIS );
353
354# get header (for debugging only)
355#
356if( $dbg ){
357 my( $magic, $archname, $blversion, $ivsize, $ptrsize, $byteorder ) =
358 get_header();
359 printf "Magic: 0x%08x\n", $magic;
360 print "Architecture: $archname\n";
361 print "Byteloader V: $blversion\n";
362 print "ivsize: $ivsize\n";
363 print "ptrsize: $ptrsize\n";
364 print "Byteorder: $byteorder\n";
365}
366
367# check by comparing files line by line
368#
369print "--- checking ---\n" if $dbg;
370$lineno = 0;
371my( $asmline, $disline );
372while( defined( $asmline = <ASM> ) ){
373 $disline = <DIS>;
374 ++$lineno;
375 last if $lineno eq $firstbadline; # bail out where errors begin
376 ok( $asmline eq $disline, $descr[$lineno] );
377 printf "%5d %s\n", $lineno, $asmline if $dbg;
378}
379close( ASM );
380close( DIS );
381
382__END__