lib/std/arch/rv64/asm/parser.rad 32.9 KiB raw
1
//! Assembler parser pass.
2
use std::mem;
3
use std::fmt;
4
use std::lang::alloc;
5
use std::lang::strings;
6
use std::lang::parser;
7
use std::lang::gen;
8
use std::collections::dict;
9
use std::arch::rv64::encode;
10
use std::arch::rv64;
11
use std::arch::rv64::atomics;
12
13
use super::emit;
14
use super::scanner;
15
16
/// Parsed memory operand with base register and signed byte offset.
17
record MemOperand: Copy {
18
    /// Base register inside the memory operand parentheses.
19
    base: gen::Reg,
20
    /// Signed byte offset preceding the base register.
21
    offset: i32,
22
}
23
24
/// Parse assembler source into the supplied assembler state.
25
export unsafe fn parseProgram 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
26
    advance(a);
27
28
    while a.scan.current.kind <> scanner::TokenKind::Eof {
29
        try parseItem(a);
30
    }
31
}
32
33
/// Align `value` upward to `alignment`, returning nil on u32 overflow.
34
fn checkedAlignUp(value: u32, alignment: u32) -> ?u32 {
35
    let padding = alignment - 1;
36
    if value > parser::U32_MAX - padding {
37
        return nil;
38
    }
39
    return mem::alignUp(value, alignment);
40
}
41
42
/// Advance the parser by one token, preserving the previous token.
43
fn advance 'parse (a: &mut super::Assembler 'parse) {
44
    set a.scan.previous = a.scan.current;
45
    set a.scan.current = scanner::next(&mut a.scan, a.pool);
46
}
47
48
/// Consume the current token when it has `kind`.
49
fn consume 'parse (a: &mut super::Assembler 'parse, kind: scanner::TokenKind) -> bool {
50
    if a.scan.current.kind == kind {
51
        advance(a);
52
        return true;
53
    }
54
    return false;
55
}
56
57
/// Create an error at the current token.
58
fn fail 'parse (a: &super::Assembler 'parse, message: *[u8]) -> super::Error {
59
    return super::Error::Invalid { offset: a.scan.current.offset, message };
60
}
61
62
/// Create an error at `tok`.
63
fn failOnToken(tok: scanner::Token, message: *[u8]) -> super::Error {
64
    return super::Error::Invalid { offset: tok.offset, message };
65
}
66
67
/// Require that a data directive appears while assembling the data section.
68
fn expectDataSection 'parse (a: &super::Assembler 'parse, tok: scanner::Token) throws (super::Error) {
69
    if a.section <> super::Section::Data {
70
        throw failOnToken(tok, "data directive is only valid in the data section");
71
    }
72
}
73
74
/// Consume `kind` or throw `message` at the current token.
75
fn expect 'parse (a: &mut super::Assembler 'parse, kind: scanner::TokenKind, message: *[u8]) throws (super::Error) {
76
    if not consume(a, kind) {
77
        throw fail(a, message);
78
    }
79
}
80
81
/// Consume `kind` and return the consumed token.
82
fn expectToken 'parse (a: &mut super::Assembler 'parse, kind: scanner::TokenKind, message: *[u8]) -> scanner::Token throws (super::Error) {
83
    try expect(a, kind, message);
84
    return a.scan.previous;
85
}
86
87
/// Require that the current item has reached its semicolon terminator.
88
fn expectTerminator 'parse (a: &super::Assembler 'parse, message: *[u8]) throws (super::Error) {
89
    if a.scan.current.kind <> scanner::TokenKind::Semicolon {
90
        throw fail(a, message);
91
    }
92
}
93
94
/// Require that `value` fits in i32.
95
fn expectI32Value 'parse (a: &super::Assembler 'parse, value: i64, message: *[u8]) -> i32 throws (super::Error) {
96
    if value < -super::I32_MIN_MAGNITUDE or value > super::I32_MAX_VALUE {
97
        throw fail(a, message);
98
    }
99
    return value as i32;
100
}
101
102
/// Require that `value` fits in a signed 12-bit immediate field.
103
fn expectSmallImmValue 'parse (a: &super::Assembler 'parse, value: i64) -> i32 throws (super::Error) {
104
    if not encode::isSmallImm64(value) {
105
        throw fail(a, "immediate out of range");
106
    }
107
    return value as i32;
108
}
109
110
/// Define a label at the current text or data offset.
111
fn defineSymbol 'parse (a: &mut super::Assembler 'parse, name: *[u8], tok: scanner::Token) throws (super::Error) {
112
    if dict::get(&a.symbolMap, name) <> nil {
113
        throw failOnToken(tok, "duplicate label");
114
    }
115
    emit::defineSymbol(a, name);
116
}
117
118
/// Emit a parsed integer data value after applying source-level range checks.
119
fn emitDataValue 'parse (a: &mut super::Assembler 'parse, value: i64, width: super::DataWidth) throws (super::Error) {
120
    match width {
121
        case super::DataWidth::Word =>
122
            try emit::emitDataValue(a, (try expectI32Value(a, value, "word literal out of range")) as i64, width),
123
        case super::DataWidth::Dword =>
124
            try emit::emitDataValue(a, value, width),
125
    }
126
}
127
128
/// Parse a possibly scoped name from one or more `::`-separated segments.
129
fn parseScopedName 'parse (
130
    a: &mut super::Assembler 'parse,
131
    kind: scanner::TokenKind,
132
    message: *[u8],
133
    trimPrefix: u32
134
) -> *[u8] throws (super::Error) {
135
    let first = try expectToken(a, kind, message);
136
    let start = first.offset + trimPrefix;
137
    let mut end = first.offset + first.source.len;
138
139
    while consume(a, scanner::TokenKind::ColonColon) {
140
        let segment = try expectToken(a, scanner::TokenKind::Ident, "expected identifier after `::`");
141
        set end = segment.offset + segment.source.len;
142
    }
143
    let source = &a.scan.source[start..end];
144
    return strings::intern(a.pool, source);
145
}
146
147
/// Parse a bare symbol name.
148
fn parseSymbolName 'parse (a: &mut super::Assembler 'parse) -> *[u8] throws (super::Error) {
149
    return try parseScopedName(a, scanner::TokenKind::Ident, "expected symbol name", 0);
150
}
151
152
/// Return `true` when [`tok`] is any label token form.
153
fn isLabel(tok: scanner::TokenKind) -> bool {
154
    return tok == scanner::TokenKind::Label or tok == scanner::TokenKind::QuotedLabel;
155
}
156
157
/// Parse the contents of a quoted label token, decoding escapes as needed.
158
unsafe fn parseQuotedLabelName 'parse (a: &mut super::Assembler 'parse) -> *[u8] throws (super::Error) {
159
    let tok = try expectToken(a, scanner::TokenKind::QuotedLabel, "expected label name");
160
    let rawStart = super::LABEL_SIGIL_LEN + super::QUOTE_DELIM_LEN;
161
    let raw = &tok.source[rawStart..tok.source.len - super::QUOTE_DELIM_LEN];
162
    let storage = try alloc::allocSlice(a.arena, 1, 1, raw.len) catch {
163
        panic "asm: out of memory allocating quoted label";
164
    } as *mut [u8];
165
    let len = fmt::unescapeString(raw, storage);
166
167
    return strings::intern(a.pool, &storage[..len]);
168
}
169
170
/// Parse a label reference or definition name.
171
unsafe fn parseLabelName 'parse (a: &mut super::Assembler 'parse) -> *[u8] throws (super::Error) {
172
    if a.scan.current.kind == scanner::TokenKind::QuotedLabel {
173
        return try parseQuotedLabelName(a);
174
    }
175
    return try parseScopedName(a, scanner::TokenKind::Label, "expected label name", super::LABEL_SIGIL_LEN);
176
}
177
178
/// Parse a directive name without its leading `.`.
179
fn parseDirectiveName 'parse (a: &mut super::Assembler 'parse) -> *[u8] throws (super::Error) {
180
    let name = try expectToken(a, scanner::TokenKind::Directive, "expected directive name");
181
    return &name.source[super::DIRECTIVE_SIGIL_LEN..];
182
}
183
184
/// Parse one top-level assembler item.
185
unsafe fn parseItem 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
186
    match a.scan.current.kind {
187
        case scanner::TokenKind::Ident => {
188
            let tok = a.scan.current;
189
            let name = try parseSymbolName(a);
190
            try parseInstruction(a, name, tok);
191
            try expect(a, scanner::TokenKind::Semicolon, "expected `;` after instruction");
192
        }
193
        case scanner::TokenKind::Number => {
194
            let tok = a.scan.current;
195
            advance(a);
196
            throw failOnToken(tok, "unexpected number at top level");
197
        }
198
        case scanner::TokenKind::Label, scanner::TokenKind::QuotedLabel => {
199
            let tok = a.scan.current;
200
            let name = try parseLabelName(a);
201
            try defineSymbol(a, name, tok);
202
        }
203
        case scanner::TokenKind::Directive => {
204
            let tok = a.scan.current;
205
            let name = try parseDirectiveName(a);
206
            try parseDirective(a, name, tok);
207
            try expect(a, scanner::TokenKind::Semicolon, "expected `;` after directive");
208
        }
209
        else => throw fail(a, "expected label, instruction, or directive"),
210
    }
211
}
212
213
/// Find `name` in a sorted descriptor table.
214
fn findSortedNameIndex(name: *[u8], len: u32, getName: fn(u32) -> *[u8]) -> ?u32 {
215
    let mut left: u32 = 0;
216
    let mut right: u32 = len;
217
218
    while left < right {
219
        let mid = left + ((right - left) / 2);
220
        let cmp = mem::cmp(name, getName(mid));
221
222
        match cmp {
223
            case -1 => set right = mid,
224
            case  1 => set left = mid + 1,
225
            else => return mid,
226
        }
227
    }
228
    return nil;
229
}
230
231
/// Adapter used by [`findSortedNameIndex`] to read an instruction mnemonic.
232
fn instructionNameAt(index: u32) -> *[u8] {
233
    return super::INSTRUCTIONS[index].name;
234
}
235
236
/// Adapter used by [`findSortedNameIndex`] to read a directive name.
237
fn directiveNameAt(index: u32) -> *[u8] {
238
    return super::DIRECTIVES[index].name;
239
}
240
241
/// Adapter used by [`findSortedNameIndex`] to read a register name.
242
fn registerNameAt(index: u32) -> *[u8] {
243
    return super::REGISTERS[index].name;
244
}
245
246
/// Adapter used by [`findSortedNameIndex`] to read a CSR name.
247
fn csrNameAt(index: u32) -> *[u8] {
248
    return super::CSRS[index].name;
249
}
250
251
/// Look up the operand parser and encoder for an instruction mnemonic.
252
fn lookupInstruction(name: *[u8]) -> ?super::InstructionEncoder {
253
    let index = findSortedNameIndex(name, super::INSTRUCTIONS.len, instructionNameAt) else {
254
        return nil;
255
    };
256
    return super::INSTRUCTIONS[index].encoder;
257
}
258
259
/// Classify a directive name.
260
fn classifyDirective(name: *[u8]) -> ?super::DirectiveKind {
261
    let index = findSortedNameIndex(name, super::DIRECTIVES.len, directiveNameAt) else {
262
        return nil;
263
    };
264
    return super::DIRECTIVES[index].kind;
265
}
266
267
/// Look up a percent-prefixed register name after the `%` has been removed.
268
fn lookupRegister(name: *[u8]) -> ?gen::Reg {
269
    let index = findSortedNameIndex(name, super::REGISTERS.len, registerNameAt) else {
270
        return nil;
271
    };
272
    return super::REGISTERS[index].reg;
273
}
274
275
/// Look up a CSR name.
276
fn lookupCsr(name: *[u8]) -> ?u32 {
277
    let index = findSortedNameIndex(name, super::CSRS.len, csrNameAt) else {
278
        return nil;
279
    };
280
    return super::CSRS[index].csr;
281
}
282
283
/// Parse an instruction after its mnemonic has already been consumed.
284
unsafe fn parseInstruction 'parse (a: &mut super::Assembler 'parse, name: *[u8], tok: scanner::Token) throws (super::Error) {
285
    if a.section <> super::Section::Text {
286
        throw failOnToken(tok, "instructions are only valid in the text section");
287
    }
288
    if let format = atomics::parse(name) {
289
        let rd = try parseRegister(a);
290
        let mut rs2 = rv64::ZERO;
291
        if format.operation <> 2 {
292
            set rs2 = try parseRegister(a);
293
        }
294
        let memory = try parseMemory(a);
295
        if memory.offset <> 0 {
296
            throw fail(a, "atomic memory offset must be zero");
297
        }
298
        try emit::emitText(a, atomics::encode(atomics::Instruction { format, rd, rs1: memory.base, rs2 }));
299
        return;
300
    }
301
    let form = lookupInstruction(name) else {
302
        throw failOnToken(tok, "unknown instruction");
303
    };
304
    match form {
305
        case super::InstructionEncoder::Fence => return try parseFence(a),
306
        case super::InstructionEncoder::NoOperand { enc } => {
307
            if a.scan.current.kind <> scanner::TokenKind::Semicolon {
308
                throw fail(a, "unexpected operand");
309
            }
310
            try emit::emitText(a, enc());
311
            return;
312
        }
313
        case super::InstructionEncoder::Li => return try parseLi(a),
314
        case super::InstructionEncoder::La => return try parseLa(a),
315
        case super::InstructionEncoder::RR { enc } => return try parseRR(a, enc),
316
        case super::InstructionEncoder::RRR { enc } => return try parseRRR(a, enc),
317
        case super::InstructionEncoder::RRI { enc } => return try parseRRI(a, enc),
318
        case super::InstructionEncoder::Shift { enc } =>
319
            return try parseShift(a, enc, super::SHIFT_LIMIT, "shift amount out of range"),
320
        case super::InstructionEncoder::WordShift { enc } =>
321
            return try parseShift(a, enc, super::WORD_SHIFT_LIMIT, "word shift amount out of range"),
322
        case super::InstructionEncoder::Load { enc } => return try parseLoad(a, enc),
323
        case super::InstructionEncoder::Store { enc } => return try parseStore(a, enc),
324
        case super::InstructionEncoder::Branch { op } => return try parseBranch(a, op),
325
        case super::InstructionEncoder::BranchZero { op } => return try parseBranchZero(a, op),
326
        case super::InstructionEncoder::Jal => return try parseJal(a),
327
        case super::InstructionEncoder::Jump { rd } => return try parseJ(a, rd),
328
        case super::InstructionEncoder::RdCsr { enc } => return try parseRdCsr(a, enc),
329
        case super::InstructionEncoder::CsrRs1 { enc } => return try parseCsrRs1(a, enc),
330
        case super::InstructionEncoder::Csrrw => return try parseCsrrw(a),
331
        case super::InstructionEncoder::Csrsi => return try parseCsrsi(a),
332
        case super::InstructionEncoder::Upper { enc } => return try parseUpper(a, enc),
333
    }
334
}
335
336
/// Parse the `li` pseudo-instruction.
337
fn parseLi 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
338
    let rd = try parseRegister(a);
339
    let value = try parseValue(a);
340
    if encode::isSmallImm64(value) {
341
        try emit::emitText(a, encode::addi(rd, rv64::ZERO, value as i32));
342
        return;
343
    }
344
    let imm = try expectI32Value(a, value, "li immediate out of range");
345
    let split = rv64::emit::splitImm(imm);
346
347
    try emit::emitText(a, encode::lui(rd, split.hi));
348
    try emit::emitText(a, encode::addi(rd, rd, split.lo));
349
}
350
351
/// Parse the `la` pseudo-instruction.
352
unsafe fn parseLa 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
353
    let rd = try parseRegister(a);
354
    let target = try parseLabelName(a);
355
    let index = a.textLen;
356
357
    try emit::recordTextFixup(a, target, super::FixupInfo::Addr { rd, index }, 2);
358
}
359
360
/// Parse a CSR read-like instruction with destination register then CSR.
361
fn parseRdCsr 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, u32) -> u32) throws (super::Error) {
362
    let rd = try parseRegister(a);
363
    let csr = try parseCsr(a);
364
365
    try emit::emitText(a, enc(rd, csr));
366
}
367
368
/// Parse a CSR write-like instruction with CSR then source register.
369
fn parseCsrRs1 'parse (a: &mut super::Assembler 'parse, enc: fn(u32, gen::Reg) -> u32) throws (super::Error) {
370
    let csr = try parseCsr(a);
371
    let rs1 = try parseRegister(a);
372
373
    try emit::emitText(a, enc(csr, rs1));
374
}
375
376
/// Parse `csrrw`.
377
fn parseCsrrw 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
378
    let rd = try parseRegister(a);
379
    let csr = try parseCsr(a);
380
    let rs1 = try parseRegister(a);
381
382
    try emit::emitText(a, encode::csrrw(rd, csr, rs1));
383
}
384
385
/// Parse a CSR immediate instruction.
386
fn parseCsrsi 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
387
    let csr = try parseCsr(a);
388
    let imm = try parseValue(a);
389
    if imm < 0 or imm >= super::CSR_IMM_LIMIT {
390
        throw fail(a, "CSR immediate out of range");
391
    }
392
    try emit::emitText(a, encode::csrsi(csr, imm as u32));
393
}
394
395
/// Parse a two-register instruction.
396
fn parseRR 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, gen::Reg) -> u32) throws (super::Error) {
397
    let rd = try parseRegister(a);
398
    let rs = try parseRegister(a);
399
400
    try emit::emitText(a, enc(rd, rs));
401
}
402
403
/// Parse a three-register instruction.
404
fn parseRRR 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, gen::Reg, gen::Reg) -> u32) throws (super::Error) {
405
    let rd = try parseRegister(a);
406
    let rs1 = try parseRegister(a);
407
    let rs2 = try parseRegister(a);
408
409
    try emit::emitText(a, enc(rd, rs1, rs2));
410
}
411
412
/// Parse a register-register-immediate instruction.
413
fn parseRRI 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, gen::Reg, i32) -> u32) throws (super::Error) {
414
    let rd = try parseRegister(a);
415
    let rs1 = try parseRegister(a);
416
    let imm = try parseSmallImm(a);
417
418
    try emit::emitText(a, enc(rd, rs1, imm));
419
}
420
421
/// Parse a shift-immediate instruction and enforce its RV64 shift bound.
422
fn parseShift 'parse (
423
    a: &mut super::Assembler 'parse,
424
    enc: fn(gen::Reg, gen::Reg, i32) -> u32,
425
    limit: i32,
426
    message: *[u8]
427
) throws (super::Error) {
428
    let rd = try parseRegister(a);
429
    let rs1 = try parseRegister(a);
430
    let shamt64 = try parseValue(a);
431
432
    if shamt64 < 0 {
433
        throw fail(a, "shift amount must be non-negative");
434
    }
435
    if shamt64 >= limit as i64 {
436
        throw fail(a, message);
437
    }
438
    let shamt = shamt64 as i32;
439
440
    try emit::emitText(a, enc(rd, rs1, shamt));
441
}
442
443
/// Parse a load instruction with a memory operand.
444
fn parseLoad 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, gen::Reg, i32) -> u32) throws (super::Error) {
445
    let rd = try parseRegister(a);
446
    let memop = try parseMemory(a);
447
448
    try emit::emitText(a, enc(rd, memop.base, memop.offset));
449
}
450
451
/// Parse a store instruction with a memory operand.
452
fn parseStore 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, gen::Reg, i32) -> u32) throws (super::Error) {
453
    let rs2 = try parseRegister(a);
454
    let memop = try parseMemory(a);
455
456
    try emit::emitText(a, enc(rs2, memop.base, memop.offset));
457
}
458
459
/// Parse a two-register branch instruction.
460
unsafe fn parseBranch 'parse (a: &mut super::Assembler 'parse, op: super::BranchOp) throws (super::Error) {
461
    let rs1 = try parseRegister(a);
462
    let rs2 = try parseRegister(a);
463
464
    try parseBranchLabel(a, op, rs1, rs2);
465
}
466
467
/// Parse an optional label operand.
468
unsafe fn parseOptionalLabel 'parse (a: &mut super::Assembler 'parse) -> ?*[u8] throws (super::Error) {
469
    if not isLabel(a.scan.current.kind) {
470
        return nil;
471
    }
472
    return try parseLabelName(a);
473
}
474
475
/// Parse a branch target as either a label fixup or immediate offset.
476
unsafe fn parseBranchLabel 'parse (a: &mut super::Assembler 'parse, op: super::BranchOp, rs1: gen::Reg, rs2: gen::Reg) throws (super::Error) {
477
    let index = a.textLen;
478
    if let target = try parseOptionalLabel(a) {
479
        try emit::recordTextFixup(a, target, super::FixupInfo::Branch { op, rs1, rs2, index }, 1);
480
        return;
481
    }
482
    let imm = try parseBranchImm(a);
483
    try emit::emitText(a, emit::encodeBranch(op, rs1, rs2, imm));
484
}
485
486
/// Parse a branch-to-zero pseudo-instruction.
487
unsafe fn parseBranchZero 'parse (a: &mut super::Assembler 'parse, op: super::BranchOp) throws (super::Error) {
488
    let rs = try parseRegister(a);
489
    try parseBranchLabel(a, op, rs, rv64::ZERO);
490
}
491
492
/// Parse `jal` with an explicit destination register.
493
unsafe fn parseJal 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
494
    let rd = try parseRegister(a);
495
    try parseJ(a, rd);
496
}
497
498
/// Parse a jump target for `jal` or a jump pseudo-instruction.
499
unsafe fn parseJ 'parse (a: &mut super::Assembler 'parse, rd: gen::Reg) throws (super::Error) {
500
    let index = a.textLen;
501
    if let target = try parseOptionalLabel(a) {
502
        try emit::recordTextFixup(a, target, super::FixupInfo::Jal { rd, index }, 1);
503
        return;
504
    }
505
    let imm = try parseJumpImm(a);
506
    try emit::emitText(a, encode::jal(rd, imm));
507
}
508
509
/// Parse an upper-immediate instruction.
510
fn parseUpper 'parse (a: &mut super::Assembler 'parse, enc: fn(gen::Reg, i32) -> u32) throws (super::Error) {
511
    let rd = try parseRegister(a);
512
    let imm64 = try parseValue(a);
513
    if imm64 < 0 or imm64 > super::UPPER_IMM_MAX_VALUE {
514
        throw fail(a, "upper immediate out of range");
515
    }
516
    try emit::emitText(a, enc(rd, imm64 as i32));
517
}
518
519
/// Parse a directive after its name has already been consumed.
520
unsafe fn parseDirective 'parse (a: &mut super::Assembler 'parse, name: *[u8], tok: scanner::Token) throws (super::Error) {
521
    let directive = classifyDirective(name) else {
522
        throw failOnToken(tok, "unknown directive");
523
    };
524
    match directive {
525
        case super::DirectiveKind::Text => {
526
            try expectTerminator(a, "unexpected operand");
527
            set a.section = super::Section::Text;
528
            return;
529
        }
530
        case super::DirectiveKind::Data => {
531
            try expectTerminator(a, "unexpected operand");
532
            set a.section = super::Section::Data;
533
            return;
534
        }
535
        case super::DirectiveKind::Align =>
536
            return try parseAlignDirective(a),
537
        case super::DirectiveKind::Ascii => {
538
            try expectDataSection(a, tok);
539
            return try parseStringDirective(a);
540
        }
541
        case super::DirectiveKind::Byte => {
542
            try expectDataSection(a, tok);
543
            return try parseByteDirective(a);
544
        }
545
        case super::DirectiveKind::Constant =>
546
            return try parseConstantDirective(a),
547
        case super::DirectiveKind::Dword => {
548
            try expectDataSection(a, tok);
549
            return try parseIntDirective(a, super::DataWidth::Dword);
550
        }
551
        case super::DirectiveKind::Export =>
552
            return try parseExportDirective(a),
553
        case super::DirectiveKind::Space => {
554
            try expectDataSection(a, tok);
555
            return try parseSpaceDirective(a);
556
        }
557
        case super::DirectiveKind::Word => {
558
            try expectDataSection(a, tok);
559
            return try parseIntDirective(a, super::DataWidth::Word);
560
        }
561
    }
562
}
563
564
/// Parse a `.constant` directive.
565
fn parseConstantDirective 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
566
    let name = try parseSymbolName(a);
567
    let value = try expectI32Value(a, try parseExpr(a), "constant out of range");
568
569
    dict::insert(&mut a.constMap, name, value);
570
}
571
572
/// Parse a `.export` directive.
573
unsafe fn parseExportDirective 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
574
    let name = try parseLabelName(a);
575
    dict::insert(&mut a.exportMap, name, 1);
576
    if let idx = dict::get(&a.symbolMap, name) {
577
        set a.symbols[idx as u32].isExported = true;
578
    }
579
}
580
581
/// Parse a `.space` directive.
582
fn parseSpaceDirective 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
583
    let count = try parseValue(a);
584
    if count < 0 {
585
        throw fail(a, "space size must be non-negative");
586
    }
587
    // The data section grows on demand; only reject sizes that cannot be
588
    // represented as a section offset.
589
    if count > super::U32_MAX_VALUE - a.dataLen as i64 {
590
        throw super::Error::DataOverflow;
591
    }
592
    for _ in 0..count as u32 {
593
        try emit::emitByte(a, 0);
594
    }
595
}
596
597
/// Parse an `.align` directive for the current section.
598
fn parseAlignDirective 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
599
    let amount64 = try parseValue(a);
600
    if amount64 <= 0 {
601
        throw fail(a, "alignment must be positive");
602
    }
603
    if amount64 > super::U32_MAX_VALUE {
604
        throw fail(a, "alignment out of range");
605
    }
606
    let amount = amount64 as u32;
607
    if (amount & (amount - 1)) <> 0 {
608
        throw fail(a, "alignment must be a power of two");
609
    }
610
    match a.section {
611
        case super::Section::Text => {
612
            if amount % rv64::INSTR_SIZE as u32 <> 0 {
613
                throw fail(a, "text alignment must be a multiple of 4");
614
            }
615
            let bytes = a.textLen * rv64::INSTR_SIZE as u32;
616
            let aligned = checkedAlignUp(bytes, amount) else {
617
                throw super::Error::TextOverflow;
618
            };
619
            let words = (aligned - bytes) / rv64::INSTR_SIZE as u32;
620
            try emit::emitTextPadding(a, words);
621
        }
622
        case super::Section::Data => {
623
            let aligned = checkedAlignUp(a.dataLen, amount) else {
624
                throw super::Error::DataOverflow;
625
            };
626
            for _ in a.dataLen..aligned {
627
                try emit::emitByte(a, 0);
628
            }
629
        }
630
    }
631
}
632
633
/// Parse a `.byte` directive.
634
fn parseByteDirective 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
635
    loop {
636
        if a.scan.current.kind == scanner::TokenKind::Char {
637
            let ch = parseCharLiteral(a.scan.current) else {
638
                throw fail(a, "invalid char literal");
639
            };
640
            try emit::emitByte(a, ch);
641
            advance(a);
642
        } else {
643
            let value = try parseValue(a);
644
            if value < 0 or value > super::U8_MAX_VALUE {
645
                throw fail(a, "byte literal out of range");
646
            }
647
            try emit::emitByte(a, value as u8);
648
        }
649
        if not consume(a, scanner::TokenKind::Comma) {
650
            return;
651
        }
652
    }
653
}
654
655
/// Parse a fixed-width integer data directive.
656
unsafe fn parseIntDirective 'parse (a: &mut super::Assembler 'parse, width: super::DataWidth) throws (super::Error) {
657
    loop {
658
        if isLabel(a.scan.current.kind) {
659
            let target = try parseLabelName(a);
660
            try emit::recordDataFixup(a, target, width);
661
        } else if a.scan.current.kind == scanner::TokenKind::Char {
662
            let ch = parseCharLiteral(a.scan.current) else {
663
                throw fail(a, "invalid char literal");
664
            };
665
            advance(a);
666
            try emitDataValue(a, ch as i64, width);
667
        } else {
668
            try emitDataValue(a, try parseValue(a), width);
669
        }
670
        if not consume(a, scanner::TokenKind::Comma) {
671
            return;
672
        }
673
    }
674
}
675
676
/// Parse a `.ascii` string literal list.
677
fn parseStringDirective 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
678
    loop {
679
        let literal = try expectToken(a, scanner::TokenKind::String, "expected string literal");
680
        try emit::emitDecodedString(a, literal.source);
681
        if not consume(a, scanner::TokenKind::Comma) {
682
            return;
683
        }
684
    }
685
}
686
687
/// Parse and resolve a register operand.
688
fn parseRegister 'parse (a: &mut super::Assembler 'parse) -> gen::Reg throws (super::Error) {
689
    let tok = try expectToken(a, scanner::TokenKind::Register, "expected register");
690
    let reg = lookupRegister(&tok.source[1..]) else {
691
        throw super::Error::Invalid { offset: tok.offset, message: "unknown register" };
692
    };
693
    return reg;
694
}
695
696
/// Parse a simple signed immediate or constant value.
697
fn parseValue 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
698
    if consume(a, scanner::TokenKind::Minus) {
699
        return -(try parseValuePrimary(a));
700
    }
701
    return try parseValuePrimary(a);
702
}
703
704
/// Parse the primary form used by simple immediate values.
705
fn parseValuePrimary 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
706
    if a.scan.current.kind == scanner::TokenKind::Number {
707
        return try parseInteger(a);
708
    }
709
    if a.scan.current.kind == scanner::TokenKind::Ident {
710
        return try parseConstantValue(a);
711
    }
712
    throw fail(a, "expected number or constant");
713
}
714
715
/// Parse an additive constant expression.
716
fn parseExpr 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
717
    let mut value = try parseExprMul(a);
718
719
    while a.scan.current.kind == scanner::TokenKind::Plus or a.scan.current.kind == scanner::TokenKind::Minus {
720
        let op = a.scan.current.kind;
721
        advance(a);
722
723
        let rhs = try parseExprMul(a);
724
        if op == scanner::TokenKind::Plus {
725
            set value += rhs;
726
        } else {
727
            set value -= rhs;
728
        }
729
    }
730
    return value;
731
}
732
733
/// Parse multiplicative expression operators.
734
fn parseExprMul 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
735
    let mut value = try parseExprUnary(a);
736
737
    while a.scan.current.kind == scanner::TokenKind::Star or a.scan.current.kind == scanner::TokenKind::Slash {
738
        let op = a.scan.current.kind;
739
        advance(a);
740
741
        let rhs = try parseExprUnary(a);
742
        if op == scanner::TokenKind::Star {
743
            set value *= rhs;
744
        } else {
745
            if rhs == 0 {
746
                throw fail(a, "division by zero");
747
            }
748
            set value /= rhs;
749
        }
750
    }
751
    return value;
752
}
753
754
/// Parse unary expression operators.
755
fn parseExprUnary 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
756
    if consume(a, scanner::TokenKind::Minus) {
757
        return -(try parseExprUnary(a));
758
    }
759
    if consume(a, scanner::TokenKind::Plus) {
760
        return try parseExprUnary(a);
761
    }
762
    return try parseExprPrimary(a);
763
}
764
765
/// Parse expression atoms.
766
fn parseExprPrimary 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
767
    if consume(a, scanner::TokenKind::LParen) {
768
        let value = try parseExpr(a);
769
        try expect(a, scanner::TokenKind::RParen, "expected `)`");
770
        return value;
771
    }
772
    if a.scan.current.kind == scanner::TokenKind::Number {
773
        return try parseInteger(a);
774
    }
775
    if a.scan.current.kind == scanner::TokenKind::Ident {
776
        return try parseConstantValue(a);
777
    }
778
    throw fail(a, "expected expression");
779
}
780
781
/// Parse and resolve a named assembler constant.
782
fn parseConstantValue 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
783
    let name = try parseSymbolName(a);
784
    let value = dict::get(&a.constMap, name) else {
785
        throw super::Error::Invalid { offset: a.scan.previous.offset, message: "undefined constant" };
786
    };
787
    return value as i64;
788
}
789
790
/// Parse and resolve a CSR operand.
791
fn parseCsr 'parse (a: &mut super::Assembler 'parse) -> u32 throws (super::Error) {
792
    let name = try parseSymbolName(a);
793
    let csr = lookupCsr(name) else {
794
        throw super::Error::Invalid { offset: a.scan.previous.offset, message: "unknown CSR" };
795
    };
796
    return csr;
797
}
798
799
/// Parse an offset(base) memory operand.
800
fn parseMemory 'parse (a: &mut super::Assembler 'parse) -> MemOperand throws (super::Error) {
801
    let mut offset: i32 = 0;
802
    if a.scan.current.kind <> scanner::TokenKind::LParen {
803
        set offset = try expectSmallImmValue(a, try parseValue(a));
804
    }
805
    try expect(a, scanner::TokenKind::LParen, "expected `(`");
806
    let base = try parseRegister(a);
807
    try expect(a, scanner::TokenKind::RParen, "expected `)`");
808
809
    return MemOperand { base, offset };
810
}
811
812
/// Parse an immediate value that fits in a signed 12-bit field.
813
fn parseSmallImm 'parse (a: &mut super::Assembler 'parse) -> i32 throws (super::Error) {
814
    return try expectSmallImmValue(a, try parseValue(a));
815
}
816
817
/// Parse and validate a branch immediate.
818
fn parseBranchImm 'parse (a: &mut super::Assembler 'parse) -> i32 throws (super::Error) {
819
    let value = try expectI32Value(a, try parseValue(a), "branch immediate out of range");
820
    if not encode::isBranchImm(value) {
821
        throw fail(a, "branch immediate out of range");
822
    }
823
    return value;
824
}
825
826
/// Parse and validate a jump immediate.
827
fn parseJumpImm 'parse (a: &mut super::Assembler 'parse) -> i32 throws (super::Error) {
828
    let value = try expectI32Value(a, try parseValue(a), "jump immediate out of range");
829
    if not encode::isJumpImm(value) {
830
        throw fail(a, "jump immediate out of range");
831
    }
832
    return value;
833
}
834
835
/// Parse an integer token as an i64.
836
fn parseInteger 'parse (a: &mut super::Assembler 'parse) -> i64 throws (super::Error) {
837
    let tok = try expectToken(a, scanner::TokenKind::Number, "expected number");
838
    let value = parseIntegerText(tok.source) else {
839
        throw failOnToken(tok, "invalid integer literal");
840
    };
841
    return value;
842
}
843
844
/// Parse integer literal text as an i64.
845
fn parseIntegerText(text: *[u8]) -> ?i64 {
846
    if text.len == 0 {
847
        return nil;
848
    }
849
    let negative = text[0] == '-';
850
    let magnitudeText = &text[1..] if negative or text[0] == '+' else text;
851
    let literal = try fmt::parseInt(magnitudeText) catch {
852
        return nil;
853
    };
854
    if negative {
855
        if literal.magnitude > parser::I64_MIN_MAGNITUDE {
856
            return nil;
857
        }
858
        if literal.magnitude == parser::I64_MIN_MAGNITUDE {
859
            return parser::I64_MIN;
860
        }
861
        return -(literal.magnitude as i64);
862
    }
863
    if literal.magnitude > parser::I64_MAX_MAGNITUDE {
864
        return nil;
865
    }
866
    return literal.magnitude as i64;
867
}
868
869
/// Parse a character literal token as one byte.
870
fn parseCharLiteral(tok: scanner::Token) -> ?u8 {
871
    return try fmt::parseChar(tok.source) catch {
872
        return nil;
873
    };
874
}
875
876
/// Parse an access-class mask without duplicate fields.
877
fn parseFenceMask 'parse (a: &mut super::Assembler 'parse) -> u32 throws (super::Error) {
878
    if a.scan.current.kind == scanner::TokenKind::Number {
879
        let value = try parseValue(a);
880
        if value <> 0 {
881
            throw fail(a, "numeric fence mask must be zero");
882
        }
883
        return 0;
884
    }
885
    let token = try expectToken(a, scanner::TokenKind::Ident, "expected fence access classes");
886
    let mut mask: u32 = 0;
887
    for ch in token.source {
888
        let mut bit: u32 = 0;
889
        match ch {
890
            case 'i' => {
891
                set bit = 8;
892
            }, case 'o' => {
893
                set bit = 4;
894
            },
895
            case 'r' => {
896
                set bit = 2;
897
            }, case 'w' => {
898
                set bit = 1;
899
            },
900
            else => throw failOnToken(token, "invalid fence access class"),
901
        }
902
        if (mask & bit) <> 0 {
903
            throw failOnToken(token, "duplicate fence access class");
904
        }
905
        set mask |= bit;
906
    }
907
    return mask;
908
}
909
910
/// Parse a full memory fence or a pair of explicit access-class masks.
911
fn parseFence 'parse (a: &mut super::Assembler 'parse) throws (super::Error) {
912
    if a.scan.current.kind == scanner::TokenKind::Semicolon {
913
        try emit::emitText(a, encode::fence());
914
        return;
915
    }
916
    let predecessor = try parseFenceMask(a);
917
    let successor = try parseFenceMask(a);
918
    try emit::emitText(a, encode::fenceOrder(predecessor, successor));
919
}