synth_backend/
elf_builder.rs

1//! ELF (Executable and Linkable Format) Builder for ARM
2//!
3//! Generates ELF32 files for ARM Cortex-M targets
4
5use synth_core::Result;
6
7/// ELF file class
8#[derive(Debug, Clone, Copy, PartialEq, Eq)]
9pub enum ElfClass {
10    /// 32-bit
11    Elf32 = 1,
12    /// 64-bit
13    Elf64 = 2,
14}
15
16/// ELF data encoding
17#[derive(Debug, Clone, Copy, PartialEq, Eq)]
18pub enum ElfData {
19    /// Little-endian
20    LittleEndian = 1,
21    /// Big-endian
22    BigEndian = 2,
23}
24
25/// ELF file type
26#[derive(Debug, Clone, Copy, PartialEq, Eq)]
27pub enum ElfType {
28    /// Relocatable file
29    Rel = 1,
30    /// Executable file
31    Exec = 2,
32    /// Shared object file
33    Dyn = 3,
34}
35
36/// ELF machine architecture
37#[derive(Debug, Clone, Copy, PartialEq, Eq)]
38pub enum ElfMachine {
39    /// ARM
40    Arm = 40,
41    /// ARM64/AArch64
42    AArch64 = 183,
43}
44
45/// Section type
46#[derive(Debug, Clone, Copy, PartialEq, Eq)]
47pub enum SectionType {
48    /// Null section
49    Null = 0,
50    /// Program data
51    ProgBits = 1,
52    /// Symbol table
53    SymTab = 2,
54    /// String table
55    StrTab = 3,
56    /// Relocation entries with addends
57    Rela = 4,
58    /// Symbol hash table
59    Hash = 5,
60    /// Dynamic linking information
61    Dynamic = 6,
62    /// Note
63    Note = 7,
64    /// No space (BSS)
65    NoBits = 8,
66    /// Relocation entries
67    Rel = 9,
68    /// ARM build attributes (`SHT_ARM_ATTRIBUTES`, #637) — the `.ARM.attributes`
69    /// section every ARM toolchain consults to auto-select the Thumb/A32 decoder.
70    ArmAttributes = 0x7000_0003,
71}
72
73/// Section flags
74#[derive(Debug, Clone, Copy)]
75pub struct SectionFlags(pub u32);
76
77impl SectionFlags {
78    /// Writable
79    pub const WRITE: u32 = 0x1;
80    /// Occupies memory during execution
81    pub const ALLOC: u32 = 0x2;
82    /// Executable
83    pub const EXEC: u32 = 0x4;
84    /// Mergeable
85    pub const MERGE: u32 = 0x10;
86    /// Contains null-terminated strings
87    pub const STRINGS: u32 = 0x20;
88}
89
90/// ELF section
91#[derive(Debug, Clone)]
92pub struct Section {
93    /// Section name (index into string table)
94    pub name: String,
95    /// Section type
96    pub section_type: SectionType,
97    /// Section flags
98    pub flags: u32,
99    /// Virtual address
100    pub addr: u32,
101    /// Section data
102    pub data: Vec<u8>,
103    /// Alignment
104    pub align: u32,
105    /// Explicit size (for NoBits sections like .bss where data is empty)
106    pub explicit_size: Option<u32>,
107}
108
109impl Section {
110    /// Create a new section
111    pub fn new(name: &str, section_type: SectionType) -> Self {
112        Self {
113            name: name.to_string(),
114            section_type,
115            flags: 0,
116            addr: 0,
117            data: Vec::new(),
118            align: 1,
119            explicit_size: None,
120        }
121    }
122
123    /// Set flags
124    pub fn with_flags(mut self, flags: u32) -> Self {
125        self.flags = flags;
126        self
127    }
128
129    /// Set address
130    pub fn with_addr(mut self, addr: u32) -> Self {
131        self.addr = addr;
132        self
133    }
134
135    /// Set alignment
136    pub fn with_align(mut self, align: u32) -> Self {
137        self.align = align;
138        self
139    }
140
141    /// Add data
142    pub fn with_data(mut self, data: Vec<u8>) -> Self {
143        self.data = data;
144        self
145    }
146
147    /// Set explicit size (for NoBits sections like .bss where data is empty)
148    pub fn with_size(mut self, size: u32) -> Self {
149        self.explicit_size = Some(size);
150        self
151    }
152
153    /// Get the effective size of the section
154    pub fn size(&self) -> u32 {
155        self.explicit_size.unwrap_or(self.data.len() as u32)
156    }
157}
158
159/// Symbol binding
160#[derive(Debug, Clone, Copy, PartialEq, Eq)]
161pub enum SymbolBinding {
162    /// Local symbol
163    Local = 0,
164    /// Global symbol
165    Global = 1,
166    /// Weak symbol
167    Weak = 2,
168}
169
170/// Symbol type
171#[derive(Debug, Clone, Copy, PartialEq, Eq)]
172pub enum SymbolType {
173    /// No type
174    NoType = 0,
175    /// Object (data)
176    Object = 1,
177    /// Function
178    Func = 2,
179    /// Section
180    Section = 3,
181    /// File name
182    File = 4,
183}
184
185/// ELF symbol
186#[derive(Debug, Clone)]
187pub struct Symbol {
188    /// Symbol name
189    pub name: String,
190    /// Value/address
191    pub value: u32,
192    /// Size
193    pub size: u32,
194    /// Binding
195    pub binding: SymbolBinding,
196    /// Type
197    pub symbol_type: SymbolType,
198    /// Section index
199    pub section: u16,
200}
201
202impl Symbol {
203    /// Create a new symbol
204    pub fn new(name: &str) -> Self {
205        Self {
206            name: name.to_string(),
207            value: 0,
208            size: 0,
209            binding: SymbolBinding::Local,
210            symbol_type: SymbolType::NoType,
211            section: 0,
212        }
213    }
214
215    /// Set value
216    pub fn with_value(mut self, value: u32) -> Self {
217        self.value = value;
218        self
219    }
220
221    /// Set size
222    pub fn with_size(mut self, size: u32) -> Self {
223        self.size = size;
224        self
225    }
226
227    /// Set binding
228    pub fn with_binding(mut self, binding: SymbolBinding) -> Self {
229        self.binding = binding;
230        self
231    }
232
233    /// Set type
234    pub fn with_type(mut self, symbol_type: SymbolType) -> Self {
235        self.symbol_type = symbol_type;
236        self
237    }
238
239    /// Set section
240    pub fn with_section(mut self, section: u16) -> Self {
241        self.section = section;
242        self
243    }
244}
245
246/// Program header type
247#[derive(Debug, Clone, Copy, PartialEq, Eq)]
248pub enum ProgramType {
249    /// Null entry
250    Null = 0,
251    /// Loadable segment
252    Load = 1,
253    /// Dynamic linking info
254    Dynamic = 2,
255    /// Interpreter path
256    Interp = 3,
257    /// Note section
258    Note = 4,
259}
260
261/// Program header flags
262pub struct ProgramFlags;
263
264impl ProgramFlags {
265    /// Executable
266    pub const EXEC: u32 = 0x1;
267    /// Writable
268    pub const WRITE: u32 = 0x2;
269    /// Readable
270    pub const READ: u32 = 0x4;
271}
272
273/// ELF program header (segment)
274#[derive(Debug, Clone)]
275pub struct ProgramHeader {
276    /// Segment type
277    pub p_type: ProgramType,
278    /// Offset in file
279    pub offset: u32,
280    /// Virtual address
281    pub vaddr: u32,
282    /// Physical address
283    pub paddr: u32,
284    /// Size in file
285    pub filesz: u32,
286    /// Size in memory
287    pub memsz: u32,
288    /// Flags (R/W/X)
289    pub flags: u32,
290    /// Alignment
291    pub align: u32,
292}
293
294impl ProgramHeader {
295    /// Create a new LOAD segment
296    pub fn load(vaddr: u32, offset: u32, size: u32, flags: u32) -> Self {
297        Self {
298            p_type: ProgramType::Load,
299            offset,
300            vaddr,
301            paddr: vaddr, // Physical = virtual for simple cases
302            filesz: size,
303            memsz: size,
304            flags,
305            align: 4,
306        }
307    }
308
309    /// Create a new LOAD segment for BSS-like regions (no file data, only memory)
310    /// Used for .bss, linear memory, and other zero-initialized regions
311    pub fn load_nobits(vaddr: u32, memsz: u32, flags: u32) -> Self {
312        Self {
313            p_type: ProgramType::Load,
314            offset: 0, // No file offset for NoBits
315            vaddr,
316            paddr: vaddr, // Physical = virtual
317            filesz: 0,    // No file data
318            memsz,        // Memory size to allocate
319            flags,
320            align: 4,
321        }
322    }
323}
324
325/// ARM relocation type
326#[derive(Debug, Clone, Copy, PartialEq, Eq)]
327pub enum ArmRelocationType {
328    /// R_ARM_THM_CALL (10) — Thumb BL/BLX instruction (Cortex-M). This is the
329    /// correct relocation for a Thumb-2 `bl` call site; `Call`/R_ARM_CALL below
330    /// is the ARM-mode form and is mis-resolved by `ld` for Thumb calls.
331    ThmCall = 10,
332    /// R_ARM_CALL (28) — BL/BLX instruction
333    Call = 28,
334    /// R_ARM_JUMP24 (29) — B/BL<cond> instruction
335    Jump24 = 29,
336    /// R_ARM_ABS32 (2) — Direct 32-bit reference
337    Abs32 = 2,
338    /// R_ARM_MOVW_ABS_NC (43) — MOVW instruction (low 16 bits)
339    MovwAbsNc = 43,
340    /// R_ARM_MOVT_ABS (44) — MOVT instruction (high 16 bits)
341    MovtAbs = 44,
342}
343
344/// A built `.rel.<name>` section: its name offset in `.shstrtab`, the target
345/// section index it relocates (`sh_info`), its file offset, and the encoded
346/// REL entries. Internal to [`ElfBuilder::build`].
347struct ExtraRelSection {
348    name_offset: usize,
349    target_idx: u32,
350    offset: usize,
351    data: Vec<u8>,
352}
353
354/// ELF relocation entry (REL format, no addend)
355#[derive(Debug, Clone)]
356pub struct Relocation {
357    /// Offset within the section where the relocation applies
358    pub offset: u32,
359    /// Symbol index in the symbol table
360    pub symbol_index: u32,
361    /// Relocation type
362    pub reloc_type: ArmRelocationType,
363}
364
365/// ARM EABI version 5 (soft-float)
366pub const EF_ARM_EABI_VER5: u32 = 0x05000000;
367/// ARM hard-float ABI flag
368pub const EF_ARM_ABI_FLOAT_HARD: u32 = 0x00000400;
369/// ARM soft-float ABI flag
370pub const EF_ARM_ABI_FLOAT_SOFT: u32 = 0x00000200;
371
372/// ARM EABI build-attribute tags and values (#637) — "Addenda to, and Errata
373/// in, the ABI for the Arm Architecture" (build attributes). Only the tags the
374/// synth ELF writer emits; consumers (objdump, gdb, `synth disasm`) use them to
375/// auto-select the Thumb vs A32 decoder without a manual `--triple`.
376pub mod aeabi {
377    /// Tag_CPU_arch (uleb value)
378    pub const TAG_CPU_ARCH: u32 = 6;
379    /// Tag_CPU_arch_profile (uleb value: 'M', 'R', 'A')
380    pub const TAG_CPU_ARCH_PROFILE: u32 = 7;
381    /// Tag_ARM_ISA_use (0 = no A32, 1 = A32 permitted)
382    pub const TAG_ARM_ISA_USE: u32 = 8;
383    /// Tag_THUMB_ISA_use (0 = none, 1 = Thumb-1 (16-bit), 2 = Thumb-2)
384    pub const TAG_THUMB_ISA_USE: u32 = 9;
385
386    /// Tag_CPU_arch value: ARMv7 (Cortex-M3 / Cortex-R profile base)
387    pub const CPU_ARCH_V7: u32 = 10;
388    /// Tag_CPU_arch value: ARMv6-M (Cortex-M0)
389    pub const CPU_ARCH_V6M: u32 = 11;
390    /// Tag_CPU_arch value: ARMv7E-M (Cortex-M4/M7)
391    pub const CPU_ARCH_V7EM: u32 = 13;
392    /// Tag_CPU_arch value: ARMv8.1-M.mainline (Cortex-M55)
393    pub const CPU_ARCH_V8_1M_MAIN: u32 = 21;
394
395    /// Tag_CPU_arch_profile value: microcontroller
396    pub const PROFILE_M: u32 = b'M' as u32;
397    /// Tag_CPU_arch_profile value: real-time
398    pub const PROFILE_R: u32 = b'R' as u32;
399
400    /// Tag_ABI_VFP_args (GI-FPU-002, #619): 0 = base (soft-float) variant,
401    /// 1 = FP args passed in VFP registers (AAPCS-VFP / hard-float).
402    pub const TAG_ABI_VFP_ARGS: u32 = 28;
403    /// Tag_FP_arch (GI-FPU-002, #619): the floating-point hardware the object
404    /// requires (tag 10 — NOT 36, which is Tag_FP_HP_extension). Value 0 = none
405    /// (soft-float; omitted by the writer).
406    pub const TAG_FP_ARCH: u32 = 10;
407    /// Tag_ABI_VFP_args value: FP args passed in VFP registers (hard-float).
408    pub const VFP_ARGS_VFP_REGS: u32 = 1;
409    /// Tag_FP_arch value: VFPv4-D16. synth's phase-1 f32 codegen uses only the
410    /// single-precision VADD/VMUL/VCVT subset shared by every Cortex-M FPU
411    /// (FPv4-SP through FPv5), so this conservative value describes the required
412    /// hardware without over-claiming (#619/#369).
413    pub const FP_ARCH_VFPV4_D16: u32 = 6;
414}
415
416/// Encode a u32 as ULEB128 (build-attribute value encoding).
417fn push_uleb128(out: &mut Vec<u8>, mut v: u32) {
418    loop {
419        let byte = (v & 0x7f) as u8;
420        v >>= 7;
421        if v == 0 {
422            out.push(byte);
423            break;
424        }
425        out.push(byte | 0x80);
426    }
427}
428
429/// Build the `.ARM.attributes` section (#637): format-version `'A'`, one
430/// `"aeabi"` vendor subsection carrying a single `Tag_File` (1) subsubsection
431/// with the given file-scope attributes. Tags with value 0 are omitted (0 is
432/// the spec default). Standard toolchains (objdump, gdb, ld) read this to
433/// auto-select the Thumb vs A32 decoder — synth objects become self-describing
434/// instead of requiring a manual `--triple=thumbv7m`.
435pub fn arm_attributes_section(
436    cpu_arch: u32,
437    cpu_arch_profile: u32,
438    arm_isa_use: u32,
439    thumb_isa_use: u32,
440    fp_arch: u32,
441    vfp_args: u32,
442) -> Section {
443    // File-scope attribute pairs (uleb tag, uleb value). Emitted in ascending
444    // tag order: CPU_arch(6), profile(7), ARM_ISA(8), THUMB_ISA(9),
445    // FP_arch(10), ABI_VFP_args(28). Tags with value 0 are omitted (spec
446    // default), so a soft-float (no-FPU) object is byte-identical to before
447    // GI-FPU-002.
448    let mut attrs = Vec::new();
449    for (tag, value) in [
450        (aeabi::TAG_CPU_ARCH, cpu_arch),
451        (aeabi::TAG_CPU_ARCH_PROFILE, cpu_arch_profile),
452        (aeabi::TAG_ARM_ISA_USE, arm_isa_use),
453        (aeabi::TAG_THUMB_ISA_USE, thumb_isa_use),
454        (aeabi::TAG_FP_ARCH, fp_arch),
455        (aeabi::TAG_ABI_VFP_ARGS, vfp_args),
456    ] {
457        if value != 0 {
458            push_uleb128(&mut attrs, tag);
459            push_uleb128(&mut attrs, value);
460        }
461    }
462
463    // Tag_File (1) subsubsection: tag byte + u32 length (self-inclusive) + attrs.
464    let file_len = (1 + 4 + attrs.len()) as u32;
465    let mut file_sub = vec![1u8]; // Tag_File
466    file_sub.extend_from_slice(&file_len.to_le_bytes());
467    file_sub.extend_from_slice(&attrs);
468
469    // "aeabi" vendor subsection: u32 length (self-inclusive) + NTBS name + data.
470    let vendor_name = b"aeabi\0";
471    let vendor_len = (4 + vendor_name.len() + file_sub.len()) as u32;
472    let mut blob = vec![b'A']; // format version
473    blob.extend_from_slice(&vendor_len.to_le_bytes());
474    blob.extend_from_slice(vendor_name);
475    blob.extend_from_slice(&file_sub);
476
477    Section::new(".ARM.attributes", SectionType::ArmAttributes)
478        .with_align(1)
479        .with_data(blob)
480}
481
482/// ELF file builder
483pub struct ElfBuilder {
484    /// File class (32 or 64 bit)
485    class: ElfClass,
486    /// Data encoding
487    data: ElfData,
488    /// File type
489    elf_type: ElfType,
490    /// Machine architecture
491    machine: ElfMachine,
492    /// Entry point address
493    entry: u32,
494    /// ELF e_flags (EABI version + float ABI)
495    e_flags: u32,
496    /// Sections
497    sections: Vec<Section>,
498    /// Symbols
499    symbols: Vec<Symbol>,
500    /// Program headers (segments)
501    program_headers: Vec<ProgramHeader>,
502    /// Relocations for .text section
503    relocations: Vec<Relocation>,
504    /// Extra per-section relocation tables, keyed by the target section's name
505    /// (e.g. `.debug_line`). Each produces a `.rel.<name>` section. Kept separate
506    /// from `relocations` (the `.text` set) so the existing `.rel.text` byte
507    /// layout is untouched: when this is empty the build is byte-identical to the
508    /// pre-generalization output (VCR-DBG-001 PR C, #394).
509    extra_relocations: Vec<(String, Vec<Relocation>)>,
510    /// #598: whether the object's functions are Thumb-encoded. Bit 0 of an
511    /// STT_FUNC `st_value` (and of `e_entry`) is the Thumb interworking bit —
512    /// it must be SET for Thumb code (Cortex-M) and CLEAR for A32 code
513    /// (cortex-r5 path). Defaults to `true` (every pre-#598 ARM object was
514    /// treated as Thumb, so Thumb outputs stay bit-identical).
515    thumb_funcs: bool,
516}
517
518impl ElfBuilder {
519    /// Create a new ELF builder for ARM32
520    pub fn new_arm32() -> Self {
521        Self {
522            class: ElfClass::Elf32,
523            data: ElfData::LittleEndian,
524            elf_type: ElfType::Exec,
525            machine: ElfMachine::Arm,
526            entry: 0,
527            e_flags: EF_ARM_EABI_VER5,
528            sections: Vec::new(),
529            symbols: Vec::new(),
530            program_headers: Vec::new(),
531            relocations: Vec::new(),
532            extra_relocations: Vec::new(),
533            thumb_funcs: true,
534        }
535    }
536
537    /// #598: mark the object's functions as A32-encoded (cortex-r5 path) —
538    /// suppresses the Thumb interworking bit on STT_FUNC `st_value`s and on
539    /// `e_entry`. Call BEFORE `with_entry`.
540    pub fn with_thumb_funcs(mut self, thumb: bool) -> Self {
541        self.thumb_funcs = thumb;
542        self
543    }
544
545    /// Set entry point
546    ///
547    /// For ARM Thumb targets, bit 0 is automatically set to indicate Thumb mode.
548    /// Cortex-M is Thumb-only, so function addresses in ELF must have bit 0 set.
549    /// A32 objects (`with_thumb_funcs(false)`, #598) keep bit 0 clear.
550    pub fn with_entry(mut self, entry: u32) -> Self {
551        self.entry = if self.machine == ElfMachine::Arm && self.thumb_funcs {
552            entry | 1 // Set Thumb bit for ARM Thumb targets
553        } else {
554            entry
555        };
556        self
557    }
558
559    /// Set ELF e_flags (e.g. to add hard-float ABI)
560    pub fn set_flags(&mut self, flags: u32) {
561        self.e_flags = flags;
562    }
563
564    /// Set file type
565    pub fn with_type(mut self, elf_type: ElfType) -> Self {
566        self.elf_type = elf_type;
567        self
568    }
569
570    /// Add a section
571    pub fn add_section(&mut self, section: Section) {
572        self.sections.push(section);
573    }
574
575    /// Add a symbol
576    pub fn add_symbol(&mut self, symbol: Symbol) {
577        self.symbols.push(symbol);
578    }
579
580    /// Add a symbol and return its 1-based index in `.symtab` (index 0 is the
581    /// reserved null symbol). Use when a later relocation must reference this
582    /// symbol — e.g. the `.text` base symbol the DWARF `.rel.debug_*` records
583    /// resolve against (VCR-DBG-001).
584    pub fn add_symbol_indexed(&mut self, symbol: Symbol) -> u32 {
585        let index = self.symbols.len() as u32 + 1;
586        self.symbols.push(symbol);
587        index
588    }
589
590    /// Add a program header (segment)
591    pub fn add_program_header(&mut self, ph: ProgramHeader) {
592        self.program_headers.push(ph);
593    }
594
595    /// Add a relocation entry for the .text section
596    pub fn add_relocation(&mut self, reloc: Relocation) {
597        self.relocations.push(reloc);
598    }
599
600    /// Add a relocation table targeting a non-`.text` section by name (e.g.
601    /// `.debug_line`). Produces a separate `.rel.<name>` section whose `sh_info`
602    /// points at the named section. The section must already have been added via
603    /// [`add_section`]; if no matching section exists at build time the table is
604    /// silently dropped. Used by VCR-DBG-001 to relocate the DWARF `.text`
605    /// references so a host linker fixes them up alongside `.text`.
606    pub fn add_section_relocations(&mut self, target_section: &str, relocs: Vec<Relocation>) {
607        if relocs.is_empty() {
608            return;
609        }
610        self.extra_relocations
611            .push((target_section.to_string(), relocs));
612    }
613
614    /// Add an undefined external symbol (e.g., __meld_dispatch_import)
615    /// Returns the symbol index (1-based, accounting for null symbol)
616    pub fn add_undefined_symbol(&mut self, name: &str) -> u32 {
617        let index = self.symbols.len() as u32 + 1; // +1 for null symbol
618        self.symbols.push(Symbol {
619            name: name.to_string(),
620            value: 0,
621            size: 0,
622            binding: SymbolBinding::Global,
623            symbol_type: SymbolType::Func,
624            section: 0, // SHN_UNDEF
625        });
626        index
627    }
628
629    /// Build the ELF file to bytes
630    pub fn build(&self) -> Result<Vec<u8>> {
631        let mut output = Vec::new();
632
633        // ELF header size (52 bytes for ELF32)
634        let header_size = 52;
635        // Program header size (32 bytes for ELF32)
636        let ph_entry_size = 32;
637        let ph_count = self.program_headers.len();
638        let ph_table_size = ph_entry_size * ph_count;
639
640        // Reserve space for ELF header + program headers
641        output.resize(header_size + ph_table_size, 0);
642
643        // Build string table for section names
644        let (shstrtab_data, section_name_offsets, extra_rel_name_offsets) =
645            self.build_section_string_table();
646
647        // Build symbol string table
648        let (strtab_data, symbol_name_offsets) = self.build_symbol_string_table();
649
650        // #656: ELF requires every STB_LOCAL symbol to precede all non-local
651        // symbols in `.symtab`, with the section's `sh_info` set to the index of
652        // the first non-local symbol. Callers add symbols in whatever order is
653        // convenient (and hold 1-based indices from `add_symbol_indexed` /
654        // `add_undefined_symbol` for their relocations), so the LOCAL/GLOBAL
655        // ordering is established here at build time: a stable locals-first
656        // permutation, plus an old→new index map every relocation is rewritten
657        // through. With zero local symbols (every pre-#656 object) the
658        // permutation is the identity and `sh_info` stays 1 — byte-identical.
659        let mut sym_order: Vec<usize> = (0..self.symbols.len()).collect();
660        sym_order.sort_by_key(|&i| self.symbols[i].binding != SymbolBinding::Local);
661        // old_to_new[old_1based] = new_1based; index 0 (the null symbol) maps to 0.
662        let mut old_to_new = vec![0u32; self.symbols.len() + 1];
663        for (new_pos, &old) in sym_order.iter().enumerate() {
664            old_to_new[old + 1] = new_pos as u32 + 1;
665        }
666        let local_count = self
667            .symbols
668            .iter()
669            .filter(|s| s.binding == SymbolBinding::Local)
670            .count();
671        // The null symbol (index 0) counts as local, so first-global = locals + 1.
672        let symtab_sh_info = local_count as u32 + 1;
673        let remap_relocs = |relocs: &[Relocation]| -> Vec<Relocation> {
674            relocs
675                .iter()
676                .map(|r| Relocation {
677                    offset: r.offset,
678                    symbol_index: old_to_new[r.symbol_index as usize],
679                    reloc_type: r.reloc_type,
680                })
681                .collect()
682        };
683
684        // Calculate section offsets (after ELF header + program headers)
685        let mut current_offset = header_size + ph_table_size;
686
687        // Section 1: .shstrtab (section name string table)
688        let shstrtab_offset = current_offset;
689        current_offset += shstrtab_data.len();
690
691        // Section 2: .strtab (symbol name string table)
692        let strtab_offset = current_offset;
693        current_offset += strtab_data.len();
694
695        // User sections
696        let mut section_offsets = Vec::new();
697        for section in &self.sections {
698            section_offsets.push(current_offset);
699            current_offset += section.data.len();
700        }
701
702        // Section 3: .symtab (symbol table), in locals-first order (#656)
703        let symtab_offset = current_offset;
704        let symtab_data = self.build_symbol_table(&symbol_name_offsets, &sym_order);
705        current_offset += symtab_data.len();
706
707        // Section 4+ (optional): .rel.text (relocations), symbol indices
708        // rewritten through the locals-first permutation (#656)
709        let rel_data = Self::encode_rel_entries(&remap_relocs(&self.relocations));
710        let rel_offset = current_offset;
711        current_offset += rel_data.len();
712
713        // Extra per-section relocation tables (.rel.<name>), laid out after
714        // .rel.text. Each entry resolves its target section index by name; a
715        // table whose target section is absent is dropped. Empty when no
716        // --debug-line ⇒ byte-identical to the pre-generalization layout.
717        let mut extra_rel: Vec<ExtraRelSection> = Vec::new();
718        for (i, (target, relocs)) in self.extra_relocations.iter().enumerate() {
719            let Some(target_idx) = self.section_index_by_name(target) else {
720                continue;
721            };
722            let data = Self::encode_rel_entries(&remap_relocs(relocs));
723            let name_offset = extra_rel_name_offsets.get(i).copied().unwrap_or(0);
724            extra_rel.push(ExtraRelSection {
725                name_offset,
726                target_idx,
727                offset: current_offset,
728                data,
729            });
730            current_offset += extra_rel.last().unwrap().data.len();
731        }
732
733        // Section header table comes at the end
734        let sh_offset = current_offset;
735
736        // Now write all the data
737        output.extend_from_slice(&shstrtab_data);
738        output.extend_from_slice(&strtab_data);
739
740        for section in &self.sections {
741            output.extend_from_slice(&section.data);
742        }
743
744        output.extend_from_slice(&symtab_data);
745        output.extend_from_slice(&rel_data);
746        for er in &extra_rel {
747            output.extend_from_slice(&er.data);
748        }
749
750        // Write section headers
751        let section_headers = self.build_section_headers_with_rel(
752            &section_name_offsets,
753            shstrtab_offset,
754            &shstrtab_data,
755            strtab_offset,
756            &strtab_data,
757            symtab_offset,
758            &symtab_data,
759            &section_offsets,
760            rel_offset,
761            &rel_data,
762            &extra_rel,
763            symtab_sh_info,
764        );
765        output.extend_from_slice(&section_headers);
766
767        // Write program headers (right after ELF header)
768        // Auto-correct p_offset for LOAD segments by matching vaddr to section addrs
769        for (i, ph) in self.program_headers.iter().enumerate() {
770            let ph_offset = header_size + i * ph_entry_size;
771            let mut corrected_ph = ph.clone();
772            if corrected_ph.filesz > 0 {
773                // Find the section whose addr matches this segment's vaddr
774                for (si, section) in self.sections.iter().enumerate() {
775                    if section.addr == corrected_ph.vaddr && si < section_offsets.len() {
776                        corrected_ph.offset = section_offsets[si] as u32;
777                        break;
778                    }
779                }
780            }
781            self.write_program_header(
782                &mut output[ph_offset..ph_offset + ph_entry_size],
783                &corrected_ph,
784            );
785        }
786
787        // Now write the actual ELF header at the beginning
788        let has_rel = !self.relocations.is_empty();
789        let num_sections = 4 + self.sections.len() + if has_rel { 1 } else { 0 } + extra_rel.len();
790        let ph_offset = if ph_count > 0 { header_size as u32 } else { 0 };
791        self.write_elf_header_with_phdrs(
792            &mut output[0..header_size],
793            ph_offset,
794            ph_count as u16,
795            sh_offset as u32,
796            num_sections as u16,
797        )?;
798
799        Ok(output)
800    }
801
802    /// Write a single program header
803    fn write_program_header(&self, output: &mut [u8], ph: &ProgramHeader) {
804        let mut cursor = 0;
805
806        // p_type (4 bytes)
807        output[cursor..cursor + 4].copy_from_slice(&(ph.p_type as u32).to_le_bytes());
808        cursor += 4;
809
810        // p_offset (4 bytes)
811        output[cursor..cursor + 4].copy_from_slice(&ph.offset.to_le_bytes());
812        cursor += 4;
813
814        // p_vaddr (4 bytes)
815        output[cursor..cursor + 4].copy_from_slice(&ph.vaddr.to_le_bytes());
816        cursor += 4;
817
818        // p_paddr (4 bytes)
819        output[cursor..cursor + 4].copy_from_slice(&ph.paddr.to_le_bytes());
820        cursor += 4;
821
822        // p_filesz (4 bytes)
823        output[cursor..cursor + 4].copy_from_slice(&ph.filesz.to_le_bytes());
824        cursor += 4;
825
826        // p_memsz (4 bytes)
827        output[cursor..cursor + 4].copy_from_slice(&ph.memsz.to_le_bytes());
828        cursor += 4;
829
830        // p_flags (4 bytes)
831        output[cursor..cursor + 4].copy_from_slice(&ph.flags.to_le_bytes());
832        cursor += 4;
833
834        // p_align (4 bytes)
835        output[cursor..cursor + 4].copy_from_slice(&ph.align.to_le_bytes());
836    }
837
838    /// Write ELF header with program header info
839    fn write_elf_header_with_phdrs(
840        &self,
841        output: &mut [u8],
842        ph_offset: u32,
843        ph_count: u16,
844        sh_offset: u32,
845        sh_count: u16,
846    ) -> Result<()> {
847        let mut cursor = 0;
848
849        // ELF magic number
850        output[cursor..cursor + 4].copy_from_slice(&[0x7f, b'E', b'L', b'F']);
851        cursor += 4;
852
853        // Class (32-bit)
854        output[cursor] = self.class as u8;
855        cursor += 1;
856
857        // Data (little-endian)
858        output[cursor] = self.data as u8;
859        cursor += 1;
860
861        // Version
862        output[cursor] = 1;
863        cursor += 1;
864
865        // OS/ABI
866        output[cursor] = 0; // System V
867        cursor += 1;
868
869        // ABI version
870        output[cursor] = 0;
871        cursor += 1;
872
873        // Padding (7 bytes)
874        output[cursor..cursor + 7].copy_from_slice(&[0; 7]);
875        cursor += 7;
876
877        // Type (little-endian u16)
878        let etype = self.elf_type as u16;
879        output[cursor..cursor + 2].copy_from_slice(&etype.to_le_bytes());
880        cursor += 2;
881
882        // Machine (little-endian u16)
883        let machine = self.machine as u16;
884        output[cursor..cursor + 2].copy_from_slice(&machine.to_le_bytes());
885        cursor += 2;
886
887        // Version (little-endian u32)
888        output[cursor..cursor + 4].copy_from_slice(&1u32.to_le_bytes());
889        cursor += 4;
890
891        // Entry point (little-endian u32)
892        output[cursor..cursor + 4].copy_from_slice(&self.entry.to_le_bytes());
893        cursor += 4;
894
895        // Program header offset (little-endian u32)
896        output[cursor..cursor + 4].copy_from_slice(&ph_offset.to_le_bytes());
897        cursor += 4;
898
899        // Section header offset (little-endian u32)
900        output[cursor..cursor + 4].copy_from_slice(&sh_offset.to_le_bytes());
901        cursor += 4;
902
903        // Flags (little-endian u32) - ARM EABI version 5 + float ABI
904        output[cursor..cursor + 4].copy_from_slice(&self.e_flags.to_le_bytes());
905        cursor += 4;
906
907        // ELF header size (little-endian u16)
908        output[cursor..cursor + 2].copy_from_slice(&52u16.to_le_bytes());
909        cursor += 2;
910
911        // Program header entry size (little-endian u16)
912        let ph_entry_size: u16 = if ph_count > 0 { 32 } else { 0 };
913        output[cursor..cursor + 2].copy_from_slice(&ph_entry_size.to_le_bytes());
914        cursor += 2;
915
916        // Program header count (little-endian u16)
917        output[cursor..cursor + 2].copy_from_slice(&ph_count.to_le_bytes());
918        cursor += 2;
919
920        // Section header entry size (little-endian u16)
921        output[cursor..cursor + 2].copy_from_slice(&40u16.to_le_bytes());
922        cursor += 2;
923
924        // Section header count (little-endian u16)
925        output[cursor..cursor + 2].copy_from_slice(&sh_count.to_le_bytes());
926        cursor += 2;
927
928        // Section header string table index (little-endian u16) - .shstrtab is section 1
929        output[cursor..cursor + 2].copy_from_slice(&1u16.to_le_bytes());
930
931        Ok(())
932    }
933
934    /// Build section name string table. Returns the bytes, the per-user-section
935    /// name offsets, and the per-extra-relocation `.rel.<name>` name offsets
936    /// (parallel to `self.extra_relocations`). The extra-rel names are appended
937    /// AFTER `.rel.text`, so when `extra_relocations` is empty the table is
938    /// byte-identical to the pre-generalization layout.
939    fn build_section_string_table(&self) -> (Vec<u8>, Vec<usize>, Vec<usize>) {
940        let mut strtab = vec![0]; // null string at offset 0
941        let mut offsets = Vec::new();
942
943        // Standard sections
944        strtab.extend_from_slice(b".shstrtab\0");
945        strtab.extend_from_slice(b".strtab\0");
946        strtab.extend_from_slice(b".symtab\0");
947
948        // User sections
949        for section in &self.sections {
950            let offset = strtab.len();
951            offsets.push(offset);
952            strtab.extend_from_slice(section.name.as_bytes());
953            strtab.push(0);
954        }
955
956        // .rel.text (if relocations exist)
957        if !self.relocations.is_empty() {
958            strtab.extend_from_slice(b".rel.text\0");
959        }
960
961        // .rel.<name> for each extra per-section relocation table.
962        let mut extra_rel_offsets = Vec::new();
963        for (target, _) in &self.extra_relocations {
964            let offset = strtab.len();
965            extra_rel_offsets.push(offset);
966            strtab.extend_from_slice(format!(".rel{target}\0").as_bytes());
967        }
968
969        (strtab, offsets, extra_rel_offsets)
970    }
971
972    /// Build symbol name string table
973    fn build_symbol_string_table(&self) -> (Vec<u8>, Vec<usize>) {
974        let mut strtab = vec![0]; // null string at offset 0
975        let mut offsets = Vec::new();
976
977        for symbol in &self.symbols {
978            let offset = strtab.len();
979            offsets.push(offset);
980            strtab.extend_from_slice(symbol.name.as_bytes());
981            strtab.push(0);
982        }
983
984        (strtab, offsets)
985    }
986
987    /// Encode a slice of relocations as ELF32 REL entries (8 bytes each). Shared
988    /// by `.rel.text` and the per-section `.rel.<name>` tables.
989    fn encode_rel_entries(relocs: &[Relocation]) -> Vec<u8> {
990        let mut rel_data = Vec::new();
991        for reloc in relocs {
992            // r_offset (4 bytes)
993            rel_data.extend_from_slice(&reloc.offset.to_le_bytes());
994            // r_info (4 bytes) = (sym_index << 8) | type
995            let r_info = (reloc.symbol_index << 8) | (reloc.reloc_type as u32);
996            rel_data.extend_from_slice(&r_info.to_le_bytes());
997        }
998        rel_data
999    }
1000
1001    /// Resolve a target section name to its ELF section index. User sections
1002    /// begin at index 4 (null=0, shstrtab=1, strtab=2, symtab=3). Returns `None`
1003    /// if no user section has that name.
1004    fn section_index_by_name(&self, name: &str) -> Option<u32> {
1005        self.sections
1006            .iter()
1007            .position(|s| s.name == name)
1008            .map(|pos| 4 + pos as u32)
1009    }
1010
1011    /// Build symbol table. `order` is the locals-first permutation of
1012    /// `self.symbols` computed in [`build`] (#656): entry `k` of the emitted
1013    /// table (after the null symbol) is `self.symbols[order[k]]`.
1014    fn build_symbol_table(&self, name_offsets: &[usize], order: &[usize]) -> Vec<u8> {
1015        let mut symtab = Vec::new();
1016
1017        // First entry is always null symbol
1018        symtab.extend_from_slice(&[0u8; 16]); // 16 bytes per symbol in ELF32
1019
1020        // User symbols, locals first (#656)
1021        for &i in order {
1022            let symbol = &self.symbols[i];
1023            let name_offset = if i < name_offsets.len() {
1024                name_offsets[i] as u32
1025            } else {
1026                0
1027            };
1028
1029            // st_name (4 bytes)
1030            symtab.extend_from_slice(&name_offset.to_le_bytes());
1031
1032            // st_value (4 bytes)
1033            // For ARM THUMB targets, STT_FUNC symbols must have bit 0 set (Thumb
1034            // interworking). #598: A32 objects (cortex-r5 path) must NOT set it —
1035            // bit 0 on an A32 function address is wrong metadata (harnesses had
1036            // to mask it; an interworking-aware consumer would mis-classify the
1037            // function as Thumb).
1038            let value = if self.machine == ElfMachine::Arm
1039                && self.thumb_funcs
1040                && symbol.symbol_type == SymbolType::Func
1041            {
1042                symbol.value | 1
1043            } else {
1044                symbol.value
1045            };
1046            symtab.extend_from_slice(&value.to_le_bytes());
1047
1048            // st_size (4 bytes)
1049            symtab.extend_from_slice(&symbol.size.to_le_bytes());
1050
1051            // st_info (1 byte) = (binding << 4) | (type & 0xf)
1052            let info = ((symbol.binding as u8) << 4) | (symbol.symbol_type as u8 & 0xf);
1053            symtab.push(info);
1054
1055            // st_other (1 byte)
1056            symtab.push(0);
1057
1058            // st_shndx (2 bytes)
1059            symtab.extend_from_slice(&symbol.section.to_le_bytes());
1060        }
1061
1062        symtab
1063    }
1064
1065    /// Build section headers (with optional .rel.text)
1066    #[allow(clippy::too_many_arguments)]
1067    fn build_section_headers_with_rel(
1068        &self,
1069        section_name_offsets: &[usize],
1070        shstrtab_offset: usize,
1071        shstrtab_data: &[u8],
1072        strtab_offset: usize,
1073        strtab_data: &[u8],
1074        symtab_offset: usize,
1075        symtab_data: &[u8],
1076        section_offsets: &[usize],
1077        rel_offset: usize,
1078        rel_data: &[u8],
1079        extra_rel: &[ExtraRelSection],
1080        // #656: `.symtab` sh_info = index of the first non-LOCAL symbol
1081        // (1 + number of local symbols; the null symbol counts as local).
1082        symtab_sh_info: u32,
1083    ) -> Vec<u8> {
1084        let mut headers = Vec::new();
1085
1086        // Section header size is 40 bytes for ELF32
1087
1088        // Section 0: null section
1089        headers.extend_from_slice(&[0u8; 40]);
1090
1091        // Section 1: .shstrtab
1092        self.write_section_header(
1093            &mut headers,
1094            1,
1095            SectionType::StrTab as u32,
1096            0,
1097            0,
1098            shstrtab_offset as u32,
1099            shstrtab_data.len() as u32,
1100            0,
1101            0,
1102            1,
1103            0,
1104        );
1105
1106        // Section 2: .strtab
1107        let strtab_name_offset = ".shstrtab\0".len();
1108        self.write_section_header(
1109            &mut headers,
1110            strtab_name_offset as u32,
1111            SectionType::StrTab as u32,
1112            0,
1113            0,
1114            strtab_offset as u32,
1115            strtab_data.len() as u32,
1116            0,
1117            0,
1118            1,
1119            0,
1120        );
1121
1122        // Section 3: .symtab (links to .strtab which is section 2)
1123        let symtab_name_offset = ".shstrtab\0.strtab\0".len();
1124        self.write_section_header(
1125            &mut headers,
1126            symtab_name_offset as u32,
1127            SectionType::SymTab as u32,
1128            0,
1129            0,
1130            symtab_offset as u32,
1131            symtab_data.len() as u32,
1132            2,
1133            // #656: was hardcoded 1 (the #430 blocker) — with local symbols
1134            // present that under-reports, and `ld` then treats every local as
1135            // global-bindable. Now the real first-non-local index.
1136            symtab_sh_info,
1137            4,
1138            16,
1139        );
1140
1141        // User sections
1142        for (i, section) in self.sections.iter().enumerate() {
1143            let name_offset = if i < section_name_offsets.len() {
1144                section_name_offsets[i] as u32
1145            } else {
1146                0
1147            };
1148            let offset = if i < section_offsets.len() {
1149                section_offsets[i] as u32
1150            } else {
1151                0
1152            };
1153
1154            self.write_section_header(
1155                &mut headers,
1156                name_offset,
1157                section.section_type as u32,
1158                section.flags,
1159                section.addr,
1160                offset,
1161                section.size(),
1162                0,
1163                0,
1164                section.align,
1165                0,
1166            );
1167        }
1168
1169        // .rel.text section (if relocations exist)
1170        if !rel_data.is_empty() {
1171            let rel_name_offset = self.rel_text_shstrtab_offset();
1172            // sh_link = symtab section index (3), sh_info = .text section index (4, first user section)
1173            let text_section_idx = 4u32; // null(0) + shstrtab(1) + strtab(2) + symtab(3) + .text(4)
1174            self.write_section_header(
1175                &mut headers,
1176                rel_name_offset as u32,
1177                SectionType::Rel as u32,
1178                0,
1179                0,
1180                rel_offset as u32,
1181                rel_data.len() as u32,
1182                3,                // sh_link = .symtab section index
1183                text_section_idx, // sh_info = section to which relocations apply
1184                4,
1185                8, // Each REL entry is 8 bytes
1186            );
1187        }
1188
1189        // Extra .rel.<name> sections (e.g. .rel.debug_line). Same shape as
1190        // .rel.text but sh_info points at the named target section.
1191        for er in extra_rel {
1192            self.write_section_header(
1193                &mut headers,
1194                er.name_offset as u32,
1195                SectionType::Rel as u32,
1196                0,
1197                0,
1198                er.offset as u32,
1199                er.data.len() as u32,
1200                3,             // sh_link = .symtab section index
1201                er.target_idx, // sh_info = relocated section
1202                4,
1203                8, // Each REL entry is 8 bytes
1204            );
1205        }
1206
1207        headers
1208    }
1209
1210    /// Compute the shstrtab offset where .rel.text name begins
1211    fn rel_text_shstrtab_offset(&self) -> usize {
1212        // Layout: \0 .shstrtab\0 .strtab\0 .symtab\0 [user sections...] .rel.text\0
1213        let mut offset = 1 + ".shstrtab\0".len() + ".strtab\0".len() + ".symtab\0".len();
1214        for section in &self.sections {
1215            offset += section.name.len() + 1;
1216        }
1217        offset
1218    }
1219
1220    /// Write a single section header
1221    #[allow(clippy::too_many_arguments)]
1222    fn write_section_header(
1223        &self,
1224        output: &mut Vec<u8>,
1225        name: u32,
1226        sh_type: u32,
1227        flags: u32,
1228        addr: u32,
1229        offset: u32,
1230        size: u32,
1231        link: u32,
1232        info: u32,
1233        align: u32,
1234        entsize: u32,
1235    ) {
1236        output.extend_from_slice(&name.to_le_bytes());
1237        output.extend_from_slice(&sh_type.to_le_bytes());
1238        output.extend_from_slice(&flags.to_le_bytes());
1239        output.extend_from_slice(&addr.to_le_bytes());
1240        output.extend_from_slice(&offset.to_le_bytes());
1241        output.extend_from_slice(&size.to_le_bytes());
1242        output.extend_from_slice(&link.to_le_bytes());
1243        output.extend_from_slice(&info.to_le_bytes());
1244        output.extend_from_slice(&align.to_le_bytes());
1245        output.extend_from_slice(&entsize.to_le_bytes());
1246    }
1247
1248    /// Write ELF header (legacy method for tests)
1249    #[allow(dead_code)]
1250    fn write_elf_header(&self, output: &mut Vec<u8>) -> Result<()> {
1251        // ELF magic number
1252        output.extend_from_slice(&[0x7f, b'E', b'L', b'F']);
1253
1254        // Class (32-bit)
1255        output.push(self.class as u8);
1256
1257        // Data (little-endian)
1258        output.push(self.data as u8);
1259
1260        // Version
1261        output.push(1);
1262
1263        // OS/ABI
1264        output.push(0); // System V
1265
1266        // ABI version
1267        output.push(0);
1268
1269        // Padding
1270        output.extend_from_slice(&[0; 7]);
1271
1272        // Type (little-endian u16)
1273        let etype = self.elf_type as u16;
1274        output.extend_from_slice(&etype.to_le_bytes());
1275
1276        // Machine (little-endian u16)
1277        let machine = self.machine as u16;
1278        output.extend_from_slice(&machine.to_le_bytes());
1279
1280        // Version (little-endian u32)
1281        output.extend_from_slice(&1u32.to_le_bytes());
1282
1283        // Entry point (little-endian u32)
1284        output.extend_from_slice(&self.entry.to_le_bytes());
1285
1286        // Program header offset (little-endian u32)
1287        output.extend_from_slice(&0u32.to_le_bytes());
1288
1289        // Section header offset (little-endian u32)
1290        output.extend_from_slice(&0u32.to_le_bytes());
1291
1292        // Flags (little-endian u32)
1293        output.extend_from_slice(&0u32.to_le_bytes());
1294
1295        // ELF header size (little-endian u16)
1296        output.extend_from_slice(&52u16.to_le_bytes());
1297
1298        // Program header entry size (little-endian u16)
1299        output.extend_from_slice(&0u16.to_le_bytes());
1300
1301        // Program header count (little-endian u16)
1302        output.extend_from_slice(&0u16.to_le_bytes());
1303
1304        // Section header entry size (little-endian u16)
1305        output.extend_from_slice(&40u16.to_le_bytes());
1306
1307        // Section header count (little-endian u16)
1308        output.extend_from_slice(&0u16.to_le_bytes());
1309
1310        // Section header string table index (little-endian u16)
1311        output.extend_from_slice(&0u16.to_le_bytes());
1312
1313        Ok(())
1314    }
1315}
1316
1317#[cfg(test)]
1318mod tests {
1319    use super::*;
1320
1321    #[test]
1322    fn test_elf_builder_creation() {
1323        let builder = ElfBuilder::new_arm32();
1324        assert_eq!(builder.class, ElfClass::Elf32);
1325        assert_eq!(builder.data, ElfData::LittleEndian);
1326        assert_eq!(builder.machine, ElfMachine::Arm);
1327    }
1328
1329    #[test]
1330    fn test_section_creation() {
1331        let section = Section::new(".text", SectionType::ProgBits)
1332            .with_flags(SectionFlags::ALLOC | SectionFlags::EXEC)
1333            .with_addr(0x8000)
1334            .with_align(4);
1335
1336        assert_eq!(section.name, ".text");
1337        assert_eq!(section.section_type, SectionType::ProgBits);
1338        assert_eq!(section.addr, 0x8000);
1339        assert_eq!(section.align, 4);
1340    }
1341
1342    #[test]
1343    fn test_symbol_creation() {
1344        let symbol = Symbol::new("main")
1345            .with_value(0x8000)
1346            .with_size(128)
1347            .with_binding(SymbolBinding::Global)
1348            .with_type(SymbolType::Func)
1349            .with_section(1);
1350
1351        assert_eq!(symbol.name, "main");
1352        assert_eq!(symbol.value, 0x8000);
1353        assert_eq!(symbol.size, 128);
1354        assert_eq!(symbol.binding, SymbolBinding::Global);
1355        assert_eq!(symbol.symbol_type, SymbolType::Func);
1356    }
1357
1358    #[test]
1359    fn test_elf_header_generation() {
1360        let builder = ElfBuilder::new_arm32().with_entry(0x8000);
1361        let elf = builder.build().unwrap();
1362
1363        // Check magic number
1364        assert_eq!(&elf[0..4], &[0x7f, b'E', b'L', b'F']);
1365
1366        // Check class (32-bit)
1367        assert_eq!(elf[4], 1);
1368
1369        // Check data (little-endian)
1370        assert_eq!(elf[5], 1);
1371
1372        // Check version
1373        assert_eq!(elf[6], 1);
1374    }
1375
1376    #[test]
1377    fn test_add_sections() {
1378        let mut builder = ElfBuilder::new_arm32();
1379
1380        let text = Section::new(".text", SectionType::ProgBits)
1381            .with_flags(SectionFlags::ALLOC | SectionFlags::EXEC);
1382
1383        let data = Section::new(".data", SectionType::ProgBits)
1384            .with_flags(SectionFlags::ALLOC | SectionFlags::WRITE);
1385
1386        builder.add_section(text);
1387        builder.add_section(data);
1388
1389        assert_eq!(builder.sections.len(), 2);
1390    }
1391
1392    #[test]
1393    fn test_add_symbols() {
1394        let mut builder = ElfBuilder::new_arm32();
1395
1396        let main_sym = Symbol::new("main")
1397            .with_binding(SymbolBinding::Global)
1398            .with_type(SymbolType::Func);
1399
1400        let data_sym = Symbol::new("data")
1401            .with_binding(SymbolBinding::Local)
1402            .with_type(SymbolType::Object);
1403
1404        builder.add_symbol(main_sym);
1405        builder.add_symbol(data_sym);
1406
1407        assert_eq!(builder.symbols.len(), 2);
1408    }
1409
1410    #[test]
1411    fn test_complete_elf_generation() {
1412        // Create a complete ELF file with sections and symbols
1413        let mut builder = ElfBuilder::new_arm32()
1414            .with_entry(0x8000)
1415            .with_type(ElfType::Exec);
1416
1417        // Add .text section with some ARM code
1418        let text_code = vec![
1419            0x00, 0x48, 0x2d, 0xe9, // push {fp, lr}
1420            0x04, 0xb0, 0x8d, 0xe2, // add fp, sp, #4
1421            0x00, 0x00, 0xa0, 0xe3, // mov r0, #0
1422            0x00, 0x88, 0xbd, 0xe8, // pop {fp, pc}
1423        ];
1424        let text = Section::new(".text", SectionType::ProgBits)
1425            .with_flags(SectionFlags::ALLOC | SectionFlags::EXEC)
1426            .with_addr(0x8000)
1427            .with_align(4)
1428            .with_data(text_code);
1429
1430        builder.add_section(text);
1431
1432        // Add .data section
1433        let data_content = vec![0x01, 0x02, 0x03, 0x04];
1434        let data = Section::new(".data", SectionType::ProgBits)
1435            .with_flags(SectionFlags::ALLOC | SectionFlags::WRITE)
1436            .with_addr(0x8100)
1437            .with_align(4)
1438            .with_data(data_content);
1439
1440        builder.add_section(data);
1441
1442        // Add .bss section (no data)
1443        let bss = Section::new(".bss", SectionType::NoBits)
1444            .with_flags(SectionFlags::ALLOC | SectionFlags::WRITE)
1445            .with_addr(0x8200)
1446            .with_align(4);
1447
1448        builder.add_section(bss);
1449
1450        // Add symbols
1451        let main_sym = Symbol::new("main")
1452            .with_value(0x8000)
1453            .with_size(16)
1454            .with_binding(SymbolBinding::Global)
1455            .with_type(SymbolType::Func)
1456            .with_section(4); // .text is section 4 (0=null, 1=shstrtab, 2=strtab, 3=symtab, 4=.text)
1457
1458        builder.add_symbol(main_sym);
1459
1460        let data_var = Symbol::new("global_var")
1461            .with_value(0x8100)
1462            .with_size(4)
1463            .with_binding(SymbolBinding::Global)
1464            .with_type(SymbolType::Object)
1465            .with_section(5); // .data is section 5
1466
1467        builder.add_symbol(data_var);
1468
1469        // Build the ELF file
1470        let elf = builder.build().unwrap();
1471
1472        // Validate ELF header
1473        assert_eq!(&elf[0..4], &[0x7f, b'E', b'L', b'F']);
1474        assert_eq!(elf[4], 1); // 32-bit
1475        assert_eq!(elf[5], 1); // little-endian
1476        assert_eq!(elf[6], 1); // version
1477
1478        // Check that we have a reasonable file size
1479        assert!(elf.len() > 52); // At least header size
1480        assert!(elf.len() < 10000); // Reasonable upper bound
1481
1482        // Validate entry point is set correctly (Thumb bit set for ARM)
1483        let entry_bytes = &elf[24..28];
1484        let entry = u32::from_le_bytes([
1485            entry_bytes[0],
1486            entry_bytes[1],
1487            entry_bytes[2],
1488            entry_bytes[3],
1489        ]);
1490        assert_eq!(entry, 0x8001); // 0x8000 | 1 (Thumb bit)
1491
1492        // Validate section header offset is non-zero
1493        let sh_off_bytes = &elf[32..36];
1494        let sh_off = u32::from_le_bytes([
1495            sh_off_bytes[0],
1496            sh_off_bytes[1],
1497            sh_off_bytes[2],
1498            sh_off_bytes[3],
1499        ]);
1500        assert!(sh_off > 0);
1501
1502        // Validate section count (null + shstrtab + strtab + symtab + .text + .data + .bss = 7)
1503        let sh_num_bytes = &elf[48..50];
1504        let sh_num = u16::from_le_bytes([sh_num_bytes[0], sh_num_bytes[1]]);
1505        assert_eq!(sh_num, 7);
1506
1507        // Validate string table index points to .shstrtab (section 1)
1508        let shstrndx_bytes = &elf[50..52];
1509        let shstrndx = u16::from_le_bytes([shstrndx_bytes[0], shstrndx_bytes[1]]);
1510        assert_eq!(shstrndx, 1);
1511    }
1512
1513    #[test]
1514    fn test_string_table_generation() {
1515        let mut builder = ElfBuilder::new_arm32();
1516
1517        builder.add_section(Section::new(".text", SectionType::ProgBits));
1518        builder.add_section(Section::new(".data", SectionType::ProgBits));
1519
1520        let (strtab, offsets, _extra_rel_offsets) = builder.build_section_string_table();
1521
1522        // Should have null byte at start
1523        assert_eq!(strtab[0], 0);
1524
1525        // Should contain .shstrtab, .strtab, .symtab, .text, .data
1526        let strtab_str = String::from_utf8_lossy(&strtab);
1527        assert!(strtab_str.contains(".shstrtab"));
1528        assert!(strtab_str.contains(".strtab"));
1529        assert!(strtab_str.contains(".symtab"));
1530        assert!(strtab_str.contains(".text"));
1531        assert!(strtab_str.contains(".data"));
1532
1533        // Should have offsets for user sections
1534        assert_eq!(offsets.len(), 2);
1535    }
1536
1537    #[test]
1538    fn test_relocation_support() {
1539        let mut builder = ElfBuilder::new_arm32()
1540            .with_entry(0x8000)
1541            .with_type(ElfType::Rel);
1542
1543        // Add .text section with a BL placeholder
1544        let text_code = vec![0x00u8; 16]; // 4 instructions of placeholder
1545        let text = Section::new(".text", SectionType::ProgBits)
1546            .with_flags(SectionFlags::ALLOC | SectionFlags::EXEC)
1547            .with_addr(0x8000)
1548            .with_align(4)
1549            .with_data(text_code);
1550        builder.add_section(text);
1551
1552        // Add undefined external symbol
1553        let sym_idx = builder.add_undefined_symbol("__meld_dispatch_import");
1554        assert!(sym_idx > 0);
1555
1556        // Add relocation for the BL at offset 4
1557        builder.add_relocation(Relocation {
1558            offset: 4,
1559            symbol_index: sym_idx,
1560            reloc_type: ArmRelocationType::Call,
1561        });
1562
1563        let elf = builder.build().unwrap();
1564
1565        // Verify ELF is valid
1566        assert_eq!(&elf[0..4], &[0x7f, b'E', b'L', b'F']);
1567
1568        // Section count should include .rel.text
1569        // null(1) + shstrtab(1) + strtab(1) + symtab(1) + .text(1) + .rel.text(1) = 6
1570        let sh_num = u16::from_le_bytes([elf[48], elf[49]]);
1571        assert_eq!(sh_num, 6);
1572
1573        // Verify the symbol table contains the undefined symbol
1574        // (section = 0 for SHN_UNDEF)
1575        let has_undef = elf
1576            .windows(b"__meld_dispatch_import".len())
1577            .any(|w| w == b"__meld_dispatch_import");
1578        assert!(
1579            has_undef,
1580            "ELF should contain __meld_dispatch_import symbol name"
1581        );
1582    }
1583
1584    #[test]
1585    fn test_symbol_table_encoding() {
1586        let mut builder = ElfBuilder::new_arm32();
1587
1588        let sym = Symbol::new("test_func")
1589            .with_value(0x1000)
1590            .with_size(64)
1591            .with_binding(SymbolBinding::Global)
1592            .with_type(SymbolType::Func)
1593            .with_section(1);
1594
1595        builder.add_symbol(sym);
1596
1597        let (_strtab, offsets) = builder.build_symbol_string_table();
1598        let symtab = builder.build_symbol_table(&offsets, &[0]);
1599
1600        // Should have null symbol (16 bytes) + 1 symbol (16 bytes) = 32 bytes
1601        assert_eq!(symtab.len(), 32);
1602
1603        // First symbol should be all zeros
1604        assert!(symtab[0..16].iter().all(|&b| b == 0));
1605
1606        // Second symbol should have correct encoding
1607        // Check st_value (bytes 4-7 of second entry)
1608        // For ARM STT_FUNC symbols, bit 0 is set for Thumb interworking
1609        let value_bytes = &symtab[20..24];
1610        let value = u32::from_le_bytes([
1611            value_bytes[0],
1612            value_bytes[1],
1613            value_bytes[2],
1614            value_bytes[3],
1615        ]);
1616        assert_eq!(value, 0x1001); // 0x1000 | 1 (Thumb bit)
1617
1618        // Check st_size (bytes 8-11 of second entry)
1619        let size_bytes = &symtab[24..28];
1620        let size = u32::from_le_bytes([size_bytes[0], size_bytes[1], size_bytes[2], size_bytes[3]]);
1621        assert_eq!(size, 64);
1622
1623        // Check st_info (byte 12 of second entry)
1624        let info = symtab[28];
1625        let binding = info >> 4;
1626        let sym_type = info & 0xf;
1627        assert_eq!(binding, SymbolBinding::Global as u8);
1628        assert_eq!(sym_type, SymbolType::Func as u8);
1629    }
1630
1631    /// #598: an A32 object (`with_thumb_funcs(false)`, cortex-r5 path) must
1632    /// NOT set the Thumb interworking bit on STT_FUNC symbols or `e_entry` —
1633    /// bit 0 on an A32 code address is wrong metadata (harnesses had to mask
1634    /// it). Non-func symbols never carried the bit; that stays true.
1635    #[test]
1636    fn test_a32_symbols_have_no_thumb_bit_598() {
1637        let mut builder = ElfBuilder::new_arm32().with_thumb_funcs(false);
1638
1639        let func_sym = Symbol::new("a32_func")
1640            .with_value(0x1000)
1641            .with_size(64)
1642            .with_binding(SymbolBinding::Global)
1643            .with_type(SymbolType::Func)
1644            .with_section(1);
1645        builder.add_symbol(func_sym);
1646
1647        let (_strtab, offsets) = builder.build_symbol_string_table();
1648        let symtab = builder.build_symbol_table(&offsets, &[0]);
1649        let value = u32::from_le_bytes(symtab[20..24].try_into().unwrap());
1650        assert_eq!(value, 0x1000, "A32 STT_FUNC st_value must keep bit 0 clear");
1651
1652        // e_entry stays clear too (was `0 | 1` on the A32 relocatable path).
1653        let builder = ElfBuilder::new_arm32()
1654            .with_thumb_funcs(false)
1655            .with_entry(0x8000);
1656        assert_eq!(builder.entry, 0x8000, "A32 e_entry must keep bit 0 clear");
1657
1658        // The default (Thumb) behavior is unchanged: bit 0 set on both.
1659        let builder = ElfBuilder::new_arm32().with_entry(0x8000);
1660        assert_eq!(builder.entry, 0x8001, "Thumb e_entry keeps the bit");
1661    }
1662
1663    /// Minimal ELF32 section-header reader for the tests below:
1664    /// (sh_type, sh_offset, sh_size, sh_link, sh_info) per section.
1665    fn read_section_headers(elf: &[u8]) -> Vec<(u32, u32, u32, u32, u32)> {
1666        let e_shoff = u32::from_le_bytes(elf[32..36].try_into().unwrap()) as usize;
1667        let e_shnum = u16::from_le_bytes(elf[48..50].try_into().unwrap()) as usize;
1668        (0..e_shnum)
1669            .map(|i| {
1670                let base = e_shoff + i * 40;
1671                let f = |off: usize| {
1672                    u32::from_le_bytes(elf[base + off..base + off + 4].try_into().unwrap())
1673                };
1674                (f(4), f(16), f(20), f(24), f(28))
1675            })
1676            .collect()
1677    }
1678
1679    /// Read symtab entries as (st_value, st_info, st_shndx).
1680    fn read_symtab(elf: &[u8]) -> (Vec<(u32, u8, u16)>, u32) {
1681        let headers = read_section_headers(elf);
1682        let &(_, off, size, _, sh_info) = headers
1683            .iter()
1684            .find(|h| h.0 == SectionType::SymTab as u32)
1685            .expect("symtab present");
1686        let syms = (0..size as usize / 16)
1687            .map(|i| {
1688                let base = off as usize + i * 16;
1689                (
1690                    u32::from_le_bytes(elf[base + 4..base + 8].try_into().unwrap()),
1691                    elf[base + 12],
1692                    u16::from_le_bytes(elf[base + 14..base + 16].try_into().unwrap()),
1693                )
1694            })
1695            .collect();
1696        (syms, sh_info)
1697    }
1698
1699    /// #656: local symbols must be emitted BEFORE globals (stable within each
1700    /// class), `.symtab` `sh_info` must be the first-non-local index, and every
1701    /// relocation's symbol index must be rewritten through the permutation.
1702    #[test]
1703    fn test_locals_sorted_first_sh_info_and_reloc_reindex_656() {
1704        let mut builder = ElfBuilder::new_arm32()
1705            .with_entry(0)
1706            .with_type(ElfType::Rel);
1707        let text = Section::new(".text", SectionType::ProgBits)
1708            .with_flags(SectionFlags::ALLOC | SectionFlags::EXEC)
1709            .with_align(4)
1710            .with_data(vec![0u8; 16]);
1711        builder.add_section(text);
1712
1713        // Added out of ELF order: global export first, then a LOCAL internal
1714        // helper, then a global undefined external.
1715        builder.add_symbol(
1716            Symbol::new("exported")
1717                .with_value(0)
1718                .with_binding(SymbolBinding::Global)
1719                .with_type(SymbolType::Func)
1720                .with_section(4),
1721        ); // pre-build index 1
1722        builder.add_symbol(
1723            Symbol::new("func_2")
1724                .with_value(8)
1725                .with_binding(SymbolBinding::Local)
1726                .with_type(SymbolType::Func)
1727                .with_section(4),
1728        ); // pre-build index 2
1729        let undef_idx = builder.add_undefined_symbol("external"); // pre-build index 3
1730        assert_eq!(undef_idx, 3);
1731
1732        // BL at offset 0 → the LOCAL func_2 (pre-build index 2);
1733        // BL at offset 4 → the undefined external (pre-build index 3).
1734        builder.add_relocation(Relocation {
1735            offset: 0,
1736            symbol_index: 2,
1737            reloc_type: ArmRelocationType::ThmCall,
1738        });
1739        builder.add_relocation(Relocation {
1740            offset: 4,
1741            symbol_index: undef_idx,
1742            reloc_type: ArmRelocationType::ThmCall,
1743        });
1744
1745        let elf = builder.build().unwrap();
1746        let (syms, sh_info) = read_symtab(&elf);
1747
1748        // Order: null, func_2 (LOCAL), exported (GLOBAL), external (GLOBAL undef).
1749        assert_eq!(syms.len(), 4);
1750        assert_eq!(syms[0], (0, 0, 0), "null symbol first");
1751        let bind = |info: u8| info >> 4;
1752        assert_eq!(bind(syms[1].1), SymbolBinding::Local as u8, "local first");
1753        assert_eq!(syms[1].0, 8 | 1, "func_2 st_value (thumb bit)");
1754        assert_eq!(bind(syms[2].1), SymbolBinding::Global as u8);
1755        assert_eq!(syms[2].0, 1, "exported st_value 0 | thumb bit");
1756        assert_eq!(bind(syms[3].1), SymbolBinding::Global as u8);
1757        assert_eq!(syms[3].2, 0, "external is SHN_UNDEF");
1758        assert_eq!(sh_info, 2, "sh_info = index of first non-local symbol");
1759
1760        // Relocations rewritten: func_2 is now index 1, external index 3.
1761        let headers = read_section_headers(&elf);
1762        let &(_, rel_off, rel_size, _, rel_info) = headers
1763            .iter()
1764            .find(|h| h.0 == SectionType::Rel as u32)
1765            .expect(".rel.text present");
1766        assert_eq!(rel_info, 4, ".rel.text still targets .text");
1767        assert_eq!(rel_size, 16);
1768        let r_info = |i: usize| {
1769            u32::from_le_bytes(
1770                elf[rel_off as usize + i * 8 + 4..rel_off as usize + i * 8 + 8]
1771                    .try_into()
1772                    .unwrap(),
1773            )
1774        };
1775        assert_eq!(r_info(0) >> 8, 1, "BL func_2 reloc remapped to new index 1");
1776        assert_eq!(r_info(0) & 0xff, ArmRelocationType::ThmCall as u32);
1777        assert_eq!(r_info(1) >> 8, 3, "BL external reloc keeps index 3");
1778    }
1779
1780    /// #656 freeze guard: with zero LOCAL symbols the permutation is the
1781    /// identity and `sh_info` stays 1 — the pre-#656 layout, byte-identical.
1782    #[test]
1783    fn test_all_global_symtab_unchanged_sh_info_1_656() {
1784        let mut builder = ElfBuilder::new_arm32()
1785            .with_entry(0)
1786            .with_type(ElfType::Rel);
1787        builder.add_section(
1788            Section::new(".text", SectionType::ProgBits)
1789                .with_flags(SectionFlags::ALLOC | SectionFlags::EXEC)
1790                .with_data(vec![0u8; 8]),
1791        );
1792        for (name, val) in [("a", 0u32), ("b", 4u32)] {
1793            builder.add_symbol(
1794                Symbol::new(name)
1795                    .with_value(val)
1796                    .with_binding(SymbolBinding::Global)
1797                    .with_type(SymbolType::Func)
1798                    .with_section(4),
1799            );
1800        }
1801        let elf = builder.build().unwrap();
1802        let (syms, sh_info) = read_symtab(&elf);
1803        assert_eq!(sh_info, 1, "no locals ⇒ sh_info stays 1 (pre-#656 layout)");
1804        assert_eq!(syms[1].0, 1, "a first (insertion order preserved)");
1805        assert_eq!(syms[2].0, 5, "b second");
1806    }
1807
1808    /// #637: `.ARM.attributes` blob structure — format version 'A', "aeabi"
1809    /// vendor subsection, Tag_File subsubsection with the uleb tag pairs, and
1810    /// zero-valued tags omitted (spec default).
1811    #[test]
1812    fn test_arm_attributes_section_bytes_637() {
1813        // Cortex-M3: v7, profile M, no A32, Thumb-2.
1814        let sec = arm_attributes_section(aeabi::CPU_ARCH_V7, aeabi::PROFILE_M, 0, 2, 0, 0);
1815        assert_eq!(sec.name, ".ARM.attributes");
1816        assert_eq!(sec.section_type, SectionType::ArmAttributes);
1817        let d = &sec.data;
1818        assert_eq!(d[0], b'A', "format version");
1819        let vendor_len = u32::from_le_bytes(d[1..5].try_into().unwrap()) as usize;
1820        assert_eq!(vendor_len, d.len() - 1, "vendor subsection length");
1821        assert_eq!(&d[5..11], b"aeabi\0");
1822        assert_eq!(d[11], 1, "Tag_File");
1823        let file_len = u32::from_le_bytes(d[12..16].try_into().unwrap()) as usize;
1824        assert_eq!(file_len, d.len() - 11, "Tag_File length");
1825        // Attribute pairs (all values < 128 ⇒ one uleb byte each).
1826        let attrs = &d[16..];
1827        assert_eq!(
1828            attrs,
1829            &[
1830                6, 10, // Tag_CPU_arch = v7
1831                7, b'M', // Tag_CPU_arch_profile = M
1832                9, 2, // Tag_THUMB_ISA_use = Thumb-2 (Tag_ARM_ISA_use=0 omitted)
1833            ],
1834        );
1835
1836        // Cortex-R5: v7, profile R, A32 permitted, Thumb-2 permitted.
1837        let sec = arm_attributes_section(aeabi::CPU_ARCH_V7, aeabi::PROFILE_R, 1, 2, 0, 0);
1838        assert_eq!(&sec.data[16..], &[6, 10, 7, b'R', 8, 1, 9, 2]);
1839
1840        // GI-FPU-002: a hard-float FPU target adds Tag_FP_arch(10)=VFPv4-D16(6)
1841        // and Tag_ABI_VFP_args(28)=1, in ascending tag order.
1842        let sec = arm_attributes_section(
1843            aeabi::CPU_ARCH_V7EM,
1844            aeabi::PROFILE_M,
1845            0,
1846            2,
1847            aeabi::FP_ARCH_VFPV4_D16,
1848            aeabi::VFP_ARGS_VFP_REGS,
1849        );
1850        assert_eq!(&sec.data[16..], &[6, 13, 7, b'M', 9, 2, 10, 6, 28, 1]);
1851    }
1852}
synth_backend/elf_builder.rs

synth_backend/
elf_builder.rs
