elf: move ownership of symbols into objects

Author: kubkon

src/Elf.zig

@@ -11,14 +11,14 @@ file: File.Index = 0,
 
 /// Atom containing this symbol if any.
 /// Use `getAtom` to get the pointer to the atom.
-atom_ref: Elf.Ref = .{},
+ref: Elf.Ref = .{},
 
 /// Assigned output section index for this symbol.
 shndx: u32 = 0,
 
 /// Index of the source symbol this symbol references.
-/// Use `getSourceSymbol` to pull the source symbol from the relevant file.
-sym_idx: Index = 0,
+/// Use `getElfSym` to pull the source symbol from the relevant file.
+esym_idx: Index = 0,
 
 /// Index of the source version symbol this symbol references if any.
 /// If the symbol is unversioned it will have either VER_NDX_LOCAL or VER_NDX_GLOBAL.
@@ -31,12 +31,12 @@ extra: u32 = 0,
 
 pub fn isAbs(symbol: Symbol, elf_file: *Elf) bool {
     const file = symbol.getFile(elf_file).?;
-    if (file == .shared) return symbol.getSourceSymbol(elf_file).st_shndx == elf.SHN_ABS;
+    if (file == .shared) return symbol.getElfSym(elf_file).st_shndx == elf.SHN_ABS;
     return !symbol.flags.import and symbol.getAtom(elf_file) == null and symbol.getMergeSubsection(elf_file) == null and symbol.shndx == 0 and file != .internal;
 }
 
 pub fn isLocal(symbol: Symbol, elf_file: *Elf) bool {
-    if (elf_file.options.relocatable) return symbol.getSourceSymbol(elf_file).st_bind() == elf.STB_LOCAL;
+    if (elf_file.options.relocatable) return symbol.getElfSym(elf_file).st_bind() == elf.STB_LOCAL;
     return !(symbol.flags.import or symbol.flags.@"export");
 }
 
@@ -46,39 +46,42 @@ pub inline fn isIFunc(symbol: Symbol, elf_file: *Elf) bool {
 
 pub fn getType(symbol: Symbol, elf_file: *Elf) u4 {
     const file = symbol.getFile(elf_file).?;
-    const s_sym = symbol.getSourceSymbol(elf_file);
+    const s_sym = symbol.getElfSym(elf_file);
     if (s_sym.st_type() == elf.STT_GNU_IFUNC and file == .shared) return elf.STT_FUNC;
     return s_sym.st_type();
 }
 
 pub fn getName(symbol: Symbol, elf_file: *Elf) [:0]const u8 {
-    return elf_file.string_intern.getAssumeExists(symbol.name);
+    return switch (symbol.getFile(elf_file).?) {
+        inline else => |x| x.getString(symbol.name),
+    };
 }
 
 pub fn getAtom(symbol: Symbol, elf_file: *Elf) ?*Atom {
-    return elf_file.getAtom(symbol.atom_ref);
+    if (symbol.flags.merge_subsection) return null;
+    return elf_file.getAtom(symbol.ref);
 }
 
 pub fn getMergeSubsection(symbol: Symbol, elf_file: *Elf) ?*MergeSubsection {
     if (!symbol.flags.merge_subsection) return null;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     return elf_file.getMergeSubsection(extra.subsection);
 }
 
-pub inline fn getFile(symbol: Symbol, elf_file: *Elf) ?File {
+pub fn getFile(symbol: Symbol, elf_file: *Elf) ?File {
     return elf_file.getFile(symbol.file);
 }
 
-pub fn getSourceSymbol(symbol: Symbol, elf_file: *Elf) elf.Elf64_Sym {
+pub fn getElfSym(symbol: Symbol, elf_file: *Elf) elf.Elf64_Sym {
     const file = symbol.getFile(elf_file).?;
     return switch (file) {
-        inline else => |x| x.symtab.items[symbol.sym_idx],
+        inline else => |x| x.symtab.items[symbol.esym_idx],
     };
 }
 
 pub fn getSymbolRank(symbol: Symbol, elf_file: *Elf) u32 {
     const file = symbol.getFile(elf_file) orelse return std.math.maxInt(u32);
-    const sym = symbol.getSourceSymbol(elf_file);
+    const sym = symbol.getElfSym(elf_file);
     const in_archive = switch (file) {
         .object => |x| !x.alive,
         else => false,
@@ -116,7 +119,7 @@ pub fn getAddress(symbol: Symbol, opts: struct {
                 if (mem.startsWith(u8, sym_name, "__EH_FRAME_BEGIN__") or
                     mem.startsWith(u8, sym_name, "__EH_FRAME_LIST__") or
                     mem.startsWith(u8, sym_name, ".eh_frame_seg") or
-                    symbol.getSourceSymbol(elf_file).st_type() == elf.STT_SECTION)
+                    symbol.getElfSym(elf_file).st_type() == elf.STT_SECTION)
                 {
                     return @intCast(sh_addr);
                 }
@@ -143,36 +146,36 @@ pub fn getOutputSymtabIndex(symbol: Symbol, elf_file: *Elf) ?u32 {
     const symtab_ctx = switch (file) {
         inline else => |x| x.output_symtab_ctx,
     };
-    const idx = symbol.getExtra(elf_file).?.symtab;
+    const idx = symbol.getExtra(elf_file).symtab;
     return if (symbol.isLocal(elf_file)) idx + symtab_ctx.ilocal else idx + symtab_ctx.iglobal;
 }
 
 pub fn getGotAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.got) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const entry = elf_file.got.entries.items[extra.got];
     return entry.getAddress(elf_file);
 }
 
 pub fn getPltGotAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!(symbol.flags.plt and symbol.flags.got)) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const shdr = elf_file.sections.items(.shdr)[elf_file.plt_got_sect_index.?];
     const cpu_arch = elf_file.options.cpu_arch.?;
     return @intCast(shdr.sh_addr + extra.plt_got * PltGotSection.entrySize(cpu_arch));
 }
 
 pub fn getPltAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.plt) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const shdr = elf_file.sections.items(.shdr)[elf_file.plt_sect_index.?];
     const cpu_arch = elf_file.options.cpu_arch.?;
     return @intCast(shdr.sh_addr + extra.plt * PltSection.entrySize(cpu_arch) + PltSection.preambleSize(cpu_arch));
 }
 
 pub fn getGotPltAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.plt) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const shdr = elf_file.sections.items(.shdr)[elf_file.got_plt_sect_index.?];
     return @intCast(shdr.sh_addr + extra.plt * 8 + GotPltSection.preamble_size);
 }
@@ -185,29 +188,29 @@ pub fn getCopyRelAddress(symbol: Symbol, elf_file: *Elf) i64 {
 
 pub fn getTlsGdAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.tlsgd) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const entry = elf_file.got.entries.items[extra.tlsgd];
     return entry.getAddress(elf_file);
 }
 
 pub fn getGotTpAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.gottp) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const entry = elf_file.got.entries.items[extra.gottp];
     return entry.getAddress(elf_file);
 }
 
 pub fn getTlsDescAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.tlsdesc) return 0;
-    const extra = symbol.getExtra(elf_file).?;
+    const extra = symbol.getExtra(elf_file);
     const entry = elf_file.got.entries.items[extra.tlsdesc];
     return entry.getAddress(elf_file);
 }
 
 pub fn getAlignment(symbol: Symbol, elf_file: *Elf) !u64 {
     const file = symbol.getFile(elf_file) orelse return 0;
     const shared = file.shared;
-    const s_sym = symbol.getSourceSymbol(elf_file);
+    const s_sym = symbol.getElfSym(elf_file);
     const shdr = shared.shdrs.items[s_sym.st_shndx];
     const alignment = @max(1, shdr.sh_addralign);
     return if (s_sym.st_value == 0)
@@ -229,11 +232,8 @@ const AddExtraOpts = struct {
     subsection: ?u32 = null,
 };
 
-pub fn addExtra(symbol: *Symbol, opts: AddExtraOpts, elf_file: *Elf) !void {
-    if (symbol.getExtra(elf_file) == null) {
-        symbol.extra = try elf_file.addSymbolExtra(.{});
-    }
-    var extra = symbol.getExtra(elf_file).?;
+pub fn addExtra(symbol: *Symbol, opts: AddExtraOpts, elf_file: *Elf) void {
+    var extra = symbol.getExtra(elf_file);
     inline for (@typeInfo(@TypeOf(opts)).@"struct".fields) |field| {
         if (@field(opts, field.name)) |x| {
             @field(extra, field.name) = x;
@@ -242,17 +242,21 @@ pub fn addExtra(symbol: *Symbol, opts: AddExtraOpts, elf_file: *Elf) !void {
     symbol.setExtra(extra, elf_file);
 }
 
-pub inline fn getExtra(symbol: Symbol, elf_file: *Elf) ?Extra {
-    return elf_file.getSymbolExtra(symbol.extra);
+pub fn getExtra(symbol: Symbol, elf_file: *Elf) Extra {
+    return switch (symbol.getFile(elf_file).?) {
+        inline else => |x| x.getSymbolExtra(symbol.extra),
+    };
 }
 
-pub inline fn setExtra(symbol: Symbol, extra: Extra, elf_file: *Elf) void {
-    elf_file.setSymbolExtra(symbol.extra, extra);
+pub fn setExtra(symbol: Symbol, extra: Extra, elf_file: *Elf) void {
+    return switch (symbol.getFile(elf_file).?) {
+        inline else => |x| x.setSymbolExtra(symbol.extra, extra),
+    };
 }
 
 pub fn setOutputSym(symbol: Symbol, elf_file: *Elf, out: *elf.Elf64_Sym) void {
     const file = symbol.getFile(elf_file).?;
-    const s_sym = symbol.getSourceSymbol(elf_file);
+    const s_sym = symbol.getElfSym(elf_file);
     const st_type = symbol.getType(elf_file);
     const st_bind: u8 = blk: {
         if (symbol.isLocal(elf_file)) break :blk 0;
@@ -347,10 +351,10 @@ fn format2(
     _ = options;
     _ = unused_fmt_string;
     const symbol = ctx.symbol;
-    try writer.print("%{d} : {s} : @{x}", .{ symbol.sym_idx, symbol.fmtName(ctx.elf_file), symbol.getAddress(.{}, ctx.elf_file) });
+    try writer.print("%{d} : {s} : @{x}", .{ symbol.esym_idx, symbol.fmtName(ctx.elf_file), symbol.getAddress(.{}, ctx.elf_file) });
     if (symbol.getFile(ctx.elf_file)) |file| {
         if (symbol.isAbs(ctx.elf_file)) {
-            if (symbol.getSourceSymbol(ctx.elf_file).st_shndx == elf.SHN_UNDEF) {
+            if (symbol.getElfSym(ctx.elf_file).st_shndx == elf.SHN_UNDEF) {
                 try writer.writeAll(" : undef");
             } else {
                 try writer.writeAll(" : absolute");

src/Elf/Atom.zig

@@ -1,6 +1,6 @@
 value: i64 = 0,
 out_shndx: u32 = 0,
-symbols: std.AutoArrayHashMapUnmanaged(Symbol.Index, void) = .{},
+symbols: std.AutoArrayHashMapUnmanaged(Elf.Ref, void) = .empty,
 output_symtab_ctx: Elf.SymtabCtx = .{},
 
 pub fn deinit(thunk: *Thunk, allocator: Allocator) void {
@@ -17,9 +17,9 @@ pub fn getAddress(thunk: Thunk, elf_file: *Elf) i64 {
     return @as(i64, @intCast(shdr.sh_addr)) + thunk.value;
 }
 
-pub fn getTargetAddress(thunk: Thunk, sym_index: Symbol.Index, elf_file: *Elf) i64 {
+pub fn getTargetAddress(thunk: Thunk, ref: Elf.Ref, elf_file: *Elf) i64 {
     const cpu_arch = elf_file.options.cpu_arch.?;
-    return thunk.getAddress(elf_file) + @as(i64, @intCast(thunk.symbols.getIndex(sym_index).? * trampolineSize(cpu_arch)));
+    return thunk.getAddress(elf_file) + @as(i64, @intCast(thunk.symbols.getIndex(ref).? * trampolineSize(cpu_arch)));
 }
 
 pub fn isReachable(atom: *const Atom, rel: elf.Elf64_Rela, elf_file: *Elf) bool {
@@ -52,8 +52,8 @@ pub fn calcSymtabSize(thunk: *Thunk, elf_file: *Elf) void {
     if (elf_file.options.strip_all) return;
 
     thunk.output_symtab_ctx.nlocals = @as(u32, @intCast(thunk.symbols.keys().len));
-    for (thunk.symbols.keys()) |sym_index| {
-        const sym = elf_file.getSymbol(sym_index);
+    for (thunk.symbols.keys()) |ref| {
+        const sym = elf_file.getSymbol(ref).?;
         thunk.output_symtab_ctx.strsize += @as(u32, @intCast(sym.getName(elf_file).len + "$thunk".len + 1));
     }
 }
@@ -62,8 +62,8 @@ pub fn writeSymtab(thunk: Thunk, elf_file: *Elf) void {
     if (elf_file.options.strip_all) return;
     const cpu_arch = elf_file.options.cpu_arch.?;
 
-    for (thunk.symbols.keys(), thunk.output_symtab_ctx.ilocal..) |sym_index, ilocal| {
-        const sym = elf_file.getSymbol(sym_index);
+    for (thunk.symbols.keys(), thunk.output_symtab_ctx.ilocal..) |ref, ilocal| {
+        const sym = elf_file.getSymbol(ref).?;
         const st_name = @as(u32, @intCast(elf_file.strtab.items.len));
         elf_file.strtab.appendSliceAssumeCapacity(sym.getName(elf_file));
         elf_file.strtab.appendSliceAssumeCapacity("$thunk");
@@ -73,7 +73,7 @@ pub fn writeSymtab(thunk: Thunk, elf_file: *Elf) void {
             .st_info = elf.STT_FUNC,
             .st_other = 0,
             .st_shndx = @intCast(thunk.out_shndx),
-            .st_value = @intCast(thunk.getTargetAddress(sym_index, elf_file)),
+            .st_value = @intCast(thunk.getTargetAddress(ref, elf_file)),
             .st_size = trampolineSize(cpu_arch),
         };
     }
@@ -123,9 +123,9 @@ fn format2(
     const thunk = ctx.thunk;
     const elf_file = ctx.elf_file;
     try writer.print("@{x} : size({x})\n", .{ thunk.value, thunk.size(elf_file) });
-    for (thunk.symbols.keys()) |index| {
-        const sym = elf_file.getSymbol(index);
-        try writer.print("  %{d} : {s} : @{x}\n", .{ index, sym.getName(elf_file), sym.value });
+    for (thunk.symbols.keys()) |ref| {
+        const sym = elf_file.getSymbol(ref).?;
+        try writer.print("  {} : {s} : @{x}\n", .{ ref, sym.getName(elf_file), sym.value });
     }
 }
 
@@ -136,7 +136,8 @@ const aarch64 = struct {
         const r_type: elf.R_AARCH64 = @enumFromInt(rel.r_type());
         if (r_type != .CALL26 and r_type != .JUMP26) return true;
         const object = atom.getObject(elf_file);
-        const target = object.getSymbol(rel.r_sym(), elf_file);
+        const target_ref = object.resolveSymbol(rel.r_sym(), elf_file);
+        const target = elf_file.getSymbol(target_ref).?;
         if (target.flags.plt) return false;
         if (atom.out_shndx != target.shndx) return false;
         const target_atom = target.getAtom(elf_file).?;
@@ -148,8 +149,8 @@ const aarch64 = struct {
     }
 
     fn write(thunk: Thunk, elf_file: *Elf, writer: anytype) !void {
-        for (thunk.symbols.keys(), 0..) |sym_index, i| {
-            const sym = elf_file.getSymbol(sym_index);
+        for (thunk.symbols.keys(), 0..) |ref, i| {
+            const sym = elf_file.getSymbol(ref).?;
             const saddr = thunk.getAddress(elf_file) + @as(i64, @intCast(i * trampoline_size));
             const taddr = sym.getAddress(.{}, elf_file);
             const pages = try util.calcNumberOfPages(saddr, taddr);

src/Elf/InternalObject.zig

@@ -152,8 +152,8 @@ pub const Cie = struct {
             if (cie_rel.r_type() != other_rel.r_type()) return false;
             if (cie_rel.r_addend != other_rel.r_addend) return false;
 
-            const cie_sym = cie.getObject(elf_file).getSymbol(cie_rel.r_sym(), elf_file);
-            const other_sym = other.getObject(elf_file).getSymbol(other_rel.r_sym(), elf_file);
+            const cie_sym = cie.getObject(elf_file).symbols.items[cie_rel.r_sym()];
+            const other_sym = other.getObject(elf_file).symbols.items[other_rel.r_sym()];
             if (!std.mem.eql(u8, std.mem.asBytes(&cie_sym), std.mem.asBytes(&other_sym))) return false;
         }
         return true;
@@ -361,7 +361,8 @@ pub fn writeEhFrame(elf_file: *Elf, writer: anytype) !void {
             const data = cie.getData(elf_file);
 
             for (cie.getRelocs(elf_file)) |rel| {
-                const sym = object.getSymbol(rel.r_sym(), elf_file);
+                const sym_ref = object.resolveSymbol(rel.r_sym(), elf_file);
+                const sym = elf_file.getSymbol(sym_ref).?;
                 resolveReloc(cie, sym, rel, elf_file, data) catch |err| switch (err) {
                     error.RelocError => has_reloc_errors = true,
                     else => |e| return e,
@@ -388,7 +389,8 @@ pub fn writeEhFrame(elf_file: *Elf, writer: anytype) !void {
             );
 
             for (fde.getRelocs(elf_file)) |rel| {
-                const sym = object.getSymbol(rel.r_sym(), elf_file);
+                const sym_ref = object.resolveSymbol(rel.r_sym(), elf_file);
+                const sym = elf_file.getSymbol(sym_ref).?;
                 resolveReloc(fde, sym, rel, elf_file, data) catch |err| switch (err) {
                     error.RelocError => has_reloc_errors = true,
                     else => |e| return e,
@@ -437,7 +439,7 @@ pub fn writeEhFrameRelocatable(elf_file: *Elf, writer: anytype) !void {
     }
 }
 
-fn emitReloc(elf_file: *Elf, rec: anytype, sym: *const Symbol, rel: elf.Elf64_Rela) elf.Elf64_Rela {
+fn emitReloc(elf_file: *Elf, rec: anytype, sym: Symbol, rel: elf.Elf64_Rela) elf.Elf64_Rela {
     const tracy = trace(@src());
     defer tracy.end();
 
@@ -483,7 +485,7 @@ pub fn writeEhFrameRelocs(elf_file: *Elf, writer: anytype) !void {
         for (object.cies.items) |cie| {
             if (!cie.alive) continue;
             for (cie.getRelocs(elf_file)) |rel| {
-                const sym = object.getSymbol(rel.r_sym(), elf_file);
+                const sym = object.symbols.items[rel.r_sym()];
                 const out_rel = emitReloc(elf_file, cie, sym, rel);
                 try writer.writeStruct(out_rel);
             }
@@ -492,7 +494,7 @@ pub fn writeEhFrameRelocs(elf_file: *Elf, writer: anytype) !void {
         for (object.fdes.items) |fde| {
             if (!fde.alive) continue;
             for (fde.getRelocs(elf_file)) |rel| {
-                const sym = object.getSymbol(rel.r_sym(), elf_file);
+                const sym = object.symbols.items[rel.r_sym()];
                 const out_rel = emitReloc(elf_file, fde, sym, rel);
                 try writer.writeStruct(out_rel);
             }
@@ -544,7 +546,8 @@ pub fn writeEhFrameHdr(elf_file: *Elf, writer: anytype) !void {
             const relocs = fde.getRelocs(elf_file);
             assert(relocs.len > 0); // Should this be an error? Things are completely broken anyhow if this trips...
             const rel = relocs[0];
-            const sym = object.getSymbol(rel.r_sym(), elf_file);
+            const sym_ref = object.resolveSymbol(rel.r_sym(), elf_file);
+            const sym = elf_file.getSymbol(sym_ref).?;
             const P = @as(i64, @intCast(fde.getAddress(elf_file)));
             const S = @as(i64, @intCast(sym.getAddress(.{}, elf_file)));
             const A = rel.r_addend;