Merge pull request ziglang#19714 from ziglang/elf-merge-strings

link/elf: implement string merging

Author: kubkon

src/link/Elf.zig

@@ -60,10 +60,10 @@ pub fn updateSymtabSize(self: *LinkerDefined, elf_file: *Elf) !void {
         if (file_ptr.index() != self.index) continue;
         global.flags.output_symtab = true;
         if (global.isLocal(elf_file)) {
-            try global.setOutputSymtabIndex(self.output_symtab_ctx.nlocals, elf_file);
+            try global.addExtra(.{ .symtab = self.output_symtab_ctx.nlocals }, elf_file);
             self.output_symtab_ctx.nlocals += 1;
         } else {
-            try global.setOutputSymtabIndex(self.output_symtab_ctx.nglobals, elf_file);
+            try global.addExtra(.{ .symtab = self.output_symtab_ctx.nglobals }, elf_file);
             self.output_symtab_ctx.nglobals += 1;
         }
         self.output_symtab_ctx.strsize += @as(u32, @intCast(global.name(elf_file).len)) + 1;

src/link/Elf/Atom.zig

@@ -231,7 +231,7 @@ pub fn resolveSymbols(self: *SharedObject, elf_file: *Elf) void {
 
         const global = elf_file.symbol(index);
         if (self.asFile().symbolRank(this_sym, false) < global.symbolRank(elf_file)) {
-            global.value = this_sym.st_value;
+            global.value = @intCast(this_sym.st_value);
             global.atom_index = 0;
             global.esym_index = esym_index;
             global.version_index = self.versyms.items[esym_index];
@@ -269,7 +269,7 @@ pub fn updateSymtabSize(self: *SharedObject, elf_file: *Elf) !void {
         if (file_ptr.index() != self.index) continue;
         if (global.isLocal(elf_file)) continue;
         global.flags.output_symtab = true;
-        try global.setOutputSymtabIndex(self.output_symtab_ctx.nglobals, elf_file);
+        try global.addExtra(.{ .symtab = self.output_symtab_ctx.nglobals }, elf_file);
         self.output_symtab_ctx.nglobals += 1;
         self.output_symtab_ctx.strsize += @as(u32, @intCast(global.name(elf_file).len)) + 1;
     }

src/link/Elf/LinkerDefined.zig

@@ -1,7 +1,7 @@
 //! Represents a defined symbol.
 
 /// Allocated address value of this symbol.
-value: u64 = 0,
+value: i64 = 0,
 
 /// Offset into the linker's string table.
 name_offset: u32 = 0,
@@ -14,7 +14,7 @@ file_index: File.Index = 0,
 /// Use `atom` to get the pointer to the atom.
 atom_index: Atom.Index = 0,
 
-/// Assigned output section index for this atom.
+/// Assigned output section index for this symbol.
 output_section_index: u32 = 0,
 
 /// Index of the source symbol this symbol references.
@@ -33,7 +33,8 @@ extra_index: u32 = 0,
 pub fn isAbs(symbol: Symbol, elf_file: *Elf) bool {
     const file_ptr = symbol.file(elf_file).?;
     if (file_ptr == .shared_object) return symbol.elfSym(elf_file).st_shndx == elf.SHN_ABS;
-    return !symbol.flags.import and symbol.atom(elf_file) == null and symbol.outputShndx() == null and
+    return !symbol.flags.import and symbol.atom(elf_file) == null and
+        symbol.mergeSubsection(elf_file) == null and symbol.outputShndx() == null and
         file_ptr != .linker_defined;
 }
 
@@ -70,6 +71,12 @@ pub fn atom(symbol: Symbol, elf_file: *Elf) ?*Atom {
     return elf_file.atom(symbol.atom_index);
 }
 
+pub fn mergeSubsection(symbol: Symbol, elf_file: *Elf) ?*MergeSubsection {
+    if (!symbol.flags.merge_subsection) return null;
+    const extras = symbol.extra(elf_file).?;
+    return elf_file.mergeSubsection(extras.subsection);
+}
+
 pub fn file(symbol: Symbol, elf_file: *Elf) ?File {
     return elf_file.file(symbol.file_index);
 }
@@ -92,7 +99,11 @@ pub fn symbolRank(symbol: Symbol, elf_file: *Elf) u32 {
     return file_ptr.symbolRank(sym, in_archive);
 }
 
-pub fn address(symbol: Symbol, opts: struct { plt: bool = true }, elf_file: *Elf) u64 {
+pub fn address(symbol: Symbol, opts: struct { plt: bool = true }, elf_file: *Elf) i64 {
+    if (symbol.mergeSubsection(elf_file)) |msub| {
+        if (!msub.alive) return 0;
+        return msub.address(elf_file) + symbol.value;
+    }
     if (symbol.flags.has_copy_rel) {
         return symbol.copyRelAddress(elf_file);
     }
@@ -108,19 +119,23 @@ pub fn address(symbol: Symbol, opts: struct { plt: bool = true }, elf_file: *Elf
         if (!atom_ptr.flags.alive) {
             if (mem.eql(u8, atom_ptr.name(elf_file), ".eh_frame")) {
                 const sym_name = symbol.name(elf_file);
+                const sh_addr, const sh_size = blk: {
+                    const shndx = elf_file.eh_frame_section_index orelse break :blk .{ 0, 0 };
+                    const shdr = elf_file.shdrs.items[shndx];
+                    break :blk .{ shdr.sh_addr, shdr.sh_size };
+                };
                 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.elfSym(elf_file).st_type() == elf.STT_SECTION)
                 {
-                    return elf_file.shdrs.items[elf_file.eh_frame_section_index.?].sh_addr;
+                    return @intCast(sh_addr);
                 }
 
                 if (mem.startsWith(u8, sym_name, "__FRAME_END__") or
                     mem.startsWith(u8, sym_name, "__EH_FRAME_LIST_END__"))
                 {
-                    const shdr = elf_file.shdrs.items[elf_file.eh_frame_section_index.?];
-                    return shdr.sh_addr + shdr.sh_size;
+                    return @intCast(sh_addr + sh_size);
                 }
 
                 // TODO I think we potentially should error here
@@ -143,65 +158,57 @@ pub fn outputSymtabIndex(symbol: Symbol, elf_file: *Elf) ?u32 {
     return if (symbol.isLocal(elf_file)) idx + symtab_ctx.ilocal else idx + symtab_ctx.iglobal;
 }
 
-pub fn setOutputSymtabIndex(symbol: *Symbol, index: u32, elf_file: *Elf) !void {
-    if (symbol.extra(elf_file)) |extras| {
-        var new_extras = extras;
-        new_extras.symtab = index;
-        symbol.setExtra(new_extras, elf_file);
-    } else try symbol.addExtra(.{ .symtab = index }, elf_file);
-}
-
-pub fn gotAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn gotAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_got) return 0;
     const extras = symbol.extra(elf_file).?;
     const entry = elf_file.got.entries.items[extras.got];
     return entry.address(elf_file);
 }
 
-pub fn pltGotAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn pltGotAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!(symbol.flags.has_plt and symbol.flags.has_got)) return 0;
     const extras = symbol.extra(elf_file).?;
     const shdr = elf_file.shdrs.items[elf_file.plt_got_section_index.?];
     const cpu_arch = elf_file.getTarget().cpu.arch;
-    return shdr.sh_addr + extras.plt_got * PltGotSection.entrySize(cpu_arch);
+    return @intCast(shdr.sh_addr + extras.plt_got * PltGotSection.entrySize(cpu_arch));
 }
 
-pub fn pltAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn pltAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_plt) return 0;
     const extras = symbol.extra(elf_file).?;
     const shdr = elf_file.shdrs.items[elf_file.plt_section_index.?];
     const cpu_arch = elf_file.getTarget().cpu.arch;
-    return shdr.sh_addr + extras.plt * PltSection.entrySize(cpu_arch) + PltSection.preambleSize(cpu_arch);
+    return @intCast(shdr.sh_addr + extras.plt * PltSection.entrySize(cpu_arch) + PltSection.preambleSize(cpu_arch));
 }
 
-pub fn gotPltAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn gotPltAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_plt) return 0;
     const extras = symbol.extra(elf_file).?;
     const shdr = elf_file.shdrs.items[elf_file.got_plt_section_index.?];
-    return shdr.sh_addr + extras.plt * 8 + GotPltSection.preamble_size;
+    return @intCast(shdr.sh_addr + extras.plt * 8 + GotPltSection.preamble_size);
 }
 
-pub fn copyRelAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn copyRelAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_copy_rel) return 0;
     const shdr = elf_file.shdrs.items[elf_file.copy_rel_section_index.?];
-    return shdr.sh_addr + symbol.value;
+    return @as(i64, @intCast(shdr.sh_addr)) + symbol.value;
 }
 
-pub fn tlsGdAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn tlsGdAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_tlsgd) return 0;
     const extras = symbol.extra(elf_file).?;
     const entry = elf_file.got.entries.items[extras.tlsgd];
     return entry.address(elf_file);
 }
 
-pub fn gotTpAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn gotTpAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_gottp) return 0;
     const extras = symbol.extra(elf_file).?;
     const entry = elf_file.got.entries.items[extras.gottp];
     return entry.address(elf_file);
 }
 
-pub fn tlsDescAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn tlsDescAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_tlsdesc) return 0;
     const extras = symbol.extra(elf_file).?;
     const entry = elf_file.got.entries.items[extras.tlsdesc];
@@ -221,7 +228,7 @@ pub fn getOrCreateZigGotEntry(symbol: *Symbol, symbol_index: Index, elf_file: *E
     return .{ .found_existing = false, .index = index };
 }
 
-pub fn zigGotAddress(symbol: Symbol, elf_file: *Elf) u64 {
+pub fn zigGotAddress(symbol: Symbol, elf_file: *Elf) i64 {
     if (!symbol.flags.has_zig_got) return 0;
     const extras = symbol.extra(elf_file).?;
     return elf_file.zig_got.entryAddress(extras.zig_got, elf_file);
@@ -240,8 +247,31 @@ pub fn dsoAlignment(symbol: Symbol, elf_file: *Elf) !u64 {
         @min(alignment, try std.math.powi(u64, 2, @ctz(esym.st_value)));
 }
 
-pub fn addExtra(symbol: *Symbol, extras: Extra, elf_file: *Elf) !void {
-    symbol.extra_index = try elf_file.addSymbolExtra(extras);
+const AddExtraOpts = struct {
+    got: ?u32 = null,
+    plt: ?u32 = null,
+    plt_got: ?u32 = null,
+    dynamic: ?u32 = null,
+    symtab: ?u32 = null,
+    copy_rel: ?u32 = null,
+    tlsgd: ?u32 = null,
+    gottp: ?u32 = null,
+    tlsdesc: ?u32 = null,
+    zig_got: ?u32 = null,
+    subsection: ?u32 = null,
+};
+
+pub fn addExtra(symbol: *Symbol, opts: AddExtraOpts, elf_file: *Elf) !void {
+    if (symbol.extra(elf_file) == null) {
+        symbol.extra_index = try elf_file.addSymbolExtra(.{});
+    }
+    var extras = symbol.extra(elf_file).?;
+    inline for (@typeInfo(@TypeOf(opts)).Struct.fields) |field| {
+        if (@field(opts, field.name)) |x| {
+            @field(extras, field.name) = x;
+        }
+    }
+    symbol.setExtra(extras, elf_file);
 }
 
 pub fn extra(symbol: Symbol, elf_file: *Elf) ?Extra {
@@ -266,6 +296,7 @@ pub fn setOutputSym(symbol: Symbol, elf_file: *Elf, out: *elf.Elf64_Sym) void {
         if (symbol.flags.has_copy_rel) break :blk @intCast(elf_file.copy_rel_section_index.?);
         if (file_ptr == .shared_object or esym.st_shndx == elf.SHN_UNDEF) break :blk elf.SHN_UNDEF;
         if (elf_file.base.isRelocatable() and esym.st_shndx == elf.SHN_COMMON) break :blk elf.SHN_COMMON;
+        if (symbol.mergeSubsection(elf_file)) |msub| break :blk @intCast(msub.mergeSection(elf_file).output_section_index);
         if (symbol.atom(elf_file) == null and file_ptr != .linker_defined) break :blk elf.SHN_ABS;
         break :blk @intCast(symbol.outputShndx() orelse elf.SHN_UNDEF);
     };
@@ -284,7 +315,7 @@ pub fn setOutputSym(symbol: Symbol, elf_file: *Elf, out: *elf.Elf64_Sym) void {
     out.st_info = (st_bind << 4) | st_type;
     out.st_other = esym.st_other;
     out.st_shndx = st_shndx;
-    out.st_value = st_value;
+    out.st_value = @intCast(st_value);
     out.st_size = esym.st_size;
 }
 
@@ -436,6 +467,9 @@ pub const Flags = packed struct {
     /// TODO this is really not needed if only we operated on esyms between
     /// codegen and ZigObject.
     is_tls: bool = false,
+
+    /// Whether the symbol is a merge subsection.
+    merge_subsection: bool = false,
 };
 
 pub const Extra = struct {
@@ -449,6 +483,7 @@ pub const Extra = struct {
     gottp: u32 = 0,
     tlsdesc: u32 = 0,
     zig_got: u32 = 0,
+    subsection: u32 = 0,
 };
 
 pub const Index = u32;
@@ -465,6 +500,7 @@ const File = @import("file.zig").File;
 const GotSection = synthetic_sections.GotSection;
 const GotPltSection = synthetic_sections.GotPltSection;
 const LinkerDefined = @import("LinkerDefined.zig");
+const MergeSubsection = @import("merge_section.zig").MergeSubsection;
 const Object = @import("Object.zig");
 const PltSection = synthetic_sections.PltSection;
 const PltGotSection = synthetic_sections.PltGotSection;

src/link/Elf/Object.zig

@@ -343,7 +343,7 @@ pub fn resolveSymbols(self: *ZigObject, elf_file: *Elf) void {
                 atom.outputShndx().?
             else
                 elf.SHN_UNDEF;
-            global.value = esym.st_value;
+            global.value = @intCast(esym.st_value);
             global.atom_index = atom_index;
             global.esym_index = esym_index;
             global.file_index = self.index;
@@ -566,7 +566,7 @@ pub fn updateSymtabSize(self: *ZigObject, elf_file: *Elf) !void {
             else => {},
         }
         local.flags.output_symtab = true;
-        try local.setOutputSymtabIndex(self.output_symtab_ctx.nlocals, elf_file);
+        try local.addExtra(.{ .symtab = self.output_symtab_ctx.nlocals }, elf_file);
         self.output_symtab_ctx.nlocals += 1;
         self.output_symtab_ctx.strsize += @as(u32, @intCast(local.name(elf_file).len)) + 1;
     }
@@ -578,10 +578,10 @@ pub fn updateSymtabSize(self: *ZigObject, elf_file: *Elf) !void {
         if (global.atom(elf_file)) |atom| if (!atom.flags.alive) continue;
         global.flags.output_symtab = true;
         if (global.isLocal(elf_file)) {
-            try global.setOutputSymtabIndex(self.output_symtab_ctx.nlocals, elf_file);
+            try global.addExtra(.{ .symtab = self.output_symtab_ctx.nlocals }, elf_file);
             self.output_symtab_ctx.nlocals += 1;
         } else {
-            try global.setOutputSymtabIndex(self.output_symtab_ctx.nglobals, elf_file);
+            try global.addExtra(.{ .symtab = self.output_symtab_ctx.nglobals }, elf_file);
             self.output_symtab_ctx.nglobals += 1;
         }
         self.output_symtab_ctx.strsize += @as(u32, @intCast(global.name(elf_file).len)) + 1;
@@ -631,7 +631,7 @@ pub fn codeAlloc(self: ZigObject, elf_file: *Elf, atom_index: Atom.Index) ![]u8
         return code;
     }
 
-    const file_offset = shdr.sh_offset + atom.value;
+    const file_offset = shdr.sh_offset + @as(u64, @intCast(atom.value));
     const size = std.math.cast(usize, atom.size) orelse return error.Overflow;
     const code = try gpa.alloc(u8, size);
     errdefer gpa.free(code);
@@ -659,7 +659,7 @@ pub fn getDeclVAddr(
         .r_info = (@as(u64, @intCast(this_sym.esym_index)) << 32) | r_type,
         .r_addend = reloc_info.addend,
     });
-    return vaddr;
+    return @intCast(vaddr);
 }
 
 pub fn getAnonDeclVAddr(
@@ -678,7 +678,7 @@ pub fn getAnonDeclVAddr(
         .r_info = (@as(u64, @intCast(sym.esym_index)) << 32) | r_type,
         .r_addend = reloc_info.addend,
     });
-    return vaddr;
+    return @intCast(vaddr);
 }
 
 pub fn lowerAnonDecl(
@@ -929,7 +929,7 @@ fn updateDeclCode(
 
     if (old_size > 0 and elf_file.base.child_pid == null) {
         const capacity = atom_ptr.capacity(elf_file);
-        const need_realloc = code.len > capacity or !required_alignment.check(atom_ptr.value);
+        const need_realloc = code.len > capacity or !required_alignment.check(@intCast(atom_ptr.value));
         if (need_realloc) {
             try atom_ptr.grow(elf_file);
             log.debug("growing {} from 0x{x} to 0x{x}", .{ decl_name.fmt(&mod.intern_pool), old_vaddr, atom_ptr.value });
@@ -984,7 +984,7 @@ fn updateDeclCode(
 
     const shdr = elf_file.shdrs.items[shdr_index];
     if (shdr.sh_type != elf.SHT_NOBITS) {
-        const file_offset = shdr.sh_offset + atom_ptr.value;
+        const file_offset = shdr.sh_offset + @as(u64, @intCast(atom_ptr.value));
         try elf_file.base.file.?.pwriteAll(code, file_offset);
     }
 }
@@ -1107,7 +1107,7 @@ pub fn updateFunc(
         try self.dwarf.?.commitDeclState(
             mod,
             decl_index,
-            sym.address(.{}, elf_file),
+            @intCast(sym.address(.{}, elf_file)),
             sym.atom(elf_file).?.size,
             ds,
         );
@@ -1186,7 +1186,7 @@ pub fn updateDecl(
         try self.dwarf.?.commitDeclState(
             mod,
             decl_index,
-            sym.address(.{}, elf_file),
+            @intCast(sym.address(.{}, elf_file)),
             sym.atom(elf_file).?.size,
             ds,
         );
@@ -1275,7 +1275,7 @@ fn updateLazySymbol(
     }
 
     const shdr = elf_file.shdrs.items[output_section_index];
-    const file_offset = shdr.sh_offset + atom_ptr.value;
+    const file_offset = shdr.sh_offset + @as(u64, @intCast(atom_ptr.value));
     try elf_file.base.file.?.pwriteAll(code, file_offset);
 }
 
@@ -1373,7 +1373,7 @@ fn lowerConst(
     local_esym.st_value = 0;
 
     const shdr = elf_file.shdrs.items[output_section_index];
-    const file_offset = shdr.sh_offset + atom_ptr.value;
+    const file_offset = shdr.sh_offset + @as(u64, @intCast(atom_ptr.value));
     try elf_file.base.file.?.pwriteAll(code, file_offset);
 
     return .{ .ok = sym_index };
@@ -1457,7 +1457,7 @@ pub fn updateExports(
 
         const actual_esym_index = global_esym_index & symbol_mask;
         const global_esym = &self.global_esyms.items(.elf_sym)[actual_esym_index];
-        global_esym.st_value = elf_file.symbol(sym_index).value;
+        global_esym.st_value = @intCast(elf_file.symbol(sym_index).value);
         global_esym.st_shndx = esym.st_shndx;
         global_esym.st_info = (stb_bits << 4) | stt_bits;
         global_esym.st_name = name_off;