Commit 0701646beb

Jakub Konka <kubkon@jakubkonka.com>
2024-07-28 10:24:26
elf: move merge subsections ownership into merge sections
1 parent 24126f5
src/link/Elf/LinkerDefined.zig
@@ -47,7 +47,7 @@ pub fn resolveSymbols(self: *LinkerDefined, elf_file: *Elf) void {
         const global = elf_file.symbol(index);
         if (self.asFile().symbolRank(this_sym, false) < global.symbolRank(elf_file)) {
             global.value = 0;
-            global.atom_ref = .{ .index = 0, .file = 0 };
+            global.ref = .{ .index = 0, .file = 0 };
             global.file_index = self.index;
             global.esym_index = sym_idx;
             global.version_index = elf_file.default_sym_version;
src/link/Elf/merge_section.zig
@@ -10,12 +10,14 @@ pub const MergeSection = struct {
         IndexContext,
         std.hash_map.default_max_load_percentage,
     ) = .{},
-    subsections: std.ArrayListUnmanaged(MergeSubsection.Index) = .{},
+    subsections: std.ArrayListUnmanaged(MergeSubsection) = .{},
+    finalized_subsections: std.ArrayListUnmanaged(MergeSubsection.Index) = .{},
 
     pub fn deinit(msec: *MergeSection, allocator: Allocator) void {
         msec.bytes.deinit(allocator);
         msec.table.deinit(allocator);
         msec.subsections.deinit(allocator);
+        msec.finalized_subsections.deinit(allocator);
     }
 
     pub fn name(msec: MergeSection, elf_file: *Elf) [:0]const u8 {
@@ -60,23 +62,25 @@ pub const MergeSection = struct {
     /// Sorts all owned subsections.
     pub fn finalize(msec: *MergeSection, elf_file: *Elf) !void {
         const gpa = elf_file.base.comp.gpa;
-        try msec.subsections.ensureTotalCapacityPrecise(gpa, msec.table.count());
+        try msec.finalized_subsections.ensureTotalCapacityPrecise(gpa, msec.subsections.items.len);
 
         var it = msec.table.iterator();
         while (it.next()) |entry| {
-            const msub = elf_file.mergeSubsection(entry.value_ptr.*);
+            const msub = msec.mergeSubsection(entry.value_ptr.*);
             if (!msub.alive) continue;
-            msec.subsections.appendAssumeCapacity(entry.value_ptr.*);
+            msec.finalized_subsections.appendAssumeCapacity(entry.value_ptr.*);
         }
         msec.table.clearAndFree(gpa);
 
         const sortFn = struct {
-            pub fn sortFn(ctx: *Elf, lhs: MergeSubsection.Index, rhs: MergeSubsection.Index) bool {
+            pub fn sortFn(ctx: *MergeSection, lhs: MergeSubsection.Index, rhs: MergeSubsection.Index) bool {
                 const lhs_msub = ctx.mergeSubsection(lhs);
                 const rhs_msub = ctx.mergeSubsection(rhs);
                 if (lhs_msub.alignment.compareStrict(.eq, rhs_msub.alignment)) {
                     if (lhs_msub.size == rhs_msub.size) {
-                        return mem.order(u8, lhs_msub.getString(ctx), rhs_msub.getString(ctx)) == .lt;
+                        const lhs_string = ctx.bytes.items[lhs_msub.string_index..][0..lhs_msub.size];
+                        const rhs_string = ctx.bytes.items[rhs_msub.string_index..][0..rhs_msub.size];
+                        return mem.order(u8, lhs_string, rhs_string) == .lt;
                     }
                     return lhs_msub.size < rhs_msub.size;
                 }
@@ -84,7 +88,19 @@ pub const MergeSection = struct {
             }
         }.sortFn;
 
-        std.mem.sort(MergeSubsection.Index, msec.subsections.items, elf_file, sortFn);
+        std.mem.sort(MergeSubsection.Index, msec.finalized_subsections.items, msec, sortFn);
+    }
+
+    pub fn addMergeSubsection(msec: *MergeSection, allocator: Allocator) !MergeSubsection.Index {
+        const index: MergeSubsection.Index = @intCast(msec.subsections.items.len);
+        const msub = try msec.subsections.addOne(allocator);
+        msub.* = .{};
+        return index;
+    }
+
+    pub fn mergeSubsection(msec: *MergeSection, index: MergeSubsection.Index) *MergeSubsection {
+        assert(index < msec.subsections.items.len);
+        return &msec.subsections.items[index];
     }
 
     pub const IndexContext = struct {
@@ -154,8 +170,8 @@ pub const MergeSection = struct {
             msec.type,
             msec.flags,
         });
-        for (msec.subsections.items) |index| {
-            try writer.print("   {}\n", .{elf_file.mergeSubsection(index).fmt(elf_file)});
+        for (msec.subsections.items) |msub| {
+            try writer.print("   {}\n", .{msub.fmt(elf_file)});
         }
     }
 
@@ -250,18 +266,26 @@ pub const InputMergeSection = struct {
         // TODO: imsec.strings.clearAndFree(allocator);
     }
 
-    pub fn findSubsection(imsec: InputMergeSection, offset: u32) ?struct { MergeSubsection.Index, u32 } {
+    const FindSubsectionResult = struct {
+        msub_index: MergeSubsection.Index,
+        offset: u32,
+    };
+
+    pub fn findSubsection(imsec: InputMergeSection, offset: u32) ?FindSubsectionResult {
         // TODO: binary search
         for (imsec.offsets.items, 0..) |off, index| {
             if (offset < off) return .{
-                imsec.subsections.items[index - 1],
-                offset - imsec.offsets.items[index - 1],
+                .msub_index = imsec.subsections.items[index - 1],
+                .offset = offset - imsec.offsets.items[index - 1],
             };
         }
         const last = imsec.offsets.items.len - 1;
         const last_off = imsec.offsets.items[last];
         const last_len = imsec.strings.items[last].len;
-        if (offset < last_off + last_len) return .{ imsec.subsections.items[last], offset - last_off };
+        if (offset < last_off + last_len) return .{
+            .msub_index = imsec.subsections.items[last],
+            .offset = offset - last_off,
+        };
         return null;
     }
 
src/link/Elf/Object.zig
@@ -388,7 +388,7 @@ fn initSymtab(self: *Object, allocator: Allocator, elf_file: *Elf) !void {
         sym_ptr.esym_index = @as(u32, @intCast(i));
         sym_ptr.file_index = self.index;
         if (sym.st_shndx != elf.SHN_ABS) {
-            sym_ptr.atom_ref = .{ .index = self.atoms_indexes.items[sym.st_shndx], .file = self.index };
+            sym_ptr.ref = .{ .index = self.atoms_indexes.items[sym.st_shndx], .file = self.index };
         }
     }
 
@@ -575,7 +575,7 @@ pub fn resolveSymbols(self: *Object, elf_file: *Elf) void {
         if (self.asFile().symbolRank(esym, !self.alive) < global.symbolRank(elf_file)) {
             switch (esym.st_shndx) {
                 elf.SHN_ABS, elf.SHN_COMMON => {},
-                else => global.atom_ref = .{
+                else => global.ref = .{
                     .index = self.atoms_indexes.items[esym.st_shndx],
                     .file = self.index,
                 },
@@ -609,7 +609,7 @@ pub fn claimUnresolved(self: *Object, elf_file: *Elf) void {
         };
 
         global.value = 0;
-        global.atom_ref = .{ .index = 0, .file = 0 };
+        global.ref = .{ .index = 0, .file = 0 };
         global.esym_index = esym_index;
         global.file_index = self.index;
         global.version_index = if (is_import) elf.VER_NDX_LOCAL else elf_file.default_sym_version;
@@ -630,7 +630,7 @@ pub fn claimUnresolvedObject(self: *Object, elf_file: *Elf) void {
         }
 
         global.value = 0;
-        global.atom_ref = .{ .index = 0, .file = 0 };
+        global.ref = .{ .index = 0, .file = 0 };
         global.esym_index = esym_index;
         global.file_index = self.index;
     }
@@ -785,8 +785,8 @@ pub fn resolveMergeSubsections(self: *Object, elf_file: *Elf) !void {
             const string = imsec.bytes.items[str.pos..][0..str.len];
             const res = try msec.insert(gpa, string);
             if (!res.found_existing) {
-                const msub_index = try elf_file.addMergeSubsection();
-                const msub = elf_file.mergeSubsection(msub_index);
+                const msub_index = try msec.addMergeSubsection(gpa);
+                const msub = msec.mergeSubsection(msub_index);
                 msub.merge_section_index = imsec.merge_section_index;
                 msub.string_index = res.key.pos;
                 msub.alignment = atom_ptr.alignment;
@@ -810,7 +810,7 @@ pub fn resolveMergeSubsections(self: *Object, elf_file: *Elf) !void {
         const imsec_index = self.input_merge_sections_indexes.items[esym.st_shndx];
         const imsec = self.inputMergeSection(imsec_index) orelse continue;
         if (imsec.offsets.items.len == 0) continue;
-        const msub_index, const offset = imsec.findSubsection(@intCast(esym.st_value)) orelse {
+        const res = imsec.findSubsection(@intCast(esym.st_value)) orelse {
             var err = try elf_file.base.addErrorWithNotes(2);
             try err.addMsg("invalid symbol value: {x}", .{esym.st_value});
             try err.addNote("for symbol {s}", .{sym.name(elf_file)});
@@ -818,9 +818,9 @@ pub fn resolveMergeSubsections(self: *Object, elf_file: *Elf) !void {
             return error.MalformedObject;
         };
 
-        try sym.addExtra(.{ .subsection = msub_index }, elf_file);
+        sym.ref = .{ .index = res.msub_index, .file = imsec.merge_section_index };
         sym.flags.merge_subsection = true;
-        sym.value = offset;
+        sym.value = res.offset;
     }
 
     for (self.atoms_indexes.items) |atom_index| {
@@ -835,28 +835,27 @@ pub fn resolveMergeSubsections(self: *Object, elf_file: *Elf) !void {
             const imsec_index = self.input_merge_sections_indexes.items[esym.st_shndx];
             const imsec = self.inputMergeSection(imsec_index) orelse continue;
             if (imsec.offsets.items.len == 0) continue;
-            const msub_index, const offset = imsec.findSubsection(@intCast(@as(i64, @intCast(esym.st_value)) + rel.r_addend)) orelse {
+            const msec = elf_file.mergeSection(imsec.merge_section_index);
+            const res = imsec.findSubsection(@intCast(@as(i64, @intCast(esym.st_value)) + rel.r_addend)) orelse {
                 var err = try elf_file.base.addErrorWithNotes(1);
                 try err.addMsg("invalid relocation at offset 0x{x}", .{rel.r_offset});
                 try err.addNote("in {}:{s}", .{ self.fmtPath(), atom_ptr.name(elf_file) });
                 return error.MalformedObject;
             };
-            const msub = elf_file.mergeSubsection(msub_index);
-            const msec = msub.mergeSection(elf_file);
 
             const out_sym_idx: u64 = @intCast(self.symbols.items.len);
             try self.symbols.ensureUnusedCapacity(gpa, 1);
-            const name = try std.fmt.allocPrint(gpa, "{s}$subsection{d}", .{ msec.name(elf_file), msub_index });
+            const name = try std.fmt.allocPrint(gpa, "{s}$subsection{d}", .{ msec.name(elf_file), res.msub_index });
             defer gpa.free(name);
             const sym_index = try elf_file.addSymbol();
             const sym = elf_file.symbol(sym_index);
             sym.* = .{
-                .value = @bitCast(@as(i64, @intCast(offset)) - rel.r_addend),
+                .value = @bitCast(@as(i64, @intCast(res.offset)) - rel.r_addend),
                 .name_offset = try self.addString(gpa, name),
                 .esym_index = rel.r_sym(),
                 .file_index = self.index,
             };
-            try sym.addExtra(.{ .subsection = msub_index }, elf_file);
+            sym.ref = .{ .index = res.msub_index, .file = imsec.merge_section_index };
             sym.flags.merge_subsection = true;
             self.symbols.addOneAssumeCapacity().* = sym_index;
             rel.r_info = (out_sym_idx << 32) | rel.r_type();
@@ -920,7 +919,7 @@ pub fn convertCommonSymbols(self: *Object, elf_file: *Elf) !void {
         try self.atoms_indexes.append(gpa, atom_index);
 
         global.value = 0;
-        global.atom_ref = .{ .index = atom_index, .file = self.index };
+        global.ref = .{ .index = atom_index, .file = self.index };
         global.flags.weak = false;
     }
 }
src/link/Elf/SharedObject.zig
@@ -232,7 +232,7 @@ 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 = @intCast(this_sym.st_value);
-            global.atom_ref = .{ .index = 0, .file = 0 };
+            global.ref = .{ .index = 0, .file = 0 };
             global.esym_index = esym_index;
             global.version_index = self.versyms.items[esym_index];
             global.file_index = self.index;
src/link/Elf/Symbol.zig
@@ -9,9 +9,9 @@ name_offset: u32 = 0,
 /// Index of file where this symbol is defined.
 file_index: File.Index = 0,
 
-/// Reference to Atom containing this symbol if any.
-/// Use `atom` to get the pointer to the atom.
-atom_ref: Elf.Ref = .{ .index = 0, .file = 0 },
+/// Reference to Atom or merge subsection containing this symbol if any.
+/// Use `atom` or `mergeSubsection` to get the pointer to the atom.
+ref: Elf.Ref = .{ .index = 0, .file = 0 },
 
 /// Assigned output section index for this symbol.
 output_section_index: u32 = 0,
@@ -71,14 +71,15 @@ pub fn name(symbol: Symbol, elf_file: *Elf) [:0]const u8 {
 }
 
 pub fn atom(symbol: Symbol, elf_file: *Elf) ?*Atom {
-    const file_ptr = elf_file.file(symbol.atom_ref.file) orelse return null;
-    return file_ptr.atom(symbol.atom_ref.index);
+    if (symbol.flags.merge_subsection) return null;
+    const file_ptr = elf_file.file(symbol.ref.file) orelse return null;
+    return file_ptr.atom(symbol.ref.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);
+    const msec = elf_file.mergeSection(symbol.ref.file);
+    return msec.mergeSubsection(symbol.ref.index);
 }
 
 pub fn file(symbol: Symbol, elf_file: *Elf) ?File {
@@ -262,7 +263,6 @@ const AddExtraOpts = struct {
     gottp: ?u32 = null,
     tlsdesc: ?u32 = null,
     zig_got: ?u32 = null,
-    subsection: ?u32 = null,
 };
 
 pub fn addExtra(symbol: *Symbol, opts: AddExtraOpts, elf_file: *Elf) !void {
@@ -488,7 +488,7 @@ pub const Extra = struct {
     gottp: u32 = 0,
     tlsdesc: u32 = 0,
     zig_got: u32 = 0,
-    subsection: u32 = 0,
+    merge_section: u32 = 0,
 };
 
 pub const Index = u32;
src/link/Elf/ZigObject.zig
@@ -291,7 +291,7 @@ pub fn newAtom(self: *ZigObject, elf_file: *Elf) !Symbol.Index {
 
     const symbol_ptr = elf_file.symbol(symbol_index);
     symbol_ptr.file_index = self.index;
-    symbol_ptr.atom_ref = .{ .index = atom_index, .file = self.index };
+    symbol_ptr.ref = .{ .index = atom_index, .file = self.index };
 
     self.local_esyms.items(.shndx)[esym_index] = atom_index;
     self.local_esyms.items(.elf_sym)[esym_index].st_shndx = SHN_ATOM;
@@ -342,7 +342,7 @@ pub fn resolveSymbols(self: *ZigObject, elf_file: *Elf) void {
                 else => unreachable,
             };
             global.value = @intCast(esym.st_value);
-            global.atom_ref = .{ .index = atom_index, .file = self.index };
+            global.ref = .{ .index = atom_index, .file = self.index };
             global.esym_index = esym_index;
             global.file_index = self.index;
             global.version_index = elf_file.default_sym_version;
@@ -371,7 +371,7 @@ pub fn claimUnresolved(self: ZigObject, elf_file: *Elf) void {
         };
 
         global.value = 0;
-        global.atom_ref = .{ .index = 0, .file = 0 };
+        global.ref = .{ .index = 0, .file = 0 };
         global.esym_index = esym_index;
         global.file_index = self.index;
         global.version_index = if (is_import) elf.VER_NDX_LOCAL else elf_file.default_sym_version;
@@ -392,7 +392,7 @@ pub fn claimUnresolvedObject(self: ZigObject, elf_file: *Elf) void {
         }
 
         global.value = 0;
-        global.atom_ref = .{ .index = 0, .file = 0 };
+        global.ref = .{ .index = 0, .file = 0 };
         global.esym_index = esym_index;
         global.file_index = self.index;
     }
src/link/Elf.zig
@@ -208,9 +208,6 @@ thunks: std.ArrayListUnmanaged(Thunk) = .{},
 
 /// List of output merge sections with deduped contents.
 merge_sections: std.ArrayListUnmanaged(MergeSection) = .{},
-/// List of output merge subsections.
-/// Each subsection is akin to Atom but belongs to a MergeSection.
-merge_subsections: std.ArrayListUnmanaged(MergeSubsection) = .{},
 
 /// Table of last atom index in a section and matching atom free list if any.
 last_atom_and_free_list_table: LastAtomAndFreeListTable = .{},
@@ -497,7 +494,6 @@ pub fn deinit(self: *Elf) void {
         sect.deinit(gpa);
     }
     self.merge_sections.deinit(gpa);
-    self.merge_subsections.deinit(gpa);
     for (self.last_atom_and_free_list_table.values()) |*value| {
         value.free_list.deinit(gpa);
     }
@@ -3288,12 +3284,13 @@ fn checkDuplicates(self: *Elf) !void {
 }
 
 pub fn addCommentString(self: *Elf) !void {
+    const gpa = self.base.comp.gpa;
     const msec_index = try self.getOrCreateMergeSection(".comment", elf.SHF_MERGE | elf.SHF_STRINGS, elf.SHT_PROGBITS);
     const msec = self.mergeSection(msec_index);
-    const res = try msec.insertZ(self.base.comp.gpa, "zig " ++ builtin.zig_version_string);
+    const res = try msec.insertZ(gpa, "zig " ++ builtin.zig_version_string);
     if (res.found_existing) return;
-    const msub_index = try self.addMergeSubsection();
-    const msub = self.mergeSubsection(msub_index);
+    const msub_index = try msec.addMergeSubsection(gpa);
+    const msub = msec.mergeSubsection(msub_index);
     msub.merge_section_index = msec_index;
     msub.string_index = res.key.pos;
     msub.alignment = .@"1";
@@ -3340,8 +3337,8 @@ pub fn finalizeMergeSections(self: *Elf) !void {
 pub fn updateMergeSectionSizes(self: *Elf) !void {
     for (self.merge_sections.items) |*msec| {
         const shdr = &self.shdrs.items[msec.output_section_index];
-        for (msec.subsections.items) |msub_index| {
-            const msub = self.mergeSubsection(msub_index);
+        for (msec.finalized_subsections.items) |msub_index| {
+            const msub = msec.mergeSubsection(msub_index);
             assert(msub.alive);
             const offset = msub.alignment.forward(shdr.sh_size);
             const padding = offset - shdr.sh_size;
@@ -3357,14 +3354,14 @@ pub fn writeMergeSections(self: *Elf) !void {
     var buffer = std.ArrayList(u8).init(gpa);
     defer buffer.deinit();
 
-    for (self.merge_sections.items) |msec| {
+    for (self.merge_sections.items) |*msec| {
         const shdr = self.shdrs.items[msec.output_section_index];
         const size = math.cast(usize, shdr.sh_size) orelse return error.Overflow;
         try buffer.ensureTotalCapacity(size);
         buffer.appendNTimesAssumeCapacity(0, size);
 
-        for (msec.subsections.items) |msub_index| {
-            const msub = self.mergeSubsection(msub_index);
+        for (msec.finalized_subsections.items) |msub_index| {
+            const msub = msec.mergeSubsection(msub_index);
             assert(msub.alive);
             const string = msub.getString(self);
             const off = math.cast(usize, msub.value) orelse return error.Overflow;
@@ -3384,7 +3381,7 @@ fn initOutputSections(self: *Elf) !void {
 
 pub fn initMergeSections(self: *Elf) !void {
     for (self.merge_sections.items) |*msec| {
-        if (msec.subsections.items.len == 0) continue;
+        if (msec.finalized_subsections.items.len == 0) continue;
         const name = msec.name(self);
         const shndx = self.sectionByName(name) orelse try self.addSection(.{
             .name = name,
@@ -3393,9 +3390,9 @@ pub fn initMergeSections(self: *Elf) !void {
         });
         msec.output_section_index = shndx;
 
-        var entsize = self.mergeSubsection(msec.subsections.items[0]).entsize;
-        for (msec.subsections.items) |index| {
-            const msub = self.mergeSubsection(index);
+        var entsize = msec.mergeSubsection(msec.finalized_subsections.items[0]).entsize;
+        for (msec.finalized_subsections.items) |msub_index| {
+            const msub = msec.mergeSubsection(msub_index);
             entsize = @min(entsize, msub.entsize);
         }
         const shdr = &self.shdrs.items[shndx];
@@ -5706,18 +5703,6 @@ pub fn zigObjectPtr(self: *Elf) ?*ZigObject {
     return self.file(index).?.zig_object;
 }
 
-pub fn addMergeSubsection(self: *Elf) !MergeSubsection.Index {
-    const index: MergeSubsection.Index = @intCast(self.merge_subsections.items.len);
-    const msec = try self.merge_subsections.addOne(self.base.comp.gpa);
-    msec.* = .{};
-    return index;
-}
-
-pub fn mergeSubsection(self: *Elf, index: MergeSubsection.Index) *MergeSubsection {
-    assert(index < self.merge_subsections.items.len);
-    return &self.merge_subsections.items[index];
-}
-
 pub fn getOrCreateMergeSection(self: *Elf, name: []const u8, flags: u64, @"type": u32) !MergeSection.Index {
     const gpa = self.base.comp.gpa;
     const out_name = name: {