authorgravatar for kubkon@jakubkonka.comJakub Konka <kubkon@jakubkonka.com> 2020-09-07 07:18:44+02:00
committergravatar for kubkon@jakubkonka.comJakub Konka <kubkon@jakubkonka.com> 2020-09-08 20:55:06+02:00
log9306dbd6194251e816f11bc8f420fac0d1ca8835
treed77007ecc023cd179070e8c3795f2a3972305bb1
parent42b1b6be90fc1034875a6e16cf3cbe1c9d6030ca
signaturelock-open Commit is signed but in an unrecognized format.

Fix bug where __text section would get overwritten

Fixes a bug where the last written load command would accidentally override the beginning of the __text section. Also defines missing MachO constants and relocation structs/enums. Signed-off-by: Jakub Konka <kubkon@jakubkonka.com>

2 files changed, 132 insertions(+), 20 deletions(-)

lib/std/macho.zig+110-7
...@@ -647,6 +647,32 @@ pub const nlist_64 = extern struct {...@@ -647,6 +647,32 @@ pub const nlist_64 = extern struct {
647 n_value: u64,647 n_value: u64,
648};648};
649649
650/// Format of a relocation entry of a Mach-O file. Modified from the 4.3BSD
651/// format. The modifications from the original format were changing the value
652/// of the r_symbolnum field for "local" (r_extern == 0) relocation entries.
653/// This modification is required to support symbols in an arbitrary number of
654/// sections not just the three sections (text, data and bss) in a 4.3BSD file.
655/// Also the last 4 bits have had the r_type tag added to them.
656pub const relocation_info = packed struct {
657 /// offset in the section to what is being relocated
658 r_address: i32,
659
660 /// symbol index if r_extern == 1 or section ordinal if r_extern == 0
661 r_symbolnum: u24,
662
663 /// was relocated pc relative already
664 r_pcrel: u1,
665
666 /// 0=byte, 1=word, 2=long, 3=quad
667 r_length: u2,
668
669 /// does not include value of sym referenced
670 r_extern: u1,
671
672 /// if not 0, machine specific relocation type
673 r_type: u4,
674};
675
650/// After MacOS X 10.1 when a new load command is added that is required to be676/// After MacOS X 10.1 when a new load command is added that is required to be
651/// understood by the dynamic linker for the image to execute properly the677/// understood by the dynamic linker for the image to execute properly the
652/// LC_REQ_DYLD bit will be or'ed into the load command constant. If the dynamic678/// LC_REQ_DYLD bit will be or'ed into the load command constant. If the dynamic
...@@ -1086,13 +1112,58 @@ pub const N_ECOML = 0xe8;...@@ -1086,13 +1112,58 @@ pub const N_ECOML = 0xe8;
1086/// second stab entry with length information1112/// second stab entry with length information
1087pub const N_LENG = 0xfe;1113pub const N_LENG = 0xfe;
10881114
1089/// If a segment contains any sections marked with S_ATTR_DEBUG then all1115// For the two types of symbol pointers sections and the symbol stubs section
1090/// sections in that segment must have this attribute. No section other than1116// they have indirect symbol table entries. For each of the entries in the
1091/// a section marked with this attribute may reference the contents of this1117// section the indirect symbol table entries, in corresponding order in the
1092/// section. A section with this attribute may contain no symbols and must have1118// indirect symbol table, start at the index stored in the reserved1 field
1093/// a section type S_REGULAR. The static linker will not copy section contents1119// of the section structure. Since the indirect symbol table entries
1094/// from sections with this attribute into its output file. These sections1120// correspond to the entries in the section the number of indirect symbol table
1095/// generally contain DWARF debugging info.1121// entries is inferred from the size of the section divided by the size of the
1122// entries in the section. For symbol pointers sections the size of the entries
1123// in the section is 4 bytes and for symbol stubs sections the byte size of the
1124// stubs is stored in the reserved2 field of the section structure.
1125
1126/// section with only non-lazy symbol pointers
1127pub const S_NON_LAZY_SYMBOL_POINTERS = 0x6;
1128
1129/// section with only lazy symbol pointers
1130pub const S_LAZY_SYMBOL_POINTERS = 0x7;
1131
1132/// section with only symbol stubs, byte size of stub in the reserved2 field
1133pub const S_SYMBOL_STUBS = 0x8;
1134
1135/// section with only function pointers for initialization
1136pub const S_MOD_INIT_FUNC_POINTERS = 0x9;
1137
1138/// section with only function pointers for termination
1139pub const S_MOD_TERM_FUNC_POINTERS = 0xa;
1140
1141/// section contains symbols that are to be coalesced
1142pub const S_COALESCED = 0xb;
1143
1144/// zero fill on demand section (that can be larger than 4 gigabytes)
1145pub const S_GB_ZEROFILL = 0xc;
1146
1147/// section with only pairs of function pointers for interposing
1148pub const S_INTERPOSING = 0xd;
1149
1150/// section with only 16 byte literals
1151pub const S_16BYTE_LITERALS = 0xe;
1152
1153/// section contains DTrace Object Format
1154pub const S_DTRACE_DOF = 0xf;
1155
1156/// section with only lazy symbol pointers to lazy loaded dylibs
1157pub const S_LAZY_DYLIB_SYMBOL_POINTERS = 0x10;
1158
1159// If a segment contains any sections marked with S_ATTR_DEBUG then all
1160// sections in that segment must have this attribute. No section other than
1161// a section marked with this attribute may reference the contents of this
1162// section. A section with this attribute may contain no symbols and must have
1163// a section type S_REGULAR. The static linker will not copy section contents
1164// from sections with this attribute into its output file. These sections
1165// generally contain DWARF debugging info.
1166
1096/// a debug section1167/// a debug section
1097pub const S_ATTR_DEBUG = 0x02000000;1168pub const S_ATTR_DEBUG = 0x02000000;
10981169
...@@ -1154,3 +1225,35 @@ pub const VM_PROT_WRITE: vm_prot_t = 0x2;...@@ -1154,3 +1225,35 @@ pub const VM_PROT_WRITE: vm_prot_t = 0x2;
11541225
1155/// VM execute permission1226/// VM execute permission
1156pub const VM_PROT_EXECUTE: vm_prot_t = 0x4;1227pub const VM_PROT_EXECUTE: vm_prot_t = 0x4;
1228
1229pub const reloc_type_x86_64 = packed enum(u4) {
1230 /// for absolute addresses
1231 X86_64_RELOC_UNSIGNED = 0,
1232
1233 /// for signed 32-bit displacement
1234 X86_64_RELOC_SIGNED,
1235
1236 /// a CALL/JMP instruction with 32-bit displacement
1237 X86_64_RELOC_BRANCH,
1238
1239 /// a MOVQ load of a GOT entry
1240 X86_64_RELOC_GOT_LOAD,
1241
1242 /// other GOT references
1243 X86_64_RELOC_GOT,
1244
1245 /// must be followed by a X86_64_RELOC_UNSIGNED
1246 X86_64_RELOC_SUBTRACTOR,
1247
1248 /// for signed 32-bit displacement with a -1 addend
1249 X86_64_RELOC_SIGNED_1,
1250
1251 /// for signed 32-bit displacement with a -2 addend
1252 X86_64_RELOC_SIGNED_2,
1253
1254 /// for signed 32-bit displacement with a -4 addend
1255 X86_64_RELOC_SIGNED_4,
1256
1257 /// for thread local variables
1258 X86_64_RELOC_TLV,
1259};
src-self-hosted/link/MachO.zig+22-13
...@@ -32,6 +32,20 @@ const LoadCommand = union(enum) {...@@ -32,6 +32,20 @@ const LoadCommand = union(enum) {
32 .Dysymtab => |x| x.cmdsize,32 .Dysymtab => |x| x.cmdsize,
33 };33 };
34 }34 }
35
36 pub fn write(self: LoadCommand, file: *fs.File, offset: u64) !void {
37 return switch (self) {
38 .Segment => |cmd| writeGeneric(cmd, file, offset),
39 .LinkeditData => |cmd| writeGeneric(cmd, file, offset),
40 .Symtab => |cmd| writeGeneric(cmd, file, offset),
41 .Dysymtab => |cmd| writeGeneric(cmd, file, offset),
42 };
43 }
44
45 fn writeGeneric(cmd: anytype, file: *fs.File, offset: u64) !void {
46 const slice = [1]@TypeOf(cmd){cmd};
47 return file.pwriteAll(mem.sliceAsBytes(slice[0..1]), offset);
48 }
35};49};
3650
37base: File,51base: File,
...@@ -258,8 +272,7 @@ pub fn flush(self: *MachO, module: *Module) !void {...@@ -258,8 +272,7 @@ pub fn flush(self: *MachO, module: *Module) !void {
258272
259 var last_cmd_offset: usize = @sizeOf(macho.mach_header_64);273 var last_cmd_offset: usize = @sizeOf(macho.mach_header_64);
260 for (self.load_commands.items) |cmd| {274 for (self.load_commands.items) |cmd| {
261 const cmd_to_write = [1]@TypeOf(cmd){cmd};275 try cmd.write(&self.base.file.?, last_cmd_offset);
262 try self.base.file.?.pwriteAll(mem.sliceAsBytes(cmd_to_write[0..1]), last_cmd_offset);
263 last_cmd_offset += cmd.cmdsize();276 last_cmd_offset += cmd.cmdsize();
264 }277 }
265 const off = @sizeOf(macho.mach_header_64) + @sizeOf(macho.segment_command_64);278 const off = @sizeOf(macho.mach_header_64) + @sizeOf(macho.segment_command_64);
...@@ -346,19 +359,18 @@ pub fn updateDecl(self: *MachO, module: *Module, decl: *Module.Decl) !void {...@@ -346,19 +359,18 @@ pub fn updateDecl(self: *MachO, module: *Module, decl: *Module.Decl) !void {
346 .n_desc = 0,359 .n_desc = 0,
347 .n_value = addr,360 .n_value = addr,
348 };361 };
349 self.offset_table.items[decl.link.macho.offset_table_index.?] = addr;
350362
363 // Since we updated the vaddr and the size, each corresponding export symbol also needs to be updated.
364 const decl_exports = module.decl_exports.get(decl) orelse &[0]*Module.Export{};
365 try self.updateDeclExports(module, decl, decl_exports);
351 try self.writeSymbol(decl.link.macho.symbol_table_index.?);366 try self.writeSymbol(decl.link.macho.symbol_table_index.?);
352367
353 const text_section = self.sections.items[self.text_section_index.?];368 const text_section = self.sections.items[self.text_section_index.?];
354 const section_offset = symbol.n_value - text_section.addr;369 const section_offset = symbol.n_value - text_section.addr;
355 const file_offset = text_section.offset + section_offset;370 const file_offset = text_section.offset + section_offset;
356 log.debug("file_offset 0x{x}\n", .{file_offset});371 log.debug("file_offset 0x{x}\n", .{file_offset});
357 try self.base.file.?.pwriteAll(code, file_offset);
358372
359 // Since we updated the vaddr and the size, each corresponding export symbol also needs to be updated.373 try self.base.file.?.pwriteAll(code, file_offset);
360 const decl_exports = module.decl_exports.get(decl) orelse &[0]*Module.Export{};
361 return self.updateDeclExports(module, decl, decl_exports);
362}374}
363375
364pub fn updateDeclLineNumber(self: *MachO, module: *Module, decl: *const Module.Decl) !void {}376pub fn updateDeclLineNumber(self: *MachO, module: *Module, decl: *const Module.Decl) !void {}
...@@ -374,7 +386,7 @@ pub fn updateDeclExports(...@@ -374,7 +386,7 @@ pub fn updateDeclExports(
374386
375 if (decl.link.macho.symbol_table_index == null) return;387 if (decl.link.macho.symbol_table_index == null) return;
376388
377 var decl_sym = self.symbol_table.items[decl.link.macho.symbol_table_index.?];389 const decl_sym = &self.symbol_table.items[decl.link.macho.symbol_table_index.?];
378 // TODO implement390 // TODO implement
379 if (exports.len == 0) return;391 if (exports.len == 0) return;
380392
...@@ -504,10 +516,7 @@ fn allocateTextBlock(self: *MachO, text_block: *TextBlock, new_block_size: u64,...@@ -504,10 +516,7 @@ fn allocateTextBlock(self: *MachO, text_block: *TextBlock, new_block_size: u64,
504 const text_capacity = self.allocatedSize(text_section.offset);516 const text_capacity = self.allocatedSize(text_section.offset);
505 const needed_size = (addr + new_block_size) - text_section.addr;517 const needed_size = (addr + new_block_size) - text_section.addr;
506 log.debug("text capacity 0x{x}, needed size 0x{x}\n", .{ text_capacity, needed_size });518 log.debug("text capacity 0x{x}, needed size 0x{x}\n", .{ text_capacity, needed_size });
507519 assert(needed_size <= text_capacity); // TODO handle growth
508 if (needed_size > text_capacity) {
509 // TODO handle growth
510 }
511520
512 self.last_text_block = text_block;521 self.last_text_block = text_block;
513 text_section.size = needed_size;522 text_section.size = needed_size;
...@@ -659,7 +668,7 @@ fn writeSymbol(self: *MachO, index: usize) !void {...@@ -659,7 +668,7 @@ fn writeSymbol(self: *MachO, index: usize) !void {
659 defer tracy.end();668 defer tracy.end();
660669
661 const symtab = &self.load_commands.items[self.symtab_cmd_index.?].Symtab;670 const symtab = &self.load_commands.items[self.symtab_cmd_index.?].Symtab;
662 var sym = [1]macho.nlist_64{self.symbol_table.items[index]};671 const sym = [1]macho.nlist_64{self.symbol_table.items[index]};
663 const off = symtab.symoff + @sizeOf(macho.nlist_64) * index;672 const off = symtab.symoff + @sizeOf(macho.nlist_64) * index;
664 log.debug("writing symbol {} at 0x{x}\n", .{ sym[0], off });673 log.debug("writing symbol {} at 0x{x}\n", .{ sym[0], off });
665 try self.base.file.?.pwriteAll(mem.sliceAsBytes(sym[0..1]), off);674 try self.base.file.?.pwriteAll(mem.sliceAsBytes(sym[0..1]), off);