It looks like ld64 requires it. With this we seem to be able to bootstrap using
llvm-ar+/usr/bin/true instead of ar+ranlib (currently on stage2).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@241842
91177308-0d34-0410-b5e6-
96231b3b80d8
print32(Out, Kind, StringTable.size()); // byte count of the string table
Out << StringTable;
- if (Out.tell() % 2)
- Out << '\0';
+ // ld64 requires the next member header to start at an offset that is
+ // 4 bytes aligned.
+ unsigned Pad = OffsetToAlignment(Out.tell(), 4);
+ while (Pad--)
+ Out.write(uint8_t(0));
// Patch up the size of the symbol table now that we know how big it is.
unsigned Pos = Out.tell();
MACHO-NEXT: 0000000000000000 t _bar
MACHO-NEXT: 0000000000000001 T _foo
MACHO-NEXT: 0000000000000002 T _main
+
+Test that we pad the symbol table so that it ends in a multiple of 4 bytes:
+8 + 60 + 36 == 104
+RUN: rm -f %t.a
+RUN: llvm-ar --format=bsd rcs %t.a %p/Inputs/trivial-object-test.macho-x86-64
+RUN: FileCheck --check-prefix=MACHO-SYMTAB-ALIGN %s < %t.a
+MACHO-SYMTAB-ALIGN: !<arch>
+MACHO-SYMTAB-ALIGN-NEXT: #1/12 {{..........}} 0 0 0 36 `