[lld/mac] Make sure __thread_ptrs is in front of __thread_bss

[thakis]

Summary

The exact location doesn't matter, but it should be in front
of __thread_bss. We put it right in front of __thread_data
which is where ld64 seems to put it as well.

Fixes PR50769.

(As mentioned on the bug, there is probably a more structural
fix too, see comment 5. If we don't address this, it's likely
we'll run into this again with other synthetic sections. But
for now, let's fix the immediate breakage.)

[thakis]

The reason it's important that zerofill sections are at the end is because dyld does

bool ImageLoaderMachO::segHasTrailingZeroFill(unsigned int segIndex)
{
	return ( segWriteable(segIndex) && (segSize(segIndex) > segFileSize(segIndex)) );
}

to detect zerofill sections. It doesn't look at section flags but only on where and how large the section is (and then it maps with VM_PROT_ZF when needed).

[int3]

Thanks! I see why you were checking for the GOT in tlv-dylib.s now...

lld/MachO/OutputSegment.cpp

if (segname == segment_names::text) {
     // For each thread spawned, dyld will initialize its TLVs by copying the
     // address range from the start of the first thread-local data section to
     // the end of the last one. We therefore arrange these sections contiguously
     // to minimize the amount of memory used. Additionally, since zerofill
     // sections must be at the end of their segments, and since TLV data
     // sections can be zerofills, we end up putting all TLV data sections at the
     // end of the segment.
     switch (sectionType(osec->flags)) {
+    case S_THREAD_LOCAL_VARIABLE_POINTERS:
+      return std::numeric_limits<int>::max() - 3;
     case S_THREAD_LOCAL_REGULAR:
       return std::numeric_limits<int>::max() - 2;
     case S_THREAD_LOCAL_ZEROFILL:
       return std::numeric_limits<int>::max() - 1;
     case S_ZEROFILL:
       return std::numeric_limits<int>::max();
     default:
       return StringSwitch<int>(osec->name)

lld/test/MachO/tlv-dylib.s

 # CHECK-NEXT: movq [[#]](%rip), %rax # [[#%x, BAR:]]
 # CHECK-NEXT: movq [[#]](%rip), %rax # [[#%x, BAZ:]]
 
 # CHECK-LABEL: Bind table:
 # CHECK-DAG: __DATA       __thread_ptrs  0x{{0*}}[[#%x, FOO]] pointer 0   libtlv   _foo
 # CHECK-DAG: __DATA       __thread_ptrs  0x{{0*}}[[#%x, BAR]] pointer 0   libtlv   _bar
 # CHECK-DAG: __DATA_CONST __got          0x{{0*}}[[#%x, BAZ]] pointer 0   libtlv   _baz
 
+## Check `type` on the various TLV sections, and check that
+## nothing's after S_THREAD_LOCAL_ZEROFILL.
 # RUN: llvm-otool -lv %t/test | FileCheck --check-prefix=FLAGS %s
 # FLAGS:       sectname __got
 # FLAGS-NEXT:   segname __DATA_CONST
 # FLAGS-NEXT:      addr
 # FLAGS-NEXT:      size 0x0000000000000008
 # FLAGS-NEXT:    offset
 # FLAGS-NEXT:     align 2^3 (8)
 # FLAGS-NEXT:    reloff 0
 # FLAGS-NEXT:    nreloc 0
 # FLAGS-NEXT:      type S_NON_LAZY_SYMBOL_POINTERS
+# FLAGS:       sectname __thread_vars
+# FLAGS-NEXT:   segname __DATA
+# FLAGS-NEXT:      addr
+# FLAGS-NEXT:      size 0x0000000000000030
+# FLAGS-NEXT:    offset
+# FLAGS-NEXT:     align
+# FLAGS-NEXT:    reloff 0
+# FLAGS-NEXT:    nreloc 0
+# FLAGS-NEXT:      type S_THREAD_LOCAL_VARIABLES
 # FLAGS:       sectname __thread_ptrs
 # FLAGS-NEXT:   segname __DATA
 # FLAGS-NEXT:      addr
 # FLAGS-NEXT:      size 0x0000000000000010
 # FLAGS-NEXT:    offset
 # FLAGS-NEXT:     align 2^3 (8)
 # FLAGS-NEXT:    reloff 0
 # FLAGS-NEXT:    nreloc 0
 # FLAGS-NEXT:      type S_THREAD_LOCAL_VARIABLE_POINTERS
+# FLAGS:       sectname __thread_data
+# FLAGS-NEXT:   segname __DATA
+# FLAGS-NEXT:      addr
+# FLAGS-NEXT:      size 0x0000000000000008
+# FLAGS-NEXT:    offset
+# FLAGS-NEXT:     align
+# FLAGS-NEXT:    reloff 0
+# FLAGS-NEXT:    nreloc 0
+# FLAGS-NEXT:      type S_THREAD_LOCAL_REGULAR
+# FLAGS:       sectname __thread_bss
+# FLAGS-NEXT:   segname __DATA
+# FLAGS-NEXT:      addr
+# FLAGS-NEXT:      size 0x0000000000000008
+# FLAGS-NEXT:    offset
+# FLAGS-NEXT:     align 2^3 (8)
+# FLAGS-NEXT:    reloff 0
+# FLAGS-NEXT:    nreloc 0
+# FLAGS-NEXT:      type S_THREAD_LOCAL_ZEROFILL
 
 #--- libtlv.s
 .section __DATA,__thread_vars,thread_local_variables
 .globl _foo, _bar, _baz
 _foo:
 _bar:
 
 .text
 _baz:
 
 #--- test.s
 .globl _main
 _main:
   mov _foo@TLVP(%rip), %rax
   mov _bar@TLVP(%rip), %rax
 ## Add a GOT entry to make sure we don't mix it up with TLVs
   mov _baz@GOTPCREL(%rip), %rax
   ret
+
+## Add some TLVs to test too, so that we can test the ordering
+## of __thread_ptrs, __thread_data, and __thread_bss.
+.section __DATA,__thread_data,thread_local_regular
+_tfoo$tlv$init:
+  .quad 123
+
+.tbss _tbaz$tlv$init, 8, 3
+
+.section __DATA,__thread_vars,thread_local_variables
+.globl  _tfoo, _tbar
+_tfoo:
+  .quad  __tlv_bootstrap
+  .quad  0
+  .quad  _tfoo$tlv$init
+_tbaz:
+  .quad  __tlv_bootstrap
+  .quad  0
+  .quad  _tbaz$tlv$init