[ELF] --emit-relocs: fix a crash if .rela.dyn is an empty output section

[MaskRay]

Summary

Fix PR48357: If .rela.dyn appears as an output section description, its type may
be SHT_RELA (due to the empty synthetic .rela.plt) while there is no input
section. The empty .rela.dyn may be retained due to a reference in a linker
script. Don't crash.

[MaskRay]

Improve test

[grimar]

lld/ELF/OutputSections.cpp

406

I wonder if the same crash can happen here with an empty .ARM.exidx?

[MaskRay]

lld/ELF/OutputSections.cpp

406

It can't (I believe flags & SHF_LINK_ORDER ensures there is at least one input section but it is very late for me and I don't think carefully though:( )

There are many arm-exidx-* tests and someone may need to check whether this is missing test coverage.

[grimar]

lld/ELF/OutputSections.cpp

406

SHF_LINK_ORDER case seems is fine. We don't keep any special flags for empty output
sections:

// We do not want to keep any special flags for output section
// in case it is empty.
bool isEmpty = (getFirstInputSection(sec) == nullptr);
if (isEmpty)
  sec->flags = flags & ((sec->nonAlloc ? 0 : (uint64_t)SHF_ALLOC) |
                        SHF_WRITE | SHF_EXECINSTR);

[grimar]

LGTM.

lld/ELF/OutputSections.cpp

 void OutputSection::finalize() {
   InputSection *first = getFirstInputSection(this);
 
   if (flags & SHF_LINK_ORDER) {
     // We must preserve the link order dependency of sections with the
     // SHF_LINK_ORDER flag. The dependency is indicated by the sh_link field. We
     // need to translate the InputSection sh_link to the OutputSection sh_link,
     // all InputSections in the OutputSection have the same dependency.
     if (auto *ex = dyn_cast<ARMExidxSyntheticSection>(first))
       link = ex->getLinkOrderDep()->getParent()->sectionIndex;
     else if (first->flags & SHF_LINK_ORDER)
       if (auto *d = first->getLinkOrderDep())
         link = d->getParent()->sectionIndex;
   }
 
   if (type == SHT_GROUP) {
     finalizeShtGroup(this, first);
     return;
   }
 
   if (!config->copyRelocs || (type != SHT_RELA && type != SHT_REL))
     return;
 
-  if (isa<SyntheticSection>(first))
+  // Skip if 'first' is synthetic, i.e. not a section created by --emit-relocs.
+  // Normally 'type' was changed by 'first' so 'first' should be non-null.
+  // However, if the output section is .rela.dyn, 'type' can be set by the empty
+  // synthetic .rela.plt and first can be null.
+  if (!first || isa<SyntheticSection>(first))
     return;
 
   link = in.symTab->getParent()->sectionIndex;
   // sh_info for SHT_REL[A] sections should contain the section header index of
   // the section to which the relocation applies.
   InputSectionBase *s = first->getRelocatedSection();
   info = s->getOutputSection()->sectionIndex;
   flags |= SHF_INFO_LINK;

lld/test/ELF/linkerscript/emit-relocs-rela-dyn.s

+# REQUIRES: x86
+## PR48357: If .rela.dyn appears as an output section description, its type may
+## be SHT_RELA (due to the empty synthetic .rela.plt) while there is no input
+## section. The empty .rela.dyn may be retained due to a reference. Don't crash.
+
+# RUN: llvm-mc -filetype=obj -triple=x86_64 /dev/null -o %t.o
+# RUN: ld.lld -shared --emit-relocs -T %s %t.o -o %t
+# RUN: llvm-readelf -S %t | FileCheck %s
+
+## Note, sh_link of such an empty .rela.dyn is 0.
+# CHECK: Name      Type Address          Off    Size   ES Flg Lk Inf Al
+# CHECK: .rela.dyn RELA 0000000000000000 001000 000000 18   A  0   0  8
+
+SECTIONS {
+  .rela.dyn : { *(.rela*) }
+  __rela_offset = ABSOLUTE(ADDR(.rela.dyn));
+}