RISC-V: Clarify the addends of pc-relative access.

The original discussion was here,
https://github.com/riscv/riscv-elf-psabi-doc/issues/184

After discussing with Kito Cheng, I think the addends of %pcrel_hi
and %pcrel_lo are both allowed in GNU toolchain.  However, both of
the them mean the offset of symbols, rather than the pc address.
But the addends of %got_pcrel_hi and it's %pcrel_lo do not look
reasonable.  I believe gcc won't generate the got patterns with
addends, so linker should report dangerous relocation errors,
in case the assembly code use them.

Another issue was here,
https://sourceware.org/pipermail/binutils/2021-June/116983.html

At the beginnig, I suppose %pcrel_hi and %pcrel_lo are valid only
when they are in the same input section.  But Jim Wilson points out
that gcc may generate %hi and %lo in the different input sections,
when -freorder-blocks-and-partition option is used.  So that a memory
references for a loop may have the %hi outside the loop, but the %lo
remain in the loop.  However, it is hard to create the testcases,
to see if %pcrel_hi and %pcrel_lo have the same behavior.

Unfortunately, I notice that the current pcrel resolver cannot
work for the above case.  For now we build a hash table for pcrel
at the start of riscv_elf_relocate_section, and then free the hash
at the end.  But riscv_elf_relocate_section only handles an input
section at a time, so that means we can only resolve the %pcrel_hi
and %pcrel_lo which are in the same input section.  Otherwise, like
the above case, we will report "%pcrel_lo missing matching %pcrel_hi"
for them.  I have no plan to improve this in the short-term, so maybe
we can wait until someone meets the problem before we deal with it.

bfd/
    * elfnn-riscv.c (riscv_pcrel_hi_reloc): Added field to store
    the original relocation type, in case the type is converted to
    R_RISCV_HI20.
    (riscv_pcrel_lo_reloc): Removed unused name field.
    (riscv_pcrel_relocs): Added comments.
    (riscv_zero_pcrel_hi_reloc): Removed unused input_bfd.
    (riscv_record_pcrel_hi_reloc): Updated.
    (riscv_record_pcrel_lo_reloc): Likewise.
    (riscv_resolve_pcrel_lo_relocs): Likewise.  Check the original
    type of auipc, to make sure the %pcrel_lo without any addends.
    Otherwise, report dangerous relocation error.
    (riscv_elf_relocate_section): Updated above functions are changed.
    For R_RISCV_GOT_HI20, report dangerous relocation error when addend
    isn't zero.
ld/
    * testsuite/ld-riscv-elf/ld-riscv-elf.exp: Updated.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3a.d: New testcase.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3a.s: Likewise.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3b.d: New testcase.
    Should report error since the %pcrel_lo with addend refers to
    %got_pcrel_hi.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3b.s: Likewise.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3c.d: New testcase.
    Should report error since the %got_pcrel_hi with addend.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3c.s: Likewise.
    * testsuite/ld-riscv-elf/pcrel-lo-addend-3.ld: Likewise.

1 files changed, 92 insertions, 68 deletions

diff --git a/bfd/elfnn-riscv.c b/bfd/elfnn-riscv.c
index eef1e80..f206708 100644
--- a/bfd/elfnn-riscv.c
+++ b/bfd/elfnn-riscv.c
@@ -1739,25 +1739,41 @@ perform_relocation (const reloc_howto_type *howto,
 
 typedef struct
 {
+  /* PC value.  */
   bfd_vma address;
+  /* Relocation value with addend.  */
   bfd_vma value;
+  /* Original reloc type.  */
+  int type;
 } riscv_pcrel_hi_reloc;
 
 typedef struct riscv_pcrel_lo_reloc
 {
+  /* PC value of auipc.  */
+  bfd_vma address;
+  /* Internal relocation.  */
+  const Elf_Internal_Rela *reloc;
+  /* Record the following information helps to resolve the %pcrel
+     which cross different input section.  For now we build a hash
+     for pcrel at the start of riscv_elf_relocate_section, and then
+     free the hash at the end.  But riscv_elf_relocate_section only
+     handles an input section at a time, so that means we can only
+     resolve the %pcrel_hi and %pcrel_lo which are in the same input
+     section.  Otherwise, we will report dangerous relocation errors
+     for those %pcrel which are not in the same input section.  */
   asection *input_section;
   struct bfd_link_info *info;
   reloc_howto_type *howto;
-  const Elf_Internal_Rela *reloc;
-  bfd_vma addr;
-  const char *name;
   bfd_byte *contents;
+  /* The next riscv_pcrel_lo_reloc.  */
   struct riscv_pcrel_lo_reloc *next;
 } riscv_pcrel_lo_reloc;
 
 typedef struct
 {
+  /* Hash table for riscv_pcrel_hi_reloc.  */
   htab_t hi_relocs;
+  /* Linked list for riscv_pcrel_lo_reloc.  */
   riscv_pcrel_lo_reloc *lo_relocs;
 } riscv_pcrel_relocs;
 
@@ -1805,8 +1821,7 @@ riscv_zero_pcrel_hi_reloc (Elf_Internal_Rela *rel,
 			   bfd_vma pc,
 			   bfd_vma addr,
 			   bfd_byte *contents,
-			   const reloc_howto_type *howto,
-			   bfd *input_bfd ATTRIBUTE_UNUSED)
+			   const reloc_howto_type *howto)
 {
   /* We may need to reference low addreses in PC-relative modes even when the
      PC is far away from these addresses.  For example, undefweak references
@@ -1839,11 +1854,14 @@ riscv_zero_pcrel_hi_reloc (Elf_Internal_Rela *rel,
 }
 
 static bool
-riscv_record_pcrel_hi_reloc (riscv_pcrel_relocs *p, bfd_vma addr,
-			     bfd_vma value, bool absolute)
+riscv_record_pcrel_hi_reloc (riscv_pcrel_relocs *p,
+			     bfd_vma addr,
+			     bfd_vma value,
+			     int type,
+			     bool absolute)
 {
   bfd_vma offset = absolute ? value : value - addr;
-  riscv_pcrel_hi_reloc entry = {addr, offset};
+  riscv_pcrel_hi_reloc entry = {addr, offset, type};
   riscv_pcrel_hi_reloc **slot =
     (riscv_pcrel_hi_reloc **) htab_find_slot (p->hi_relocs, &entry, INSERT);
 
@@ -1857,20 +1875,19 @@ riscv_record_pcrel_hi_reloc (riscv_pcrel_relocs *p, bfd_vma addr,
 
 static bool
 riscv_record_pcrel_lo_reloc (riscv_pcrel_relocs *p,
+			     bfd_vma addr,
+			     const Elf_Internal_Rela *reloc,
 			     asection *input_section,
 			     struct bfd_link_info *info,
 			     reloc_howto_type *howto,
-			     const Elf_Internal_Rela *reloc,
-			     bfd_vma addr,
-			     const char *name,
 			     bfd_byte *contents)
 {
   riscv_pcrel_lo_reloc *entry;
   entry = (riscv_pcrel_lo_reloc *) bfd_malloc (sizeof (riscv_pcrel_lo_reloc));
   if (entry == NULL)
     return false;
-  *entry = (riscv_pcrel_lo_reloc) {input_section, info, howto, reloc, addr,
-				   name, contents, p->lo_relocs};
+  *entry = (riscv_pcrel_lo_reloc) {addr, reloc, input_section, info,
+				   howto, contents, p->lo_relocs};
   p->lo_relocs = entry;
   return true;
 }
@@ -1884,26 +1901,34 @@ riscv_resolve_pcrel_lo_relocs (riscv_pcrel_relocs *p)
     {
       bfd *input_bfd = r->input_section->owner;
 
-      riscv_pcrel_hi_reloc search = {r->addr, 0};
+      riscv_pcrel_hi_reloc search = {r->address, 0, 0};
       riscv_pcrel_hi_reloc *entry = htab_find (p->hi_relocs, &search);
-      if (entry == NULL
-	  /* Check the overflow when adding reloc addend.  */
-	  || (RISCV_CONST_HIGH_PART (entry->value)
-	      != RISCV_CONST_HIGH_PART (entry->value + r->reloc->r_addend)))
+      /* There may be a risk if the %pcrel_lo with addend refers to
+	 an IFUNC symbol.  The %pcrel_hi has been relocated to plt,
+	 so the corresponding %pcrel_lo with addend looks wrong.  */
+      char *string = NULL;
+      if (entry == NULL)
+	string = _("%pcrel_lo missing matching %pcrel_hi");
+      else if (entry->type == R_RISCV_GOT_HI20
+	       && r->reloc->r_addend != 0)
+	string = _("%pcrel_lo with addend isn't allowed for R_RISCV_GOT_HI20");
+      else if (RISCV_CONST_HIGH_PART (entry->value)
+	       != RISCV_CONST_HIGH_PART (entry->value + r->reloc->r_addend))
 	{
-	  char *string;
-	  if (entry == NULL)
-	    string = _("%pcrel_lo missing matching %pcrel_hi");
-	  else if (asprintf (&string,
-			     _("%%pcrel_lo overflow with an addend, the "
-			       "value of %%pcrel_hi is 0x%" PRIx64 " without "
-			       "any addend, but may be 0x%" PRIx64 " after "
-			       "adding the %%pcrel_lo addend"),
-			     (int64_t) RISCV_CONST_HIGH_PART (entry->value),
-			     (int64_t) RISCV_CONST_HIGH_PART
+	  /* Check the overflow when adding reloc addend.  */
+	  if (asprintf (&string,
+			_("%%pcrel_lo overflow with an addend, the "
+			  "value of %%pcrel_hi is 0x%" PRIx64 " without "
+			  "any addend, but may be 0x%" PRIx64 " after "
+			  "adding the %%pcrel_lo addend"),
+			(int64_t) RISCV_CONST_HIGH_PART (entry->value),
+			(int64_t) RISCV_CONST_HIGH_PART
 				(entry->value + r->reloc->r_addend)) == -1)
 	    string = _("%pcrel_lo overflow with an addend");
+	}
 
+      if (string != NULL)
+	{
 	  (*r->info->callbacks->reloc_dangerous)
 	    (r->info, string, input_bfd, r->input_section, r->reloc->r_offset);
 	  return true;
@@ -2225,12 +2250,9 @@ riscv_elf_relocate_section (bfd *output_bfd,
 		relocation = base_got->output_section->vma
 			     + base_got->output_offset + off;
 
-		r_type = ELFNN_R_TYPE (rel->r_info);
-		howto = riscv_elf_rtype_to_howto (input_bfd, r_type);
-		if (howto == NULL)
-		  r = bfd_reloc_notsupported;
-		else if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
-						       relocation, false))
+		if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
+						  relocation, r_type,
+						  false))
 		  r = bfd_reloc_overflow;
 		goto do_relocation;
 
@@ -2242,12 +2264,9 @@ riscv_elf_relocate_section (bfd *output_bfd,
 		goto do_relocation;
 
 	      case R_RISCV_PCREL_HI20:
-		r_type = ELFNN_R_TYPE (rel->r_info);
-		howto = riscv_elf_rtype_to_howto (input_bfd, r_type);
-		if (howto == NULL)
-		  r = bfd_reloc_notsupported;
-		else if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
-						       relocation, false))
+		if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
+						  relocation, r_type,
+						  false))
 		  r = bfd_reloc_overflow;
 		goto do_relocation;
 
@@ -2384,21 +2403,29 @@ riscv_elf_relocate_section (bfd *output_bfd,
 		  local_got_offsets[r_symndx] |= 1;
 		}
 	    }
-	  relocation = sec_addr (htab->elf.sgot) + off;
-	  absolute = riscv_zero_pcrel_hi_reloc (rel,
-						info,
-						pc,
-						relocation,
-						contents,
-						howto,
-						input_bfd);
-	  r_type = ELFNN_R_TYPE (rel->r_info);
-	  howto = riscv_elf_rtype_to_howto (input_bfd, r_type);
-	  if (howto == NULL)
-	    r = bfd_reloc_notsupported;
-	  else if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
-						 relocation, absolute))
-	    r = bfd_reloc_overflow;
+
+	  if (rel->r_addend != 0)
+	    {
+	      msg = _("The addend isn't allowed for R_RISCV_GOT_HI20");
+	      r = bfd_reloc_dangerous;
+	    }
+	  else
+	    {
+	      /* Address of got entry.  */
+	      relocation = sec_addr (htab->elf.sgot) + off;
+	      absolute = riscv_zero_pcrel_hi_reloc (rel, info, pc,
+						    relocation, contents,
+						    howto);
+	      /* Update howto if relocation is changed.  */
+	      howto = riscv_elf_rtype_to_howto (input_bfd,
+						ELFNN_R_TYPE (rel->r_info));
+	      if (howto == NULL)
+		r = bfd_reloc_notsupported;
+	      else if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
+						     relocation, r_type,
+						     absolute))
+		r = bfd_reloc_overflow;
+	    }
 	  break;
 
 	case R_RISCV_ADD8:
@@ -2499,20 +2526,16 @@ riscv_elf_relocate_section (bfd *output_bfd,
 	  }
 
 	case R_RISCV_PCREL_HI20:
-	  absolute = riscv_zero_pcrel_hi_reloc (rel,
-						info,
-						pc,
-						relocation,
-						contents,
-						howto,
-						input_bfd);
-	  r_type = ELFNN_R_TYPE (rel->r_info);
-	  howto = riscv_elf_rtype_to_howto (input_bfd, r_type);
+	  absolute = riscv_zero_pcrel_hi_reloc (rel, info, pc, relocation,
+						contents, howto);
+	  /* Update howto if relocation is changed.  */
+	  howto = riscv_elf_rtype_to_howto (input_bfd,
+					    ELFNN_R_TYPE (rel->r_info));
 	  if (howto == NULL)
 	    r = bfd_reloc_notsupported;
 	  else if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
 						 relocation + rel->r_addend,
-						 absolute))
+						 r_type, absolute))
 	    r = bfd_reloc_overflow;
 	  break;
 
@@ -2532,8 +2555,8 @@ riscv_elf_relocate_section (bfd *output_bfd,
 	      break;
 	    }
 
-	  if (riscv_record_pcrel_lo_reloc (&pcrel_relocs, input_section, info,
-					   howto, rel, relocation, name,
+	  if (riscv_record_pcrel_lo_reloc (&pcrel_relocs, relocation, rel,
+					   input_section, info, howto,
 					   contents))
 	    continue;
 	  r = bfd_reloc_overflow;
@@ -2726,7 +2749,8 @@ riscv_elf_relocate_section (bfd *output_bfd,
 	  BFD_ASSERT (off < (bfd_vma) -2);
 	  relocation = sec_addr (htab->elf.sgot) + off + (is_ie ? ie_off : 0);
 	  if (!riscv_record_pcrel_hi_reloc (&pcrel_relocs, pc,
-					    relocation, false))
+					    relocation, r_type,
+					    false))
 	    r = bfd_reloc_overflow;
 	  unresolved_reloc = false;
 	  break;