From 74edb473c9ecf5e2053ecf8e429ee608feafb9e1 Mon Sep 17 00:00:00 2001
From: Tamar Christina <tamar.christina@arm.com>
Date: Thu, 1 Apr 2021 17:10:38 +0100
Subject: [PATCH] PE/Windows x86_64: Fix weak undef symbols after image base
 change

The change in PR19011 changed the image load address from being in the lower
32-bit address space to the higher 64-bit address space.

However when you have a weak undef symbol which stays undef at the end of
linking the linker has to resolve this (Windows loader does not support undef
symbols).  As such typically these would resolve to 0.

The relocation used for these weak symbols are the normal 32-bit PC_REL call
relocs.  So when doing the overflow check LD checks if the distance between the
symbol and the call is within range.  However now that the load address is
> 32-bits and the symbol val is 0 this overflow check will always fail.

As such the linker gives a bogus error.  This patch makes the linker not emit
the overflow failure but chooses to still let the check be performed (as it's
mid-end code).

One down side of this is that it does break the common convention that the call
be to sym at 0x0. i.e. before you'd get

      401015:   74 05                   je     40101c
      401017:   e8 e4 ef bf ff          callq  0

and now you get

   140001015:   74 05                   je     14000101c
   140001017:   e8 e4 ef ff bf          call   100000000

since the call is PC_REL there's no way to get the range large enough to
resolve to 0.  As such I have chosen to leave it as the furthest simple range
that we can still represent.

By only ignoring the error we leave the symbol value itself to still be 0
such that the if(<symbol>) checks still work correctly.

bfd/ChangeLog:

2021-04-01  Tamar Christina  <tamar.christina@arm.com>

	PR ld/26659
	* cofflink.c (_bfd_coff_generic_relocate_section): Ignore overflow.

ld/ChangeLog:

2021-04-01  Tamar Christina  <tamar.christina@arm.com>

	PR ld/26659
	* testsuite/ld-pe/pe.exp: Add test.
	* testsuite/ld-pe/pr26659-weak-undef-sym.d: New test.
	* testsuite/ld-pe/pr26659-weak-undef-sym.s: New test.
---
 bfd/ChangeLog                               |  5 +++
 bfd/cofflink.c                              | 15 ++++++++
 ld/ChangeLog                                |  7 ++++
 ld/testsuite/ld-pe/pe.exp                   |  1 +
 ld/testsuite/ld-pe/pr26659-weak-undef-sym.d | 32 +++++++++++++++++
 ld/testsuite/ld-pe/pr26659-weak-undef-sym.s | 38 +++++++++++++++++++++
 6 files changed, 98 insertions(+)
 create mode 100644 ld/testsuite/ld-pe/pr26659-weak-undef-sym.d
 create mode 100644 ld/testsuite/ld-pe/pr26659-weak-undef-sym.s

diff --git a/bfd/cofflink.c b/bfd/cofflink.c
index c6b65c761d7..dd3e8dd4bf6 100644
--- a/bfd/cofflink.c
+++ b/bfd/cofflink.c
@@ -3129,6 +3129,21 @@ _bfd_coff_generic_relocate_section (bfd *output_bfd,
 	  return false;
 	case bfd_reloc_overflow:
 	  {
+
+	    /* Ignore any weak undef symbols that may have overflowed.  Due to
+	       PR ld/19011 the base address is now in the upper 64-bit address
+	       range.  This means that when _bfd_final_link_relocate calculates
+	       the overlow it takes the distance between the symbol and the VMA
+	       which will now always overflow as 0 - 64-bit addr > 32-bit range
+	       of the relocation.  This ends up creating PR ld/26659.  */
+	    if (val == 0
+		/* Reverse the hack where 4 is subtracted from the addend.  */
+		&& (addend + 4) == 0
+		&& sym->n_sclass == C_NT_WEAK
+		&& bfd_coff_classify_symbol (output_bfd, sym)
+		     == COFF_SYMBOL_UNDEFINED)
+	      break;
+
 	    const char *name;
 	    char buf[SYMNMLEN + 1];
 
diff --git a/ld/testsuite/ld-pe/pe.exp b/ld/testsuite/ld-pe/pe.exp
index b1a200180be..a536b51c3e6 100644
--- a/ld/testsuite/ld-pe/pe.exp
+++ b/ld/testsuite/ld-pe/pe.exp
@@ -81,6 +81,7 @@ run_dump_test "reloc"
 run_dump_test "weakdef-1"
 
 run_dump_test "pr19803"
+run_dump_test "pr26659-weak-undef-sym"
 set pr19803_dll {
     {   "PR 19803: not exporting swept symbols"
 	"-shared --out-implib dx.dll --gc-sections"
diff --git a/ld/testsuite/ld-pe/pr26659-weak-undef-sym.d b/ld/testsuite/ld-pe/pr26659-weak-undef-sym.d
new file mode 100644
index 00000000000..0b48994877d
--- /dev/null
+++ b/ld/testsuite/ld-pe/pr26659-weak-undef-sym.d
@@ -0,0 +1,32 @@
+#source: pr26659-weak-undef-sym.s
+#target: x86_64-*-cygwin* x86_64-*-pe x86_64-*-mingw*
+#ld: -e0
+#objdump: -d
+
+#...
+0000000140001000 <foo>:
+   140001000:	55                   	push   %rbp
+   140001001:	48 89 e5             	mov    %rsp,%rbp
+   140001004:	48 83 ec 20          	sub    \$0x20,%rsp
+   140001008:	89 4d 10             	mov    %ecx,0x10\(%rbp\)
+   14000100b:	48 8b 05 ee 0f 00 00 	mov    0xfee\(%rip\),%rax        # 140002000 <__data_end__>
+   140001012:	48 85 c0             	test   %rax,%rax
+   140001015:	74 05                	je     14000101c <foo\+0x1c>
+   140001017:	e8 e4 ef ff bf       	call   100000000 <__size_of_stack_reserve__\+0xffe00000>
+   14000101c:	48 8b 05 ed 0f 00 00 	mov    0xfed\(%rip\),%rax        # 140002010 <.refptr.bar2>
+   140001023:	48 85 c0             	test   %rax,%rax
+   140001026:	74 05                	je     14000102d <foo\+0x2d>
+   140001028:	e8 d3 ef ff bf       	call   100000000 <__size_of_stack_reserve__\+0xffe00000>
+   14000102d:	8b 45 10             	mov    0x10\(%rbp\),%eax
+   140001030:	0f af c0             	imul   %eax,%eax
+   140001033:	48 83 c4 20          	add    \$0x20,%rsp
+   140001037:	5d                   	pop    %rbp
+   140001038:	c3                   	ret    
+   140001039:	90                   	nop
+   14000103a:	90                   	nop
+   14000103b:	90                   	nop
+   14000103c:	90                   	nop
+   14000103d:	90                   	nop
+   14000103e:	90                   	nop
+   14000103f:	90                   	nop
+#pass
diff --git a/ld/testsuite/ld-pe/pr26659-weak-undef-sym.s b/ld/testsuite/ld-pe/pr26659-weak-undef-sym.s
new file mode 100644
index 00000000000..7a42759e752
--- /dev/null
+++ b/ld/testsuite/ld-pe/pr26659-weak-undef-sym.s
@@ -0,0 +1,38 @@
+        .text
+        .globl  foo
+        .def    foo;    .scl    2;      .type   32;     .endef
+foo:
+        pushq   %rbp
+        movq    %rsp, %rbp
+        subq    $32, %rsp
+        movl    %ecx, 16(%rbp)
+        movq    .refptr.bar1(%rip), %rax
+        testq   %rax, %rax
+        je      .L2
+        call    bar1
+.L2:
+        movq    .refptr.bar2(%rip), %rax
+        testq   %rax, %rax
+        je      .L3
+        call    bar2
+.L3:
+        movl    16(%rbp), %eax
+        imull   %eax, %eax
+        addq    $32, %rsp
+        popq    %rbp
+        ret
+        .weak   bar2
+        .weak   bar1
+        .def    bar1;   .scl    2;      .type   32;     .endef
+        .def    bar2;   .scl    2;      .type   32;     .endef
+        .section        .rdata$.refptr.bar2, "dr"
+        .globl  .refptr.bar2
+        .linkonce       discard
+.refptr.bar2:
+        .quad   bar2
+        .section        .rdata$.refptr.bar1, "dr"
+        .globl  .refptr.bar1
+        .linkonce       discard
+.refptr.bar1:
+        .quad   bar1
+
-- 
2.27.0