[binutils-gdb] [gdb/symtab] Fix data race in cooked_index_functions::expand_symtabs_matching

[binutils-gdb] [gdb/symtab] Fix data race in cooked_index_functions::expand_symtabs_matching

[Tom de Vries]

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=7d1a572d6b5194d36a96f36b3d28ce591341deb6

commit 7d1a572d6b5194d36a96f36b3d28ce591341deb6
Author: Tom de Vries <tdevries@suse.de>
Date:   Thu Jul 14 20:47:54 2022 +0200

    [gdb/symtab] Fix data race in cooked_index_functions::expand_symtabs_matching
    
    When building gdb with -fsanitize-threads and running test-case
    gdb.ada/char_enum_unicode.exp, I run into:
    ...
    WARNING: ThreadSanitizer: data race (pid=21301)^M
      Write of size 8 at 0x7b2000008080 by main thread:^M
        #0 free <null> (libtsan.so.2+0x4c5e2)^M
        #1 _dl_close_worker <null> (ld-linux-x86-64.so.2+0x4b7b)^M
        #2 convert_between_encodings() charset.c:584^M
      ...
        #21 cooked_index_functions::expand_symtabs_matching() read.c:18606
    ...
    
    This is fixed by making cooked_index_functions::expand_symtabs_matching wait
    for the cooked index finalization to be done.
    
    Tested on x86_64-linux.
    
    https://sourceware.org/bugzilla/show_bug.cgi?id=29311
    https://sourceware.org/bugzilla/show_bug.cgi?id=29286

Diff:
---
 gdb/dwarf2/cooked-index.h | 11 +++++++++--
 gdb/dwarf2/read.c         |  8 +++++---
 2 files changed, 14 insertions(+), 5 deletions(-)

diff --git a/gdb/dwarf2/cooked-index.h b/gdb/dwarf2/cooked-index.h
index 439cbb19fa7..f3c26480a81 100644
--- a/gdb/dwarf2/cooked-index.h
+++ b/gdb/dwarf2/cooked-index.h
@@ -295,6 +295,14 @@ public:
   explicit cooked_index_vector (vec_type &&vec);
   DISABLE_COPY_AND_ASSIGN (cooked_index_vector);
 
+  /* Wait until the finalization of the entire cooked_index_vector is
+     done.  */
+  void wait ()
+  {
+    for (auto &item : m_vector)
+      item->wait ();
+  }
+
   ~cooked_index_vector ()
   {
     /* The 'finalize' methods may be run in a different thread.  If
@@ -303,8 +311,7 @@ public:
        complete avoids this problem; and the cost seems ignorable
        because creating and immediately destroying the debug info is a
        relatively rare thing to do.  */
-    for (auto &item : m_vector)
-      item->wait ();
+    wait ();
   }
 
   /* A range over a vector of subranges.  */
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index 5ca787cc095..bcd01107377 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -18569,6 +18569,11 @@ cooked_index_functions::expand_symtabs_matching
   if (per_objfile->per_bfd->index_table == nullptr)
     return true;
 
+  cooked_index_vector *table
+    = (static_cast<cooked_index_vector *>
+       (per_objfile->per_bfd->index_table.get ()));
+  table->wait ();
+
   dw_expand_symtabs_matching_file_matcher (per_objfile, file_matcher);
 
   /* This invariant is documented in quick-functions.h.  */
@@ -18604,9 +18609,6 @@ cooked_index_functions::expand_symtabs_matching
     language_ada
   };
 
-  cooked_index_vector *table
-    = (static_cast<cooked_index_vector *>
-       (per_objfile->per_bfd->index_table.get ()));
   for (enum language lang : unique_styles)
     {
       std::vector<gdb::string_view> name_vec