Pre-read DWZ file in DWARF reader

While working on background reading of DWARF, I came across the
DWZ-reading code.  This code can query the user (via the debuginfod
support) -- something that cannot be done off the main thread.

Looking into it, I realized that this code can be run much earlier,
avoiding this problem.  Digging a bit deeper, I also found a
discrepancy here between how the DWARF reader works in "readnow" mode
as compared to the normal modes.

This patch cleans this up by trying to read the DWZ file earlier, and
also by having the DWARF reader convert any exception here into a
warning.  This unifies the various cases, but also makes it so that
errors do not prevent gdb from continuing on to the extent possible.

Regression tested on x86-64 Fedora 38.

diff --git a/gdb/dwarf2/dwz.c b/gdb/dwarf2/dwz.c
index bd2bcb91d83f3566cfd533bb7d54251b52165f32..edd5b56d7786809448b8e9b9faee8908fcbac3a3 100644 (file)
--- a/gdb/dwarf2/dwz.c
+++ b/gdb/dwarf2/dwz.c
@@ -29,6 +29,7 @@
 #include "gdbcore.h"
 #include "gdbsupport/pathstuff.h"
 #include "gdbsupport/scoped_fd.h"
+#include "run-on-main-thread.h"
 
 const char *
 dwz_file::read_string (struct objfile *objfile, LONGEST str_offset)
@@ -196,8 +197,20 @@ dwarf2_get_dwz_file (dwarf2_per_bfd *per_bfd, bool require)
   size_t buildid_len;
   bfd_byte *buildid;
 
-  if (per_bfd->dwz_file != NULL)
-    return per_bfd->dwz_file.get ();
+  if (per_bfd->dwz_file.has_value ())
+    {
+      dwz_file *result = per_bfd->dwz_file->get ();
+      if (require && result == nullptr)
+       error (_("could not read '.gnu_debugaltlink' section"));
+      return result;
+    }
+
+  /* This may query the user via the debuginfod support, so it may
+     only be run in the main thread.  */
+  gdb_assert (is_main_thread ());
+
+  /* Set this early, so that on error it remains NULL.  */
+  per_bfd->dwz_file.emplace (nullptr);
 
   bfd_set_error (bfd_error_no_error);
   gdb::unique_xmalloc_ptr<char> data
@@ -283,5 +296,5 @@ dwarf2_get_dwz_file (dwarf2_per_bfd *per_bfd, bool require)
 
   gdb_bfd_record_inclusion (per_bfd->obfd, result->dwz_bfd.get ());
   per_bfd->dwz_file = std::move (result);
-  return per_bfd->dwz_file.get ();
+  return per_bfd->dwz_file->get ();
 }

diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index c4948164fb1620a8c4c1c664a137538e42e34eb7..7a3b9c68279262732e5d44662c2c44c891816d50 100644 (file)
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -3213,6 +3213,17 @@ dwarf2_initialize_objfile (struct objfile *objfile)
 
   dwarf_read_debug_printf ("called");
 
+  /* Try to fetch any potential dwz file early, while still on the
+     main thread.  */
+  try
+    {
+      dwarf2_get_dwz_file (per_bfd);
+    }
+  catch (const gdb_exception_error &err)
+    {
+      warning (_("%s"), err.what ());
+    }
+
   /* If we're about to read full symbols, don't bother with the
      indices.  In this case we also don't care if some other debug
      format is making psymtabs, because they are all about to be
@@ -5120,16 +5131,7 @@ create_all_units (dwarf2_per_objfile *per_objfile)
                                  &per_objfile->per_bfd->abbrev, 0,
                                  types_htab, rcuh_kind::TYPE);
 
-  dwz_file *dwz;
-  try
-    {
-      dwz = dwarf2_get_dwz_file (per_objfile->per_bfd);
-    }
-  catch (const gdb_exception_error &)
-    {
-      per_objfile->per_bfd->all_units.clear ();
-      throw;
-    }
+  dwz_file *dwz = dwarf2_get_dwz_file (per_objfile->per_bfd);
   if (dwz != NULL)
     {
       /* Pre-read the sections we'll need to construct an index.  */

diff --git a/gdb/dwarf2/read.h b/gdb/dwarf2/read.h
index 542b57426aa39b83a6fa439e390309130b7c9e08..da907729320381d5f905f91e048fe21b3615beb6 100644 (file)
--- a/gdb/dwarf2/read.h
+++ b/gdb/dwarf2/read.h
@@ -524,7 +524,7 @@ public:
 
   /* The shared '.dwz' file, if one exists.  This is used when the
      original data was compressed using 'dwz -m'.  */
-  std::unique_ptr<struct dwz_file> dwz_file;
+  gdb::optional<std::unique_ptr<struct dwz_file>> dwz_file;
 
   /* Whether copy relocations are supported by this object format.  */
   bool can_copy;

diff --git a/gdb/testsuite/gdb.dwarf2/dwzbuildid.exp b/gdb/testsuite/gdb.dwarf2/dwzbuildid.exp
index 76e3a7d6e4260d261f5be958069f5598a3f3ff83..3814323926e7fc2999db3570138197564e365a61 100644 (file)
--- a/gdb/testsuite/gdb.dwarf2/dwzbuildid.exp
+++ b/gdb/testsuite/gdb.dwarf2/dwzbuildid.exp
@@ -142,13 +142,6 @@ proc do_test {} {
 
     gdb_load ${::binfile}-${::testname}
 
-    if { $::testname == "mismatch" && [readnow] } {
-       # Main is found in the minimal symbols.  When using readnow, a
-       # failure to read the dwarf also causes the minimal symbols to be
-       # unavailable.
-       # Setup a kfail for "FAIL: gdb_breakpoint: set breakpoint at main".
-       setup_kfail "symtab/26797" *-*-*
-    }
     if {![runto_main]} {
        return
     }

diff --git a/gdb/testsuite/gdb.dwarf2/no-gnu-debuglink.exp b/gdb/testsuite/gdb.dwarf2/no-gnu-debuglink.exp
index 538004ec95a6ba66a53ac56b0721b680adc4cf4e..a76136fdfd80fe256af70429fa86767bda88064b 100644 (file)
--- a/gdb/testsuite/gdb.dwarf2/no-gnu-debuglink.exp
+++ b/gdb/testsuite/gdb.dwarf2/no-gnu-debuglink.exp
@@ -37,8 +37,8 @@ if { [build_executable $testfile.exp $testfile [list $srcfile $asm_file]] } {
 
 clean_restart
 
-set msg "\r\ncould not find '\.gnu_debugaltlink' file for \[^\r\n\]*"
+set msg "\r\nwarning: could not find '\.gnu_debugaltlink' file for \[^\r\n\]*"
 gdb_test "file $binfile" "$msg" "file command"
 
 set question "Load new symbol table from .*\? .y or n. "
-gdb_test "file $binfile" "$msg" "file command, again" $question "y"
+gdb_test "file $binfile" "" "file command, again" $question "y"