[PATCH 0/7] Changes to gdb-index creation

[PATCH 0/7] Changes to gdb-index creation

[Andrew Burgess]

Patches #1 and #2 are pretty straight forward improvements; don't
think they will be controversial.

Patch #4 is an interesting bug fix.

Patches #6 and #7 are what this series is really all about.  Making
the indexes created by 'save gdb-index' be determenistic as the number
of worker-threads that GDB uses changes.


---

Andrew Burgess (7):
  gdb: allow use of ~ in 'save gdb-index' command
  gdb: option completion for 'save gdb-index' command
  gdb/testsuite: small refactor in selftest-support.exp
  gdb: reduce size of generated gdb-index file
  gdb: C++-ify mapped_symtab from dwarf2/index-write.c
  gdb: generate gdb-index identically regardless of work thread count
  gdb: generate dwarf-5 index identically as worker-thread count changes

 gdb/dwarf2/index-write.c                     | 315 ++++++++++++++-----
 gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp |  91 ++++++
 gdb/testsuite/gdb.gdb/index-file.exp         | 162 ++++++++++
 gdb/testsuite/lib/gdb.exp                    |  15 +
 gdb/testsuite/lib/selftest-support.exp       |  32 +-
 5 files changed, 536 insertions(+), 79 deletions(-)
 create mode 100644 gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp
 create mode 100644 gdb/testsuite/gdb.gdb/index-file.exp


base-commit: 935dc9ff652ca256c10672412c1df3da95cadbfb
-- 
2.25.4

[PATCH 1/7] gdb: allow use of ~ in 'save gdb-index' command

[Andrew Burgess]

Add a call to gdb_tilde_expand in the save_gdb_index_command function,
this means that we can now do:

  (gdb) save gdb-index ~/blah/

Previous this wouldn't work.
---
 gdb/dwarf2/index-write.c                     |  7 +-
 gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp | 91 ++++++++++++++++++++
 2 files changed, 96 insertions(+), 2 deletions(-)
 create mode 100644 gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp

diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index d1b10a28823..8ee5e420936 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -39,6 +39,7 @@
 #include "objfiles.h"
 #include "ada-lang.h"
 #include "dwarf2/tag.h"
+#include "gdbsupport/gdb_tilde_expand.h"
 
 #include <algorithm>
 #include <cmath>
@@ -1548,6 +1549,8 @@ save_gdb_index_command (const char *arg, int from_tty)
   if (!*arg)
     error (_("usage: save gdb-index [-dwarf-5] DIRECTORY"));
 
+  std::string directory (gdb_tilde_expand (arg));
+
   for (objfile *objfile : current_program_space->objfiles ())
     {
       /* If the objfile does not correspond to an actual file, skip it.  */
@@ -1567,8 +1570,8 @@ save_gdb_index_command (const char *arg, int from_tty)
 	      if (dwz != NULL)
 		dwz_basename = lbasename (dwz->filename ());
 
-	      write_dwarf_index (per_objfile->per_bfd, arg, basename,
-				 dwz_basename, index_kind);
+	      write_dwarf_index (per_objfile->per_bfd, directory.c_str (),
+				 basename, dwz_basename, index_kind);
 	    }
 	  catch (const gdb_exception_error &except)
 	    {
diff --git a/gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp b/gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp
new file mode 100644
index 00000000000..d30d0e86cf2
--- /dev/null
+++ b/gdb/testsuite/gdb.dwarf2/gdb-index-tilde.exp
@@ -0,0 +1,91 @@
+# Copyright 2023 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# Test that tilde expansion works for the 'save gdb-index' command.
+
+# This test relies on using the $HOME directory.  We could make this
+# work for remote hosts, but right now, this isn't supported.
+require {!is_remote host}
+
+# Can't save an index with readnow.
+require !readnow
+
+standard_testfile main.c
+
+# Create a directory to generate an index file into.
+set full_dir [standard_output_file "index_files"]
+remote_exec host "mkdir -p ${full_dir}"
+
+# The users home directory.
+set home $::env(HOME)
+
+# Check that FULL_DIR is within the $HOME directory.  If it's not, then
+# that's fine, but we can't test tilde expansion in this case.
+if { [string compare -length [string length $home] $full_dir $home] != 0 } {
+    unsupported "test not run within home directory"
+    return -1
+}
+
+# Convert the $HOME prefix in to ~.
+set dir "~[string range $full_dir [string length $home] end]"
+
+# Build the test executable.
+if { [prepare_for_testing "failed to prepare" "${testfile}" ${srcfile}] } {
+    return -1
+}
+
+# Start GDB and load in the executable.
+clean_restart ${binfile}
+
+# If the executable was built with an index, or lacks the debug
+# information required to create an index, then we'll not be able to
+# generate an index, so lets not even try.
+set has_index false
+set can_dump_index false
+gdb_test_multiple "maint print objfile $binfile" "check we can generate an index" {
+    -re "\r\n\\.gdb_index: version ${decimal}(?=\r\n)" {
+	set has_index true
+	gdb_test_lines "" $gdb_test_name ".*"
+    }
+    -re "\r\n\\.debug_names: exists(?=\r\n)" {
+	set has_index true
+	gdb_test_lines "" $gdb_test_name ".*"
+    }
+    -re "\r\n(Cooked index in use:|Psymtabs)(?=\r\n)" {
+	set can_dump_index true
+	gdb_test_lines "" $gdb_test_name ".*"
+    }
+    -re -wrap "" {
+    }
+}
+
+if { $has_index } {
+    unsupported "already have an index"
+    return -1
+}
+
+if { !$can_dump_index } {
+    unsupported "lacks debug information needed to dump index"
+    return -1
+}
+
+# Generate an index file.
+gdb_test_no_output "save gdb-index $dir"
+gdb_exit
+
+# Confirm that the index file exists.
+set index_filename "${full_dir}/${gdb_test_file_name}.gdb-index"
+gdb_assert { [remote_file host exists $index_filename] } \
+    "confirm the index file exists"
-- 
2.25.4

[PATCH 2/7] gdb: option completion for 'save gdb-index' command

[Andrew Burgess]

Add proper support for option completion to the 'save gdb-index'
command.  Update save_gdb_index_command function to make use of the
new option_def data structures for parsing the '-dwarf-5' option.
---
 gdb/dwarf2/index-write.c | 69 ++++++++++++++++++++++++++++++----------
 1 file changed, 52 insertions(+), 17 deletions(-)

diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index 8ee5e420936..c0867799f6d 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -1523,6 +1523,48 @@ write_dwarf_index (dwarf2_per_bfd *per_bfd, const char *dir,
     dwz_index_wip->finalize ();
 }
 
+/* Options structure for the 'save gdb-index' command.  */
+
+struct save_gdb_index_options
+{
+  bool dwarf_5 = false;
+};
+
+/* The option_def list for the 'save gdb-index' command.  */
+
+static const gdb::option::option_def save_gdb_index_options_defs[] = {
+  gdb::option::boolean_option_def<save_gdb_index_options> {
+    "dwarf-5",
+    [] (save_gdb_index_options *opt) { return &opt->dwarf_5; },
+    nullptr, /* show_cmd_cb */
+    nullptr /* set_doc */
+  }
+};
+
+/* Create an options_def_group for the 'save gdb-index' command.  */
+
+static gdb::option::option_def_group
+make_gdb_save_index_options_def_group (save_gdb_index_options *opts)
+{
+  return {{save_gdb_index_options_defs}, opts};
+}
+
+/* Completer for the "save gdb-index" command.  */
+
+static void
+gdb_save_index_cmd_completer (struct cmd_list_element *ignore,
+			      completion_tracker &tracker,
+			      const char *text, const char *word)
+{
+  auto grp = make_gdb_save_index_options_def_group (nullptr);
+  if (gdb::option::complete_options
+      (tracker, &text, gdb::option::PROCESS_OPTIONS_UNKNOWN_IS_OPERAND, grp))
+    return;
+
+  word = advance_to_filename_complete_word_point (tracker, text);
+  filename_completer (ignore, tracker, text, word);
+}
+
 /* Implementation of the `save gdb-index' command.
 
    Note that the .gdb_index file format used by this command is
@@ -1530,26 +1572,19 @@ write_dwarf_index (dwarf2_per_bfd *per_bfd, const char *dir,
    there.  */
 
 static void
-save_gdb_index_command (const char *arg, int from_tty)
+save_gdb_index_command (const char *args, int from_tty)
 {
-  const char dwarf5space[] = "-dwarf-5 ";
-  dw_index_kind index_kind = dw_index_kind::GDB_INDEX;
-
-  if (!arg)
-    arg = "";
-
-  arg = skip_spaces (arg);
-  if (strncmp (arg, dwarf5space, strlen (dwarf5space)) == 0)
-    {
-      index_kind = dw_index_kind::DEBUG_NAMES;
-      arg += strlen (dwarf5space);
-      arg = skip_spaces (arg);
-    }
+  save_gdb_index_options opts;
+  const auto group = make_gdb_save_index_options_def_group (&opts);
+  gdb::option::process_options
+    (&args, gdb::option::PROCESS_OPTIONS_UNKNOWN_IS_OPERAND, group);
 
-  if (!*arg)
+  if (args == nullptr || *args == '\0')
     error (_("usage: save gdb-index [-dwarf-5] DIRECTORY"));
 
-  std::string directory (gdb_tilde_expand (arg));
+  std::string directory (gdb_tilde_expand (args));
+  dw_index_kind index_kind
+    = (opts.dwarf_5 ? dw_index_kind::DEBUG_NAMES : dw_index_kind::GDB_INDEX);
 
   for (objfile *objfile : current_program_space->objfiles ())
     {
@@ -1675,5 +1710,5 @@ No options create one file with .gdb-index extension for pre-DWARF-5\n\
 compatible .gdb_index section.  With -dwarf-5 creates two files with\n\
 extension .debug_names and .debug_str for DWARF-5 .debug_names section."),
 	       &save_cmdlist);
-  set_cmd_completer (c, filename_completer);
+  set_cmd_completer_handle_brkchars (c, gdb_save_index_cmd_completer);
 }
-- 
2.25.4

Re: [PATCH 2/7] gdb: option completion for 'save gdb-index' command

[Tom Tromey]

>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

Andrew> Add proper support for option completion to the 'save gdb-index'
Andrew> command.  Update save_gdb_index_command function to make use of the
Andrew> new option_def data structures for parsing the '-dwarf-5' option.

I thought there was some reason this was difficult... the reason the
'palves/filename-options' branch exists (maybe only in Pedro's github).

However if it works then it seems like a good improvement to me.

Tom

Re: [PATCH 2/7] gdb: option completion for 'save gdb-index' command

[Andrew Burgess]

Tom Tromey <tom@tromey.com> writes:

>>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:
>
> Andrew> Add proper support for option completion to the 'save gdb-index'
> Andrew> command.  Update save_gdb_index_command function to make use of the
> Andrew> new option_def data structures for parsing the '-dwarf-5' option.
>
> I thought there was some reason this was difficult... the reason the
> 'palves/filename-options' branch exists (maybe only in Pedro's github).

I've only skimmed that branch, but I think that branch is adding options
that are themselves filenames, e.g. '-option FILENAME', but that's not
the case here.  The '-dwarf-5' is the option, and that's separate from
FILENAME.

We already have two other commands that follow the same pattern as I'm
proposing here 'maint print c-tdesc' and 'compile file', admittedly,
neither of these are likely high usage commands, so maybe they are
broken in some subtle way that has never been spotted ...

... anyway, at least for the basic usage I've tested, everything still
seems fine with this change in place.

>
> However if it works then it seems like a good improvement to me.

Thanks,
Andrew

[PATCH 3/7] gdb/testsuite: small refactor in selftest-support.exp

[Andrew Burgess]

Split out the code that makes a copy of the GDB executable ready for
self testing into a new proc.  A later commit in this series wants to
load the GDB executable into GDB (for creating an on-disk debug
index), but doesn't need to make use of the full do_self_tests proc.

There should be no changes in what is tested after this commit.
---
 gdb/testsuite/lib/selftest-support.exp | 32 ++++++++++++++++++--------
 1 file changed, 23 insertions(+), 9 deletions(-)

diff --git a/gdb/testsuite/lib/selftest-support.exp b/gdb/testsuite/lib/selftest-support.exp
index 1e7a2cea8b7..abb1f43f1b0 100644
--- a/gdb/testsuite/lib/selftest-support.exp
+++ b/gdb/testsuite/lib/selftest-support.exp
@@ -92,11 +92,13 @@ proc selftest_setup { executable function } {
     return 0
 }
 
-# A simple way to run some self-tests.
-
-proc do_self_tests {function body} {
-    global GDB tool
-
+# Prepare for running a self-test by moving the GDB executable to a
+# location where we can use it as the inferior.  Return the filename
+# of the new location.
+#
+# If the current testing setup is not suitable for running a
+# self-test, then return an empty string.
+proc selftest_prepare {} {
     # Are we testing with a remote board?  In that case, the target
     # won't have access to the GDB's auxilliary data files
     # (data-directory, etc.).  It's simpler to just skip.
@@ -120,19 +122,31 @@ proc do_self_tests {function body} {
     # Run the test with self.  Copy the file executable file in case
     # this OS doesn't like to edit its own text space.
 
-    set GDB_FULLPATH [find_gdb $GDB]
+    set gdb_fullpath [find_gdb $::GDB]
 
     if {[is_remote host]} {
-	set xgdb x$tool
+	set xgdb x$::tool
     } else {
-	set xgdb [standard_output_file x$tool]
+	set xgdb [standard_output_file x$::tool]
     }
 
     # Remove any old copy lying around.
     remote_file host delete $xgdb
 
+    set filename [remote_download host $gdb_fullpath $xgdb]
+
+    return $filename
+}
+
+# A simple way to run some self-tests.
+
+proc do_self_tests {function body} {
+    set file [selftest_prepare]
+    if { $file eq "" } {
+	return
+    }
+
     gdb_start
-    set file [remote_download host $GDB_FULLPATH $xgdb]
 
     # When debugging GDB with GDB, some operations can take a relatively long
     # time, especially if the build is non-optimized.  Bump the timeout for the
-- 
2.25.4

[PATCH 4/7] gdb: reduce size of generated gdb-index file

[Andrew Burgess]

I noticed in passing that out algorithm for generating the gdb-index
file is incorrect.  When building the hash table in add_index_entry we
count every incoming entry rehash when the number of entries gets too
large.  However, some of the incoming entries will be duplicates,
which don't actually result in new items being added to the hash
table.

As a result, we grow the gdb-index hash table far too often.

With an unmodified GDB, generating a gdb-index for GDB, I see a file
size of 90M, with a hash usage (in the generated index file) of just
2.6%.

With a patched GDB, generating a gdb-index for the _same_ GDB binary,
I now see a gdb-index file size of 30M, with a hash usage of 41.9%.

This is a 67% reduction in gdb-index file size.

Obviously, not every gdb-index file is going to see such big savings,
however, the larger a program, and the more symbols that are
duplicated between compilation units, the more GDB would over count,
and so, over-grow the index.

The gdb-index hash table we create has a minimum size of 1024, and
then we grow the hash when it is 75% full, doubling the hash table at
that time.  Given this, then we expect that either:

  a. The hash table is size 1024, and less than 75% full, or
  b. The hash table is between 37.5% and 75% full.

I've include a test that checks some of these constraints -- I've not
bothered to check the upper limit, and over full hash table isn't
really a problem here, but if the fill percentage is less than 37.5%
then this indicates that we've done something wrong (obviously, I also
check for the 1024 minimum size).
---
 gdb/dwarf2/index-write.c             |  29 ++++---
 gdb/testsuite/gdb.gdb/index-file.exp | 115 +++++++++++++++++++++++++++
 2 files changed, 134 insertions(+), 10 deletions(-)
 create mode 100644 gdb/testsuite/gdb.gdb/index-file.exp

diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index c0867799f6d..5960bacf8fb 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -256,20 +256,29 @@ add_index_entry (struct mapped_symtab *symtab, const char *name,
 		 int is_static, gdb_index_symbol_kind kind,
 		 offset_type cu_index)
 {
-  offset_type cu_index_and_attrs;
+  symtab_index_entry *slot = &find_slot (symtab, name);
+  if (slot->name == NULL)
+    {
+      /* This is a new element in the hash table.  */
+      ++symtab->n_elements;
 
-  ++symtab->n_elements;
-  if (4 * symtab->n_elements / 3 >= symtab->data.size ())
-    hash_expand (symtab);
+      /* We might need to grow the hash table.  */
+      if (4 * symtab->n_elements / 3 >= symtab->data.size ())
+	{
+	  hash_expand (symtab);
 
-  symtab_index_entry &slot = find_slot (symtab, name);
-  if (slot.name == NULL)
-    {
-      slot.name = name;
+	  /* This element will have a different slot in the new table.  */
+	  slot = &find_slot (symtab, name);
+
+	  /* But it should still be a new element in the hash table.  */
+	  gdb_assert (slot->name == nullptr);
+	}
+
+      slot->name = name;
       /* index_offset is set later.  */
     }
 
-  cu_index_and_attrs = 0;
+  offset_type cu_index_and_attrs = 0;
   DW2_GDB_INDEX_CU_SET_VALUE (cu_index_and_attrs, cu_index);
   DW2_GDB_INDEX_SYMBOL_STATIC_SET_VALUE (cu_index_and_attrs, is_static);
   DW2_GDB_INDEX_SYMBOL_KIND_SET_VALUE (cu_index_and_attrs, kind);
@@ -281,7 +290,7 @@ add_index_entry (struct mapped_symtab *symtab, const char *name,
      the last entry pushed), but a symbol could have multiple kinds in one CU.
      To keep things simple we don't worry about the duplication here and
      sort and uniquify the list after we've processed all symbols.  */
-  slot.cu_indices.push_back (cu_index_and_attrs);
+  slot->cu_indices.push_back (cu_index_and_attrs);
 }
 
 /* See symtab_index_entry.  */
diff --git a/gdb/testsuite/gdb.gdb/index-file.exp b/gdb/testsuite/gdb.gdb/index-file.exp
new file mode 100644
index 00000000000..c6edd286fb9
--- /dev/null
+++ b/gdb/testsuite/gdb.gdb/index-file.exp
@@ -0,0 +1,115 @@
+# Copyright 2023 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# Load the GDB executable, and then 'save gdb-index', and make some
+# checks of the generated index file.
+
+load_lib selftest-support.exp
+
+# Can't save an index with readnow.
+require !readnow
+
+# A multiplier used to ensure slow tasks are less likely to timeout.
+set timeout_factor 20
+
+set filename [selftest_prepare]
+if { $filename eq "" } {
+    unsupported "${gdb_test_file_name}.exp"
+    return -1
+}
+
+with_timeout_factor $timeout_factor {
+    # Start GDB, load FILENAME.
+    clean_restart $filename
+}
+
+# Generate an index file.
+set dir1 [standard_output_file "index_1"]
+remote_exec host "mkdir -p ${dir1}"
+with_timeout_factor $timeout_factor {
+    gdb_test_no_output "save gdb-index $dir1" \
+	"create gdb-index file"
+}
+
+# Close GDB.
+gdb_exit
+
+# Validate that the index-file FILENAME has made efficient use of its
+# symbol hash table.  Calculate the number of symbols in the hash
+# table and the total hash table size.  The hash table starts with
+# 1024 entries, and then doubles each time it is filled to 75%.  At
+# 75% filled, doubling the size takes it to 37.5% filled.
+#
+# Thus, the hash table is correctly filled if:
+#  1. Its size is 1024 (i.e. it has not yet had its first doubling), or
+#  2. Its filled percentage is over 37%
+#
+# We could check that it is not over filled, but I don't as that's not
+# really an issue.  But we did once have a bug where the table was
+# doubled incorrectly, in which case we'd see a filled percentage of
+# around 2% in some cases, which is a huge waste of disk space.
+proc check_symbol_table_usage { filename } {
+    # Open the file in binary mode and read-only mode.
+    set fp [open $filename rb]
+
+    # Configure the channel to use binary translation.
+    fconfigure $fp -translation binary
+
+    # Read the first 8 bytes of the file, which contain the header of
+    # the index section.
+    set header [read $fp [expr 7 * 4]]
+
+    # Scan the header to get the version, the CU list offset, and the
+    # types CU list offset.
+    binary scan $header iiiiii version \
+	_ _ _ symbol_table_offset shortcut_offset
+
+    # The length of the symbol hash table (in entries).
+    set len [expr ($shortcut_offset - $symbol_table_offset) / 8]
+
+    # Now walk the hash table and count how many entries are in use.
+    set offset $symbol_table_offset
+    set count 0
+    while { $offset < $shortcut_offset } {
+	seek $fp $offset
+	set entry [read $fp 8]
+	binary scan $entry ii name_ptr flags
+	if { $name_ptr != 0 } {
+	    incr count
+	}
+
+	incr offset 8
+    }
+
+    # Close the file.
+    close $fp
+
+    # Calculate how full the cache is.
+    set pct [expr (100 * double($count)) / $len]
+
+    # Write our results out to the gdb.log.
+    verbose -log "Hash table size: $len"
+    verbose -log "Hash table entries: $count"
+    verbose -log "Percentage usage: $pct%"
+
+    # The minimum fill percentage is actually 37.5%, but we give TCL a
+    # little flexibility in case the FP maths give a result a little
+    # off.
+    gdb_assert { $len == 1024 || $pct > 37 } \
+	"symbol hash table usage"
+}
+
+set index_filename_base [file tail $filename]
+check_symbol_table_usage "$dir1/${index_filename_base}.gdb-index"
-- 
2.25.4

[PATCH 5/7] gdb: C++-ify mapped_symtab from dwarf2/index-write.c

[Andrew Burgess]

Make static the functions add_index_entry, find_slot, and hash_expand,
member functions of the mapped_symtab class.

Fold an additional snippet of code from write_gdbindex into
mapped_symtab::minimize, this code relates to minimisation, so this
seems like a good home for it.

Make the n_elements, data, and m_string_obstack member variables of
mapped_symtab private.  Provide a new obstack() member function to
provide access to the obstack when needed, and also add member
functions begin(), end(), cbegin(), and cend() so that the
mapped_symtab class can be treated like a contained and iterated
over.

I've also taken this opportunity to split out the logic for whether
the hash table (m_data) needs expanding, this is the new function
hash_needs_expanding.  This will be useful in a later commit.

There should be no user visible changes after this commit.
---
 gdb/dwarf2/index-write.c | 138 ++++++++++++++++++++++++++-------------
 1 file changed, 92 insertions(+), 46 deletions(-)

diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index 5960bacf8fb..e1c343e8790 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -189,86 +189,135 @@ struct mapped_symtab
 {
   mapped_symtab ()
   {
-    data.resize (1024);
+    m_data.resize (1024);
   }
 
-  /* Minimize each entry in the symbol table, removing duplicates.  */
+  /* If there are no elements in the symbol table, then reduce the table
+     size to zero.  Otherwise call symtab_index_entry::minimize each entry
+     in the symbol table.  */
+
   void minimize ()
   {
-    for (symtab_index_entry &item : data)
+    if (m_element_count == 0)
+      m_data.resize (0);
+
+    for (symtab_index_entry &item : m_data)
       item.minimize ();
   }
 
-  offset_type n_elements = 0;
-  std::vector<symtab_index_entry> data;
+  /* Add an entry to SYMTAB.  NAME is the name of the symbol.  CU_INDEX is
+     the index of the CU in which the symbol appears.  IS_STATIC is one if
+     the symbol is static, otherwise zero (global).  */
+
+  void add_index_entry (const char *name, int is_static,
+			gdb_index_symbol_kind kind, offset_type cu_index);
+
+  /* Access the obstack.  */
+  auto_obstack *obstack ()
+  { return &m_string_obstack; }
+
+private:
+
+  /* Find a slot in SYMTAB for the symbol NAME.  Returns a reference to
+     the slot.
+
+     Function is used only during write_hash_table so no index format
+     backward compatibility is needed.  */
+
+  symtab_index_entry &find_slot (const char *name);
+
+  /* Expand SYMTAB's hash table.  */
+
+  void hash_expand ();
+
+  /* Return true if the hash table in data needs to grow.  */
+
+  bool hash_needs_expanding () const
+  { return 4 * m_element_count / 3 >= m_data.size (); }
+
+  /* A vector that is used as a hash table.  */
+  std::vector<symtab_index_entry> m_data;
+
+  /* The number of elements stored in the m_data hash.  */
+  offset_type m_element_count = 0;
 
   /* Temporary storage for names.  */
   auto_obstack m_string_obstack;
-};
 
-/* Find a slot in SYMTAB for the symbol NAME.  Returns a reference to
-   the slot.
+public:
+  using iterator = decltype (m_data)::iterator;
+  using const_iterator = decltype (m_data)::const_iterator;
 
-   Function is used only during write_hash_table so no index format backward
-   compatibility is needed.  */
+  iterator begin ()
+  { return m_data.begin (); }
 
-static symtab_index_entry &
-find_slot (struct mapped_symtab *symtab, const char *name)
+  iterator end ()
+  { return m_data.end (); }
+
+  const_iterator cbegin ()
+  { return m_data.cbegin (); }
+
+  const_iterator cend ()
+  { return m_data.cend (); }
+};
+
+/* See class definition.  */
+
+symtab_index_entry &
+mapped_symtab::find_slot (const char *name)
 {
   offset_type index, step, hash = mapped_index_string_hash (INT_MAX, name);
 
-  index = hash & (symtab->data.size () - 1);
-  step = ((hash * 17) & (symtab->data.size () - 1)) | 1;
+  index = hash & (m_data.size () - 1);
+  step = ((hash * 17) & (m_data.size () - 1)) | 1;
 
   for (;;)
     {
-      if (symtab->data[index].name == NULL
-	  || strcmp (name, symtab->data[index].name) == 0)
-	return symtab->data[index];
-      index = (index + step) & (symtab->data.size () - 1);
+      if (m_data[index].name == NULL
+	  || strcmp (name, m_data[index].name) == 0)
+	return m_data[index];
+      index = (index + step) & (m_data.size () - 1);
     }
 }
 
-/* Expand SYMTAB's hash table.  */
+/* See class definition.  */
 
-static void
-hash_expand (struct mapped_symtab *symtab)
+void
+mapped_symtab::hash_expand ()
 {
-  auto old_entries = std::move (symtab->data);
+  auto old_entries = std::move (m_data);
 
-  symtab->data.clear ();
-  symtab->data.resize (old_entries.size () * 2);
+  gdb_assert (m_data.size () == 0);
+  m_data.resize (old_entries.size () * 2);
 
   for (auto &it : old_entries)
     if (it.name != NULL)
       {
-	auto &ref = find_slot (symtab, it.name);
+	auto &ref = this->find_slot (it.name);
 	ref = std::move (it);
       }
 }
 
-/* Add an entry to SYMTAB.  NAME is the name of the symbol.
-   CU_INDEX is the index of the CU in which the symbol appears.
-   IS_STATIC is one if the symbol is static, otherwise zero (global).  */
+/* See class definition.  */
 
-static void
-add_index_entry (struct mapped_symtab *symtab, const char *name,
-		 int is_static, gdb_index_symbol_kind kind,
-		 offset_type cu_index)
+void
+mapped_symtab::add_index_entry (const char *name, int is_static,
+				gdb_index_symbol_kind kind,
+				offset_type cu_index)
 {
-  symtab_index_entry *slot = &find_slot (symtab, name);
+  symtab_index_entry *slot = &this->find_slot (name);
   if (slot->name == NULL)
     {
       /* This is a new element in the hash table.  */
-      ++symtab->n_elements;
+      ++this->m_element_count;
 
       /* We might need to grow the hash table.  */
-      if (4 * symtab->n_elements / 3 >= symtab->data.size ())
+      if (this->hash_needs_expanding ())
 	{
-	  hash_expand (symtab);
+	  this->hash_expand ();
 
 	  /* This element will have a different slot in the new table.  */
-	  slot = &find_slot (symtab, name);
+	  slot = &this->find_slot (name);
 
 	  /* But it should still be a new element in the hash table.  */
 	  gdb_assert (slot->name == nullptr);
@@ -389,7 +438,7 @@ write_hash_table (mapped_symtab *symtab, data_buf &output, data_buf &cpool)
 
     /* We add all the index vectors to the constant pool first, to
        ensure alignment is ok.  */
-    for (symtab_index_entry &entry : symtab->data)
+    for (symtab_index_entry &entry : *symtab)
       {
 	if (entry.name == NULL)
 	  continue;
@@ -418,7 +467,7 @@ write_hash_table (mapped_symtab *symtab, data_buf &output, data_buf &cpool)
 
   /* Now write out the hash table.  */
   std::unordered_map<c_str_view, offset_type, c_str_view_hasher> str_table;
-  for (const auto &entry : symtab->data)
+  for (const auto &entry : *symtab)
     {
       offset_type str_off, vec_off;
 
@@ -1159,7 +1208,7 @@ write_cooked_index (cooked_index *table,
       const auto it = cu_index_htab.find (entry->per_cu);
       gdb_assert (it != cu_index_htab.cend ());
 
-      const char *name = entry->full_name (&symtab->m_string_obstack);
+      const char *name = entry->full_name (symtab->obstack ());
 
       if (entry->per_cu->lang () == language_ada)
 	{
@@ -1167,8 +1216,7 @@ write_cooked_index (cooked_index *table,
 	     gdb, it has to use the encoded name, with any
 	     suffixes stripped.  */
 	  std::string encoded = ada_encode (name, false);
-	  name = obstack_strdup (&symtab->m_string_obstack,
-				 encoded.c_str ());
+	  name = obstack_strdup (symtab->obstack (), encoded.c_str ());
 	}
       else if (entry->per_cu->lang () == language_cplus
 	       && (entry->flags & IS_LINKAGE) != 0)
@@ -1199,8 +1247,8 @@ write_cooked_index (cooked_index *table,
       else
 	kind = GDB_INDEX_SYMBOL_KIND_TYPE;
 
-      add_index_entry (symtab, name, (entry->flags & IS_STATIC) != 0,
-		       kind, it->second);
+      symtab->add_index_entry (name, (entry->flags & IS_STATIC) != 0,
+			       kind, it->second);
     }
 }
 
@@ -1303,8 +1351,6 @@ write_gdbindex (dwarf2_per_bfd *per_bfd, cooked_index *table,
   symtab.minimize ();
 
   data_buf symtab_vec, constant_pool;
-  if (symtab.n_elements == 0)
-    symtab.data.resize (0);
 
   write_hash_table (&symtab, symtab_vec, constant_pool);
 
-- 
2.25.4

Re: [PATCH 5/7] gdb: C++-ify mapped_symtab from dwarf2/index-write.c

[Tom Tromey]

>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

Andrew> mapped_symtab class can be treated like a contained and iterated

s/contained/container

Andrew> +  /* Access the obstack.  */
Andrew> +  auto_obstack *obstack ()

A real nit, but it is probably better to just return 'struct obstack *'
here.  Returning an auto_obstack makes it seem like the special auto-
behavior is important somewhere, but it isn't and/or shouldn't be.

Tom

Re: [PATCH 5/7] gdb: C++-ify mapped_symtab from dwarf2/index-write.c

[Andrew Burgess]

Tom Tromey <tom@tromey.com> writes:

>>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:
>
> Andrew> mapped_symtab class can be treated like a contained and iterated
>
> s/contained/container
>
> Andrew> +  /* Access the obstack.  */
> Andrew> +  auto_obstack *obstack ()
>
> A real nit, but it is probably better to just return 'struct obstack *'
> here.  Returning an auto_obstack makes it seem like the special auto-
> behavior is important somewhere, but it isn't and/or shouldn't be.

I'll change this and re-test.

Thanks,
Andrew

[PATCH 6/7] gdb: generate gdb-index identically regardless of work thread count

[Andrew Burgess]

It was observed that changing the number of worker threads that GDB
uses (maintenance set worker-threads NUM) would have an impact on the
layout of the generated gdb-index.

The cause seems to be how the CU are distributed between threads, and
then symbols that appear in multiple CU can be encountered earlier or
later depending on whether a particular CU moves between threads.

I certainly found this behaviour was reproducible when generating an
index for GDB itself, like:

  gdb -q -nx -nh -batch \
      -eiex 'maint set worker-threads NUM' \
      -ex 'save gdb-index /tmp/'

And then setting different values for NUM will change the generated
index.

Now, the question is: does this matter?

I would like to suggest that yes, this does matter.  At Red Hat we
generate a gdb-index as part of the build process, and we would
ideally like to have reproducible builds: for the same source,
compiled with the same tool-chain, we should get the exact same output
binary.  And we do .... except for the index.

Now we could simply force GDB to only use a single worker thread when
we build the index, but, I don't think the idea of reproducible builds
is that strange, so I think we should ensure that our generated
indexes are always reproducible.

To achieve this, I propose that we add an extra step when building the
gdb-index file.  After constructing the initial symbol hash table
contents, we will pull all the symbols out of the hash, sort them,
then re-insert them in sorted order.  This will ensure that the
structure of the generated hash will remain consistent (given the same
set of symbols).

I've extended the existing index-file test to check that the generated
index doesn't change if we adjust the number of worker threads used.
Given that this test is already rather slow, I've only made one change
to the worker-thread count.  Maybe this test should be changed to use
a smaller binary, which is quicker to load, and for which we could
then try many different worker thread counts.
---
 gdb/dwarf2/index-write.c             | 69 ++++++++++++++++++++++++++++
 gdb/testsuite/gdb.gdb/index-file.exp | 41 +++++++++++++++++
 gdb/testsuite/lib/gdb.exp            | 15 ++++++
 3 files changed, 125 insertions(+)

diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index e1c343e8790..bc07bcb8f4c 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -212,6 +212,13 @@ struct mapped_symtab
   void add_index_entry (const char *name, int is_static,
 			gdb_index_symbol_kind kind, offset_type cu_index);
 
+  /* When entries are originally added into the data hash the order will
+     vary based on the number of worker threads GDB is configured to use.
+     This function will rebuild the hash such that the final layout will be
+     deterministic regardless of the number of worker threads used.  */
+
+  void sort ();
+
   /* Access the obstack.  */
   auto_obstack *obstack ()
   { return &m_string_obstack; }
@@ -298,6 +305,65 @@ mapped_symtab::hash_expand ()
       }
 }
 
+/* See mapped_symtab class declaration.  */
+
+void mapped_symtab::sort ()
+{
+  /* Move contents out of this->data vector.  */
+  std::vector<symtab_index_entry> original_data = std::move (m_data);
+
+  /* Restore the size of m_data, this will avoid having to expand the hash
+     table (and rehash all elements) when we reinsert after sorting.
+     However, we do reset the element count, this allows for some sanity
+     checking asserts during the reinsert phase.  */
+  gdb_assert (m_data.size () == 0);
+  m_data.resize (original_data.size ());
+  m_element_count = 0;
+
+  /* Remove empty entries from ORIGINAL_DATA, this makes sorting quicker.  */
+  auto it = std::remove_if (original_data.begin (), original_data.end (),
+			    [] (const symtab_index_entry &entry) -> bool
+			    {
+			      return entry.name == nullptr;
+			    });
+  original_data.erase (it, original_data.end ());
+
+  /* Sort the existing contents.  */
+  std::sort (original_data.begin (), original_data.end (),
+	     [] (const symtab_index_entry &a,
+		 const symtab_index_entry &b) -> bool
+	     {
+	       /* Return true if A is before B.  */
+	       gdb_assert (a.name != nullptr);
+	       gdb_assert (b.name != nullptr);
+
+	       return strcmp (a.name, b.name) < 0;
+	     });
+
+  /* Re-insert each item from the sorted list.  */
+  for (auto &entry : original_data)
+    {
+      /* We know that ORIGINAL_DATA contains no duplicates, this data was
+	 taken from a hash table that de-duplicated entries for us, so
+	 count this as a new item.
+
+	 As we retained the original size of m_data (see above) then we
+	 should never need to grow m_data_ during this re-insertion phase,
+	 assert that now.  */
+      ++m_element_count;
+      gdb_assert (!this->hash_needs_expanding ());
+
+      /* Lookup a slot.  */
+      symtab_index_entry &slot = this->find_slot (entry.name);
+
+      /* As discussed above, we should not find duplicates.  */
+      gdb_assert (slot.name == nullptr);
+
+      /* Move this item into the slot we found.  */
+      slot = std::move (entry);
+    }
+}
+
 /* See class definition.  */
 
 void
@@ -1346,6 +1412,9 @@ write_gdbindex (dwarf2_per_bfd *per_bfd, cooked_index *table,
   for (auto map : table->get_addrmaps ())
     write_address_map (map, addr_vec, cu_index_htab);
 
+  /* Ensure symbol hash is built domestically.  */
+  symtab.sort ();
+
   /* Now that we've processed all symbols we can shrink their cu_indices
      lists.  */
   symtab.minimize ();
diff --git a/gdb/testsuite/gdb.gdb/index-file.exp b/gdb/testsuite/gdb.gdb/index-file.exp
index c6edd286fb9..08415920061 100644
--- a/gdb/testsuite/gdb.gdb/index-file.exp
+++ b/gdb/testsuite/gdb.gdb/index-file.exp
@@ -35,6 +35,9 @@ with_timeout_factor $timeout_factor {
     clean_restart $filename
 }
 
+# Record how many worker threads GDB is using.
+set worker_threads [gdb_get_worker_threads]
+
 # Generate an index file.
 set dir1 [standard_output_file "index_1"]
 remote_exec host "mkdir -p ${dir1}"
@@ -113,3 +116,41 @@ proc check_symbol_table_usage { filename } {
 
 set index_filename_base [file tail $filename]
 check_symbol_table_usage "$dir1/${index_filename_base}.gdb-index"
+
+# If GDB is using more than 1 worker thread then reduce the number of
+# worker threads, regenerate the index, and check that we get the same
+# index file back.  At one point the layout of the index would vary
+# based on the number of worker threads used.
+if { $worker_threads > 1 } {
+    # Start GDB, but don't load a file yet.
+    clean_restart
+
+    # Adjust the number of threads to use.
+    set reduced_threads [expr $worker_threads / 2]
+    gdb_test_no_output "maint set worker-threads $reduced_threads"
+
+    with_timeout_factor $timeout_factor {
+	# Now load the test binary.
+	gdb_file_cmd $filename
+    }
+
+    # Generate an index file.
+    set dir2 [standard_output_file "index_2"]
+    remote_exec host "mkdir -p ${dir2}"
+    with_timeout_factor $timeout_factor {
+	gdb_test_no_output "save gdb-index $dir2" \
+	    "create second gdb-index file"
+    }
+
+    # Close GDB.
+    gdb_exit
+
+    # Now check that the index files are identical.
+    foreach suffix { gdb-index  } {
+	set result \
+	    [remote_exec host \
+		 "cmp -s \"$dir1/${index_filename_base}.${suffix}\" \"$dir2/${index_filename_base}.${suffix}\""]
+	gdb_assert { [lindex $result 0] == 0 } \
+	    "$suffix files are identical"
+    }
+}
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 63885860795..b534a61d066 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -10026,6 +10026,21 @@ proc is_target_non_stop { {testname ""} } {
     return $is_non_stop
 }
 
+# Return the number of worker threads that GDB is currently using.
+
+proc gdb_get_worker_threads { {testname ""} } {
+    set worker_threads "UNKNOWN"
+    gdb_test_multiple "maintenance show worker-threads" $testname {
+	-wrap -re "^The number of worker threads GDB can use is unlimited \\(currently ($::decimal)\\)\\." {
+	    set worker_threads $expect_out(1,string)
+	}
+	-wrap -re "The number of worker threads GDB can use is ($::decimal)\\." {
+	    set worker_threads $expect_out(1,string)
+	}
+    }
+    return $worker_threads
+}
+
 # Check if the compiler emits epilogue information associated
 # with the closing brace or with the last statement line.
 #
-- 
2.25.4

[PATCH 7/7] gdb: generate dwarf-5 index identically as worker-thread count changes

[Andrew Burgess]

Similar to the previous commit, this commit ensures that the dwarf-5
index files are generated identically as the number of worker-threads
changes.

Building the dwarf-5 index makes use of a closed hash table, the
bucket_hash local within debug_names::build().  Entries are added to
bucket_hash from m_name_to_value_set, which, in turn, is populated
by calls to debug_names::insert() in write_debug_names.  The insert
calls are ordered based on the entries within the cooked_index, and
the ordering within cooked_index depends on the number of worker
threads that GDB is using.

My proposal is to sort each chain within the bucket_hash closed hash
table prior to using this to build the dwarf-5 index.

The buckets within bucket_hash will always have the same ordering (for
a given GDB build with a given executable), and by sorting the chains
within each bucket, we can be sure that GDB will see each entry in a
deterministic order.

I've extended the index creation test to cover this case.
---
 gdb/dwarf2/index-write.c             | 17 +++++++++++++++--
 gdb/testsuite/gdb.gdb/index-file.exp |  8 +++++++-
 2 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index bc07bcb8f4c..1bc0c8ab57e 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -454,6 +454,11 @@ class c_str_view
     return strcmp (m_cstr, other.m_cstr) == 0;
   }
 
+  bool operator< (const c_str_view &other) const
+  {
+    return strcmp (m_cstr, other.m_cstr) < 0;
+  }
+
   /* Return the underlying C string.  Note, the returned string is
      only a reference with lifetime of this object.  */
   const char *c_str () const
@@ -773,10 +778,18 @@ class debug_names
       }
     for (size_t bucket_ix = 0; bucket_ix < bucket_hash.size (); ++bucket_ix)
       {
-	const std::forward_list<hash_it_pair> &hashitlist
-	  = bucket_hash[bucket_ix];
+	std::forward_list<hash_it_pair> &hashitlist = bucket_hash[bucket_ix];
 	if (hashitlist.empty ())
 	  continue;
+
+	/* Sort the items within each bucket.  This ensures that the
+	   generated index files will be the same no matter the order in
+	   which symbols were added into the index.  */
+	hashitlist.sort ([] (const hash_it_pair &a, const hash_it_pair &b)
+	{
+	  return a.it->first < b.it->first;
+	});
+
 	uint32_t &bucket_slot = m_bucket_table[bucket_ix];
 	/* The hashes array is indexed starting at 1.  */
 	store_unsigned_integer (reinterpret_cast<gdb_byte *> (&bucket_slot),
diff --git a/gdb/testsuite/gdb.gdb/index-file.exp b/gdb/testsuite/gdb.gdb/index-file.exp
index 08415920061..7154d234dd5 100644
--- a/gdb/testsuite/gdb.gdb/index-file.exp
+++ b/gdb/testsuite/gdb.gdb/index-file.exp
@@ -44,6 +44,9 @@ remote_exec host "mkdir -p ${dir1}"
 with_timeout_factor $timeout_factor {
     gdb_test_no_output "save gdb-index $dir1" \
 	"create gdb-index file"
+
+    gdb_test_no_output "save gdb-index -dwarf-5 $dir1" \
+	"create dwarf-index files"
 }
 
 # Close GDB.
@@ -140,13 +143,16 @@ if { $worker_threads > 1 } {
     with_timeout_factor $timeout_factor {
 	gdb_test_no_output "save gdb-index $dir2" \
 	    "create second gdb-index file"
+
+	gdb_test_no_output "save gdb-index -dwarf-5 $dir2" \
+	    "create second dwarf-index files"
     }
 
     # Close GDB.
     gdb_exit
 
     # Now check that the index files are identical.
-    foreach suffix { gdb-index  } {
+    foreach suffix { gdb-index debug_names debug_str } {
 	set result \
 	    [remote_exec host \
 		 "cmp -s \"$dir1/${index_filename_base}.${suffix}\" \"$dir2/${index_filename_base}.${suffix}\""]
-- 
2.25.4

Re: [PATCH 0/7] Changes to gdb-index creation

[Tom Tromey]

>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

Andrew> Patches #1 and #2 are pretty straight forward improvements; don't
Andrew> think they will be controversial.

Andrew> Patch #4 is an interesting bug fix.

Andrew> Patches #6 and #7 are what this series is really all about.  Making
Andrew> the indexes created by 'save gdb-index' be determenistic as the number
Andrew> of worker-threads that GDB uses changes.

I read through this.  I was a bit worried about new gdb.gdb tests (I
feel those are often more trouble than they're worth), but this one
avoids the problem stuff and anyway there doesn't seem to be a better
way to do it.

Anyway, I sent a nit or two, everything looks good to me.
Thanks for doing this, and nice find on the hash table bug.

Approved-By: Tom Tromey <tom@tromey.com>

Tom

Re: [PATCH 0/7] Changes to gdb-index creation

[Andrew Burgess]

Tom Tromey <tom@tromey.com> writes:

>>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:
>
> Andrew> Patches #1 and #2 are pretty straight forward improvements; don't
> Andrew> think they will be controversial.
>
> Andrew> Patch #4 is an interesting bug fix.
>
> Andrew> Patches #6 and #7 are what this series is really all about.  Making
> Andrew> the indexes created by 'save gdb-index' be determenistic as the number
> Andrew> of worker-threads that GDB uses changes.
>
> I read through this.  I was a bit worried about new gdb.gdb tests (I
> feel those are often more trouble than they're worth), but this one
> avoids the problem stuff and anyway there doesn't seem to be a better
> way to do it.
>
> Anyway, I sent a nit or two, everything looks good to me.
> Thanks for doing this, and nice find on the hash table bug.
>
> Approved-By: Tom Tromey <tom@tromey.com>

I pushed this series.

Thanks,
Andrew