[PATCH 0/4] Improve Ada name lookup performance

[Tom Tromey <tromey@adacore.com>]

This series started when a user noticed that gdb was very slow to
print Ada records in a large program.  I tracked this down and fixed
it in patch #1.  However, then I noticed that, with a little work, an
entire method could be removed from quick_symbol_functions.  This is
desirable because this method was only used by Ada.

I regression tested this on x86-64 Fedora 38.  I also used regression
tested using the .debug_names and .gdb_index target boards.  It's also
been run through the AdaCore internal test suite.

---
Tom Tromey (4):
      Improve performance of Ada name searches
      Always use expand_symtabs_matching in ada-lang.c
      Remove split_style::UNDERSCORE
      Remove quick_symbol_functions::expand_matching_symbols

 gdb/ada-lang.c                | 111 ++++--------------------
 gdb/ada-lang.h                |  14 +++-
 gdb/dwarf2/cooked-index.c     |   6 +-
 gdb/dwarf2/read-debug-names.c |  46 ----------
 gdb/dwarf2/read-gdb-index.c   | 190 ------------------------------------------
 gdb/dwarf2/read.c             |  54 ------------
 gdb/objfiles.h                |   6 --
 gdb/psymtab.c                 | 104 -----------------------
 gdb/psymtab.h                 |   7 --
 gdb/quick-symbol.h            |  27 ------
 gdb/split-name.c              |  13 ---
 gdb/split-name.h              |   2 -
 gdb/symfile-debug.c           |  18 ----
 gdb/symtab.h                  |  13 ++-
 14 files changed, 40 insertions(+), 571 deletions(-)
---
base-commit: 97b29a61f7f001550fe985bf2deea1550e3c20dd
change-id: 20231121-ada-lookup-perf-67f46d937484

Best regards,
-- 
Tom Tromey <tromey@adacore.com>