[PATCH v2 0/2] Call target_can_do_single_step from maybe_software_singlestep

[Tom Tromey]

This series implements a suggestion that Pedro made back in 2018.  The
idea has been kicking around internally here at AdaCore since then,
and I finally got around to writing the patch.

Tested on x86-64 Fedora 41.

Note, though, that the btrace tests have been flaky for me since I got
a new machine.  This patch seems to cause some changes in the results
here, but I can't really be sure whether or not those are regressions.

---
Changes in v2:
- Changed record-full to call maybe_software_singlestep
- Combined maybe_software_singlestep and insert_single_step_breakpoints
- Link to v1: https://inbox.sourceware.org/gdb-patches/20230612-sw-single-step-v1-0-c06d648e121b@adacore.com

---
Tom Tromey (2):
      Call target_can_do_single_step from maybe_software_singlestep
      Disabling hardware single step in gdbserver

 gdb/arm-linux-tdep.c      |  5 -----
 gdb/breakpoint.c          | 25 ++++++++++++++++++++-----
 gdb/breakpoint.h          |  8 ++++----
 gdb/gdbarch-gen.h         | 12 +++---------
 gdb/gdbarch_components.py | 12 +++---------
 gdb/infrun.c              | 19 -------------------
 gdb/record-full.c         |  4 ++--
 gdbserver/server.cc       | 12 ++++++++----
 gdbserver/server.h        |  1 +
 9 files changed, 41 insertions(+), 57 deletions(-)
---
base-commit: 8b9c9b26e18a63aae5b7fcac6b68cede5b61aaf3
change-id: 20230612-sw-single-step-946fd4d856f2

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