From: Bruno Larsen <blarsen@redhat.com>
To: Simon Marchi <simon.marchi@polymtl.ca>, gdb-patches@sourceware.org
Subject: Re: [PATCH v2 00/13] Make frame_info_ptr automatic
Date: Tue, 20 Dec 2022 17:57:43 +0100 [thread overview]
Message-ID: <c1cd5bab-744d-aa39-df3e-ff3b4d74e325@redhat.com> (raw)
In-Reply-To: <20221214033441.499512-1-simon.marchi@polymtl.ca>
On 14/12/2022 04:34, Simon Marchi via Gdb-patches wrote:
> This is v2 of:
>
> https://inbox.sourceware.org/gdb-patches/20221202180052.212745-1-simon.marchi@polymtl.ca/
>
> The first main change is that I dropped the patch removing the user of
> frame_info_ptr at many places, following Tom's comment. If we end up
> trying to reinflate a frame at a moment where it's not possible (because
> the frame's id is being computed), we'll hit an assert in
> frame_info_ptr. This is fine, because it indicates some logic problem
> in GDB.
>
> The second main change, found after doing the first one, is to fix a bug
> with the user-created frame reinflation. In v1, if we had two
> frame_info_ptr wrapping the same user-created frame_info object, they
> would each create their own frame_info instance on reinflation. And
> that confused GDB down the line. There are a few new patches to deal
> with that, by adding user-created frames to the frame stash, so they can
> be found by frame_info_ptr when reinflating.
>
> At the end of the series, frame_info_ptr::frame_info_ptr (at the end of
> the series) accesses some frame_info internals (the frame's id, without
> calling get_frame_id), so I decided to move frame_info_ptr to
> frame.{c,h} (patch 6). I think that makes sense anyway, as the logic in
> the implementation of frame_info_ptr is tightly coupled to the rest of
> the frame stuff. However, to do so, we need to break some inclusion
> cycles. This is done in the few patches before. Patches 1 and 2 are
> not strictly necessary, but fix oddities I found along the way.
>
> Regression-tested on Ubuntu 22.04 x86-64.
This series doesn't add any regressions and is where I wanted to take
frame_info_ptr all along. I have only one nit with patch 6 (sent as a
direct reply), but other than that:
Reviewed-By: Bruno Larsen <blarsen@redhat.com>
--
Cheers,
Bruno
>
> Simon Marchi (13):
> gdb: move type_map_instance to compile/compile.c
> gdb: move compile_instance to compile/compile.h
> gdb: remove language.h include from frame.h
> gdb: move sect_offset and cu_offset to dwarf2/types.h
> gdb: move call site types to call-site.h
> gdb: move frame_info_ptr to frame.{c,h}
> gdb: add frame_id::user_created_p
> gdb: add user-created frames to stash
> gdb: add create_new_frame(frame_id) overload
> gdb: make it possible to restore selected user-created frames
> gdb: make user-created frames reinflatable
> gdb: make frame_info_ptr grab frame level and id on construction
> gdb: make frame_info_ptr auto-reinflatable
>
> gdb/Makefile.in | 2 +-
> gdb/aarch64-tdep.c | 1 +
> gdb/amd64-tdep.c | 1 +
> gdb/arm-tdep.c | 1 +
> gdb/c-lang.c | 1 -
> gdb/c-lang.h | 2 +
> gdb/compile/compile-c.h | 1 +
> gdb/compile/compile-cplus.h | 1 +
> gdb/compile/compile-internal.h | 129 ------------
> gdb/compile/compile.c | 13 ++
> gdb/compile/compile.h | 116 +++++++++++
> gdb/cp-abi.c | 1 +
> gdb/cp-support.c | 1 +
> gdb/dwarf2/abbrev-cache.h | 1 -
> gdb/dwarf2/attribute.h | 2 +-
> gdb/dwarf2/call-site.h | 244 +++++++++++++++++++++++
> gdb/dwarf2/comp-unit-head.h | 5 +-
> gdb/dwarf2/cooked-index.h | 2 +-
> gdb/dwarf2/expr.h | 2 +-
> gdb/dwarf2/line-header.h | 2 -
> gdb/dwarf2/types.h | 40 ++++
> gdb/expop.h | 1 +
> gdb/f-lang.h | 1 +
> gdb/frame-id.h | 4 +
> gdb/frame-info.c | 65 ------
> gdb/frame-info.h | 211 --------------------
> gdb/frame.c | 127 ++++++++++--
> gdb/frame.h | 206 ++++++++++++++++++-
> gdb/gdbtypes.h | 229 ---------------------
> gdb/gnu-v3-abi.c | 1 +
> gdb/go-lang.h | 1 +
> gdb/infcmd.c | 2 -
> gdb/language.c | 1 -
> gdb/m2-typeprint.c | 1 +
> gdb/mi/mi-cmd-stack.c | 2 -
> gdb/ppc-sysv-tdep.c | 1 +
> gdb/python/py-disasm.c | 1 +
> gdb/python/py-frame.c | 1 +
> gdb/stack.c | 16 --
> gdb/symtab.c | 1 +
> gdb/symtab.h | 1 +
> gdb/testsuite/gdb.base/frame-view.c | 80 ++++++++
> gdb/testsuite/gdb.base/frame-view.exp | 109 ++++++++++
> gdb/testsuite/gdb.base/frame-view.py | 41 ++++
> gdb/thread.c | 1 +
> gdb/unittests/frame_info_ptr-selftests.c | 76 +++++++
> 46 files changed, 1065 insertions(+), 683 deletions(-)
> create mode 100644 gdb/dwarf2/call-site.h
> create mode 100644 gdb/dwarf2/types.h
> delete mode 100644 gdb/frame-info.c
> delete mode 100644 gdb/frame-info.h
> create mode 100644 gdb/testsuite/gdb.base/frame-view.c
> create mode 100644 gdb/testsuite/gdb.base/frame-view.exp
> create mode 100644 gdb/testsuite/gdb.base/frame-view.py
> create mode 100644 gdb/unittests/frame_info_ptr-selftests.c
>
>
> base-commit: a4b83845ded201e7d46220213d3e38950c30dbb5
next prev parent reply other threads:[~2022-12-20 16:57 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-12-14 3:34 Simon Marchi
2022-12-14 3:34 ` [PATCH v2 01/13] gdb: move type_map_instance to compile/compile.c Simon Marchi
2022-12-14 3:34 ` [PATCH v2 02/13] gdb: move compile_instance to compile/compile.h Simon Marchi
2022-12-14 3:34 ` [PATCH v2 03/13] gdb: remove language.h include from frame.h Simon Marchi
2022-12-14 3:34 ` [PATCH v2 04/13] gdb: move sect_offset and cu_offset to dwarf2/types.h Simon Marchi
2022-12-14 3:34 ` [PATCH v2 05/13] gdb: move call site types to call-site.h Simon Marchi
2022-12-14 3:34 ` [PATCH v2 06/13] gdb: move frame_info_ptr to frame.{c,h} Simon Marchi
2022-12-20 17:01 ` Bruno Larsen
2023-01-03 18:59 ` Simon Marchi
2022-12-14 3:34 ` [PATCH v2 07/13] gdb: add frame_id::user_created_p Simon Marchi
2022-12-14 3:34 ` [PATCH v2 08/13] gdb: add user-created frames to stash Simon Marchi
2022-12-14 3:34 ` [PATCH v2 09/13] gdb: add create_new_frame(frame_id) overload Simon Marchi
2022-12-14 3:34 ` [PATCH v2 10/13] gdb: make it possible to restore selected user-created frames Simon Marchi
2022-12-14 3:34 ` [PATCH v2 11/13] gdb: make user-created frames reinflatable Simon Marchi
2023-01-23 12:57 ` Tom de Vries
2023-01-23 14:34 ` Luis Machado
2023-01-24 3:55 ` Simon Marchi
2023-01-24 8:22 ` Luis Machado
2023-01-25 3:45 ` Simon Marchi
2023-01-30 8:49 ` Luis Machado
2023-01-30 16:20 ` Simon Marchi
2022-12-14 3:34 ` [PATCH v2 12/13] gdb: make frame_info_ptr grab frame level and id on construction Simon Marchi
2022-12-14 3:34 ` [PATCH v2 13/13] gdb: make frame_info_ptr auto-reinflatable Simon Marchi
2022-12-20 16:57 ` Bruno Larsen [this message]
2023-01-03 19:00 ` [PATCH v2 00/13] Make frame_info_ptr automatic Simon Marchi
2023-01-03 19:09 ` Simon Marchi
2023-01-18 18:10 ` Tom Tromey
2023-01-19 3:40 ` Simon Marchi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c1cd5bab-744d-aa39-df3e-ff3b4d74e325@redhat.com \
--to=blarsen@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=simon.marchi@polymtl.ca \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).