From: "Christian Biesinger via gdb-patches" <gdb-patches@sourceware.org>
To: Tom Tromey <tom@tromey.com>
Cc: gdb-patches <gdb-patches@sourceware.org>
Subject: Re: [PATCH 4/8] Store the comp_unit instead of the FDE table
Date: Sun, 09 Feb 2020 23:56:00 -0000 [thread overview]
Message-ID: <CAPTJ0XFpfZ_sCy-o4gkmE1cy1qVk+eW2Q6yNQSXMj7QCdF2NPQ@mail.gmail.com> (raw)
In-Reply-To: <20200208152758.29385-5-tom@tromey.com>
On Sat, Feb 8, 2020, 10:28 Tom Tromey <tom@tromey.com> wrote:
> This changes the DWARF frame code to store the comp_unit on the
> objfile, rather than storing the FDE table. It also changes the
> comp_unit to be heap-allocated using "new".
>
> This change makes it simpler for a later patch to add a field to the
> comp_unit, and to have deallaction work properly. This in turn is
> important for making the frame data be independent of the objfile.
>
> gdb/ChangeLog
> 2020-02-08 Tom Tromey <tom@tromey.com>
>
> * dwarf2/frame.c (struct comp_unit): Add initializers.
> (dwarf2_frame_objfile_data): Store a comp_unit.
> (dwarf2_frame_find_fde): Update.
> (dwarf2_build_frame_info): Use "new".
>
> Change-Id: Id339892824caf47e7909ed5ee7e2f0556d645dba
> ---
> gdb/ChangeLog | 7 +++++++
> gdb/dwarf2/frame.c | 43 ++++++++++++++++++++-----------------------
> 2 files changed, 27 insertions(+), 23 deletions(-)
>
> diff --git a/gdb/dwarf2/frame.c b/gdb/dwarf2/frame.c
> index 6d87e598345..0e74b8e7e68 100644
> --- a/gdb/dwarf2/frame.c
> +++ b/gdb/dwarf2/frame.c
> @@ -137,24 +137,27 @@ typedef std::vector<dwarf2_fde *> dwarf2_fde_table;
> struct comp_unit
> {
> /* Keep the bfd convenient. */
> - bfd *abfd;
> + bfd *abfd = nullptr;
>
> - struct objfile *objfile;
> + struct objfile *objfile = nullptr;
>
> /* Pointer to the .debug_frame section loaded into memory. */
> - const gdb_byte *dwarf_frame_buffer;
> + const gdb_byte *dwarf_frame_buffer = nullptr;
>
> /* Length of the loaded .debug_frame section. */
> - bfd_size_type dwarf_frame_size;
> + bfd_size_type dwarf_frame_size = 0;
>
> /* Pointer to the .debug_frame section. */
> - asection *dwarf_frame_section;
> + asection *dwarf_frame_section = nullptr;
>
> /* Base for DW_EH_PE_datarel encodings. */
> - bfd_vma dbase;
> + bfd_vma dbase = 0;
>
> /* Base for DW_EH_PE_textrel encodings. */
> - bfd_vma tbase;
> + bfd_vma tbase = 0;
> +
> + /* The FDE table. */
> + dwarf2_fde_table fde_table;
> };
>
> static struct dwarf2_fde *dwarf2_frame_find_fde (CORE_ADDR *pc,
> @@ -1467,7 +1470,7 @@ dwarf2_frame_cfa (struct frame_info *this_frame)
> return get_frame_base (this_frame);
> }
>
> -const struct objfile_key<dwarf2_fde_table> dwarf2_frame_objfile_data;
> +const struct objfile_key<comp_unit> dwarf2_frame_objfile_data;
>
>
>
> @@ -1630,18 +1633,18 @@ dwarf2_frame_find_fde (CORE_ADDR *pc, CORE_ADDR
> *out_offset)
> {
> for (objfile *objfile : current_program_space->objfiles ())
> {
> - dwarf2_fde_table *fde_table;
> CORE_ADDR offset;
> CORE_ADDR seek_pc;
>
> - fde_table = dwarf2_frame_objfile_data.get (objfile);
> - if (fde_table == NULL)
> + comp_unit *unit = dwarf2_frame_objfile_data.get (objfile);
> + if (unit == NULL)
> {
> dwarf2_build_frame_info (objfile);
> - fde_table = dwarf2_frame_objfile_data.get (objfile);
> + unit = dwarf2_frame_objfile_data.get (objfile);
> }
> - gdb_assert (fde_table != NULL);
> + gdb_assert (unit != NULL);
>
> + dwarf2_fde_table *fde_table = &unit->fde_table;
> if (fde_table->empty ())
> continue;
>
> @@ -2120,14 +2123,11 @@ dwarf2_build_frame_info (struct objfile *objfile)
> const gdb_byte *frame_ptr;
> dwarf2_cie_table cie_table;
> dwarf2_fde_table fde_table;
> - dwarf2_fde_table *fde_table2;
>
> /* Build a minimal decoding of the DWARF2 compilation unit. */
> - unit = XOBNEW (&objfile->objfile_obstack, comp_unit);
> + unit = new comp_unit;
>
Move the declaration here?
unit->abfd = objfile->obfd;
> unit->objfile = objfile;
>
These two could be constructor arguments?
- unit->dbase = 0;
> - unit->tbase = 0;
>
> if (objfile->separate_debug_objfile_backlink == NULL)
> {
> @@ -2202,9 +2202,6 @@ dwarf2_build_frame_info (struct objfile *objfile)
> }
> }
>
> - /* Copy fde_table to obstack: it is needed at runtime. */
> - fde_table2 = new dwarf2_fde_table;
> -
> if (!fde_table.empty ())
> {
> struct dwarf2_fde *fde_prev = NULL;
> @@ -2249,13 +2246,13 @@ dwarf2_build_frame_info (struct objfile *objfile)
> && fde_prev->initial_location == fde->initial_location)
> continue;
>
> - fde_table2->push_back (fde);
> + unit->fde_table.push_back (fde);
> fde_prev = fde;
> }
> - fde_table2->shrink_to_fit ();
> + unit->fde_table.shrink_to_fit ();
> }
>
> - dwarf2_frame_objfile_data.set (objfile, fde_table2);
> + dwarf2_frame_objfile_data.set (objfile, unit);
> }
>
> /* Handle 'maintenance show dwarf unwinders'. */
> --
> 2.17.2
>
>
next prev parent reply other threads:[~2020-02-09 23:56 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-08 15:28 [PATCH 0/8] Share DWARF frame information across inferiors Tom Tromey
2020-02-08 15:28 ` [PATCH 6/8] Remove a use of the comp_unit backlink Tom Tromey
2020-02-08 15:28 ` [PATCH 5/8] Add per-unit obstack Tom Tromey
2020-02-11 10:34 ` Luis Machado
2020-02-12 3:53 ` Simon Marchi
2020-02-12 22:41 ` Tom Tromey
2020-02-12 22:48 ` Simon Marchi
2020-02-12 22:51 ` Tom Tromey
2020-02-08 15:28 ` [PATCH 2/8] Dont' allow copying of auto_obstack Tom Tromey
2020-02-09 23:56 ` Christian Biesinger via gdb-patches
2020-02-12 0:16 ` Tom Tromey
2020-02-12 0:53 ` Simon Marchi
2020-02-12 1:01 ` Tom Tromey
2020-02-08 15:28 ` [PATCH 1/8] Don't forward-declare struct objfile in frame.h Tom Tromey
2020-02-12 0:51 ` Simon Marchi
2020-02-12 1:59 ` Tom Tromey
2020-02-08 15:28 ` [PATCH 8/8] Move the frame data to the BFD when possible Tom Tromey
2020-02-11 10:45 ` Luis Machado
2020-02-12 0:31 ` Tom Tromey
2020-11-14 3:31 ` Simon Marchi
2020-02-08 15:28 ` [PATCH 4/8] Store the comp_unit instead of the FDE table Tom Tromey
2020-02-09 23:56 ` Christian Biesinger via gdb-patches [this message]
2020-02-12 0:19 ` Tom Tromey
2020-02-11 10:32 ` Luis Machado
2020-02-12 0:20 ` Tom Tromey
2020-02-12 3:36 ` Simon Marchi
2020-02-12 22:20 ` Tom Tromey
2020-02-08 15:28 ` [PATCH 3/8] Change fde table to a vector Tom Tromey
[not found] ` <5a373a74-9283-5d82-a22d-7a2606a4d3f5@linaro.org>
2020-02-12 0:29 ` Tom Tromey
2020-02-12 3:28 ` Simon Marchi
2020-02-12 3:33 ` Simon Marchi
2020-02-12 22:36 ` Tom Tromey
2020-02-12 22:47 ` Simon Marchi
2020-02-08 15:28 ` [PATCH 7/8] Remove the objfile backlink from comp_unit Tom Tromey
2020-02-09 23:56 ` Christian Biesinger via gdb-patches
2020-02-12 0:26 ` Tom Tromey
2020-02-11 10:40 ` Luis Machado
2020-02-08 15:29 ` [PATCH 0/8] Share DWARF frame information across inferiors Tom Tromey
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=CAPTJ0XFpfZ_sCy-o4gkmE1cy1qVk+eW2Q6yNQSXMj7QCdF2NPQ@mail.gmail.com \
--to=gdb-patches@sourceware.org \
--cc=cbiesinger@google.com \
--cc=tom@tromey.com \
/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