From: Tom Tromey <tom@tromey.com>
To: gdb-patches@sourceware.org
Cc: Tom Tromey <tom@tromey.com>
Subject: [PATCH 12/30] Introduce DWARF abbrev cache
Date: Wed, 25 Aug 2021 20:19:19 -0600 [thread overview]
Message-ID: <20210826021937.1490292-13-tom@tromey.com> (raw)
In-Reply-To: <20210826021937.1490292-1-tom@tromey.com>
The replacement for the DWARF psymbol reader works in a somewhat
different way. The current reader reads and stores all the DIEs that
might be interesting. Then, if it is missing a DIE, it re-scans the
CU and reads them all. This approach is used for both intra- and
inter-CU references.
I instrumented the partial DIE hash to see how frequently it was used:
[ 0] -> 1538165
[ 1] -> 4912
[ 2] -> 96102
[ 3] -> 175
[ 4] -> 244
That is, most DIEs are never used, and some are looked up twice -- but
this is just an artifact of the implementation of
partial_die_info::fixup, which may do two lookups.
Based on this, the new implementation doesn't try to store any DIEs,
but instead just re-scans them on demand. In order to do this,
though, it is convenient to have a cache of DWARF abbrevs. This way,
if a second CU is needed to resolve an inter-CU reference, the abbrevs
for that CU need only be computed a single time.
---
gdb/Makefile.in | 1 +
gdb/dwarf2/abbrev-cache.c | 65 +++++++++++++++++++++++++++++++++++++++
gdb/dwarf2/abbrev-cache.h | 58 ++++++++++++++++++++++++++++++++++
gdb/dwarf2/abbrev.c | 5 +--
gdb/dwarf2/abbrev.h | 4 ++-
5 files changed, 130 insertions(+), 3 deletions(-)
create mode 100644 gdb/dwarf2/abbrev-cache.c
create mode 100644 gdb/dwarf2/abbrev-cache.h
diff --git a/gdb/Makefile.in b/gdb/Makefile.in
index 45fc6bdcd05..f120c0357ff 100644
--- a/gdb/Makefile.in
+++ b/gdb/Makefile.in
@@ -1028,6 +1028,7 @@ COMMON_SFILES = \
displaced-stepping.c \
dummy-frame.c \
dwarf2/abbrev.c \
+ dwarf2/abbrev-cache.c \
dwarf2/attribute.c \
dwarf2/comp-unit-head.c \
dwarf2/cu.c \
diff --git a/gdb/dwarf2/abbrev-cache.c b/gdb/dwarf2/abbrev-cache.c
new file mode 100644
index 00000000000..f0705065a3c
--- /dev/null
+++ b/gdb/dwarf2/abbrev-cache.c
@@ -0,0 +1,65 @@
+/* DWARF 2 abbrev table cache
+
+ Copyright (C) 2021 Free Software Foundation, Inc.
+
+ This file is part of GDB.
+
+ 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/>. */
+
+#include "defs.h"
+#include "dwarf2/read.h"
+#include "dwarf2/abbrev-cache.h"
+
+/* Hash function for an abbrev table. */
+
+hashval_t
+abbrev_cache::hash_table (const void *item)
+{
+ const struct abbrev_table *table = (const struct abbrev_table *) item;
+ return to_underlying (table->sect_off);
+}
+
+/* Comparison function for abbrev table. */
+
+int
+abbrev_cache::eq_table (const void *lhs, const void *rhs)
+{
+ const struct abbrev_table *l_table = (const struct abbrev_table *) lhs;
+ const search_key *key = (const search_key *) rhs;
+ return (l_table->section == key->section
+ && l_table->sect_off == key->offset);
+}
+
+abbrev_cache::abbrev_cache ()
+ : m_tables (htab_create_alloc (20, hash_table, eq_table,
+ htab_delete_entry<abbrev_table>,
+ xcalloc, xfree))
+{
+}
+
+void
+abbrev_cache::add (abbrev_table_up table)
+{
+ /* We allow this as a convenience to the caller. */
+ if (table == nullptr)
+ return;
+
+ search_key key = { table->section, table->sect_off };
+ void **slot = htab_find_slot_with_hash (m_tables.get (), &key,
+ to_underlying (table->sect_off),
+ INSERT);
+ /* If this one already existed, then it should have been reused. */
+ gdb_assert (*slot == nullptr);
+ *slot = (void *) table.release ();
+}
diff --git a/gdb/dwarf2/abbrev-cache.h b/gdb/dwarf2/abbrev-cache.h
new file mode 100644
index 00000000000..a651fcc0ec0
--- /dev/null
+++ b/gdb/dwarf2/abbrev-cache.h
@@ -0,0 +1,58 @@
+/* DWARF abbrev table cache
+
+ Copyright (C) 2020 Free Software Foundation, Inc.
+
+ This file is part of GDB.
+
+ 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/>. */
+
+#ifndef GDB_DWARF2_ABBREV_CACHE_H
+#define GDB_DWARF2_ABBREV_CACHE_H
+
+#include "dwarf2/abbrev.h"
+#include <unordered_set>
+#include "gdbtypes.h"
+
+class abbrev_cache
+{
+public:
+ abbrev_cache ();
+ DISABLE_COPY_AND_ASSIGN (abbrev_cache);
+
+ abbrev_table *find (struct dwarf2_section_info *section, sect_offset offset)
+ {
+ search_key key = { section, offset };
+
+ return (abbrev_table *) htab_find_with_hash (m_tables.get (), &key,
+ to_underlying (offset));
+ }
+
+ void add (abbrev_table_up table);
+
+private:
+
+ static hashval_t hash_table (const void *item);
+ static int eq_table (const void *lhs, const void *rhs);
+
+ struct search_key
+ {
+ struct dwarf2_section_info *section;
+ sect_offset offset;
+ };
+
+ /* Hash table of abbrev tables. */
+ htab_up m_tables;
+};
+
+#endif /* GDB_DWARF2_ABBREV_CACHE_H */
diff --git a/gdb/dwarf2/abbrev.c b/gdb/dwarf2/abbrev.c
index a8bdf7182de..c84f21256fd 100644
--- a/gdb/dwarf2/abbrev.c
+++ b/gdb/dwarf2/abbrev.c
@@ -58,8 +58,9 @@ eq_abbrev (const void *lhs, const void *rhs)
dies from a section we read in all abbreviations and install them
in a hash table. */
-abbrev_table::abbrev_table (sect_offset off)
+abbrev_table::abbrev_table (sect_offset off, struct dwarf2_section_info *sect)
: sect_off (off),
+ section (sect),
m_abbrevs (htab_create_alloc (20, hash_abbrev, eq_abbrev,
nullptr, xcalloc, xfree))
{
@@ -85,7 +86,7 @@ abbrev_table::read (struct dwarf2_section_info *section,
const gdb_byte *abbrev_ptr;
struct abbrev_info *cur_abbrev;
- abbrev_table_up abbrev_table (new struct abbrev_table (sect_off));
+ abbrev_table_up abbrev_table (new struct abbrev_table (sect_off, section));
struct obstack *obstack = &abbrev_table->m_abbrev_obstack;
/* Caller must ensure this. */
diff --git a/gdb/dwarf2/abbrev.h b/gdb/dwarf2/abbrev.h
index 50f2ed46083..0a1ca4d39b4 100644
--- a/gdb/dwarf2/abbrev.h
+++ b/gdb/dwarf2/abbrev.h
@@ -85,9 +85,11 @@ struct abbrev_table
This is used as a sanity check when the table is used. */
const sect_offset sect_off;
+ struct dwarf2_section_info *section;
+
private:
- explicit abbrev_table (sect_offset off);
+ abbrev_table (sect_offset off, struct dwarf2_section_info *sect);
DISABLE_COPY_AND_ASSIGN (abbrev_table);
--
2.31.1
next prev parent reply other threads:[~2021-08-26 2:19 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-26 2:19 [PATCH 00/30] Rewrite the DWARF "partial" reader Tom Tromey
2021-08-26 2:19 ` [PATCH 01/30] Introduce make_unique_xstrndup Tom Tromey
2021-08-26 2:19 ` [PATCH 02/30] Split create_addrmap_from_aranges Tom Tromey
2021-08-26 2:19 ` [PATCH 03/30] Add dwarf2_per_cu_data::addresses_seen Tom Tromey
2021-08-26 2:19 ` [PATCH 04/30] Refactor dwarf2_get_pc_bounds Tom Tromey
2021-08-26 2:19 ` [PATCH 05/30] Allow ada_decode not to decode operators Tom Tromey
2021-08-26 2:19 ` [PATCH 06/30] Let skip_one_die not skip children Tom Tromey
2021-08-26 2:19 ` [PATCH 07/30] Add name splitting Tom Tromey
2021-08-26 2:19 ` [PATCH 08/30] Add new overload of dwarf5_djb_hash Tom Tromey
2021-08-26 2:19 ` [PATCH 09/30] Refactor build_type_psymtabs_reader Tom Tromey
2021-08-26 2:19 ` [PATCH 10/30] Add batching parameter to parallel_for_each Tom Tromey
2021-08-26 2:19 ` [PATCH 11/30] Return vector of results from parallel_for_each Tom Tromey
2021-08-27 6:20 ` Tom de Vries
2021-08-28 19:20 ` Tom Tromey
2021-08-26 2:19 ` Tom Tromey [this message]
2021-08-26 2:19 ` [PATCH 13/30] Statically examine abbrev properties Tom Tromey
2021-09-06 22:31 ` Lancelot SIX
2021-11-04 18:00 ` Tom Tromey
2021-08-26 2:19 ` [PATCH 14/30] Update skip_one_die for new " Tom Tromey
2021-08-26 2:19 ` [PATCH 15/30] Introduce the new DWARF index class Tom Tromey
2021-09-09 23:32 ` Lancelot SIX
2021-11-04 18:03 ` Tom Tromey
2021-08-26 2:19 ` [PATCH 16/30] The new DWARF indexer Tom Tromey
2021-08-26 2:19 ` [PATCH 17/30] Implement quick_symbol_functions for cooked DWARF index Tom Tromey
2021-08-26 2:19 ` [PATCH 18/30] Wire in the new DWARF indexer Tom Tromey
2021-08-26 2:19 ` [PATCH 19/30] Pre-read DWARF section data Tom Tromey
2021-08-26 2:19 ` [PATCH 20/30] Parallelize DWARF indexing Tom Tromey
2021-08-26 2:19 ` [PATCH 21/30] "Finalize" the DWARF index in the background Tom Tromey
2021-08-26 2:19 ` [PATCH 22/30] Rename write_psymtabs_to_index Tom Tromey
2021-08-26 2:19 ` [PATCH 23/30] Change the key type in psym_index_map Tom Tromey
2021-08-26 2:19 ` [PATCH 24/30] Change parameters to write_address_map Tom Tromey
2021-08-26 2:19 ` [PATCH 25/30] Genericize addrmap handling in the DWARF index writer Tom Tromey
2021-08-26 2:19 ` [PATCH 26/30] Adapt .gdb_index writer to new DWARF scanner Tom Tromey
2021-08-26 2:19 ` [PATCH 27/30] Adapt .debug_names " Tom Tromey
2021-08-26 2:19 ` [PATCH 28/30] Enable the new DWARF indexer Tom Tromey
2021-08-26 2:19 ` [PATCH 29/30] Delete DWARF psymtab code Tom Tromey
2021-08-26 2:19 ` [PATCH 30/30] Remove dwarf2_per_cu_data::v Tom Tromey
2021-08-26 20:32 ` [PATCH 00/30] Rewrite the DWARF "partial" reader Tom de Vries
2021-08-26 21:29 ` Tom Tromey
2021-08-27 7:31 ` Tom de Vries
2021-08-30 15:04 ` Tom Tromey
2021-09-06 19:46 ` Tom Tromey
2021-09-07 10:58 ` Tom de Vries
2021-09-07 12:16 ` Tom de Vries
2021-10-29 23:06 ` Tom Tromey
2021-09-09 19:00 ` Wei-min Pan
2021-09-11 21:08 ` Tom Tromey
2021-09-13 16:50 ` Weimin Pan
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=20210826021937.1490292-13-tom@tromey.com \
--to=tom@tromey.com \
--cc=gdb-patches@sourceware.org \
/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).