From: Tom Tromey <tom@tromey.com>
To: gdb-patches@sourceware.org
Cc: Tom Tromey <tom@tromey.com>
Subject: [PATCH v2 03/28] Emit some type declarations in .gdb_index
Date: Wed, 02 Apr 2025 17:45:02 -0600 [thread overview]
Message-ID: <20250402-search-in-psyms-v2-3-ea91704487cb@tromey.com> (raw)
In-Reply-To: <20250402-search-in-psyms-v2-0-ea91704487cb@tromey.com>
If you run struct-decl.exp with the .gdb_index board, you will see
that "the_type" is not emitted in the index. This would cause a
failure in this series. The fix is to ensure that certain necessary
type declarations are emitted.
However, a naive fix here will regress stub-array-size.exp, where a
type declaration and a type definition are both seen -- but the
declaration is seen first and causes a failure. This is handled by
adding some code (including a mild hack) to filter out type
declarations when a corresponding type definition is seen.
---
gdb/dwarf2/index-write.c | 79 +++++++++++++++++++++++++++++++++++++++++++++---
1 file changed, 74 insertions(+), 5 deletions(-)
diff --git a/gdb/dwarf2/index-write.c b/gdb/dwarf2/index-write.c
index da1f6cd1e0869620a99959639718a554882bab2c..389fcaa727ce15706a735ce82c047374fbc57d2b 100644
--- a/gdb/dwarf2/index-write.c
+++ b/gdb/dwarf2/index-write.c
@@ -55,7 +55,7 @@
#define DW2_GDB_INDEX_SYMBOL_KIND_SET_VALUE(cu_index, value) \
do { \
gdb_assert ((value) >= GDB_INDEX_SYMBOL_KIND_TYPE \
- && (value) <= GDB_INDEX_SYMBOL_KIND_OTHER); \
+ && (value) <= GDB_INDEX_SYMBOL_KIND_UNUSED5); \
GDB_INDEX_SYMBOL_KIND_SET_VALUE((cu_index), (value)); \
} while (0)
@@ -422,10 +422,53 @@ symtab_index_entry::minimize ()
if (name == nullptr || cu_indices.empty ())
return;
- std::sort (cu_indices.begin (), cu_indices.end ());
+ /* We sort the indexes in a funny way: GDB_INDEX_SYMBOL_KIND_UNUSED5
+ is always sorted last; then otherwise we sort by numeric value.
+ This ensures that we prefer the definition when both a definition
+ and a declaration (stub type) are seen. */
+ std::sort (cu_indices.begin (), cu_indices.end (),
+ [] (offset_type vala, offset_type valb)
+ {
+ gdb_index_symbol_kind kinda
+ = GDB_INDEX_SYMBOL_KIND_VALUE (vala);
+ gdb_index_symbol_kind kindb
+ = GDB_INDEX_SYMBOL_KIND_VALUE (valb);
+ if (kinda == GDB_INDEX_SYMBOL_KIND_UNUSED5)
+ {
+ if (kindb == GDB_INDEX_SYMBOL_KIND_UNUSED5)
+ return vala < valb;
+ /* Declaration sorts last. */
+ return false;
+ }
+ else if (kindb == GDB_INDEX_SYMBOL_KIND_UNUSED5)
+ {
+ /* Declaration sorts last. */
+ return true;
+ }
+ return vala < valb;
+ });
auto from = std::unique (cu_indices.begin (), cu_indices.end ());
cu_indices.erase (from, cu_indices.end ());
+ /* Rewrite GDB_INDEX_SYMBOL_KIND_UNUSED5. This ensures that a type
+ declaration will be deleted by the subsequent squashing step, if
+ warranted. */
+ for (auto &val : cu_indices)
+ {
+ gdb_index_symbol_kind kind = GDB_INDEX_SYMBOL_KIND_VALUE (val);
+ if (kind != GDB_INDEX_SYMBOL_KIND_UNUSED5)
+ continue;
+
+ offset_type newval = 0;
+ DW2_GDB_INDEX_CU_SET_VALUE (newval, GDB_INDEX_CU_VALUE (val));
+ DW2_GDB_INDEX_SYMBOL_STATIC_SET_VALUE
+ (newval, GDB_INDEX_SYMBOL_STATIC_VALUE (val));
+ DW2_GDB_INDEX_SYMBOL_KIND_SET_VALUE (newval,
+ GDB_INDEX_SYMBOL_KIND_TYPE);
+
+ val = newval;
+ }
+
/* We don't want to enter a type more than once, so
remove any such duplicates from the list as well. When doing
this, we want to keep the entry from the first CU -- but this is
@@ -1212,6 +1255,22 @@ write_cooked_index (cooked_index *table,
const cu_index_map &cu_index_htab,
struct mapped_symtab *symtab)
{
+ /* A type declaration that is used as a (non-trivial) parent entry
+ must be written out. */
+ gdb::unordered_set<const cooked_index_entry *> required_entries;
+ for (const cooked_index_entry *entry : table->all_entries ())
+ {
+ if ((entry->flags & IS_TYPE_DECLARATION) == 0
+ && entry->get_parent () != nullptr)
+ {
+ for (const cooked_index_entry *parent = entry->get_parent ();
+ parent != nullptr;
+ parent = parent->get_parent ())
+ if ((parent->flags & IS_TYPE_DECLARATION) != 0)
+ required_entries.insert (parent);
+ }
+ }
+
for (const cooked_index_entry *entry : table->all_entries ())
{
const auto it = cu_index_htab.find (entry->per_cu);
@@ -1239,8 +1298,9 @@ write_cooked_index (cooked_index *table,
}
else if ((entry->flags & IS_TYPE_DECLARATION) != 0)
{
- /* Don't add type declarations to the index. */
- continue;
+ /* Don't add most type declarations to the index. */
+ if (!required_entries.contains (entry))
+ continue;
}
gdb_index_symbol_kind kind;
@@ -1252,7 +1312,16 @@ write_cooked_index (cooked_index *table,
|| entry->tag == DW_TAG_enumerator)
kind = GDB_INDEX_SYMBOL_KIND_VARIABLE;
else if (tag_is_type (entry->tag))
- kind = GDB_INDEX_SYMBOL_KIND_TYPE;
+ {
+ /* If we added a type declaration, we want to note this
+ fact for later, because we don't want a type declaration
+ to cause the real definition to be omitted from the
+ index. GDB_INDEX_SYMBOL_KIND_UNUSED5 is used here, but
+ rewritten later before the index is written. */
+ kind = ((entry->flags & IS_TYPE_DECLARATION) == 0
+ ? GDB_INDEX_SYMBOL_KIND_TYPE
+ : GDB_INDEX_SYMBOL_KIND_UNUSED5);
+ }
else
kind = GDB_INDEX_SYMBOL_KIND_OTHER;
--
2.46.1
next prev parent reply other threads:[~2025-04-02 23:48 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-02 23:44 [PATCH v2 00/28] Search symbols via quick API Tom Tromey
2025-04-02 23:45 ` [PATCH v2 01/28] Add another minor hack to cooked_index_entry::full_name Tom Tromey
2025-04-02 23:45 ` [PATCH v2 02/28] Change ada_decode to preserve upper-case in some situations Tom Tromey
2025-04-02 23:45 ` Tom Tromey [this message]
2025-04-21 2:50 ` [PATCH v2 03/28] Emit some type declarations in .gdb_index Simon Marchi
2025-04-21 14:50 ` Tom Tromey
2025-04-23 4:11 ` Simon Marchi
2025-04-23 20:54 ` Tom Tromey
2025-04-02 23:45 ` [PATCH v2 04/28] Ada import functions not in index Tom Tromey
2025-04-02 23:45 ` [PATCH v2 05/28] Fix index's handling of DW_TAG_imported_declaration Tom Tromey
2025-04-02 23:45 ` [PATCH v2 06/28] Put all CTF symbols in global scope Tom Tromey
2025-04-02 23:45 ` [PATCH v2 07/28] Restore "ingestion" of .debug_str when writing .debug_names Tom Tromey
2025-04-02 23:45 ` [PATCH v2 08/28] Entries from anon-struct.exp not in cooked index Tom Tromey
2025-04-02 23:45 ` [PATCH v2 09/28] Remove dwarf2_per_cu_data::mark Tom Tromey
2025-04-21 3:09 ` Simon Marchi
2025-04-21 15:38 ` Tom Tromey
2025-04-23 4:12 ` Simon Marchi
2025-04-02 23:45 ` [PATCH v2 10/28] Have expand_symtabs_matching work for already-expanded CUs Tom Tromey
2025-04-23 15:53 ` Simon Marchi
2025-04-23 20:39 ` Tom Tromey
2025-04-23 20:57 ` Tom Tromey
2025-04-02 23:45 ` [PATCH v2 11/28] Rewrite the .gdb_index reader Tom Tromey
2025-04-23 17:22 ` Simon Marchi
2025-04-23 20:50 ` Tom Tromey
2025-04-24 14:37 ` Pedro Alves
2025-04-02 23:45 ` [PATCH v2 12/28] Convert default_collect_symbol_completion_matches_break_on Tom Tromey
2025-04-02 23:45 ` [PATCH v2 13/28] Convert gdbpy_lookup_static_symbols Tom Tromey
2025-04-02 23:45 ` [PATCH v2 14/28] Convert ada_add_global_exceptions Tom Tromey
2025-04-02 23:45 ` [PATCH v2 15/28] Convert ada_language_defn::collect_symbol_completion_matches Tom Tromey
2025-04-02 23:45 ` [PATCH v2 16/28] Convert ada-lang.c:map_matching_symbols Tom Tromey
2025-04-02 23:45 ` [PATCH v2 17/28] Remove expand_symtabs_matching Tom Tromey
2025-04-02 23:45 ` [PATCH v2 18/28] Simplify basic_lookup_transparent_type Tom Tromey
2025-04-02 23:45 ` [PATCH v2 19/28] Remove objfile::expand_symtabs_for_function Tom Tromey
2025-04-02 23:45 ` [PATCH v2 20/28] Convert linespec.c:iterate_over_all_matching_symtabs Tom Tromey
2025-04-02 23:45 ` [PATCH v2 21/28] Simplify block_lookup_symbol_primary Tom Tromey
2025-04-02 23:45 ` [PATCH v2 22/28] Pass lookup_name_info to block_lookup_symbol_primary Tom Tromey
2025-04-02 23:45 ` [PATCH v2 23/28] Simplify block_lookup_symbol Tom Tromey
2025-04-02 23:45 ` [PATCH v2 24/28] Add best_symbol_tracker Tom Tromey
2025-04-02 23:45 ` [PATCH v2 25/28] Convert lookup_symbol_via_quick_fns Tom Tromey
2025-04-02 23:45 ` [PATCH v2 26/28] Convert lookup_symbol_in_objfile Tom Tromey
2025-04-02 23:45 ` [PATCH v2 27/28] Make dw_expand_symtabs_matching_file_matcher static Tom Tromey
2025-04-23 20:00 ` Simon Marchi
2025-04-23 20:09 ` Tom Tromey
2025-04-23 20:44 ` Tom Tromey
2025-04-02 23:45 ` [PATCH v2 28/28] Remove enter_symbol_lookup Tom Tromey
2025-04-23 20:09 ` [PATCH v2 00/28] Search symbols via quick API Simon Marchi
2025-04-24 21:09 ` Tom Tromey
2025-04-28 14:07 ` Guinevere Larsen
2025-04-28 22:06 ` Tom Tromey
2025-04-29 19:31 ` Guinevere Larsen
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=20250402-search-in-psyms-v2-3-ea91704487cb@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