public inbox for gdb-cvs@sourceware.org help / color / mirror / Atom feed
From: Tom Tromey <tromey@sourceware.org> To: gdb-cvs@sourceware.org Subject: [binutils-gdb] Remove addrmap_fixed::set_entry Date: Mon, 15 Jan 2024 18:54:27 +0000 (GMT) [thread overview] Message-ID: <20240115185427.E22C33858C2F@sourceware.org> (raw) https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=a73afeff18db7304495bf7c42b25f3b93645b46f commit a73afeff18db7304495bf7c42b25f3b93645b46f Author: Tom Tromey <tom@tromey.com> Date: Sun Jan 14 11:20:39 2024 -0700 Remove addrmap_fixed::set_entry It occurred to me that there is no reason for addrmap_fixed::set_entry to exist. This patch removes it and removes the abstract virtual function from the base class. This then required a few minor changes in the DWARF reader. I consider this a type-safety improvement. Tested by rebuilding. Reviewed-By: Tom de Vries <tdevries@suse.de> Diff: --- gdb/addrmap.c | 9 ------ gdb/addrmap.h | 81 ++++++++++++++++++++++++---------------------------- gdb/dwarf2/aranges.c | 2 +- gdb/dwarf2/aranges.h | 4 +-- gdb/dwarf2/read.c | 8 +++--- 5 files changed, 45 insertions(+), 59 deletions(-) diff --git a/gdb/addrmap.c b/gdb/addrmap.c index 2753f85e0a6..162ab84763b 100644 --- a/gdb/addrmap.c +++ b/gdb/addrmap.c @@ -30,15 +30,6 @@ static_assert (sizeof (splay_tree_value) >= sizeof (void *)); \f /* Fixed address maps. */ -void -addrmap_fixed::set_empty (CORE_ADDR start, CORE_ADDR end_inclusive, - void *obj) -{ - internal_error ("addrmap_fixed_set_empty: " - "fixed addrmaps can't be changed\n"); -} - - void * addrmap_fixed::do_find (CORE_ADDR addr) const { diff --git a/gdb/addrmap.h b/gdb/addrmap.h index c6417169443..ba83607ad8c 100644 --- a/gdb/addrmap.h +++ b/gdb/addrmap.h @@ -46,46 +46,6 @@ struct addrmap { virtual ~addrmap () = default; - /* In the mutable address map MAP, associate the addresses from START - to END_INCLUSIVE that are currently associated with NULL with OBJ - instead. Addresses mapped to an object other than NULL are left - unchanged. - - As the name suggests, END_INCLUSIVE is also mapped to OBJ. This - convention is unusual, but it allows callers to accurately specify - ranges that abut the top of the address space, and ranges that - cover the entire address space. - - This operation seems a bit complicated for a primitive: if it's - needed, why not just have a simpler primitive operation that sets a - range to a value, wiping out whatever was there before, and then - let the caller construct more complicated operations from that, - along with some others for traversal? - - It turns out this is the mutation operation we want to use all the - time, at least for now. Our immediate use for address maps is to - represent lexical blocks whose address ranges are not contiguous. - We walk the tree of lexical blocks present in the debug info, and - only create 'struct block' objects after we've traversed all a - block's children. If a lexical block declares no local variables - (and isn't the lexical block for a function's body), we omit it - from GDB's data structures entirely. - - However, this means that we don't decide to create a block (and - thus record it in the address map) until after we've traversed its - children. If we do decide to create the block, we do so at a time - when all its children have already been recorded in the map. So - this operation --- change only those addresses left unset --- is - actually the operation we want to use every time. - - It seems simpler to let the code which operates on the - representation directly deal with the hair of implementing these - semantics than to provide an interface which allows it to be - implemented efficiently, but doesn't reveal too much of the - representation. */ - virtual void set_empty (CORE_ADDR start, CORE_ADDR end_inclusive, - void *obj) = 0; - /* Return the object associated with ADDR in MAP. */ const void *find (CORE_ADDR addr) const { return this->do_find (addr); } @@ -127,8 +87,6 @@ public: addrmap_fixed (struct obstack *obstack, addrmap_mutable *mut); DISABLE_COPY_AND_ASSIGN (addrmap_fixed); - void set_empty (CORE_ADDR start, CORE_ADDR end_inclusive, - void *obj) override; void relocate (CORE_ADDR offset) override; private: @@ -165,8 +123,45 @@ public: ~addrmap_mutable (); DISABLE_COPY_AND_ASSIGN (addrmap_mutable); + /* In the mutable address map MAP, associate the addresses from START + to END_INCLUSIVE that are currently associated with NULL with OBJ + instead. Addresses mapped to an object other than NULL are left + unchanged. + + As the name suggests, END_INCLUSIVE is also mapped to OBJ. This + convention is unusual, but it allows callers to accurately specify + ranges that abut the top of the address space, and ranges that + cover the entire address space. + + This operation seems a bit complicated for a primitive: if it's + needed, why not just have a simpler primitive operation that sets a + range to a value, wiping out whatever was there before, and then + let the caller construct more complicated operations from that, + along with some others for traversal? + + It turns out this is the mutation operation we want to use all the + time, at least for now. Our immediate use for address maps is to + represent lexical blocks whose address ranges are not contiguous. + We walk the tree of lexical blocks present in the debug info, and + only create 'struct block' objects after we've traversed all a + block's children. If a lexical block declares no local variables + (and isn't the lexical block for a function's body), we omit it + from GDB's data structures entirely. + + However, this means that we don't decide to create a block (and + thus record it in the address map) until after we've traversed its + children. If we do decide to create the block, we do so at a time + when all its children have already been recorded in the map. So + this operation --- change only those addresses left unset --- is + actually the operation we want to use every time. + + It seems simpler to let the code which operates on the + representation directly deal with the hair of implementing these + semantics than to provide an interface which allows it to be + implemented efficiently, but doesn't reveal too much of the + representation. */ void set_empty (CORE_ADDR start, CORE_ADDR end_inclusive, - void *obj) override; + void *obj); void relocate (CORE_ADDR offset) override; private: diff --git a/gdb/dwarf2/aranges.c b/gdb/dwarf2/aranges.c index 0a4ae63801d..4287a5a884b 100644 --- a/gdb/dwarf2/aranges.c +++ b/gdb/dwarf2/aranges.c @@ -26,7 +26,7 @@ bool read_addrmap_from_aranges (dwarf2_per_objfile *per_objfile, dwarf2_section_info *section, - addrmap *mutable_map, + addrmap_mutable *mutable_map, deferred_warnings *warn) { /* Caller must ensure that the section has already been read. */ diff --git a/gdb/dwarf2/aranges.h b/gdb/dwarf2/aranges.h index d923e2d4905..2ba8a05620d 100644 --- a/gdb/dwarf2/aranges.h +++ b/gdb/dwarf2/aranges.h @@ -22,7 +22,7 @@ class dwarf2_per_objfile; class dwarf2_section_info; -class addrmap; +class addrmap_mutable; /* Read the address map data from DWARF-5 .debug_aranges, and use it to populate given addrmap. Returns true on success, false on @@ -30,7 +30,7 @@ class addrmap; extern bool read_addrmap_from_aranges (dwarf2_per_objfile *per_objfile, dwarf2_section_info *section, - addrmap *mutable_map, + addrmap_mutable *mutable_map, deferred_warnings *warn); #endif /* GDB_DWARF2_ARANGES_H */ diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c index a50248c4d56..1b5e566bb01 100644 --- a/gdb/dwarf2/read.c +++ b/gdb/dwarf2/read.c @@ -907,7 +907,7 @@ static enum pc_bounds_kind dwarf2_get_pc_bounds (struct die_info *, unrelocated_addr *, unrelocated_addr *, struct dwarf2_cu *, - addrmap *, + addrmap_mutable *, void *); static void get_scope_pc_bounds (struct die_info *, @@ -11013,7 +11013,7 @@ dwarf2_ranges_process (unsigned offset, struct dwarf2_cu *cu, dwarf_tag tag, static int dwarf2_ranges_read (unsigned offset, unrelocated_addr *low_return, unrelocated_addr *high_return, struct dwarf2_cu *cu, - addrmap *map, void *datum, dwarf_tag tag) + addrmap_mutable *map, void *datum, dwarf_tag tag) { dwarf2_per_objfile *per_objfile = cu->per_objfile; int low_set = 0; @@ -11123,7 +11123,7 @@ dwarf2_get_pc_bounds_entry_point (die_info *die, unrelocated_addr *low, static pc_bounds_kind dwarf_get_pc_bounds_ranges_or_highlow_pc (die_info *die, unrelocated_addr *low, unrelocated_addr *high, dwarf2_cu *cu, - addrmap *map, void *datum) + addrmap_mutable *map, void *datum) { gdb_assert (low != nullptr); gdb_assert (high != nullptr); @@ -11192,7 +11192,7 @@ dwarf_get_pc_bounds_ranges_or_highlow_pc (die_info *die, unrelocated_addr *low, static enum pc_bounds_kind dwarf2_get_pc_bounds (struct die_info *die, unrelocated_addr *lowpc, unrelocated_addr *highpc, struct dwarf2_cu *cu, - addrmap *map, void *datum) + addrmap_mutable *map, void *datum) { dwarf2_per_objfile *per_objfile = cu->per_objfile;
reply other threads:[~2024-01-15 18:54 UTC|newest] Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
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=20240115185427.E22C33858C2F@sourceware.org \ --to=tromey@sourceware.org \ --cc=gdb-cvs@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: linkBe 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).