[PATCH] Add initializers to comp_unit_head

[PATCH] Add initializers to comp_unit_head

[Tom Tromey]

PR symtab/29343 points out that it would be beneficial if
comp_unit_head had a constructor and used initializers.  This patch
implements this.  I'm unsure if this is sufficient to close the bug,
but at least it's a step.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29343
---
 gdb/dwarf2/comp-unit-head.h | 24 ++++++++++++------------
 gdb/dwarf2/cu.h             |  2 +-
 gdb/dwarf2/read.c           |  2 --
 gdb/dwarf2/read.h           |  2 +-
 4 files changed, 14 insertions(+), 16 deletions(-)

diff --git a/gdb/dwarf2/comp-unit-head.h b/gdb/dwarf2/comp-unit-head.h
index a7ee3e64546..7579fe7fa29 100644
--- a/gdb/dwarf2/comp-unit-head.h
+++ b/gdb/dwarf2/comp-unit-head.h
@@ -34,36 +34,36 @@
    translation, looks like this.  */
 struct comp_unit_head
 {
-  unsigned int length;
-  unsigned char version;
-  unsigned char addr_size;
-  unsigned char signed_addr_p;
-  sect_offset abbrev_sect_off;
+  unsigned int length = 0;
+  unsigned char version = 0;
+  unsigned char addr_size = 0;
+  unsigned char signed_addr_p = 0;
+  sect_offset abbrev_sect_off {};
 
   /* Size of file offsets; either 4 or 8.  */
-  unsigned int offset_size;
+  unsigned int offset_size = 0;
 
   /* Size of the length field; either 4 or 12.  */
-  unsigned int initial_length_size;
+  unsigned int initial_length_size = 0;
 
-  enum dwarf_unit_type unit_type;
+  enum dwarf_unit_type unit_type {};
 
   /* Offset to first die in this cu from the start of the cu.
      This will be the first byte following the compilation unit header.  */
-  cu_offset first_die_cu_offset;
+  cu_offset first_die_cu_offset {};
 
   /* Offset to the first byte of this compilation unit header in the
      .debug_info section, for resolving relative reference dies.  */
-  sect_offset sect_off;
+  sect_offset sect_off {};
 
   /* For types, offset in the type's DIE of the type defined by this TU.  */
-  cu_offset type_cu_offset_in_tu;
+  cu_offset type_cu_offset_in_tu {};
 
   /* 64-bit signature of this unit. For type units, it denotes the signature of
      the type (DW_UT_type in DWARF 4, additionally DW_UT_split_type in DWARF 5).
      Also used in DWARF 5, to denote the dwo id when the unit type is
      DW_UT_skeleton or DW_UT_split_compile.  */
-  ULONGEST signature;
+  ULONGEST signature = 0;
 
   /* Return the total length of the CU described by this header.  */
   unsigned int get_length () const
diff --git a/gdb/dwarf2/cu.h b/gdb/dwarf2/cu.h
index 52975095871..b126acab7b2 100644
--- a/gdb/dwarf2/cu.h
+++ b/gdb/dwarf2/cu.h
@@ -98,7 +98,7 @@ struct dwarf2_cu
   void add_dependence (struct dwarf2_per_cu_data *ref_per_cu);
 
   /* The header of the compilation unit.  */
-  struct comp_unit_head header {};
+  struct comp_unit_head header;
 
   /* Base address of this compilation unit.  */
   gdb::optional<CORE_ADDR> base_address;
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index 7bd12c107e2..28fadb886aa 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -23580,8 +23580,6 @@ dwarf2_per_cu_data::get_header () const
       const gdb_byte *info_ptr
 	= this->section->buffer + to_underlying (this->sect_off);
 
-      memset (&m_header, 0, sizeof (m_header));
-
       read_comp_unit_head (&m_header, info_ptr, this->section,
 			   rcuh_kind::COMPILE);
 
diff --git a/gdb/dwarf2/read.h b/gdb/dwarf2/read.h
index dd718266883..0dc4e2e885c 100644
--- a/gdb/dwarf2/read.h
+++ b/gdb/dwarf2/read.h
@@ -206,7 +206,7 @@ struct dwarf2_per_cu_data
 
      Don't access this field directly, use the get_header method instead.  It
      should be private, but we can't make it private at the moment.  */
-  mutable comp_unit_head m_header {};
+  mutable comp_unit_head m_header;
 
   /* The file and directory for this CU.  This is cached so that we
      don't need to re-examine the DWO in some situations.  This may be
-- 
2.38.1

Re: [PATCH] Add initializers to comp_unit_head

[tdevries]

[-- Attachment #1: Type: text/plain, Size: 477 bytes --]

On 2022-12-24 22:44, Tom Tromey wrote:
> PR symtab/29343 points out that it would be beneficial if
> comp_unit_head had a constructor and used initializers.  This patch
> implements this.

LGTM.

> I'm unsure if this is sufficient to close the bug,
> but at least it's a step.
> 

I think it's sufficient.

FWIW, I've rebased the patch attached to the PR on your patch, and I'm 
left with the patch below, which makes the length field of 
comp_unit_head private.

Thanks,
- Tom

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-gdb-symtab-Make-comp_unit_head.length-private.patch --]
[-- Type: text/x-patch; name=0001-gdb-symtab-Make-comp_unit_head.length-private.patch, Size: 3328 bytes --]

From b04c791586da2916ca87521eca3c82ba71cd22fb Mon Sep 17 00:00:00 2001
From: Tom de Vries <tdevries@suse.de>
Date: Sun, 25 Dec 2022 08:53:43 +0100
Subject: [PATCH] [gdb/symtab] Make comp_unit_head.length private

Make comp_unit_head.length private, to enforce using accessor function
get_length.

Tested on x86_64-linux.

PR symtab/29343
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29343
---
 gdb/dwarf2/comp-unit-head.c |  4 ++--
 gdb/dwarf2/comp-unit-head.h | 16 +++++++++++++---
 gdb/dwarf2/read.c           |  4 ++--
 3 files changed, 17 insertions(+), 7 deletions(-)

diff --git a/gdb/dwarf2/comp-unit-head.c b/gdb/dwarf2/comp-unit-head.c
index edb2a8aa55a..128d62fcbb1 100644
--- a/gdb/dwarf2/comp-unit-head.c
+++ b/gdb/dwarf2/comp-unit-head.c
@@ -44,7 +44,7 @@ read_comp_unit_head (struct comp_unit_head *cu_header,
   const char *filename = section->get_file_name ();
   bfd *abfd = section->get_bfd_owner ();
 
-  cu_header->length = read_initial_length (abfd, info_ptr, &bytes_read);
+  cu_header->set_length (read_initial_length (abfd, info_ptr, &bytes_read));
   cu_header->initial_length_size = bytes_read;
   cu_header->offset_size = (bytes_read == 4) ? 4 : 8;
   info_ptr += bytes_read;
@@ -166,7 +166,7 @@ error_check_comp_unit_head (dwarf2_per_objfile *per_objfile,
       > section->size)
     error (_("Dwarf Error: bad length (0x%x) in compilation unit header "
 	   "(offset %s + 0) [in module %s]"),
-	   header->length, sect_offset_str (header->sect_off),
+	   header->get_length (false), sect_offset_str (header->sect_off),
 	   filename);
 }
 
diff --git a/gdb/dwarf2/comp-unit-head.h b/gdb/dwarf2/comp-unit-head.h
index 7579fe7fa29..56b37ae537e 100644
--- a/gdb/dwarf2/comp-unit-head.h
+++ b/gdb/dwarf2/comp-unit-head.h
@@ -34,7 +34,9 @@
    translation, looks like this.  */
 struct comp_unit_head
 {
-  unsigned int length = 0;
+private:
+  unsigned int m_length = 0;
+public:
   unsigned char version = 0;
   unsigned char addr_size = 0;
   unsigned char signed_addr_p = 0;
@@ -65,10 +67,18 @@ struct comp_unit_head
      DW_UT_skeleton or DW_UT_split_compile.  */
   ULONGEST signature = 0;
 
+  void set_length (unsigned int length)
+  {
+    m_length = length;
+  }
+
   /* Return the total length of the CU described by this header.  */
-  unsigned int get_length () const
+  unsigned int get_length (bool include_initial = true) const
   {
-    return initial_length_size + length;
+    unsigned int res = m_length;
+    if (include_initial)
+      res += initial_length_size;
+    return res;
   }
 
   /* Return TRUE if OFF is within this CU.  */
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index 28fadb886aa..50881929d47 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -7744,7 +7744,7 @@ load_full_comp_unit (dwarf2_per_cu_data *this_cu,
 
   gdb_assert (cu->die_hash == NULL);
   cu->die_hash =
-    htab_create_alloc_ex (cu->header.length / 12,
+    htab_create_alloc_ex (cu->header.get_length (false) / 12,
 			  die_hash,
 			  die_eq,
 			  NULL,
@@ -23026,7 +23026,7 @@ read_signatured_type (signatured_type *sig_type,
 
       gdb_assert (cu->die_hash == NULL);
       cu->die_hash =
-	htab_create_alloc_ex (cu->header.length / 12,
+	htab_create_alloc_ex (cu->header.get_length (false) / 12,
 			      die_hash,
 			      die_eq,
 			      NULL,
-- 
2.34.1

Re: [PATCH] Add initializers to comp_unit_head

[Tom Tromey]

>>>>> tdevries via Gdb-patches <gdb-patches@sourceware.org> writes:

> FWIW, I've rebased the patch attached to the PR on your patch, and I'm
> left with the patch below, which makes the length field of
> comp_unit_head private.

>    /* Return the total length of the CU described by this header.  */
> -  unsigned int get_length () const
> +  unsigned int get_length (bool include_initial = true) const
>    {
> -    return initial_length_size + length;
> +    unsigned int res = m_length;
> +    if (include_initial)
> +      res += initial_length_size;
> +    return res;

I think that since the callers all statically decide which length they
want, there should probably just be two methods with different names.

Tom

Re: [PATCH] Add initializers to comp_unit_head

[tdevries]

On 2022-12-25 15:33, Tom Tromey wrote:
>>>>>> tdevries via Gdb-patches <gdb-patches@sourceware.org> writes:
> 
>> FWIW, I've rebased the patch attached to the PR on your patch, and I'm
>> left with the patch below, which makes the length field of
>> comp_unit_head private.
> 
>>    /* Return the total length of the CU described by this header.  */
>> -  unsigned int get_length () const
>> +  unsigned int get_length (bool include_initial = true) const
>>    {
>> -    return initial_length_size + length;
>> +    unsigned int res = m_length;
>> +    if (include_initial)
>> +      res += initial_length_size;
>> +    return res;
> 
> I think that since the callers all statically decide which length they
> want, there should probably just be two methods with different names.
> 

OK, so how about:
- get_length_with_initial
- get_length_without_initial
?

Thanks,
- Tom

> Tom

Re: [PATCH] Add initializers to comp_unit_head

[Tom Tromey]

> OK, so how about:
> - get_length_with_initial
> - get_length_without_initial
> ?

Works for me, thank you.

Tom

[pushed] [gdb/symtab] Make comp_unit_head.length private

[Tom de Vries]

[-- Attachment #1: Type: text/plain, Size: 267 bytes --]

[was: Re: [PATCH] Add initializers to comp_unit_head ]
On 12/26/22 18:59, Tom Tromey wrote:
>> OK, so how about:
>> - get_length_with_initial
>> - get_length_without_initial
>> ?
> 
> Works for me, thank you.

Committed with that approach, as attached.

Thanks,
- Tom

[-- Attachment #2: 0001-gdb-symtab-Make-comp_unit_head.length-private.patch --]
[-- Type: text/x-patch, Size: 7449 bytes --]

From d8f52a9a9ccbf7411cf4ae487d2756826f5d0bd5 Mon Sep 17 00:00:00 2001
From: Tom de Vries <tdevries@suse.de>
Date: Fri, 30 Dec 2022 13:55:22 +0100
Subject: [pushed] [gdb/symtab] Make comp_unit_head.length private

Make comp_unit_head.length private, to enforce using accessor functions.

Replace accessor function get_length with get_length_with_initial and
get_length_without_initial, to make it explicit which variant we're using.

Tested on x86_64-linux.

PR symtab/29343
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29343
---
 gdb/dwarf2/comp-unit-head.c |  6 +++---
 gdb/dwarf2/comp-unit-head.h | 25 ++++++++++++++++++++-----
 gdb/dwarf2/read.c           | 20 ++++++++++----------
 3 files changed, 33 insertions(+), 18 deletions(-)

diff --git a/gdb/dwarf2/comp-unit-head.c b/gdb/dwarf2/comp-unit-head.c
index edb2a8aa55a..b618a674808 100644
--- a/gdb/dwarf2/comp-unit-head.c
+++ b/gdb/dwarf2/comp-unit-head.c
@@ -44,7 +44,7 @@ read_comp_unit_head (struct comp_unit_head *cu_header,
   const char *filename = section->get_file_name ();
   bfd *abfd = section->get_bfd_owner ();
 
-  cu_header->length = read_initial_length (abfd, info_ptr, &bytes_read);
+  cu_header->set_length (read_initial_length (abfd, info_ptr, &bytes_read));
   cu_header->initial_length_size = bytes_read;
   cu_header->offset_size = (bytes_read == 4) ? 4 : 8;
   info_ptr += bytes_read;
@@ -162,11 +162,11 @@ error_check_comp_unit_head (dwarf2_per_objfile *per_objfile,
 
   /* Cast to ULONGEST to use 64-bit arithmetic when possible to
      avoid potential 32-bit overflow.  */
-  if (((ULONGEST) header->sect_off + header->get_length ())
+  if (((ULONGEST) header->sect_off + header->get_length_with_initial ())
       > section->size)
     error (_("Dwarf Error: bad length (0x%x) in compilation unit header "
 	   "(offset %s + 0) [in module %s]"),
-	   header->length, sect_offset_str (header->sect_off),
+	   header->get_length_without_initial (), sect_offset_str (header->sect_off),
 	   filename);
 }
 
diff --git a/gdb/dwarf2/comp-unit-head.h b/gdb/dwarf2/comp-unit-head.h
index 7579fe7fa29..45384fac720 100644
--- a/gdb/dwarf2/comp-unit-head.h
+++ b/gdb/dwarf2/comp-unit-head.h
@@ -34,7 +34,9 @@
    translation, looks like this.  */
 struct comp_unit_head
 {
-  unsigned int length = 0;
+private:
+  unsigned int m_length = 0;
+public:
   unsigned char version = 0;
   unsigned char addr_size = 0;
   unsigned char signed_addr_p = 0;
@@ -65,17 +67,30 @@ struct comp_unit_head
      DW_UT_skeleton or DW_UT_split_compile.  */
   ULONGEST signature = 0;
 
-  /* Return the total length of the CU described by this header.  */
-  unsigned int get_length () const
+  void set_length (unsigned int length)
   {
-    return initial_length_size + length;
+    m_length = length;
+  }
+
+  /* Return the total length of the CU described by this header, including the
+     initial length field.  */
+  unsigned int get_length_with_initial () const
+  {
+    return m_length + initial_length_size;
+  }
+
+  /* Return the total length of the CU described by this header, excluding the
+     initial length field.  */
+  unsigned int get_length_without_initial () const
+  {
+    return m_length;
   }
 
   /* Return TRUE if OFF is within this CU.  */
   bool offset_in_cu_p (sect_offset off) const
   {
     sect_offset bottom = sect_off;
-    sect_offset top = sect_off + get_length ();
+    sect_offset top = sect_off + get_length_with_initial ();
     return off >= bottom && off < top;
   }
 
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index 28fadb886aa..b33ea6847e0 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -5598,7 +5598,7 @@ create_debug_type_hash_table (dwarf2_per_objfile *per_objfile,
       ptr = read_and_check_comp_unit_head (per_objfile, &header, section,
 					   abbrev_section, ptr, section_kind);
 
-      length = header.get_length ();
+      length = header.get_length_with_initial ();
 
       /* Skip dummy type units.  */
       if (ptr >= info_ptr + length
@@ -6005,7 +6005,7 @@ read_cutu_die_from_dwo (dwarf2_cu *cu,
       gdb_assert (dwo_unit->sect_off == cu->header.sect_off);
       /* For DWOs coming from DWP files, we don't know the CU length
 	 nor the type's offset in the TU until now.  */
-      dwo_unit->length = cu->header.get_length ();
+      dwo_unit->length = cu->header.get_length_with_initial ();
       dwo_unit->type_offset_in_tu = cu->header.type_cu_offset_in_tu;
 
       /* Establish the type offset that can be used to lookup the type.
@@ -6021,7 +6021,7 @@ read_cutu_die_from_dwo (dwarf2_cu *cu,
       gdb_assert (dwo_unit->sect_off == cu->header.sect_off);
       /* For DWOs coming from DWP files, we don't know the CU length
 	 until now.  */
-      dwo_unit->length = cu->header.get_length ();
+      dwo_unit->length = cu->header.get_length_with_initial ();
     }
 
   dwo_abbrev_section->read (objfile);
@@ -6289,7 +6289,7 @@ cutu_reader::cutu_reader (dwarf2_per_cu_data *this_cu,
 
 	  /* LENGTH has not been set yet for type units if we're
 	     using .gdb_index.  */
-	  this_cu->set_length (cu->header.get_length ());
+	  this_cu->set_length (cu->header.get_length_with_initial ());
 
 	  /* Establish the type offset that can be used to lookup the type.  */
 	  sig_type->type_offset_in_section =
@@ -6305,7 +6305,7 @@ cutu_reader::cutu_reader (dwarf2_per_cu_data *this_cu,
 						    rcuh_kind::COMPILE);
 
 	  gdb_assert (this_cu->sect_off == cu->header.sect_off);
-	  this_cu->set_length (cu->header.get_length ());
+	  this_cu->set_length (cu->header.get_length_with_initial ());
 	  this_cu->set_version (cu->header.version);
 	}
     }
@@ -6465,7 +6465,7 @@ cutu_reader::cutu_reader (dwarf2_per_cu_data *this_cu,
       m_new_cu->str_offsets_base = parent_cu->str_offsets_base;
       m_new_cu->addr_base = parent_cu->addr_base;
     }
-  this_cu->set_length (m_new_cu->header.get_length ());
+  this_cu->set_length (m_new_cu->header.get_length_with_initial ());
 
   /* Skip dummy compilation units.  */
   if (info_ptr >= begin_info_ptr + this_cu->length ()
@@ -7249,7 +7249,7 @@ read_comp_units_from_section (dwarf2_per_objfile *per_objfile,
 	  *slot = sig_ptr;
 	}
       this_cu->sect_off = sect_off;
-      this_cu->set_length (cu_header.get_length ());
+      this_cu->set_length (cu_header.get_length_with_initial ());
       this_cu->is_dwz = is_dwz;
       this_cu->section = section;
       /* Init this asap, to avoid a data race in the set_version in
@@ -7744,7 +7744,7 @@ load_full_comp_unit (dwarf2_per_cu_data *this_cu,
 
   gdb_assert (cu->die_hash == NULL);
   cu->die_hash =
-    htab_create_alloc_ex (cu->header.length / 12,
+    htab_create_alloc_ex (cu->header.get_length_without_initial () / 12,
 			  die_hash,
 			  die_eq,
 			  NULL,
@@ -18409,7 +18409,7 @@ cooked_indexer::index_dies (cutu_reader *reader,
 {
   const gdb_byte *end_ptr = (reader->buffer
 			     + to_underlying (reader->cu->header.sect_off)
-			     + reader->cu->header.get_length ());
+			     + reader->cu->header.get_length_with_initial ());
 
   while (info_ptr < end_ptr)
     {
@@ -23026,7 +23026,7 @@ read_signatured_type (signatured_type *sig_type,
 
       gdb_assert (cu->die_hash == NULL);
       cu->die_hash =
-	htab_create_alloc_ex (cu->header.length / 12,
+	htab_create_alloc_ex (cu->header.get_length_without_initial () / 12,
 			      die_hash,
 			      die_eq,
 			      NULL,

base-commit: a984f112b015b7d33c3c91230eb4c35695926539
-- 
2.35.3