[07/20] corpus: make get_(undefined_)?_(var|fun)_symbols use the new symtab

Message ID 20210127125853.886677-8-maennich@google.com
State Committed
Headers
Series Refactor (k)symtab reader |

Commit Message

Matthias Männich Jan. 27, 2021, 12:58 p.m. UTC
  Make the corresponding members an implementation detail of corpus::priv.
They get computed based on the new symtab whenever they are accessed
first with an atomic instantiation. That simplifies the implementation
and homogenizes the access to functions and variables. Sorting does not
need to be done as the symtab already gives a guarantee for that.

Due to improved alias detection in the new symtab reader, ensure we only
write symbol aliases to ksymtab symbols if having a ksymtab main symbol.

Test data needed to be adjusted as the new symtab reader is stricter in
regards to symbols listed in ksymtab. I.e. init_module is not an
exported symbol in the ksymtab of a kernel module.

	* src/abg-corpus-priv.h (corpus::priv::sorted_var_symbols): make
	  private, mutable and optional.
	  (corpus::sorted_undefined_var_symbols): Likewise.
	  (corpus::sorted_fun_symbols): Likewise.
	  (corpus::sorted_undefined_fun_symbols): Likewise.
	  (corpus::priv::get_sorted_fun_symbols): New method declaration.
	  (corpus::priv::get_sorted_undefined_fun_symbols): Likewise.
	  (corpus::priv::get_sorted_var_symbols): Likewise.
	  (corpus::priv::get_sorted_undefined_var_symbols): Likewise.
	* src/abg-corpus.cc
	  (corpus::elf_symbol_comp_functor): Delete struct.
	  (corpus::priv::get_sorted_fun_symbols): New method implementation.
	  (corpus::priv::get_sorted_undefined_fun_symbols): Likewise.
	  (corpus::priv::get_sorted_var_symbols): Likewise.
	  (corpus::priv::get_sorted_undefined_var_symbols): Likewise.
	  (corpus::get_sorted_fun_symbols): Proxy call to corpus::priv.
	  (corpus::get_sorted_undefined_fun_symbols): Likewise.
	  (corpus::get_sorted_var_symbols): Likewise.
	  (corpus::get_sorted_undefined_var_symbols): Likewise.
	* src/abg-writer.cc (write_elf_symbol_aliases): When emitting
	  aliases for a kernel symbol, ensure to only emit exported,
	  public aliases.
	* tests/data/test-read-dwarf/PR25007-sdhci.ko.abi: update test
	  data.

Reviewed-by: Giuliano Procida <gprocida@google.com>
Signed-off-by: Matthias Maennich <maennich@google.com>
---
 src/abg-corpus-priv.h                         |  21 +-
 src/abg-corpus.cc                             | 241 +++++++-----------
 src/abg-writer.cc                             |  39 ++-
 .../data/test-read-dwarf/PR25007-sdhci.ko.abi |   3 -
 4 files changed, 135 insertions(+), 169 deletions(-)
  

Comments

Dodji Seketeli March 15, 2021, 10:05 a.m. UTC | #1
Hello,

Matthias Maennich <maennich@google.com> a écrit:

[...]

> diff --git a/src/abg-corpus.cc b/src/abg-corpus.cc
> index 0d5849ddac5d..75f897eb72a8 100644
> --- a/src/abg-corpus.cc
> +++ b/src/abg-corpus.cc

[...]

> -/// Return a sorted vector of function symbols for this corpus.
> -///
> -/// Note that the first time this function is called, the symbols are
> -/// sorted and cached.  Subsequent invocations of this function return
> -/// the cached vector that was built previously.
> -///
> -/// @return the sorted list of function symbols.

Please keep this function comment even if it's redundant.  We need this
for the API doc to be complete.

>  const elf_symbols&
>  corpus::get_sorted_fun_symbols() const

> -{
> -  if (priv_->sorted_fun_symbols.empty()
> -      && !get_fun_symbol_map().empty())
> -    {
> -      priv_->sorted_fun_symbols.reserve(get_fun_symbol_map().size());
> -      for (string_elf_symbols_map_type::const_iterator i =
> -	     get_fun_symbol_map().begin();
> -	   i != get_fun_symbol_map().end();
> -	   ++i)
> -	for (elf_symbols::const_iterator s = i->second.begin();
> -	     s != i->second.end();
> -	     ++s)
> -	  priv_->sorted_fun_symbols.push_back(*s);
> +{ return priv_->get_sorted_fun_symbols(); }

Please remove the leading and trailing white space.

[...]

> -/// Getter for a sorted vector of the function symbols undefined in
> -/// this corpus.
> -///
> -/// @return a vector of the function symbols undefined in this corpus,
> -/// sorted by name and then version.

Likewise.

>  const elf_symbols&
>  corpus::get_sorted_undefined_fun_symbols() const
> -{
> -  if (priv_->sorted_undefined_fun_symbols.empty()
> -      && !get_undefined_fun_symbol_map().empty())
> -    {
> -      priv_->sorted_undefined_fun_symbols.reserve
> -	(get_undefined_fun_symbol_map().size());
> -      for (string_elf_symbols_map_type::const_iterator i =
> -	     get_undefined_fun_symbol_map().begin();
> -	   i != get_undefined_fun_symbol_map().end();
> -	   ++i)
> -	for (elf_symbols::const_iterator s = i->second.begin();
> -	     s != i->second.end();
> -	     ++s)
> -	  priv_->sorted_undefined_fun_symbols.push_back(*s);
> +{ return priv_->get_sorted_undefined_fun_symbols(); }

Likewise.

[...]

> +const elf_symbols&
> +corpus::get_sorted_var_symbols() const

Likewise.

> +{ return priv_->get_sorted_var_symbols(); }

Likewise.

> +const elf_symbols&
> +corpus::get_sorted_undefined_var_symbols() const

Likewise.

> +{ return priv_->get_sorted_undefined_var_symbols(); }

Likewise.

[...]

> Make the corresponding members an implementation detail of corpus::priv.
> They get computed based on the new symtab whenever they are accessed
> first with an atomic instantiation. That simplifies the implementation
> and homogenizes the access to functions and variables. Sorting does not
> need to be done as the symtab already gives a guarantee for that.
>
> Due to improved alias detection in the new symtab reader, ensure we only
> write symbol aliases to ksymtab symbols if having a ksymtab main symbol.
>
> Test data needed to be adjusted as the new symtab reader is stricter in
> regards to symbols listed in ksymtab. I.e. init_module is not an
> exported symbol in the ksymtab of a kernel module.
>
> 	* src/abg-corpus-priv.h (corpus::priv::sorted_var_symbols): make
> 	  private, mutable and optional.
> 	  (corpus::sorted_undefined_var_symbols): Likewise.
> 	  (corpus::sorted_fun_symbols): Likewise.
> 	  (corpus::sorted_undefined_fun_symbols): Likewise.
> 	  (corpus::priv::get_sorted_fun_symbols): New method declaration.
> 	  (corpus::priv::get_sorted_undefined_fun_symbols): Likewise.
> 	  (corpus::priv::get_sorted_var_symbols): Likewise.
> 	  (corpus::priv::get_sorted_undefined_var_symbols): Likewise.
> 	* src/abg-corpus.cc
> 	  (corpus::elf_symbol_comp_functor): Delete struct.
> 	  (corpus::priv::get_sorted_fun_symbols): New method implementation.
> 	  (corpus::priv::get_sorted_undefined_fun_symbols): Likewise.
> 	  (corpus::priv::get_sorted_var_symbols): Likewise.
> 	  (corpus::priv::get_sorted_undefined_var_symbols): Likewise.
> 	  (corpus::get_sorted_fun_symbols): Proxy call to corpus::priv.
> 	  (corpus::get_sorted_undefined_fun_symbols): Likewise.
> 	  (corpus::get_sorted_var_symbols): Likewise.
> 	  (corpus::get_sorted_undefined_var_symbols): Likewise.
> 	* src/abg-writer.cc (write_elf_symbol_aliases): When emitting
> 	  aliases for a kernel symbol, ensure to only emit exported,
> 	  public aliases.
> 	* tests/data/test-read-dwarf/PR25007-sdhci.ko.abi: update test
> 	  data.
>
> Reviewed-by: Giuliano Procida <gprocida@google.com>
> Signed-off-by: Matthias Maennich <maennich@google.com>

OK to apply to master with the above changes once the previous patches
have been applied.

Thanks!

[...]

Cheers,
  

Patch

diff --git a/src/abg-corpus-priv.h b/src/abg-corpus-priv.h
index 290e19c60e67..282796e1b685 100644
--- a/src/abg-corpus-priv.h
+++ b/src/abg-corpus-priv.h
@@ -683,13 +683,9 @@  struct corpus::priv
   vector<var_decl*>				vars;
   string_elf_symbols_map_sptr			var_symbol_map;
   string_elf_symbols_map_sptr			undefined_var_symbol_map;
-  elf_symbols					sorted_var_symbols;
-  elf_symbols					sorted_undefined_var_symbols;
   symtab_reader::symtab_sptr			symtab_;
   string_elf_symbols_map_sptr			fun_symbol_map;
   string_elf_symbols_map_sptr			undefined_fun_symbol_map;
-  elf_symbols					sorted_fun_symbols;
-  elf_symbols					sorted_undefined_fun_symbols;
   elf_symbols					unrefed_fun_symbols;
   elf_symbols					unrefed_var_symbols;
   // The type maps contained in this data member are populated if the
@@ -711,6 +707,11 @@  struct corpus::priv
 private:
   priv();
 
+  mutable abg_compat::optional<elf_symbols> sorted_var_symbols;
+  mutable abg_compat::optional<elf_symbols> sorted_undefined_var_symbols;
+  mutable abg_compat::optional<elf_symbols> sorted_fun_symbols;
+  mutable abg_compat::optional<elf_symbols> sorted_undefined_fun_symbols;
+
 public:
   priv(const string &	p,
        environment*	e)
@@ -730,6 +731,18 @@  public:
   const type_maps&
   get_types() const;
 
+  const elf_symbols&
+  get_sorted_fun_symbols() const;
+
+  const elf_symbols&
+  get_sorted_undefined_fun_symbols() const;
+
+  const elf_symbols&
+  get_sorted_var_symbols() const;
+
+  const elf_symbols&
+  get_sorted_undefined_var_symbols() const;
+
   unordered_set<interned_string, hash_interned_string>*
   get_public_types_pretty_representations();
 
diff --git a/src/abg-corpus.cc b/src/abg-corpus.cc
index 0d5849ddac5d..75f897eb72a8 100644
--- a/src/abg-corpus.cc
+++ b/src/abg-corpus.cc
@@ -438,6 +438,87 @@  const type_maps&
 corpus::priv::get_types() const
 {return types_;}
 
+/// Return a sorted vector of function symbols for this corpus.
+///
+/// Note that the first time this function is called, the symbols are
+/// sorted and cached.  Subsequent invocations of this function return
+/// the cached vector that was built previously.
+///
+/// @return the sorted list of function symbols.
+const elf_symbols&
+corpus::priv::get_sorted_fun_symbols() const
+{
+  if (!sorted_fun_symbols)
+    {
+      auto filter = symtab_->make_filter();
+      filter.set_functions();
+      sorted_fun_symbols = elf_symbols(symtab_->begin(filter), symtab_->end());
+    }
+  return *sorted_fun_symbols;
+}
+
+/// Getter for a sorted vector of the function symbols undefined in
+/// this corpus.
+///
+/// @return a vector of the function symbols undefined in this corpus,
+/// sorted by name and then version.
+const elf_symbols&
+corpus::priv::get_sorted_undefined_fun_symbols() const
+{
+  if (!sorted_undefined_fun_symbols)
+    {
+      auto filter = symtab_->make_filter();
+      filter.set_functions();
+      filter.set_undefined_symbols();
+      filter.set_public_symbols(false);
+
+      sorted_undefined_fun_symbols =
+	elf_symbols(symtab_->begin(filter), symtab_->end());
+    }
+  return *sorted_undefined_fun_symbols;
+}
+
+/// Getter for the sorted vector of variable symbols for this corpus.
+///
+/// Note that the first time this function is called, it computes the
+/// sorted vector, caches the result and returns it.  Subsequent
+/// invocations of this function just return the cached vector.
+///
+/// @return the sorted vector of variable symbols for this corpus.
+const elf_symbols&
+corpus::priv::get_sorted_var_symbols() const
+{
+  if (!sorted_var_symbols)
+    {
+      auto filter = symtab_->make_filter();
+      filter.set_variables();
+
+      sorted_var_symbols = elf_symbols(symtab_->begin(filter), symtab_->end());
+    }
+  return *sorted_var_symbols;
+}
+
+/// Getter for a sorted vector of the variable symbols undefined in
+/// this corpus.
+///
+/// @return a vector of the variable symbols undefined in this corpus,
+/// sorted by name and then version.
+const elf_symbols&
+corpus::priv::get_sorted_undefined_var_symbols() const
+{
+  if (!sorted_undefined_var_symbols)
+    {
+      auto filter = symtab_->make_filter();
+      filter.set_variables();
+      filter.set_undefined_symbols();
+      filter.set_public_symbols(false);
+
+      sorted_undefined_var_symbols =
+	  elf_symbols(symtab_->begin(filter), symtab_->end());
+    }
+  return *sorted_undefined_var_symbols;
+}
+
 /// Getter of the set of pretty representation of types that are
 /// reachable from public interfaces (global functions and variables).
 ///
@@ -933,104 +1014,21 @@  const string_elf_symbols_map_type&
 corpus::get_undefined_fun_symbol_map() const
 {return *get_undefined_fun_symbol_map_sptr();}
 
-/// Functor to sort instances of @ref elf_symbol.
-struct elf_symbol_comp_functor
-{
-
-  /// Return true if the first argument is less than the second one.
-  ///
-  /// @param l the first parameter to consider.
-  ///
-  /// @param r the second parameter to consider.
-  ///
-  /// @return true if @p l is less than @p r
-  bool
-  operator()(elf_symbol& l, elf_symbol& r)
-  {return (l.get_id_string() < r.get_id_string());}
-
-  /// Return true if the first argument is less than the second one.
-  ///
-  /// @param l the first parameter to consider.
-  ///
-  /// @param r the second parameter to consider.
-  ///
-  /// @return true if @p l is less than @p r
-  bool
-  operator()(elf_symbol* l, elf_symbol* r)
-  {return operator()(*l, *r);}
-
-  /// Return true if the first argument is less than the second one.
-  ///
-  /// @param l the first parameter to consider.
-  ///
-  /// @param r the second parameter to consider.
-  ///
-  /// @return true if @p l is less than @p r
-  bool
-  operator()(elf_symbol_sptr l, elf_symbol_sptr r)
-  {return operator()(*l, *r);}
-}; // end struct elf_symbol_comp_functor
-
-/// Return a sorted vector of function symbols for this corpus.
-///
-/// Note that the first time this function is called, the symbols are
-/// sorted and cached.  Subsequent invocations of this function return
-/// the cached vector that was built previously.
-///
-/// @return the sorted list of function symbols.
 const elf_symbols&
 corpus::get_sorted_fun_symbols() const
-{
-  if (priv_->sorted_fun_symbols.empty()
-      && !get_fun_symbol_map().empty())
-    {
-      priv_->sorted_fun_symbols.reserve(get_fun_symbol_map().size());
-      for (string_elf_symbols_map_type::const_iterator i =
-	     get_fun_symbol_map().begin();
-	   i != get_fun_symbol_map().end();
-	   ++i)
-	for (elf_symbols::const_iterator s = i->second.begin();
-	     s != i->second.end();
-	     ++s)
-	  priv_->sorted_fun_symbols.push_back(*s);
+{ return priv_->get_sorted_fun_symbols(); }
 
-      elf_symbol_comp_functor comp;
-      std::sort(priv_->sorted_fun_symbols.begin(),
-		priv_->sorted_fun_symbols.end(),
-		comp);
-    }
-  return priv_->sorted_fun_symbols;
-}
-
-/// Getter for a sorted vector of the function symbols undefined in
-/// this corpus.
-///
-/// @return a vector of the function symbols undefined in this corpus,
-/// sorted by name and then version.
 const elf_symbols&
 corpus::get_sorted_undefined_fun_symbols() const
-{
-  if (priv_->sorted_undefined_fun_symbols.empty()
-      && !get_undefined_fun_symbol_map().empty())
-    {
-      priv_->sorted_undefined_fun_symbols.reserve
-	(get_undefined_fun_symbol_map().size());
-      for (string_elf_symbols_map_type::const_iterator i =
-	     get_undefined_fun_symbol_map().begin();
-	   i != get_undefined_fun_symbol_map().end();
-	   ++i)
-	for (elf_symbols::const_iterator s = i->second.begin();
-	     s != i->second.end();
-	     ++s)
-	  priv_->sorted_undefined_fun_symbols.push_back(*s);
+{ return priv_->get_sorted_undefined_fun_symbols(); }
 
-      elf_symbol_comp_functor comp;
-      std::sort(priv_->sorted_undefined_fun_symbols.begin(),
-		priv_->sorted_undefined_fun_symbols.end(),
-		comp);
-    }
-  return priv_->sorted_undefined_fun_symbols;
-}
+const elf_symbols&
+corpus::get_sorted_var_symbols() const
+{ return priv_->get_sorted_var_symbols(); }
+
+const elf_symbols&
+corpus::get_sorted_undefined_var_symbols() const
+{ return priv_->get_sorted_undefined_var_symbols(); }
 
 /// Getter for the variable symbols map.
 ///
@@ -1070,65 +1068,6 @@  const string_elf_symbols_map_type&
 corpus::get_undefined_var_symbol_map() const
 {return *get_undefined_var_symbol_map_sptr();}
 
-/// Getter for the sorted vector of variable symbols for this corpus.
-///
-/// Note that the first time this function is called, it computes the
-/// sorted vector, caches the result and returns it.  Subsequent
-/// invocations of this function just return the cached vector.
-///
-/// @return the sorted vector of variable symbols for this corpus.
-const elf_symbols&
-corpus::get_sorted_var_symbols() const
-{
-  if (priv_->sorted_var_symbols.empty()
-      && !get_var_symbol_map().empty())
-    {
-      priv_->sorted_var_symbols.reserve(get_var_symbol_map().size());
-      for (string_elf_symbols_map_type::const_iterator i =
-	     get_var_symbol_map().begin();
-	   i != get_var_symbol_map().end();
-	   ++i)
-	for (elf_symbols::const_iterator s = i->second.begin();
-	     s != i->second.end(); ++s)
-	  priv_->sorted_var_symbols.push_back(*s);
-
-      elf_symbol_comp_functor comp;
-      std::sort(priv_->sorted_var_symbols.begin(),
-		priv_->sorted_var_symbols.end(),
-		comp);
-    }
-  return priv_->sorted_var_symbols;
-}
-
-/// Getter for a sorted vector of the variable symbols undefined in
-/// this corpus.
-///
-/// @return a vector of the variable symbols undefined in this corpus,
-/// sorted by name and then version.
-const elf_symbols&
-corpus::get_sorted_undefined_var_symbols() const
-{
-  if (priv_->sorted_undefined_var_symbols.empty()
-      && !get_undefined_var_symbol_map().empty())
-    {
-      priv_->sorted_undefined_var_symbols.reserve
-	(get_undefined_var_symbol_map().size());
-      for (string_elf_symbols_map_type::const_iterator i =
-	     get_undefined_var_symbol_map().begin();
-	   i != get_undefined_var_symbol_map().end();
-	   ++i)
-	for (elf_symbols::const_iterator s = i->second.begin();
-	     s != i->second.end(); ++s)
-	  priv_->sorted_undefined_var_symbols.push_back(*s);
-
-      elf_symbol_comp_functor comp;
-      std::sort(priv_->sorted_undefined_var_symbols.begin(),
-		priv_->sorted_undefined_var_symbols.end(),
-		comp);
-    }
-  return priv_->sorted_undefined_var_symbols;
-}
-
 /// Look in the function symbols map for a symbol with a given name.
 ///
 /// @param n the name of the symbol to look for.
diff --git a/src/abg-writer.cc b/src/abg-writer.cc
index d55bbbc166d5..d4d5e5531894 100644
--- a/src/abg-writer.cc
+++ b/src/abg-writer.cc
@@ -1678,26 +1678,43 @@  write_elf_symbol_visibility(elf_symbol::visibility v, ostream& o)
 ///
 /// @return true upon successful completion.
 static bool
-write_elf_symbol_aliases(const elf_symbol& sym, ostream& o)
+write_elf_symbol_aliases(const elf_symbol& sym, ostream& out)
 {
   if (!sym.is_main_symbol() || !sym.has_aliases())
     return false;
 
-  bool emitted = false;
-  o << " alias='";
-  for (elf_symbol_sptr s = sym.get_next_alias();
-       !s->is_main_symbol();
+
+  std::vector<std::string> aliases;
+  for (elf_symbol_sptr s = sym.get_next_alias(); s && !s->is_main_symbol();
        s = s->get_next_alias())
     {
-      if (s->get_next_alias()->is_main_symbol())
-	o << s->get_id_string() << "'";
-      else
-	o << s->get_id_string() << ",";
+      if (!s->is_public())
+	continue;
+
+      if (s->is_suppressed())
+	continue;
+
+      if (sym.is_in_ksymtab() != s->is_in_ksymtab())
+	continue;
 
-      emitted = true;
+      aliases.push_back(s->get_id_string());
     }
 
-  return emitted;
+  if (!aliases.empty())
+    {
+      out << " alias='";
+      std::string separator;
+      for (const auto& alias : aliases)
+	{
+	  out << separator << alias;
+	  separator = ",";
+	}
+
+      out << "'";
+      return true;
+    }
+
+  return false;
 }
 
 /// Write an XML attribute for the reference to a symbol for the
diff --git a/tests/data/test-read-dwarf/PR25007-sdhci.ko.abi b/tests/data/test-read-dwarf/PR25007-sdhci.ko.abi
index 44e2e608260b..1167e5aa478a 100644
--- a/tests/data/test-read-dwarf/PR25007-sdhci.ko.abi
+++ b/tests/data/test-read-dwarf/PR25007-sdhci.ko.abi
@@ -2,8 +2,6 @@ 
   <elf-function-symbols>
     <elf-symbol name='__sdhci_add_host' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='__sdhci_read_caps' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
-    <elf-symbol name='cleanup_module' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
-    <elf-symbol name='init_module' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='sdhci_add_host' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='sdhci_adma_write_desc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='sdhci_alloc_host' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
@@ -40,7 +38,6 @@ 
     <elf-symbol name='sdhci_suspend_host' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
   </elf-function-symbols>
   <elf-variable-symbols>
-    <elf-symbol name='__this_module' size='896' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
   </elf-variable-symbols>
   <abi-instr version='1.0' address-size='64' path='/ws/android/kernel/aosp/common-mainline/common/drivers/mmc/host/sdhci.c' comp-dir-path='/ws/android/kernel/aosp/common-mainline/out/android-mainline/common' language='LANG_C89'>
     <type-decl name='__ARRAY_SIZE_TYPE__' size-in-bits='64' id='type-id-1'/>