[PATCH 0/3] semind: Index more symbols

[PATCH 0/3] semind: Index more symbols

[Alexey Gladkov]

Greetings!

For indexing purposes, macros definitions and typedefs are added to the
semind database. Functions that are not used in the code are also indexed.

Alexey Gladkov (3):
  dissect: Allow to show all symbols
  dissect: Show macro definitions
  dissect: Show typedefs

 dissect.c      | 61 +++++++++++++++++++++++++++++++++++++++++++++++++-
 options.c      |  5 +++++
 options.h      |  2 ++
 semind.c       |  1 +
 test-dissect.c |  5 ++++-
 5 files changed, 72 insertions(+), 2 deletions(-)

-- 
2.33.0

[PATCH 1/3] dissect: Allow to show all symbols

[Alexey Gladkov]

Currently dissect sees only used symbols. For indexing purposes, it is
useful to see all declared symbols.

$ nl -s\  -w2 ./z.c
 1 struct foo {
 2         int member;
 3 };
 4 #ifdef OPT
 5 static void func1(void) {
 6         struct foo *x;
 7         return 0;
 8 }
 9 #endif
10 static inline void func2(void) { return; }
11 void func(void) { return; }

$ ./test-dissect ./z.c

FILE: ./z.c

  11:6                    def   f func                             void ( ... )

$ ./test-dissect --param=dissect-show-all-symbols ./z.c

FILE: ./z.c

   1:8                    def   s foo                              struct foo
   2:13                   def   m foo.member                       int
  10:20                   def   f func2                            void ( ... )
  11:6                    def   f func                             void ( ... )

Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
---
 dissect.c | 39 ++++++++++++++++++++++++++++++++++++++-
 options.c |  5 +++++
 options.h |  2 ++
 semind.c  |  1 +
 4 files changed, 46 insertions(+), 1 deletion(-)

diff --git a/dissect.c b/dissect.c
index 582e8fc3..0d6c3288 100644
--- a/dissect.c
+++ b/dissect.c
@@ -652,9 +652,46 @@ static void do_sym_list(struct symbol_list *list)
 	DO_LIST(list, sym, do_symbol(sym));
 }
 
+static inline bool valid_namespace(enum namespace ns)
+{
+	return (ns == NS_STRUCT || ns == NS_SYMBOL);
+}
+
+static void do_file(char *file)
+{
+	struct symbol_list *res = sparse_keep_tokens(file);
+
+	if (!dissect_show_all_symbols) {
+		do_sym_list(res);
+		goto end;
+	}
+
+	DO_LIST(file_scope->symbols, sym,
+		if (input_streams[sym->pos.stream].fd != -1 && valid_namespace(sym->namespace)) {
+			if (sym->type == SYM_STRUCT || sym->type == SYM_UNION) {
+				sym->ctype.base_type = sym;
+				examine_sym_node(sym, NULL);
+				continue;
+			}
+
+			do_symbol(sym);
+		}
+	);
+
+	DO_LIST(global_scope->symbols, sym,
+		if (input_streams[sym->pos.stream].fd != -1 && valid_namespace(sym->namespace)) {
+			do_symbol(sym);
+		}
+	);
+
+end:
+	/* Drop the tokens for this file after parsing */
+	clear_token_alloc();
+}
+
 void dissect(struct reporter *rep, struct string_list *filelist)
 {
 	reporter = rep;
 
-	DO_LIST(filelist, file, do_sym_list(__sparse(file)));
+	DO_LIST(filelist, file, do_file(file));
 }
diff --git a/options.c b/options.c
index 6704fc8d..b4684357 100644
--- a/options.c
+++ b/options.c
@@ -70,6 +70,8 @@ int dbg_postorder = 0;
 int dump_macro_defs = 0;
 int dump_macros_only = 0;
 
+int dissect_show_all_symbols = 0;
+
 unsigned long fdump_ir;
 int fhosted = 1;
 unsigned int fmax_errors = 100;
@@ -958,6 +960,9 @@ static char **handle_param(char *arg, char **next)
 	if (!value)
 		die("missing argument for --param option");
 
+	if (!strcmp(value, "dissect-show-all-symbols"))
+		dissect_show_all_symbols = 1;
+
 	return next;
 }
 
diff --git a/options.h b/options.h
index 0aec8764..c2a9551a 100644
--- a/options.h
+++ b/options.h
@@ -70,6 +70,8 @@ extern int dbg_postorder;
 extern int dump_macro_defs;
 extern int dump_macros_only;
 
+extern int dissect_show_all_symbols;
+
 extern unsigned long fdump_ir;
 extern int fhosted;
 extern unsigned int fmax_errors;
diff --git a/semind.c b/semind.c
index 911fc747..ad8003ba 100644
--- a/semind.c
+++ b/semind.c
@@ -329,6 +329,7 @@ done:
 	optind--;
 
 	sparse_initialize(argc - optind, argv + optind, &semind_filelist);
+	dissect_show_all_symbols = 1;
 }
 
 static void parse_cmdline_rm(int argc, char **argv)
-- 
2.33.0

[PATCH 2/3] dissect: Show macro definitions

[Alexey Gladkov]

Add the ability to dissect to see macro definitions. The patch does not
add full support for the usage of macros, but only their definitions.

Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
---
 dissect.c      | 13 ++++++++++++-
 test-dissect.c |  3 ++-
 2 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/dissect.c b/dissect.c
index 0d6c3288..7d5d92c9 100644
--- a/dissect.c
+++ b/dissect.c
@@ -610,6 +610,11 @@ static struct symbol *do_initializer(struct symbol *type, struct expression *exp
 	return type;
 }
 
+static inline bool is_macro(struct symbol *sym)
+{
+	return (sym->namespace == NS_MACRO || sym->namespace == NS_UNDEF);
+}
+
 static inline struct symbol *do_symbol(struct symbol *sym)
 {
 	struct symbol *type = base_type(sym);
@@ -654,7 +659,7 @@ static void do_sym_list(struct symbol_list *list)
 
 static inline bool valid_namespace(enum namespace ns)
 {
-	return (ns == NS_STRUCT || ns == NS_SYMBOL);
+	return (ns == NS_MACRO || ns == NS_UNDEF || ns == NS_STRUCT || ns == NS_SYMBOL);
 }
 
 static void do_file(char *file)
@@ -668,6 +673,12 @@ static void do_file(char *file)
 
 	DO_LIST(file_scope->symbols, sym,
 		if (input_streams[sym->pos.stream].fd != -1 && valid_namespace(sym->namespace)) {
+			if (is_macro(sym)) {
+				sym->kind = 'd';
+				reporter->r_symdef(sym);
+				continue;
+			}
+
 			if (sym->type == SYM_STRUCT || sym->type == SYM_UNION) {
 				sym->ctype.base_type = sym;
 				examine_sym_node(sym, NULL);
diff --git a/test-dissect.c b/test-dissect.c
index 58b3e633..3d870a97 100644
--- a/test-dissect.c
+++ b/test-dissect.c
@@ -57,11 +57,12 @@ static void r_symbol(unsigned mode, struct position *pos, struct symbol *sym)
 		show_typename(sym->ctype.base_type));
 
 	switch (sym->kind) {
+	case 'd':
+		break;
 	case 's':
 		if (sym->type == SYM_STRUCT || sym->type == SYM_UNION)
 			break;
 		goto err;
-
 	case 'f':
 		if (sym->type != SYM_BAD && sym->ctype.base_type->type != SYM_FN)
 			goto err;
-- 
2.33.0

[PATCH 3/3] dissect: Show typedefs

[Alexey Gladkov]

For indexing purposes, it is useful to see type definitions.

$ semind search __kernel_ulong_t
(def) include/uapi/asm-generic/posix_types.h 16 23 typedef unsigned long   __kernel_ulong_t;

Signed-off-by: Alexey Gladkov <gladkov.alexey@gmail.com>
---
 dissect.c      | 13 ++++++++++++-
 test-dissect.c |  2 ++
 2 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/dissect.c b/dissect.c
index 7d5d92c9..300d5ca9 100644
--- a/dissect.c
+++ b/dissect.c
@@ -615,6 +615,11 @@ static inline bool is_macro(struct symbol *sym)
 	return (sym->namespace == NS_MACRO || sym->namespace == NS_UNDEF);
 }
 
+static inline bool is_typedef(struct symbol *sym)
+{
+	return (sym->namespace == NS_TYPEDEF);
+}
+
 static inline struct symbol *do_symbol(struct symbol *sym)
 {
 	struct symbol *type = base_type(sym);
@@ -659,7 +664,7 @@ static void do_sym_list(struct symbol_list *list)
 
 static inline bool valid_namespace(enum namespace ns)
 {
-	return (ns == NS_MACRO || ns == NS_UNDEF || ns == NS_STRUCT || ns == NS_SYMBOL);
+	return (ns == NS_TYPEDEF || ns == NS_MACRO || ns == NS_UNDEF || ns == NS_STRUCT || ns == NS_SYMBOL);
 }
 
 static void do_file(char *file)
@@ -673,6 +678,12 @@ static void do_file(char *file)
 
 	DO_LIST(file_scope->symbols, sym,
 		if (input_streams[sym->pos.stream].fd != -1 && valid_namespace(sym->namespace)) {
+			if (is_typedef(sym)) {
+				sym->kind = 't';
+				reporter->r_symdef(sym);
+				continue;
+			}
+
 			if (is_macro(sym)) {
 				sym->kind = 'd';
 				reporter->r_symdef(sym);
diff --git a/test-dissect.c b/test-dissect.c
index 3d870a97..65b205f8 100644
--- a/test-dissect.c
+++ b/test-dissect.c
@@ -63,6 +63,8 @@ static void r_symbol(unsigned mode, struct position *pos, struct symbol *sym)
 		if (sym->type == SYM_STRUCT || sym->type == SYM_UNION)
 			break;
 		goto err;
+	case 't':
+		break;
 	case 'f':
 		if (sym->type != SYM_BAD && sym->ctype.base_type->type != SYM_FN)
 			goto err;
-- 
2.33.0

Re: [PATCH 0/3] semind: Index more symbols

[Alexey Gladkov]

On Tue, Nov 02, 2021 at 03:06:42PM +0100, Alexey Gladkov wrote:
> Greetings!
> 
> For indexing purposes, macros definitions and typedefs are added to the
> semind database. Functions that are not used in the code are also indexed.

careful ping.

> Alexey Gladkov (3):
>   dissect: Allow to show all symbols
>   dissect: Show macro definitions
>   dissect: Show typedefs
> 
>  dissect.c      | 61 +++++++++++++++++++++++++++++++++++++++++++++++++-
>  options.c      |  5 +++++
>  options.h      |  2 ++
>  semind.c       |  1 +
>  test-dissect.c |  5 ++++-
>  5 files changed, 72 insertions(+), 2 deletions(-)
> 
> -- 
> 2.33.0
> 

-- 
Rgrds, legion

Re: [PATCH 0/3] semind: Index more symbols

[Luc Van Oostenryck]

On Tue, Nov 02, 2021 at 03:06:42PM +0100, Alexey Gladkov wrote:
> Greetings!
> 
> For indexing purposes, macros definitions and typedefs are added to the
> semind database. Functions that are not used in the code are also indexed.

Thank you very much, pushed now, an ... apologies for this huge delay.

-- Luc

Re: [PATCH 0/3] semind: Index more symbols

[Alexey Gladkov]

On Sat, May 21, 2022 at 11:46:22AM +0200, Luc Van Oostenryck wrote:
> On Tue, Nov 02, 2021 at 03:06:42PM +0100, Alexey Gladkov wrote:
> > Greetings!
> > 
> > For indexing purposes, macros definitions and typedefs are added to the
> > semind database. Functions that are not used in the code are also indexed.
> 
> Thank you very much, pushed now, an ... apologies for this huge delay.

Looks like I'm the only one using this indexer at work. But in case anyone
is interested, I have a vim plugin for semind [1].

[1] https://github.com/legionus/vim-semind

-- 
Rgrds, legion