[RFA] Handle stack underflow in dbxread.c

[RFA] Handle stack underflow in dbxread.c

[Klee Dienes]

[-- Attachment #1: Type: text/plain, Size: 369 bytes --]

2002-11-15  Klee Dienes  <kdienes@apple.com>

         * buildsym.h (pop_context): Check for stack underflow.
         * dbxread.c (process_one_symbol): Complain and stop processing
         that symbol if we are already at the top of the context stack 
for
         a function-end N_FUN (this would imply an umatched RBRAC).  
Ditto
         when processing N_RBRAC.


[-- Attachment #2: pop-context.txt --]
[-- Type: text/plain, Size: 2654 bytes --]

2002-11-15  Klee Dienes  <kdienes@apple.com>
        
        * buildsym.h (pop_context): Check for stack underflow.
        * dbxread.c (process_one_symbol): Complain and stop processing
        that symbol if we are already at the top of the context stack for
        a function-end N_FUN (this would imply an umatched RBRAC).  Ditto
        when processing N_RBRAC.

Index: dbxread.c
===================================================================
RCS file: /cvs/Darwin/src/live/cygnus/src/gdb/dbxread.c,v
retrieving revision 1.38
diff -u -r1.38 dbxread.c
--- dbxread.c	2002/10/26 09:20:14	1.38
+++ dbxread.c	2002/11/15 08:56:07
@@ -2946,6 +2946,11 @@
 	      complain (&fun_end_outside_fun_complaint);
 	      break;
 	    }
+	  if (context_stack_depth <= 0)
+	    {
+	      complain (&lbrac_mismatch_complaint, symnum);
+	      break;
+	    }
 	  
 	  saw_fun_start = 0;
 	  record_line (current_subfile, 0, last_function_start + valu);
@@ -3027,6 +3032,11 @@
 	valu += last_source_start_addr;
 #endif
 
+      if (context_stack_depth <= 0)
+	{
+	  complain (&lbrac_mismatch_complaint, symnum);
+	  break;
+	}
       new = pop_context ();
       if (desc != new->depth)
 	complain (&lbrac_mismatch_complaint, symnum);
Index: buildsym.h
===================================================================
RCS file: /cvs/Darwin/src/live/cygnus/src/gdb/buildsym.h,v
retrieving revision 1.5
diff -u -r1.5 buildsym.h
--- buildsym.h	2002/11/10 05:33:33	1.5
+++ buildsym.h	2002/11/15 08:56:07
@@ -22,6 +22,8 @@
 #if !defined (BUILDSYM_H)
 #define BUILDSYM_H 1
 
+#include "gdb_assert.h"
+
 /* This module provides definitions used for creating and adding to
    the symbol table.  These routines are called from various symbol-
    file-reading routines.
@@ -173,12 +175,6 @@
 
 EXTERN int context_stack_size;
 
-/* Macro "function" for popping contexts from the stack.  Pushing is
-   done by a real function, push_context.  This returns a pointer to a
-   struct context_stack.  */
-
-#define	pop_context() (&context_stack[--context_stack_depth]);
-
 /* Non-zero if the context stack is empty.  */
 #define outermost_context_p() (context_stack_depth == 0)
 
@@ -271,6 +267,13 @@
 extern void buildsym_init (void);
 
 extern struct context_stack *push_context (int desc, CORE_ADDR valu);
+
+/* Macro "function" for popping contexts from the stack.  Pushing is
+   done by a real function, push_context.  This returns a pointer to a
+   struct context_stack.  */
+
+#define pop_context() \
+  (gdb_assert (context_stack_depth > 0), &context_stack[--context_stack_depth]);
 
 extern void record_line (struct subfile *subfile, int line, CORE_ADDR pc);
 

Re: [RFA] Handle stack underflow in dbxread.c

[Jim Blandy]

The context in your first patch hunk doesn't appear anywhere in my
sources.  Is this diff against the current sources?

Klee Dienes <klee@apple.com> writes:

> 2002-11-15  Klee Dienes  <kdienes@apple.com>
> 
>          * buildsym.h (pop_context): Check for stack underflow.
>          * dbxread.c (process_one_symbol): Complain and stop processing
>          that symbol if we are already at the top of the context stack
> for
>          a function-end N_FUN (this would imply an umatched RBRAC).
> Ditto
>          when processing N_RBRAC.
> 
> 
> 2002-11-15  Klee Dienes  <kdienes@apple.com>
>         
>         * buildsym.h (pop_context): Check for stack underflow.
>         * dbxread.c (process_one_symbol): Complain and stop processing
>         that symbol if we are already at the top of the context stack for
>         a function-end N_FUN (this would imply an umatched RBRAC).  Ditto
>         when processing N_RBRAC.
> 
> Index: dbxread.c
> ===================================================================
> RCS file: /cvs/Darwin/src/live/cygnus/src/gdb/dbxread.c,v
> retrieving revision 1.38
> diff -u -r1.38 dbxread.c
> --- dbxread.c	2002/10/26 09:20:14	1.38
> +++ dbxread.c	2002/11/15 08:56:07
> @@ -2946,6 +2946,11 @@
>  	      complain (&fun_end_outside_fun_complaint);
>  	      break;
>  	    }
> +	  if (context_stack_depth <= 0)
> +	    {
> +	      complain (&lbrac_mismatch_complaint, symnum);
> +	      break;
> +	    }
>  	  
>  	  saw_fun_start = 0;
>  	  record_line (current_subfile, 0, last_function_start + valu);
> @@ -3027,6 +3032,11 @@
>  	valu += last_source_start_addr;
>  #endif
>  
> +      if (context_stack_depth <= 0)
> +	{
> +	  complain (&lbrac_mismatch_complaint, symnum);
> +	  break;
> +	}
>        new = pop_context ();
>        if (desc != new->depth)
>  	complain (&lbrac_mismatch_complaint, symnum);
> Index: buildsym.h
> ===================================================================
> RCS file: /cvs/Darwin/src/live/cygnus/src/gdb/buildsym.h,v
> retrieving revision 1.5
> diff -u -r1.5 buildsym.h
> --- buildsym.h	2002/11/10 05:33:33	1.5
> +++ buildsym.h	2002/11/15 08:56:07
> @@ -22,6 +22,8 @@
>  #if !defined (BUILDSYM_H)
>  #define BUILDSYM_H 1
>  
> +#include "gdb_assert.h"
> +
>  /* This module provides definitions used for creating and adding to
>     the symbol table.  These routines are called from various symbol-
>     file-reading routines.
> @@ -173,12 +175,6 @@
>  
>  EXTERN int context_stack_size;
>  
> -/* Macro "function" for popping contexts from the stack.  Pushing is
> -   done by a real function, push_context.  This returns a pointer to a
> -   struct context_stack.  */
> -
> -#define	pop_context() (&context_stack[--context_stack_depth]);
> -
>  /* Non-zero if the context stack is empty.  */
>  #define outermost_context_p() (context_stack_depth == 0)
>  
> @@ -271,6 +267,13 @@
>  extern void buildsym_init (void);
>  
>  extern struct context_stack *push_context (int desc, CORE_ADDR valu);
> +
> +/* Macro "function" for popping contexts from the stack.  Pushing is
> +   done by a real function, push_context.  This returns a pointer to a
> +   struct context_stack.  */
> +
> +#define pop_context() \
> +  (gdb_assert (context_stack_depth > 0), &context_stack[--context_stack_depth]);
>  
>  extern void record_line (struct subfile *subfile, int line, CORE_ADDR pc);
>  

Re: [RFA] Handle stack underflow in dbxread.c

[Klee Dienes]

[-- Attachment #1: Type: text/plain, Size: 9 bytes --]

My bad:


[-- Attachment #2: check-context.txt --]
[-- Type: text/plain, Size: 2396 bytes --]

Index: buildsym.h
===================================================================
RCS file: /cvs/src/src/gdb/buildsym.h,v
retrieving revision 1.6
diff -u -r1.6 buildsym.h
--- buildsym.h	8 Nov 2002 20:05:10 -0000	1.6
+++ buildsym.h	15 Nov 2002 20:47:34 -0000
@@ -22,6 +22,8 @@
 #if !defined (BUILDSYM_H)
 #define BUILDSYM_H 1
 
+#include "gdb_assert.h"
+
 /* This module provides definitions used for creating and adding to
    the symbol table.  These routines are called from various symbol-
    file-reading routines.
@@ -173,12 +175,6 @@
 
 EXTERN int context_stack_size;
 
-/* Macro "function" for popping contexts from the stack.  Pushing is
-   done by a real function, push_context.  This returns a pointer to a
-   struct context_stack.  */
-
-#define	pop_context() (&context_stack[--context_stack_depth]);
-
 /* Non-zero if the context stack is empty.  */
 #define outermost_context_p() (context_stack_depth == 0)
 
@@ -271,6 +267,13 @@
 extern void buildsym_init (void);
 
 extern struct context_stack *push_context (int desc, CORE_ADDR valu);
+
+/* Macro "function" for popping contexts from the stack.  Pushing is
+   done by a real function, push_context.  This returns a pointer to a
+   struct context_stack.  */
+
+#define pop_context() \
+  (gdb_assert (context_stack_depth > 0), &context_stack[--context_stack_depth]);
 
 extern void record_line (struct subfile *subfile, int line, CORE_ADDR pc);
 
Index: dbxread.c
===================================================================
RCS file: /cvs/src/src/gdb/dbxread.c,v
retrieving revision 1.36
diff -u -r1.36 dbxread.c
--- dbxread.c	25 Oct 2002 22:25:55 -0000	1.36
+++ dbxread.c	15 Nov 2002 20:47:42 -0000
@@ -2774,6 +2774,13 @@
 	{
 	  /* This N_FUN marks the end of a function.  This closes off the
 	     current block.  */
+
+ 	  if (context_stack_depth <= 0)
+ 	    {
+ 	      complain (&lbrac_mismatch_complaint, symnum);
+ 	      break;
+ 	    }
+
 	  record_line (current_subfile, 0, function_start_offset + valu);
 	  within_function = 0;
 	  new = pop_context ();
@@ -2842,6 +2849,12 @@
 	/* On most machines, the block addresses are relative to the
 	   N_SO, the linker did not relocate them (sigh).  */
 	valu += last_source_start_addr;
+
+      if (context_stack_depth <= 0)
+	{
+	  complain (&lbrac_mismatch_complaint, symnum);
+	  break;
+	}
 
       new = pop_context ();
       if (desc != new->depth)

[-- Attachment #3: Type: text/plain, Size: 189 bytes --]



On Friday, November 15, 2002, at 03:02 PM, Jim Blandy wrote:

>
> The context in your first patch hunk doesn't appear anywhere in my
> sources.  Is this diff against the current sources?

Re: [RFA] Handle stack underflow in dbxread.c

[Jim Blandy]

Okay, that looks like a good change to me --- please commit it.

But rather than making pop_context a hairy macro, could you make it a
function in buildsym.c, right after push_context?

Klee Dienes <klee@apple.com> writes:
> My bad:
> 
> 
> 
> 
> On Friday, November 15, 2002, at 03:02 PM, Jim Blandy wrote:
> 
> >
> > The context in your first patch hunk doesn't appear anywhere in my
> > sources.  Is this diff against the current sources?

Re: [RFA] Handle stack underflow in dbxread.c

[Klee Dienes]

[-- Attachment #1: Type: text/plain, Size: 293 bytes --]

On Monday, November 18, 2002, at 12:40 PM, Jim Blandy wrote:

>
> Okay, that looks like a good change to me --- please commit it.
>
> But rather than making pop_context a hairy macro, could you make it a
> function in buildsym.c, right after push_context?


Gladly.  Committed the following:


[-- Attachment #2: check-context.txt --]
[-- Type: text/plain, Size: 3273 bytes --]

2002-11-18  Klee Dienes  <kdienes@apple.com>

        * buildsym.h (pop_context): Convert to function, defined in
        buildsym.c.
        * buildsym.c: Include gdb_assert.h.
        (pop_context): Implement as C function.  Add check for stack
        underflow.
        * dbxread.c (process_one_symbol): Complain and stop processing
        that symbol if we are already at the top of the context stack for
        a function-end N_FUN (this would imply an umatched RBRAC).  Ditto
        when processing N_RBRAC.

Index: buildsym.c
===================================================================
RCS file: /cvs/src/src/gdb/buildsym.c,v
retrieving revision 1.21
diff -u -r1.21 buildsym.c
--- buildsym.c	25 Oct 2002 22:25:55 -0000	1.21
+++ buildsym.c	18 Nov 2002 19:13:17 -0000
@@ -33,6 +33,7 @@
 #include "symfile.h"
 #include "objfiles.h"
 #include "gdbtypes.h"
+#include "gdb_assert.h"
 #include "complaints.h"
 #include "gdb_string.h"
 #include "expression.h"		/* For "enum exp_opcode" used by... */
@@ -1100,6 +1101,14 @@
 
   return new;
 }
+
+struct context_stack *
+pop_context (void)
+{
+  gdb_assert (context_stack_depth > 0);
+  return (&context_stack[--context_stack_depth]);
+}
+
 \f
 
 /* Compute a small integer hash code for the given name. */
Index: buildsym.h
===================================================================
RCS file: /cvs/src/src/gdb/buildsym.h,v
retrieving revision 1.6
diff -u -r1.6 buildsym.h
--- buildsym.h	8 Nov 2002 20:05:10 -0000	1.6
+++ buildsym.h	18 Nov 2002 19:13:17 -0000
@@ -173,12 +173,6 @@
 
 EXTERN int context_stack_size;
 
-/* Macro "function" for popping contexts from the stack.  Pushing is
-   done by a real function, push_context.  This returns a pointer to a
-   struct context_stack.  */
-
-#define	pop_context() (&context_stack[--context_stack_depth]);
-
 /* Non-zero if the context stack is empty.  */
 #define outermost_context_p() (context_stack_depth == 0)
 
@@ -271,6 +265,8 @@
 extern void buildsym_init (void);
 
 extern struct context_stack *push_context (int desc, CORE_ADDR valu);
+
+extern struct context_stack *pop_context (void);
 
 extern void record_line (struct subfile *subfile, int line, CORE_ADDR pc);
 
Index: dbxread.c
===================================================================
RCS file: /cvs/src/src/gdb/dbxread.c,v
retrieving revision 1.36
diff -u -r1.36 dbxread.c
--- dbxread.c	25 Oct 2002 22:25:55 -0000	1.36
+++ dbxread.c	18 Nov 2002 19:13:18 -0000
@@ -2774,6 +2774,13 @@
 	{
 	  /* This N_FUN marks the end of a function.  This closes off the
 	     current block.  */
+
+ 	  if (context_stack_depth <= 0)
+ 	    {
+ 	      complain (&lbrac_mismatch_complaint, symnum);
+ 	      break;
+ 	    }
+
 	  record_line (current_subfile, 0, function_start_offset + valu);
 	  within_function = 0;
 	  new = pop_context ();
@@ -2842,6 +2849,12 @@
 	/* On most machines, the block addresses are relative to the
 	   N_SO, the linker did not relocate them (sigh).  */
 	valu += last_source_start_addr;
+
+      if (context_stack_depth <= 0)
+	{
+	  complain (&lbrac_mismatch_complaint, symnum);
+	  break;
+	}
 
       new = pop_context ();
       if (desc != new->depth)

Re: [RFA] Handle stack underflow in dbxread.c

[Andrew Cagney]

> On Monday, November 18, 2002, at 12:40 PM, Jim Blandy wrote:
> 
> 
> Okay, that looks like a good change to me --- please commit it.
> 
> But rather than making pop_context a hairy macro, could you make it a
> function in buildsym.c, right after push_context?
> 
> 
> Gladly.  Committed the following:
> 
> 
> 
> 2002-11-18  Klee Dienes  <kdienes@apple.com>
> 
>         * buildsym.h (pop_context): Convert to function, defined in
>         buildsym.c.
>         * buildsym.c: Include gdb_assert.h.
>         (pop_context): Implement as C function.  Add check for stack
>         underflow.
>         * dbxread.c (process_one_symbol): Complain and stop processing
>         that symbol if we are already at the top of the context stack for
>         a function-end N_FUN (this would imply an umatched RBRAC).  Ditto
>         when processing N_RBRAC.

BTW, can you please put some sort of equivalent comment back?

> -/* Macro "function" for popping contexts from the stack.  Pushing is
> -   done by a real function, push_context.  This returns a pointer to a
> -   struct context_stack.  */
> -
> -#define	pop_context() (&context_stack[--context_stack_depth]);

Andrew

Re: [RFA] Handle stack underflow in dbxread.c

[Klee Dienes]

I added the following comment:

/* Pop a context block.  Returns the address of the context block just
    popped. */

On Monday, November 18, 2002, at 03:42 PM, Andrew Cagney wrote:
>
> BTW, can you please put some sort of equivalent comment back?
>
>> -/* Macro "function" for popping contexts from the stack.  Pushing is
>> -   done by a real function, push_context.  This returns a pointer to 
>> a
>> -   struct context_stack.  */
>> -
>> -#define	pop_context() (&context_stack[--context_stack_depth]);
>
> Andrew
>
>

Re: [RFA] Handle stack underflow in dbxread.c

[Andrew Cagney]

Thanks!  Don't forget to post the ChangeLog and patch.

Andrew
> I added the following comment:
> 
> /* Pop a context block.  Returns the address of the context block just
>    popped. */
> 
> On Monday, November 18, 2002, at 03:42 PM, Andrew Cagney wrote:
> 
> BTW, can you please put some sort of equivalent comment back?
> 
> -/* Macro "function" for popping contexts from the stack.  Pushing is
> -   done by a real function, push_context.  This returns a pointer to a
> -   struct context_stack.  */
> -
> -#define    pop_context() (&context_stack[--context_stack_depth]);
> 
> Andrew
> 
> 
> 
> 

Re: [RFA] Handle stack underflow in dbxread.c

[Klee Dienes]

[-- Attachment #1: Type: text/plain, Size: 155 bytes --]

> Thanks!  Don't forget to post the ChangeLog and patch.

2002-11-18  Klee Dienes  <kdienes@apple.com>

         * buildsym.c (pop_context): Add comment.


[-- Attachment #2: pop-context.txt --]
[-- Type: text/plain, Size: 470 bytes --]

Index: buildsym.c
===================================================================
RCS file: /cvs/src/src/gdb/buildsym.c,v
retrieving revision 1.22
retrieving revision 1.23
diff -u -r1.22 -r1.23
--- buildsym.c	18 Nov 2002 19:14:10 -0000	1.22
+++ buildsym.c	18 Nov 2002 20:57:55 -0000	1.23
@@ -1102,6 +1102,9 @@
   return new;
 }
 
+/* Pop a context block.  Returns the address of the context block just
+   popped. */
+
 struct context_stack *
 pop_context (void)
 {

Re: [RFA] Handle stack underflow in dbxread.c

[Jim Blandy]

Klee Dienes <klee@apple.com> writes:
> On Monday, November 18, 2002, at 12:40 PM, Jim Blandy wrote:
> 
> >
> > Okay, that looks like a good change to me --- please commit it.
> >
> > But rather than making pop_context a hairy macro, could you make it a
> > function in buildsym.c, right after push_context?
> 
> 
> Gladly.  Committed the following:

Great!

By the way, buildsym.o now depends on gdb_assert.h.  Could you add the
dependency to Makefile.in?

Re: [RFA] Handle stack underflow in dbxread.c

[Klee Dienes]

[-- Attachment #1: Type: text/plain, Size: 141 bytes --]

Thanks for the catch.

2002-11-18   Klee Dienes  <kdienes@apple.com>

         * Makefile.in (buildsym.o): Add dependency for gdb_assert.h.


[-- Attachment #2: make.txt --]
[-- Type: text/plain, Size: 994 bytes --]

Index: Makefile.in
===================================================================
RCS file: /cvs/src/src/gdb/Makefile.in,v
retrieving revision 1.279
diff -u -r1.279 Makefile.in
--- Makefile.in	9 Nov 2002 03:14:11 -0000	1.279
+++ Makefile.in	18 Nov 2002 21:46:15 -0000
@@ -1548,9 +1548,9 @@
 	$(objfiles_h) $(linespec_h) $(completer_h) $(gdb_h) $(ui_out_h) \
 	$(cli_script_h) $(gdb_events_h) $(source_h)
 buildsym.o: buildsym.c $(defs_h) $(bfd_h) $(gdb_obstack_h) $(symtab_h) \
-	$(symfile_h) $(objfiles_h) $(gdbtypes_h) $(complaints_h) \
-	$(gdb_string_h) $(expression_h) $(language_h) $(bcache_h) \
-	$(filenames_h) $(macrotab_h) $(demangle_h) $(buildsym_h) \
+	$(symfile_h) $(objfiles_h) $(gdbtypes_h) $(gdb_assert_h) \
+	$(complaints_h)	$(gdb_string_h) $(expression_h) $(language_h) \
+	$(bcache_h) $(filenames_h) $(macrotab_h) $(demangle_h) $(buildsym_h) \
 	$(stabsread_h)
 builtin-regs.o: builtin-regs.c $(defs_h) $(builtin_regs_h) $(gdbtypes_h) \
 	$(gdb_string_h) $(gdb_assert_h)

[-- Attachment #3: Type: text/plain, Size: 167 bytes --]




On Monday, November 18, 2002, at 04:02 PM, Jim Blandy wrote:

> By the way, buildsym.o now depends on gdb_assert.h.  Could you add the
> dependency to Makefile.in?