Re: [RFA] valprint.c / *-valprint.c: Don't lose `embedded_offset'

[Pedro Alves <pedro at codesourcery dot com>]

On Thursday 07 October 2010 19:13:10, Tom Tromey wrote:
> >>>>> "Pedro" == Pedro Alves <pedro@codesourcery.com> writes:

> Pedro> Trouble is in languages other than C/C++ where the
> Pedro> advance-embedded_offset-don't-touch-valaddr-or-address contract
> Pedro> isn't compromised in many places.
> 
> I think the embedded_offset is very confusing.  Every time I have to fix
> a bug in it, I have to re-learn what it really means.  

I used to find it very confusing as well, but I think that with this
patch, it became much clearer.  In all the changes that I have been
doing on top, I never again got confused.  :-)

> Perhaps next time
> I will plan ahead and fix the commentary in value.c to be more clear (to
> me at least).

I was planning doing that too, along with extending the comment around
val_print.  :-)

> Pedro> All in all, I think it's just better to be consistent throughout.
> Pedro> I know I got mighty confused with some functions taking adjusted
> Pedro> `valaddr' and `address', while others taking `embedded_offset'
> Pedro> into account.  Maybe some day we will not allow passing around
> Pedro> a NULL `val', and thus we will be able to get rid of `valaddr'
> Pedro> and `address' parameters throughout, and instead get those from
> Pedro> the value directly.  I don't plan to actually do that, but
> Pedro> this seems like a step in that direction.
> 
> I was going to do this last time I had to add a parameter to the whole
> val_print hierarchy, but then blinked.
> 
> I think we should just get rid of val_print entirely, and only have
> value_print, passing around values.  If that is not efficient enough
> (too much copying or something), we can change struct value to make it
> efficient.
> 
> What do you think of that?

Should be possible.  Actually, I did go one step further, because it
occured to me that I might as well add an assertion to val_print that
valaddr is in fact always equal to value->contents.  See patch below
that applies on top of yesterday's.  I don't know why that didn't occur
to me sooner.  :-)  This passes regression testing as well.

So, the steps I guess would be:

 - apply yesterday's and this patch.

 - add an assertion to val_print forbidding a NULL struct value, and
   fix all callers to make sure to construct a value.  Not sure how
   many there are, might not be that many.  I now that "info reg" is
   one case.

 - get rid of valaddr and address from all the val_print methods,
   getting at the contents of the passed in value instead.  It's also
   nice to get rid of the `address' parameter, because not all values
   actually _have_ a notion of value.  Currently, passing around an
   address is an abstraction violation.

 - investigate whether passing an offset around is cool, or whether
   we need something like a new value type that provides a view into
   another value, and pass that around instead?

> I did not read the patch extremely closely.  What I did read looks
> reasonable.  Unfortunately the nature of a patch like this is that it is
> very hard to know whether some spot was missed :-(

Yeah...

-- 
Pedro Alves

2010-10-07  Pedro Alves  <pedro@codesourcery.com>

	* ada-valprint.c (val_print_packed_array_elements): Pass the
	correct struct value to val_print.
	(ada_val_print_1): Ditto.
	* valprint.c (val_print): Add assertions.
	* value.c (value_contents_for_printing_const): New.
	* value.h (value_contents_for_printing_const): Declare.

---
 gdb/ada-valprint.c |   11 ++++++-----
 gdb/valprint.c     |    5 +++++
 gdb/value.c        |    7 +++++++
 gdb/value.h        |    2 ++
 4 files changed, 20 insertions(+), 5 deletions(-)

Index: src/gdb/ada-valprint.c
===================================================================
--- src.orig/gdb/ada-valprint.c	2010-10-07 11:53:48.000000000 +0100
+++ src/gdb/ada-valprint.c	2010-10-07 11:58:08.000000000 +0100
@@ -211,7 +211,7 @@ val_print_packed_array_elements (struct 
 	  opts.deref_ref = 0;
 	  val_print (elttype, value_contents_for_printing (v0),
 		     value_embedded_offset (v0), 0, stream,
-		     recurse + 1, val, &opts, current_language);
+		     recurse + 1, v0, &opts, current_language);
 	  annotate_elt_rep (i - i0);
 	  fprintf_filtered (stream, _(" <repeats %u times>"), i - i0);
 	  annotate_elt_rep_end ();
@@ -242,7 +242,7 @@ val_print_packed_array_elements (struct 
 		}
 	      val_print (elttype, value_contents_for_printing (v0),
 			 value_embedded_offset (v0), 0, stream,
-			 recurse + 1, val, &opts, current_language);
+			 recurse + 1, v0, &opts, current_language);
 	      annotate_elt ();
 	    }
 	}
@@ -765,12 +765,13 @@ ada_val_print_1 (struct type *type, cons
 	      return ada_val_print_1 (target_type,
 				      value_contents_for_printing (v),
 				      value_embedded_offset (v), 0,
-				      stream, recurse + 1, NULL, options);
+				      stream, recurse + 1, v, options);
 	    }
 	  else
 	    return ada_val_print_1 (TYPE_TARGET_TYPE (type),
 				    valaddr, offset,
-				    address, stream, recurse, original_value, options);
+				    address, stream, recurse,
+				    original_value, options);
 	}
       else
 	{
@@ -909,7 +910,7 @@ ada_val_print_1 (struct type *type, cons
                          value_contents_for_printing (deref_val),
                          value_embedded_offset (deref_val),
                          value_address (deref_val), stream, recurse + 1,
-			 original_value, options, current_language);
+			 deref_val, options, current_language);
             }
           else
             fputs_filtered ("(null)", stream);
Index: src/gdb/valprint.c
===================================================================
--- src.orig/gdb/valprint.c	2010-10-07 11:53:48.000000000 +0100
+++ src/gdb/valprint.c	2010-10-07 20:04:35.000000000 +0100
@@ -305,6 +305,11 @@ val_print (struct type *type, const gdb_
   struct value_print_options local_opts = *options;
   struct type *real_type = check_typedef (type);
 
+  gdb_assert (valaddr != NULL);
+  gdb_assert (embedded_offset >= 0);
+  gdb_assert (val == NULL
+	      || (value_contents_for_printing_const (val) == valaddr));
+
   if (local_opts.pretty == Val_pretty_default)
     local_opts.pretty = (local_opts.prettyprint_structs
 			 ? Val_prettyprint : Val_no_prettyprint);
Index: src/gdb/value.c
===================================================================
--- src.orig/gdb/value.c	2010-10-07 11:53:48.000000000 +0100
+++ src/gdb/value.c	2010-10-07 20:23:01.000000000 +0100
@@ -434,6 +434,13 @@ value_contents_for_printing (struct valu
 }
 
 const gdb_byte *
+value_contents_for_printing_const (const struct value *value)
+{
+  gdb_assert (!value->lazy);
+  return value->contents;
+}
+
+const gdb_byte *
 value_contents_all (struct value *value)
 {
   const gdb_byte *result = value_contents_for_printing (value);
Index: src/gdb/value.h
===================================================================
--- src.orig/gdb/value.h	2010-10-07 11:53:48.000000000 +0100
+++ src/gdb/value.h	2010-10-07 20:22:55.000000000 +0100
@@ -262,6 +262,8 @@ extern const gdb_byte *value_contents_al
    plan to check the validity manually.  */
 extern const gdb_byte *value_contents_for_printing (struct value *value);
 
+extern const gdb_byte *value_contents_for_printing_const (const struct value *value);
+
 extern int value_fetch_lazy (struct value *val);
 extern int value_contents_equal (struct value *val1, struct value *val2);