Re: warning about const multidimensional array as function parameter

[Martin Uecker <uecker at eecs dot berkeley dot edu>]

Manuel LÃpez-IbÃÃez <lopezibanez@gmail.com>:

> On 14 October 2014 01:12, Martin Uecker <uecker@eecs.berkeley.edu> wrote:
> > Converting a pointer to an array to a pointer to a constant array
> > is safe. Converting a pointer to a pointer to a pointer to a pointer
> > to a constant is not (as the CFAQ points out).
> 
> You are probably right that it is safe. Unfortunately, C considers
> invalid cases that are safe and that are allowed by C++ (as mentioned
> in that C FAQ). I updated the FAQ with comments by Joseph Myers taken
> from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=47143 and
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=33076.

Thank you. I am not sure about the full C++ semantics, but the
case of passing pointers to const arrays seems fairly straightforward.

I also raised the issue at comp.std.c and proposed a change the
following extension to the pointer conversion rule (6.3.2.3):

"A pointer to an array with non-q-qualified element type may be
converted to a pointer to an array with the q-qualified version
of the type as element type." 

Maybe somebody from the committee is reading there...


> I'm not sure how much effort would be to make the C FE of GCC follow
> the rules of C++ or whether this would conflict with other parts of
> the standard (that is, rejecting valid C programs).

I played bit at gcc/c/c-typeck.c and getting rid of the warning
seems simple (see below). But so far I don't fully understand the
code or what I did...

I also haven't figured out how to add a then-required warning about
the case where a pointer-to-constant-array is converted to a 
pointer-to-array.

> Perhaps the option could be separated into two, one only enabled by
> -Wpedantic that controls cases that are known to be safe
> (-Wpedantic-incompatible-pointer-types? -Wincompatible-but-safe?).

My preferred solution would be to simple allow the case where it is 
safe and warn only when enabled by -Wpedantic. 

> Still, someone would need to step up and do the work:
> https://gcc.gnu.org/wiki/GettingStarted#Basics:_Contributing_to_GCC_in_10_easy_steps

I will talk a look at this.

Martin
 
> Cheers,
> 
> Manuel.


Index: gcc/c/c-typeck.c
===================================================================
--- gcc/c/c-typeck.c	(Revision 216134)
+++ gcc/c/c-typeck.c	(Arbeitskopie)
@@ -1217,6 +1217,7 @@
 comp_target_types (location_t location, tree ttl, tree ttr)
 {
   int val;
+  int val2;
   tree mvl = TREE_TYPE (ttl);
   tree mvr = TREE_TYPE (ttr);
   addr_space_t asl = TYPE_ADDR_SPACE (mvl);
@@ -1228,13 +1229,20 @@
   if (!addr_space_superset (asl, asr, &as_common))
     return 0;
 
+  val2 = 1;
+
+  if ((TREE_CODE (mvl) == ARRAY_TYPE) && (TREE_CODE (mvr) == ARRAY_TYPE))
+    val2 = comptypes (mvl, mvr);
+
   /* Do not lose qualifiers on element types of array types that are
      pointer targets by taking their TYPE_MAIN_VARIANT.  */
-  if (TREE_CODE (mvl) != ARRAY_TYPE)
+
+//  if (TREE_CODE (mvl) != ARRAY_TYPE)
     mvl = (TYPE_ATOMIC (mvl)
 	   ? c_build_qualified_type (TYPE_MAIN_VARIANT (mvl), TYPE_QUAL_ATOMIC)
 	   : TYPE_MAIN_VARIANT (mvl));
-  if (TREE_CODE (mvr) != ARRAY_TYPE)
+
+//  if (TREE_CODE (mvr) != ARRAY_TYPE)
     mvr = (TYPE_ATOMIC (mvr)
 	   ? c_build_qualified_type (TYPE_MAIN_VARIANT (mvr), TYPE_QUAL_ATOMIC)
 	   : TYPE_MAIN_VARIANT (mvr));
@@ -1241,6 +1249,10 @@
   enum_and_int_p = false;
   val = comptypes_check_enum_int (mvl, mvr, &enum_and_int_p);
 
+  if ((val == 1) && (val2 != 1))
+   pedwarn (location, OPT_Wpedantic, "pointers to array have incompatible qualifiers");
+
   if (val == 2)
     pedwarn (location, OPT_Wpedantic, "types are not quite compatible");