[Bug c/31887] bad warning converting qualified void* to qualified array pointer

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31887

Andrew Pinski pinskia at gcc dot gnu.org changed:

   What|Removed |Added

 Status|UNCONFIRMED |RESOLVED
 Resolution|--- |INVALID

--- Comment #5 from Andrew Pinski pinskia at gcc dot gnu.org ---
Invalid as qualifiers for arrays are not attached to the array type but rather
to the element so there is no such thing as a const array.

[Bug c/31887] bad warning converting qualified void* to qualified array pointer

[bugs at 59A2 dot org]

--- Comment #4 from bugs at 59A2 dot org  2008-11-17 12:24 ---
There is no way to qualify an array type, but this sort of conversion is very
useful.  For instance, the optimizer produces better code for

void foo(const double a[], int m, int n) {
  const double (*b)[n] = (const double(*)[n])a;
  /* use b[i][j] in inner loops */
}

than when `a[i*n+j]' is used (without manually hoisting arithmetic out of inner
loops).  With -Wcast-qual, a warning is given since the qualifier does not
apply to the array type, only the elements.  Note that this example is very
similar to the example given in
http://www.open-std.org/jtc1/sc22/wg14/www/C99RationaleV5.10.pdf page 87
(physical page 94).  Clearly the cast above should be possible, although there
is no way to express it in a way that is compatible with C99 paragraph 6.7.3.8.
 I propose that -Wcast-qual promote qualifiers on array elements to the array
itself when determining whether a cast discards qualifiers.  That is, not warn
for casts like the one above.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31887

[Bug c/31887] bad warning converting qualified void* to qualified array pointer

[raeburn at raeburn dot org]

--- Comment #3 from raeburn at raeburn dot org  2007-08-15 06:03 ---
Section 6.7.3 says:  If the specification of an array type includes any type
qualifiers, the element type is so-qualified, not the array type.  The more I
think about it, the more I think the compiler is technically correct, though I
think this is probably a bug in the standard.

I brought this up on comp.std.c a couple weeks or so ago, though, and while
Doug Gwyn, whose interpretation on these things I generally respect, seemed to
indicate that my current interpretation is wrong (and that it should indeed be
const array of 8 chars), he didn't provide references to that effect, so I'm
not sure where that interpretation comes from.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31887

[Bug c/31887] bad warning converting qualified void* to qualified array pointer

[raeburn at raeburn dot org]

--- Comment #2 from raeburn at raeburn dot org  2007-05-11 08:05 ---
Subject: Re:  bad warning converting qualified void* to qualified array pointer

On May 10, 2007, at 19:00, pinskia at gcc dot gnu dot org wrote:
 I think only void* can be done with an implicat cast so we go  
 from const
 void* to void* and then to const block*.

ISO/IEC 9899-1999 section 6.5.16.1 (assignment constraints) includes:

  - one operand is a pointer to an object or incomplete type and the  
other is a pointer to a qualified or unqualified version of void, and  
the type pointed to by the left has all the qualifiers of the type  
pointed to by the right.

So I'm pretty sure we don't automatically lose on the const-ness in  
the general case.  What I'm less sure about is how the array type  
plays into it.  According to 6.7.3, const block means array [8] of  
const unsigned char, so is there an argument to be made that  
assigning a pointer-to-const value to an lvalue of type pointer to an  
unqualified array type (containing a type that happens to be const- 
qualified) isn't allowed?  I don't think it would be logical, but  
that doesn't guarantee that that's how the standard actually works.

Ken


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31887

[Bug c/31887] bad warning converting qualified void* to qualified array pointer

[pinskia at gcc dot gnu dot org]

--- Comment #1 from pinskia at gcc dot gnu dot org  2007-05-11 00:00 ---
I think only void* can be done with an implicat cast so we go from const
void* to void* and then to const block*.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31887