Re: Add VIEW_CONVERT_EXPR to operand_equal_p

[Richard Biener <richard.guenther@gmail.com>]

On Thu, Oct 29, 2015 at 12:31 PM, Richard Biener
<richard.guenther@gmail.com> wrote:
> On Thu, Oct 29, 2015 at 12:22 PM, Richard Biener
> <richard.guenther@gmail.com> wrote:
>> On Thu, Oct 29, 2015 at 4:39 AM, Jan Hubicka <hubicka@ucw.cz> wrote:
>>>> > Added and comitted now.
>>>>
>>>> Thanks.  Now on to the wrong code issues. :-)
>>>>
>>>> Up to the change, the useless_type_conversion_p predicate was relying on
>>>> structural equivalence via the TYPE_CANONICAL check, now it only looks at the
>>>> outermost level (size, mode).  Now some back-ends, most notably x86-64, do a
>>>> deep structural scan to determine the calling conventions (classify_argument)
>>>> instead of just looking at the size and the mode, so consistency dictates that
>>>> the type of the argument and that of the parameter be structurally equivalent
>>>> and this sometimes can only be achieved by a VCE... which is now deleted. :-(
>>>> See the call to derivedIP in the attached testcase which now fails on x86-64.
>>>>
>>>> How do we get away from here?
>>>
>>> Hmm, I noticed this in ipa-icf context and wrote checker that two functions are ABI
>>> compatile (did not pushed it out yet), but of course this is nastier.
>>>
>>> I think the problem exists before my patch with LTO - it is just matter of
>>> doing two types which will be considered equivalent by
>>> gimple_canonical_types_compatible_p but have different type conversion.  An
>>> example of such type would be:
>>>
>>> struct a {
>>>   int a[4];
>>> };
>>> struct b {
>>>   int a[4];
>>> } __attribute__ ((__aligned__(16)));
>>>
>>> I tried to turn this into an testcase, the problem is that I don't know of a way
>>> to obtain VIEW_CONVERT_EXPR between the two types out of C or C++ frontend and we
>>> don't seem to synthetize these in middle end (even in cases it would make sense).
>>> I will try to play with it more - would be nice to have a C reproducer.
>>>
>>> We may be safe before my patch from wrong code issues if there is no way to
>>> rpduce VIEW_CONVERT_EXPR between types like this in languages that support
>>> aligned attribute.
>>>
>>> I think the problem is generally similar to memory references - the gimple type
>>> compatibility should not be tied to ABI details.  Probably most consistent
>>> solution would be to extend GIMPLE_CALL to also list types of parameters and do
>>> not rely on whatever type the operand have....
>>>
>>> Richard, any ideas?
>>
>> IMHO it was always wrong/fragile for backends to look at the actual arguments to
>> decide on the calling convention.  The backends should _solely_ rely on
>> gimple_call_fntype and its TYPE_ARG_TYPES here.
>>
>> Of course then there are varargs ... (not sure if we hit this here).
>>
>> But yes, the VIEW_CONVERT "stripping" is a bit fragile and I don't remember
>> what exactly we gain from it (when not done on registers).
>>
>> But I also don't see where we do the stripping mentioned on memory references.
>> The match.pd pattern doesn't apply to memory, only in the GENERIC path
>> which is guarded with exact type equality.  So I can't see where we end up
>> stripping the V_C_E.
>>
>> There is one bogus case still in fold-const.c:
>>
>>     case VIEW_CONVERT_EXPR:
>>       if (TREE_CODE (op0) == MEM_REF)
>>         /* ???  Bogus for aligned types.  */
>>         return fold_build2_loc (loc, MEM_REF, type,
>>                                 TREE_OPERAND (op0, 0), TREE_OPERAND (op0, 1));
>>
>>       return NULL_TREE;
>>
>> that comment is only in my local tree ... (we lose alignment info that is
>> on the original MEM_REF type which may be a smaller one).
>
> Ah - tree_ssa_useless_type_conversion and callers, during gimplification.
> I'd like to get rid of it but maybe simply delete the VIEW_CONVERT_EXPR
> case from it for now (and return true unconditionally for NON_LVALUE_EXPR).
>
> Index: gcc/tree-ssa.c
> ===================================================================
> --- gcc/tree-ssa.c      (revision 229517)
> +++ gcc/tree-ssa.c      (working copy)
> @@ -1142,13 +1161,16 @@ delete_tree_ssa (struct function *fn)
>  bool
>  tree_ssa_useless_type_conversion (tree expr)
>  {
> +  /* Not strictly a conversion but this function is used to strip
> +     useless stuff from trees returned from GENERIC folding.  */
> +  if (TREE_CODE (expr) == NON_LVALUE_EXPR)
> +    return true;
> +
>    /* If we have an assignment that merely uses a NOP_EXPR to change
>       the top of the RHS to the type of the LHS and the type conversion
>       is "safe", then strip away the type conversion so that we can
>       enter LHS = RHS into the const_and_copies table.  */
> -  if (CONVERT_EXPR_P (expr)
> -      || TREE_CODE (expr) == VIEW_CONVERT_EXPR
> -      || TREE_CODE (expr) == NON_LVALUE_EXPR)
> +  if (CONVERT_EXPR_P (expr))
>      return useless_type_conversion_p
>        (TREE_TYPE (expr),
>         TREE_TYPE (TREE_OPERAND (expr, 0)));
>
> IMHO the gimplifier use should be more explicit and most remaining GIMPLE
> middle-end uses should be removed (after auditing).

The above is pre-approved if one of you does the required testing (it
fixes the Ada
testcase for me).

Richard.

> Richard.
>
>> Richard.
>>
>>> Honza
>>>>
>>>>
>>>>       * gnat.dg/discr44.adb: New test.
>>>>
>>>> --
>>>> Eric Botcazou
>>>
>>>> -- { dg-do run }
>>>> -- { dg-options "-gnatws" }
>>>>
>>>> procedure Discr44 is
>>>>
>>>>   function Ident (I : Integer) return Integer is
>>>>   begin
>>>>     return I;
>>>>   end;
>>>>
>>>>   type Int is range 1 .. 10;
>>>>
>>>>   type Str is array (Int range <>) of Character;
>>>>
>>>>   type Parent (D1, D2 : Int; B : Boolean) is record
>>>>     S : Str (D1 .. D2);
>>>>   end record;
>>>>
>>>>   type Derived (D : Int) is new Parent (D1 => D, D2 => D, B => False);
>>>>
>>>>   X1 : Derived (D => Int (Ident (7)));
>>>>
>>>> begin
>>>>   if X1.D /= 7 then
>>>>     raise Program_Error;
>>>>   end if;
>>>> end;
>>>