Re: [patch] fix wrong gimple code with ptr to forward vla

[Olivier Hainque <hainque at adacore dot com>]

Joseph S. Myers wrote:
> The pointer declaration point is in general in the middle of an 
> expression.  This patch breaks

> {
>   int a = 1;
>   if (sizeof (*(++a, (char (*)[a])0)) != 2)
>     abort ();

 Oh, indeed.

> This patch might be safer if the TYPE_DECL were only inserted in the case 
> of decl_context == NORMAL.

 Humm, I see. We at least need FIELD as well to cover the initial bug
 (20050527-1.c)

> That should avoid the TYPENAME problems like the testcase I give
> (which should be added to the gcc.c-torture/execute testsuite)

 Right. One concern, still: are we sure that the initial bug (sizes
 evaluated in a conditional branch and reused in another) couldn't
 show up in the TYPENAME case ?

> and also avoid possible problems with PARM (where in the case of
> declarations that aren't definitions, no special handling should be
> needed, and in the case of those that are definitions,
> gimplify_parameters should arrange the necessary evaluation of array
> sizes).

 OK.

 Thanks for your feedback.

 Olivier