Re: [Patch][OpenMP] Fix use_device_… with absent optional arg

[Tobias Burnus]

Hi Jakub,

thanks for the review – committed as Rev. 279004.

On 12/5/19 12:46 PM, Jakub Jelinek wrote:
> On Fri, Nov 29, 2019 at 01:03:13PM +0100, Tobias Burnus wrote:
>> --- a/gcc/fortran/trans-openmp.c
>> +	  && TREE_TYPE (decl) != pvoid_type_node
> Is it always pvoid_type_node, or could it be say const qualified version
> thereof etc. (C void const *) etc.? If the latter, then
> 	  && !VOID_TYPE_P (TREE_TYPE (TREE_TYPE (decl)))
> might be better check.  If it is always just pvoid_type_node, the above is
> fine sure.

I will use your version (twice) just to be sure. — Currently, I think it 
will always be pvoid_type_node but I don't want to rule out some const 
qualifier might be possible. (volatile is not permitted with value, 
restricted and atomic qualifiers are unlikely.)

[…]

> Shouln't the stop arguments differ from anything else already in the 
> testcase? 
Yes – it makes failure debugging easier. I thought, I had re-enumerated, 
but I missed the two.

Tobias