[Patch, Fortran] PR 38487 Spurious warning: "Actual argument might interfere..."

Mikael Morin mikael.morin@tele2.fr
Mon Dec 15 18:30:00 GMT 2008


Tobias Burnus wrote:
> Hello Mikael,
> 
> Mikael Morin wrote:
>> this disables the warning I added for PR 35681 in the pointer case as
>> there may be too many false positive.
>>
>> No regression on x86_64-unknown-linux-gnu. Ok for trunk?
>>   
> I wonder whether one might get for e->expr_type != EXPR_VARIABLE a
> spurious warning, but I don't think so.
No, because either in gfc_check_argument_var_dependency or in
gfc_check_dependency, there is a gcc_assert and a switch that assert
that we are in the EXPR_VARIABLE case for both of the arguments we check.

I added the e->expr_type check so that the function can be used later in
the non-EXPR_VARIABLE case. But it is not useful now.

> 
> Thus: OK and thanks for the patch.
Applied to trunk as revision 142766. Thanks for the review.
I'll wait a week before applying to the 4.3 branch.

Mikael



More information about the Gcc-patches mailing list