Re: [PATCH v3 10/11] c: Turn -Wincompatible-pointer-types into a permerror

[Richard Earnshaw]

On 05/12/2023 09:46, Florian Weimer wrote:
* Richard Earnshaw:

(I think it's this patch, not one of the others in the series).

This breaks building libgfortran with newlib on arm and aarch64:


/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2208:46:
error: pointer type mismatch in conditional expression
[-Wincompatible-pointer-types]
  2208 |                           dtp->common.iostat : &noiostat;
       |                                              ^
/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2208:27:
note: first expression has type ‘GFC_INTEGER_4 *’ {aka ‘long int *’}
  2208 |                           dtp->common.iostat : &noiostat;
       |                           ^~~~~~~~~~~~~~~~~~
/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2208:48:
note: second expression has type ‘int *’
  2208 |                           dtp->common.iostat : &noiostat;
       |                                                ^~~~~~~~~
/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2224:34:
error: passing argument 2 of ‘dtp->u.p.fdtio_ptr’ from incompatible
pointer type [-Wincompatible-pointer-types]
  2224 |           dtp->u.p.fdtio_ptr (p, &unit, iotype, &vlist,
       |                                  ^~~~~
       |                                  |
       |                                  int *
/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2224:34:
note: expected ‘GFC_INTEGER_4 *’ {aka ‘long int *’} but argument is of
type ‘int *’
/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2225:31:
error: passing argument 5 of ‘dtp->u.p.fdtio_ptr’ from incompatible
pointer type [-Wincompatible-pointer-types]
  2225 |                               child_iostat, child_iomsg,
       |                               ^~~~~~~~~~~~
       |                               |
       |                               int *
/work/rearnsha/gnusrc/nightly/gcc-cross/master/libgfortran/io/list_read.c:2225:31:
note: expected ‘GFC_INTEGER_4 *’ {aka ‘long int *’} but argument is of
type ‘int *’

Presumably the fixes will look like this?

diff --git a/libgfortran/io/list_read.c b/libgfortran/io/list_read.c
index db3330060ce..4fcc77dbf83 100644
--- a/libgfortran/io/list_read.c
+++ b/libgfortran/io/list_read.c
@@ -2987,13 +2987,13 @@ nml_read_obj (st_parameter_dt *dtp, namelist_info *nl, 
index_type offset,
            /* If this object has a User Defined procedure, call it.  */
            if (nl->dtio_sub != NULL)
              {
-               int unit = dtp->u.p.current_unit->unit_number;
+               GFC_INTEGER_4 unit = dtp->u.p.current_unit->unit_number;
                char iotype[] = "NAMELIST";
                gfc_charlen_type iotype_len = 8;
                char tmp_iomsg[IOMSG_LEN] = "";
                char *child_iomsg;
                gfc_charlen_type child_iomsg_len;
-               int noiostat;
+               GFC_INTEGER_4 noiostat;
                int *child_iostat = NULL;
                gfc_full_array_i4 vlist;
                formatted_dtio dtio_ptr = (formatted_dtio)nl->dtio_sub;


Apparently the targets I built define GFC_INTEGER_4 as int, so this
didn't show up.

It looks reasonable to me, but I'm not a real user of libgfortran, so 
there's possibly something more subtle that I've missed (I can't even 
guarantee I copied all of the errors from the build log).

I've copied Tobias as a fortran maintainer, but I don't know if this is 
his forte either, though perhaps he might know who's it is.

R.


Thanks,
Florian