This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch, fortran] PR29758 - Runtime segfault in RESHAPE with insufficient elements in SOURCE


Ping!

I would not do this so soon, were I not on my travels again on Sunday. If it goes to trunk this weekend, it can be committed to 4.2/4.1 the following weekend.

Cheers

Paul
:ADDPATCH fortran:

This patch checks to see if there are enough elements in reshape's source to match shape, when shape is a constant array and source has a constant shape. It is a self explanatory bit of hacking in check.c. The testcase checks two obvious errors and checks that a correct one goes through.

Regtested Suse10.1/amd64.

OK for trunk, 4.2 and 4.1?

Paul

2006-11-08 Paul Thomas <pault@gcc.gnu.org>

   PR fortran/29758
   * check.c (gfc_check_reshape): Check that there are enough
   elements in the source array as to be able to fill an array
   defined by shape, when pad is absent.

2006-11-08 Paul Thomas <pault@gcc.gnu.org>

   PR fortran/29758
   * gfortran.dg/reshape_source_size_1.f90: New test.


------------------------------------------------------------------------


Index: gcc/fortran/check.c
===================================================================
*** gcc/fortran/check.c (revision 118597)
--- gcc/fortran/check.c (working copy)
*************** gfc_check_reshape (gfc_expr * source, gf
*** 2110,2115 ****
--- 2110,2116 ----
gfc_expr * pad, gfc_expr * order)
{
mpz_t size;
+ mpz_t nelems;
int m;
if (array_check (source, 0) == FAILURE)
*************** gfc_check_reshape (gfc_expr * source, gf
*** 2149,2154 ****
--- 2150,2187 ----
if (order != NULL && array_check (order, 3) == FAILURE)
return FAILURE;
+ if (pad == NULL
+ && shape->expr_type == EXPR_ARRAY
+ && gfc_is_constant_expr (shape)
+ && !(source->expr_type == EXPR_VARIABLE
+ && source->symtree->n.sym->as
+ && source->symtree->n.sym->as->type == AS_ASSUMED_SIZE))
+ {
+ /* Check the match in size between source and destination. */
+ if (gfc_array_size (source, &nelems) == SUCCESS)
+ {
+ gfc_constructor *c;
+ bool test;
+ + c = shape->value.constructor;
+ mpz_init_set_ui (size, 1);
+ for (; c; c = c->next)
+ mpz_mul (size, size, c->expr->value.integer);
+ + test = mpz_cmp (nelems, size) < 0 && mpz_cmp_ui (size, 0) > 0;
+ mpz_clear (nelems);
+ mpz_clear (size);
+ + if (test)
+ {
+ gfc_error ("Without padding, there are not enough elements in the "
+ "intrinsic RESHAPE source at %L to match the shape",
+ &source->where);
+ return FAILURE;
+ }
+ }
+ }
+ return SUCCESS;
}
Index: gcc/testsuite/gfortran.dg/reshape_source_size_1.f90
===================================================================
*** gcc/testsuite/gfortran.dg/reshape_source_size_1.f90 (revision 0)
--- gcc/testsuite/gfortran.dg/reshape_source_size_1.f90 (revision 0)
***************
*** 0 ****
--- 1,11 ----
+ ! { dg-do compile }
+ ! Tests patch for PR29758, which arose from PR29431. There was no check that there
+ ! were enough elements in the source to match the shape.
+ !
+ ! Contributed by Paul Thomas <pault@gcc.gnu.org>
+ !
+ real :: a(2,2), b = 1.0, c(3), d(4)
+ a = reshape ([b], [2,2]) ! { dg-error "not enough elements" }
+ a = reshape (c, [2,2]) ! { dg-error "not enough elements" }
+ a = reshape (d, [2,2])
+ end


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]