[PATCH] Remove warning while compiling with CLooG isl
Jack Howarth
howarth@bromo.med.uc.edu
Fri Nov 12 20:05:00 GMT 2010
On Fri, Nov 12, 2010 at 11:30:34AM -0500, Tobias Grosser wrote:
> Hi,
>
> this patch fixes a warning that appears when compiling graphite with
> CLooG isl.
>
> * graphite-cloog-util.c (oppose_constraint,
> cloog_matrix_to_ppl_constraint,
> new_Constraint_System_from_Cloog_Matrix): Explicitly cast to int
> as CLooG isl uses unsigned integers. This triggered a warning.
>
> @Jack: This patch should fix the compile failure with CLooG isl. Can you
> verify this?
Tobi,
I can confirm that this patch allows gcc trunk to bootstrap on x86_64-apple-darwin10
with ppl 0.15.9 and cloog.org git built as cloog-isl. Also..
make -k check RUNTESTFLAGS="graphite.exp --target_board=unix'{-m32,-m64}'"
shows no regressions.
Jack
ps Are there any instructions for building cloog-ppl from cloog.org git? I would like to
test gcc trunk against that instead of legacy cloog's ppl support. I assume that cloog.org's
cloog-ppl might require ppl 0.11, no?
>
> Cheers
> Tobi
> >From 4ecdd356cd507dabeffc9c1b6c9f066a530624b7 Mon Sep 17 00:00:00 2001
> From: Tobias Grosser <grosser@fim.uni-passau.de>
> Date: Fri, 12 Nov 2010 11:22:52 -0500
> Subject: [PATCH] Remove warning because of CLooG isl/ppl difference
>
> * graphite-cloog-util.c (oppose_constraint,
> cloog_matrix_to_ppl_constraint,
> new_Constraint_System_from_Cloog_Matrix): Explicitly cast to int
> as CLooG isl uses unsigned integers. This triggered a warning.
> ---
> gcc/graphite-cloog-util.c | 15 ++++++++++++---
> 1 files changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/gcc/graphite-cloog-util.c b/gcc/graphite-cloog-util.c
> index 40c6fbc..df90b83 100644
> --- a/gcc/graphite-cloog-util.c
> +++ b/gcc/graphite-cloog-util.c
> @@ -60,7 +60,10 @@ oppose_constraint (CloogMatrix *m, int row)
> int k;
>
> /* Do not oppose the first column: it is the eq/ineq one. */
> - for (k = 1; k < m->NbColumns; k++)
> + /* Cast needed to remove warning that is generated as CLooG isl
> + is using an unsigned int for NbColumns and CLooG PPL is
> + using a signed int for NBColumns. */
> + for (k = 1; k < (int)m->NbColumns; k++)
> mpz_neg (m->p[row][k], m->p[row][k]);
> }
>
> @@ -177,7 +180,10 @@ cloog_matrix_to_ppl_constraint (CloogMatrix *matrix, int row)
> ppl_new_Coefficient (&coef);
> ppl_new_Linear_Expression_with_dimension (&expr, dim);
>
> - for (j = 1; j < matrix->NbColumns - 1; j++)
> + /* Cast needed to remove warning that is generated as CLooG isl
> + is using an unsigned int for NbColumns and CLooG PPL is
> + using a signed int for NBColumns. */
> + for (j = 1; j < (int)matrix->NbColumns - 1; j++)
> {
> ppl_assign_Coefficient_from_mpz_t (coef, matrix->p[row][j]);
> ppl_Linear_Expression_add_to_coefficient (expr, j - 1, coef);
> @@ -207,7 +213,10 @@ new_Constraint_System_from_Cloog_Matrix (ppl_Constraint_System_t *pcs,
>
> ppl_new_Constraint_System (pcs);
>
> - for (i = 0; i < matrix->NbRows; i++)
> + /* Cast needed to remove warning that is generated as CLooG isl
> + is using an unsigned int for NbColumns and CLooG PPL is
> + using a signed int for NBColumns. */
> + for (i = 0; i < (int)matrix->NbRows; i++)
> {
> ppl_Constraint_t c = cloog_matrix_to_ppl_constraint (matrix, i);
> ppl_Constraint_System_insert_Constraint (*pcs, c);
> --
> 1.7.1
>
More information about the Gcc-patches
mailing list