-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 04/29/11 10:43, Jakub Jelinek wrote: > Hi! > > The following patch fixes a bug in tree-switch-conversion.c with > signed index_expr's. build_arrays would compute index_expr - range_min > in index_expr's type and use that as index into CSWTCH.N array, > which is wrong, because in this case index_expr 98 - (-62) computed > in signed char type results in signed overflow and we end up > loading from CSWTCH.2[-96]. Apparently for the bounds checking > we perform the same index_expr - range_min computation, but in > corresponding unsigned type. This patch computes it just once in > unsigned type, so that overflow isn't undefined. > > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for > trunk/4.6/4.5/4.4? > > 2011-04-29 Jakub Jelinek <ja...@redhat.com> > > PR tree-optimization/48809 > * tree-switch-conversion.c (build_arrays): Compute tidx in unsigned > type. > (gen_inbound_check): Don't compute index_expr - range_min in utype > again, instead reuse SSA_NAME initialized in build_arrays. > Remove two useless gsi_for_stmt calls. > > * gcc.c-torture/execute/pr48809.c: New test. OK.
jeff -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJNuwpHAAoJEBRtltQi2kC71SAH/1VFBhEB7SXDEogMFx59LD1j G1D+1DBL9qQEDRXMQi+aA6H8Y8WxfUDsUTQUD3rrg9HxZDEaUz55P46nc6dd/DXS 72tFm5OW5bucJbnHocxu+PeAedk8w0cmjxdnXA38ezPhqpeqUKK/ojhPPcGSErHH 7Z9YUdDHGbAcZACjvd1qTObfp4NUG0PTZ7d6+ZOsAt7RwxgECcUJuyJhZaQdxZVV MTF96cNa205xqOIBYwGCyW5bvFhgL4kDvdUMSln2CURXyl/N+JqeGnuGEElItM+A 8ryLWBIyzuUFHNML6FK6wrROs/QPiUdan2RqDyW46WT8UdE3PgVakT/7c9E42sc= =395m -----END PGP SIGNATURE-----
