------- Comment #3 from rguenther at suse dot de 2008-06-26 12:11 -------
Subject: Re: [4.3/4.4 Regression] ICE in
vect_update_ivs_after_vectorizer
On Thu, 26 Jun 2008, irar at il dot ibm dot com wrote:
> ------- Comment #2 from irar at il dot ibm dot com 2008-06-26 11:57 -------
> For access function (short int) {(short unsigned int) i_44, +, 1}_1)
> evolution_part_in_loop_num() returns NULL, which causes the failure.
>
> I tried to peel NOP_EXPRs from POLYNOMIAL_CHRECs in
> evolution_part_in_loop_num(), but it caused lots of testsuite failures, and I
> am not sure this is the correct way.
Well, the problem is that an unsigned induction variable may validly
wrap while a signed induction variable may not. This is probably the
cause of the testsuite failures. Can't you simply handle a NULL return
from evolution_part_in_loop_num in the vectorizer?
Richard.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36630
