RE: reg_nonzero_bits_for_combine misunderstood

2013-11-28 Thread Paulo Matos
> -Original Message- > From: Eric Botcazou [mailto:ebotca...@adacore.com] > Sent: 28 November 2013 11:27 > To: Paulo Matos > Cc: gcc@gcc.gnu.org > Subject: Re: reg_nonzero_bits_for_combine misunderstood > > > Right, didn't notice nonzero_sign_valid

Re: reg_nonzero_bits_for_combine misunderstood

2013-11-28 Thread Eric Botcazou
> Right, didn't notice nonzero_sign_valid below. I think restricting mode == > last_set_mode is too restrictive. > > nonzero_bits is still valid if the new mode has less precision than the old > mode. Sure, but I'm not suggesting to restrict anything, quite the contrary: { unsigned HOS

RE: reg_nonzero_bits_for_combine misunderstood

2013-11-28 Thread Paulo Matos
> -Original Message- > From: Eric Botcazou [mailto:ebotca...@adacore.com] > Sent: 27 November 2013 18:27 > To: Paulo Matos > Cc: gcc@gcc.gnu.org > Subject: Re: reg_nonzero_bits_for_combine misunderstood > > > But the problem is that if the mode of the regis

Re: reg_nonzero_bits_for_combine misunderstood

2013-11-27 Thread Eric Botcazou
> But the problem is that if the mode of the register is larger than the mode > of the register when last set then we can't know anything about > nonzero_bits. Right, this case is handled for nonzero_sign_valid a few lines below. > Fixed the problem. Any comments? Why not duplicate the treatmen