Re: [PATCH 1/2] Check negative combined step

On Thu, 27 Jan 2022, Jiufu Guo wrote: > Hi Richard, > > Richard Biener writes: > > > On Tue, 25 Jan 2022, Richard Biener wrote: > > > >> On Tue, 25 Jan 2022, Jiufu Guo wrote: > >> > ... > >> > > > >> > > The point is that we may not change the iteration number at which > >> > > overflow occurs

Re: [PATCH 1/2] Check negative combined step

Hi Richard, Richard Biener writes: > On Tue, 25 Jan 2022, Richard Biener wrote: > >> On Tue, 25 Jan 2022, Jiufu Guo wrote: >> ... >> > > >> > > The point is that we may not change the iteration number at which >> > > overflow occurs since that alters the result of the < compare. >> > > Only if w

Re: [PATCH 1/2] Check negative combined step

On Tue, 25 Jan 2022, Richard Biener wrote: > On Tue, 25 Jan 2022, Jiufu Guo wrote: > > > Richard Biener writes: > > > > > On Tue, 25 Jan 2022, Jiufu Guo wrote: > > > > > >> Jiufu Guo writes: > > >> > > >> > Richard Biener writes: > > >> > > > >> >> On Thu, 13 Jan 2022, Jiufu Guo wrote: > > >

Re: [PATCH 1/2] Check negative combined step

On Tue, 25 Jan 2022, Jiufu Guo wrote: > Richard Biener writes: > > > On Tue, 25 Jan 2022, Jiufu Guo wrote: > > > >> Jiufu Guo writes: > >> > >> > Richard Biener writes: > >> > > >> >> On Thu, 13 Jan 2022, Jiufu Guo wrote: > >> > ... > >> >> > >> >>> - /* No need to check sign of the new

Re: [PATCH 1/2] Check negative combined step

Richard Biener writes: > On Tue, 25 Jan 2022, Jiufu Guo wrote: > >> Jiufu Guo writes: >> >> > Richard Biener writes: >> > >> >> On Thu, 13 Jan 2022, Jiufu Guo wrote: >> > ... >> >> >> >>> - /* No need to check sign of the new step since below code takes >> >>> care >> >>> - of th

Re: [PATCH 1/2] Check negative combined step

On Tue, 25 Jan 2022, Jiufu Guo wrote: > Jiufu Guo writes: > > > Richard Biener writes: > > > >> On Thu, 13 Jan 2022, Jiufu Guo wrote: > > ... > >> > >>> - /* No need to check sign of the new step since below code takes > >>> care > >>> - of this well. */ > >>> + /* Like cases shown

Re: [PATCH 1/2] Check negative combined step

Jiufu Guo writes: > Richard Biener writes: > >> On Thu, 13 Jan 2022, Jiufu Guo wrote: > ... >> >>> - /* No need to check sign of the new step since below code takes care >>> -of this well. */ >>> + /* Like cases shown in PR100740/102131, negtive step is not safe. */ >>> + if

Re: [PATCH 1/2] Check negative combined step

Richard Biener writes: > On Thu, 13 Jan 2022, Jiufu Guo wrote: ... > >> - /* No need to check sign of the new step since below code takes care >> - of this well. */ >> + /* Like cases shown in PR100740/102131, negtive step is not safe. */ >> + if (tree_int_cst_sign_bit (step)

Re: [PATCH 1/2] Check negative combined step

On Thu, 13 Jan 2022, Jiufu Guo wrote: > Hi, > > Previously, there is discussion in: > https://gcc.gnu.org/pipermail/gcc-patches/2021-December/586460.html > I seperate it as two patches. > > This first patch is to avoid negative step when combining two ivs. > The second patch is adding more accur