Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-26 Thread Richard Biener
On Fri, May 26, 2017 at 1:59 PM, Bin.Cheng wrote: > On Fri, May 26, 2017 at 12:39 PM, Richard Biener > wrote: >> On Thu, May 25, 2017 at 5:15 PM, Bin.Cheng wrote: >>> On Wed, May 24, 2017 at 11:54 AM, Richard Sandiford >>> wrote: "Bin.Cheng" writes: > On Tue, May 23, 2017 at 6:53 PM,

Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-26 Thread Bin.Cheng
On Fri, May 26, 2017 at 12:39 PM, Richard Biener wrote: > On Thu, May 25, 2017 at 5:15 PM, Bin.Cheng wrote: >> On Wed, May 24, 2017 at 11:54 AM, Richard Sandiford >> wrote: >>> "Bin.Cheng" writes: On Tue, May 23, 2017 at 6:53 PM, Richard Sandiford wrote: > AIUI, the reason the ol

Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-26 Thread Richard Biener
On Thu, May 25, 2017 at 5:15 PM, Bin.Cheng wrote: > On Wed, May 24, 2017 at 11:54 AM, Richard Sandiford > wrote: >> "Bin.Cheng" writes: >>> On Tue, May 23, 2017 at 6:53 PM, Richard Sandiford >>> wrote: AIUI, the reason the old code mishandled negative steps was that the associated seg

Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-25 Thread Bin.Cheng
On Wed, May 24, 2017 at 11:54 AM, Richard Sandiford wrote: > "Bin.Cheng" writes: >> On Tue, May 23, 2017 at 6:53 PM, Richard Sandiford >> wrote: >>> AIUI, the reason the old code mishandled negative steps was that the >>> associated segment lengths were stored as sizetype and so looked like >>>

Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-24 Thread Richard Sandiford
"Bin.Cheng" writes: > On Tue, May 23, 2017 at 6:53 PM, Richard Sandiford > wrote: >> AIUI, the reason the old code mishandled negative steps was that the >> associated segment lengths were stored as sizetype and so looked like >> big unsigned values. Those values therefore satisfied tree_fits_uh

Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-24 Thread Bin.Cheng
On Tue, May 23, 2017 at 6:53 PM, Richard Sandiford wrote: > AIUI, the reason the old code mishandled negative steps was that the > associated segment lengths were stored as sizetype and so looked like > big unsigned values. Those values therefore satisfied tree_fits_uhwi_p > even though they were

Re: [PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-23 Thread Richard Sandiford
AIUI, the reason the old code mishandled negative steps was that the associated segment lengths were stored as sizetype and so looked like big unsigned values. Those values therefore satisfied tree_fits_uhwi_p even though they were semantically negative. Is that right? Assuming yes, and quoting

[PATCH GCC][3/6]Fix PR80815 by handling negative DR_STEPs in runtime alias check

2017-05-23 Thread Bin Cheng
Hi, This patch fixes PR80815 in which negative DR_STEP is mis-handled. It does below: 1) Reorder three cases in which we merge alias checks, in order like: old_case_A -> new_case_B old_case_B -> new_case_C (and removed as described in 3)) old_case_C -> new_case_A