Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-11 Thread Ilya Enkovich
2016-05-11 15:45 GMT+03:00 H.J. Lu : > On Wed, May 11, 2016 at 2:26 AM, Ilya Enkovich wrote: >> 2016-05-10 21:13 GMT+03:00 H.J. Lu : >>> On Tue, May 10, 2016 at 9:19 AM, Ilya Enkovich >>> wrote: Hi, Curretly CSE may modify CFG and leave invalid dominance info. This patch imp

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-11 Thread H.J. Lu
On Wed, May 11, 2016 at 2:26 AM, Ilya Enkovich wrote: > 2016-05-10 21:13 GMT+03:00 H.J. Lu : >> On Tue, May 10, 2016 at 9:19 AM, Ilya Enkovich >> wrote: >>> Hi, >>> >>> Curretly CSE may modify CFG and leave invalid dominance info. This patch >>> improves track of CFG changes by CSE passes and f

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-11 Thread Ilya Enkovich
2016-05-10 21:13 GMT+03:00 H.J. Lu : > On Tue, May 10, 2016 at 9:19 AM, Ilya Enkovich wrote: >> Hi, >> >> Curretly CSE may modify CFG and leave invalid dominance info. This patch >> improves track of CFG changes by CSE passes and frees dominance info if >> required. This allows to remove corresp

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread H.J. Lu
On Tue, May 10, 2016 at 9:19 AM, Ilya Enkovich wrote: > Hi, > > Curretly CSE may modify CFG and leave invalid dominance info. This patch > improves track of CFG changes by CSE passes and frees dominance info if > required. This allows to remove corresponding workaround from STV pass. > > Does it

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread Bernd Schmidt
On 05/10/2016 07:36 PM, Richard Biener wrote: The policy for passes is to only free dominator info if they wreck it. They have to expect it being present but can also choose to free it early of course. Seems like a policy that's inviting surprises such as this one, but if this is the case, t

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread Richard Biener
On May 10, 2016 7:17:59 PM GMT+02:00, Bernd Schmidt wrote: >On 05/10/2016 07:06 PM, Richard Biener wrote: >> >> Dominators are never freed. > >Not sure what you're trying to say here. There's lots of calls to >free_dominance_info in the tree and obviously we need to do something >to >prevent the

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread Bernd Schmidt
On 05/10/2016 07:06 PM, Richard Biener wrote: Dominators are never freed. Not sure what you're trying to say here. There's lots of calls to free_dominance_info in the tree and obviously we need to do something to prevent the bug Ilya is trying to fix. Bernd

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread Richard Biener
On May 10, 2016 6:59:31 PM GMT+02:00, Bernd Schmidt wrote: >On 05/10/2016 06:19 PM, Ilya Enkovich wrote: >> Curretly CSE may modify CFG and leave invalid dominance info. This >patch >> improves track of CFG changes by CSE passes and frees dominance info >if >> required. This allows to remove cor

Re: [PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread Bernd Schmidt
On 05/10/2016 06:19 PM, Ilya Enkovich wrote: Curretly CSE may modify CFG and leave invalid dominance info. This patch improves track of CFG changes by CSE passes and frees dominance info if required. This allows to remove corresponding workaround from STV pass. Does it look OK? Better tracki

[PATCH, PR middle-end/70807] Free dominance info in CSE pass

2016-05-10 Thread Ilya Enkovich
Hi, Curretly CSE may modify CFG and leave invalid dominance info. This patch improves track of CFG changes by CSE passes and frees dominance info if required. This allows to remove corresponding workaround from STV pass. Does it look OK? Bootstrapped and regtested on x86-64-unknown-linux-gnu.