search for: d45344

...lock/function. (You can sort of get around > the problem in simple cases by adding arbitrary limits to the number of you > scan, but it doesn't generalize well.) > > -Eli > > > On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: > > As we talked in https://reviews.llvm.org/D45344, the problem was dead > stores. And I know why :D There was just -instcombine pass. I forgot to do > -dse before -instcombine so this is why I did custom "store removal" code > there. > > I would like to finish malloc + llvm.memset folding. Yes, you told you > would like...

...u can sort of get around the problem in > simple cases by adding arbitrary limits to the number of you scan, > but it doesn't generalize well.) > > -Eli > > > On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >> As we talked in https://reviews.llvm.org/D45344 >> <https://reviews.llvm.org/D45344>, the problem was dead stores. >> And I know why :D There was just -instcombine pass. I forgot to >> do -dse before -instcombine so this is why I did custom "store >> removal" code there. >> >>...

Hello, I would like to find a way to do this removal properly. I found DSE and "eliminateNoopStore" can be useful for this thing. What I mean? int *test = malloc(15 * sizeof(int)); test[10] = 12; < ----- remove this store memset(test,0,sizeof(int) * 15); I already checked the function "eliminateNoopStore". Looks good, I think I would be to get the value ("A") we

...simple cases by adding arbitrary limits to the number of you >>> scan, but it doesn't generalize well.) >>> >>> -Eli >>> >>> >>> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>> >>> As we talked in https://reviews.llvm.org/D45344, the problem was dead >>> stores. And I know why :D There was just -instcombine pass. I forgot to do >>> -dse before -instcombine so this is why I did custom "store removal" code >>> there. >>> >>> I would like to finish malloc + llvm.memset fo...

...arbitrary limits to the number of you scan, but it >> doesn't generalize well.) >> >> -Eli >> >> >> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>> As we talked in https://reviews.llvm.org/D45344 >>> <https://reviews.llvm.org/D45344>, the problem was dead >>> stores. And I know why :D There was just >>> -instcombine pass. I forgot to do -dse before >>> -instcombine so this is why I did custom "sto...

...around >> the problem in simple cases by adding arbitrary limits to the number of you >> scan, but it doesn't generalize well.) >> >> -Eli >> >> >> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >> >> As we talked in https://reviews.llvm.org/D45344, the problem was dead >> stores. And I know why :D There was just -instcombine pass. I forgot to do >> -dse before -instcombine so this is why I did custom "store removal" code >> there. >> >> I would like to finish malloc + llvm.memset folding. Yes, you told...

...it doesn't generalize well.) >>>>>> >>>>>> -Eli >>>>>> >>>>>> >>>>>> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>>>>> >>>>>> As we talked in https://reviews.llvm.org/D45344, the problem was >>>>>> dead stores. And I know why :D There was just -instcombine pass. I forgot >>>>>> to do -dse before -instcombine so this is why I did custom "store removal" >>>>>> code there. >>>>>> >>&gt...

...limits to the number of you >>>> scan, but it doesn't generalize well.) >>>> >>>> -Eli >>>> >>>> >>>> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>>> >>>> As we talked in https://reviews.llvm.org/D45344, the problem was dead >>>> stores. And I know why :D There was just -instcombine pass. I forgot to do >>>> -dse before -instcombine so this is why I did custom "store removal" code >>>> there. >>>> >>>> I would like to finish mal...

...ll.) >>>>>>> >>>>>>> -Eli >>>>>>> >>>>>>> >>>>>>> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>>>>>> >>>>>>> As we talked in https://reviews.llvm.org/D45344, the problem was >>>>>>> dead stores. And I know why :D There was just -instcombine pass. I forgot >>>>>>> to do -dse before -instcombine so this is why I did custom "store removal" >>>>>>> code there. >>>>>>...

...-Eli >>> >>> >>> On 5/17/2018 12:17 PM, Dávid Bolvanský >>> wrote: >>>> As we talked in >>>> https://reviews.llvm.org/D45344 >>>> <https://reviews.llvm.org/D45344>, the >>>> problem was dead stores. And I know why >>>> :D There was just -instcombine pass. I >>>>...

...;>>>> scan, but it doesn't generalize well.) >>>>> >>>>> -Eli >>>>> >>>>> >>>>> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>>>> >>>>> As we talked in https://reviews.llvm.org/D45344, the problem was dead >>>>> stores. And I know why :D There was just -instcombine pass. I forgot to do >>>>> -dse before -instcombine so this is why I did custom "store removal" code >>>>> there. >>>>> >>>>> I woul...

...;>> >>>>>>>> -Eli >>>>>>>> >>>>>>>> >>>>>>>> On 5/17/2018 12:17 PM, Dávid Bolvanský wrote: >>>>>>>> >>>>>>>> As we talked in https://reviews.llvm.org/D45344, the problem was >>>>>>>> dead stores. And I know why :D There was just -instcombine pass. I forgot >>>>>>>> to do -dse before -instcombine so this is why I did custom "store removal" >>>>>>>> code there. >>>...