https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92980
--- Comment #10 from rguenther at suse dot de <rguenther at suse dot de> --- On January 16, 2020 3:55:18 AM GMT+01:00, wwwhhhyyy333 at gmail dot com <gcc-bugzi...@gcc.gnu.org> wrote: >https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92980 > >--- Comment #9 from Hongyu Wang <wwwhhhyyy333 at gmail dot com> --- >(In reply to Hongtao.liu from comment #6) >> New fail by removal > >> unix/-m32: gcc.dg/tree-ssa/copy-headers-5.c scan-tree-dump ch2 "is >now >> do-while loop" >> unix/-m32: gcc.dg/tree-ssa/copy-headers-5.c scan-tree-dump-times ch2 >" if " >> 3 >> unix/-m32: gcc.dg/tree-ssa/copy-headers-7.c scan-tree-dump ch2 "is >now >> do-while loop" >> unix/-m32: gcc.dg/tree-ssa/copy-headers-7.c scan-tree-dump-times ch2 >"Will >> duplicate bb" 3 >> unix/-m32: gcc.dg/tree-ssa/pr81744.c scan-tree-dump-times pcom >"Store-stores >> chain" 2 >> unix/-m64: gcc.dg/tree-ssa/copy-headers-5.c scan-tree-dump ch2 "is >now >> do-while loop" >> unix/-m64: gcc.dg/tree-ssa/copy-headers-5.c scan-tree-dump-times ch2 >" if " >> 3 >> unix/-m64: gcc.dg/tree-ssa/copy-headers-7.c scan-tree-dump ch2 "is >now >> do-while loop" >> unix/-m64: gcc.dg/tree-ssa/copy-headers-7.c scan-tree-dump-times ch2 >"Will >> duplicate bb" 3 > >I looked into this and found it is because of current judgement for >do-while >loop: > >---------- > <bb 2> [local count: 114863532]: > goto <bb 4>; [100.00%] > > <bb 3> [local count: 1014686025]: > i.0_1 = (unsigned int) i_11; > _2 = i.0_1 * 4; > _3 = a_14(D) + _2; > _4 = *_3; > _5 = i_11 + 1; > _6 = (unsigned int) _5; > _7 = _6 * 4; > _8 = a_14(D) + _7; > _9 = *_8; > if (_4 > _9) > goto <bb 5>; [5.50%] > else > goto <bb 6>; [94.50%] > > <bb 6> [local count: 958878294]: > > <bb 4> [local count: 1073741824]: > # i_11 = PHI <0(2), _5(6)> > _10 = n_13(D) + -1; > if (_10 > i_11) > goto <bb 3>; [94.50%] > else > goto <bb 5>; [5.50%] > > <bb 5> [local count: 114863532]: > # _12 = PHI <0(3), 1(4)> > return _12; > >This is regarded as a do-while loop because the it satisfies all the >condition >for do-while loop judgement: > >1) Loop latch is empty. >2) The latch has single predecessor. >3) The latch predecessor will exit loop. > >But I don't think the loop above is a do-while loop, so does the >conditions >need to be modified? It has two exits which makes it difficult Or impossible to make it truly do-while. But it's close enough and further rotating the loop doesn't make it better.
