> On 10 Jul 2026, at 16:18, Alfie Richards <[email protected]> wrote:
> 
> On 10/07/2026 15:09, Kyrylo Tkachov wrote:
>>> On 10 Jul 2026, at 15:57, Alfie Richards <[email protected]> wrote:
>>> 
>>> On 10/07/2026 11:42, Kyrylo Tkachov wrote:
>>>>> On 9 Jul 2026, at 14:36, Kyrylo Tkachov <[email protected]> wrote:
>>>>> 
>>>>> Hi Alfie,
>>>>> 
>>>>>> On 9 Jul 2026, at 10:24, Alfie Richards <[email protected]> wrote:
>>>>>> 
>>>>>> On 06/07/2026 09:40, Kyrylo Tkachov wrote:
>>>>>>>> On 3 Jul 2026, at 13:56, Kyrylo Tkachov <[email protected]> wrote:
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>>> On 2 Jul 2026, at 10:56, Richard Biener <[email protected]> wrote:
>>>>>>>>> 
>>>>>>>>> On Thu, 2 Jul 2026, [email protected] wrote:
>>>>>>>>> 
>>>>>>>>>> From: Kyrylo Tkachov <[email protected]>
>>>>>>>>>> 
>>>>>>>>>> pass_split_paths duplicates the join block of an IF-THEN-ELSE that 
>>>>>>>>>> feeds a
>>>>>>>>>> loop latch, splitting the two paths to the backedge.  It runs only 
>>>>>>>>>> at -O3.
>>>>>>>>>> In practice it interacts badly with later optimizations: it 
>>>>>>>>>> duplicates the
>>>>>>>>>> loop body before loads have been commoned and before if-conversion 
>>>>>>>>>> runs, so
>>>>>>>>>> it can block both loop unrolling (PR120892) and if-conversion of the
>>>>>>>>>> duplicated diamond, while its own heuristic already declines about 
>>>>>>>>>> half of
>>>>>>>>>> all candidate blocks, most often to avoid spoiling if-conversion.
>>>>>>>>>> 
>>>>>>>>>> Remove the pass and deprecate the -fsplit-paths option.  The option 
>>>>>>>>>> is kept
>>>>>>>>>> accepted for backward compatibility via the Ignore flag and now does 
>>>>>>>>>> nothing,
>>>>>>>>>> matching how other optimization options have been retired (for 
>>>>>>>>>> example
>>>>>>>>>> -ftree-lrs).  param_max_jump_thread_duplication_stmts is retained as 
>>>>>>>>>> it is
>>>>>>>>>> shared with the jump-threading passes.
>>>>>>>>>> 
>>>>>>>>>> Statistics from the pass on SPEC CPU 2026 (intrate + fprate, counted 
>>>>>>>>>> from the
>>>>>>>>>> split-paths dump):
>>>>>>>>>> 
>>>>>>>>>>                  candidates   splits   declined   to protect 
>>>>>>>>>> if-conversion
>>>>>>>>>> -O3                  122894    62050     60844           37166
>>>>>>>>>> -O3 -flto=auto        52423    21257     31166           21822
>>>>>>>>>> 
>>>>>>>>>> The pass splits about half of the blocks it considers and declines 
>>>>>>>>>> the rest,
>>>>>>>>>> most often to avoid spoiling if-conversion.  The duplication grows 
>>>>>>>>>> .text by
>>>>>>>>>> 0.32% at -O3 and 0.24% at -O3 -flto=auto.
>>>>>>>>>> 
>>>>>>>>>> Andrea and Jeff indicated in PR120892 that removing -fsplit-paths 
>>>>>>>>>> may be
>>>>>>>>>> the way to go there.
>>>>>>>>>> 
>>>>>>>>>> -fsplit-paths also complicates the control-flow and defeats the
>>>>>>>>>> load-commoning necessary to get good if-conversion of the hot loop 
>>>>>>>>>> from
>>>>>>>>>> Snappy from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125557#c13 .
>>>>>>>>>> 
>>>>>>>>>> Bootstrapped and tested on aarch64-none-linux-gnu and x86_64-linux.
>>>>>>>>> 
>>>>>>>>> OK.
>>>>>>>>> 
>>>>>>>>> Please leave others a day or so to chime in.
>>>>>>>> 
>>>>>>>> Thanks, I’ll push next wee once the USA have had their holidays.
>>>>>>>> Kyrill
>>>>>>> Pushed now as g:5c23bb074af23f00dd3fe1745b9dd99245fa4bba
>>>>>> Hi,
>>>>>> 
>>>>>> Just to note we see about a 3% regression on 
>>>>>> spec2006.intrate.462.libquantum from this change on Neoverse-V1.
>>>>> 
>>>>> I’m having a look. I don’t have access to a Neoverse V1 but I’ll try to 
>>>>> reproduce it on the machines that I have access to and see if I can raise 
>>>>> a GCC PR to record whatever missed optimisation this may be exposing so 
>>>>> that we can decide if we want to implement it in some other way.
>>>>> 
>>>> So I did reproduce the slowdown, but it looks like an unfortunate code 
>>>> alignment change. libquantum has some very hot paths but the split paths 
>>>> pass only triggered on cold paths/functions, but due to code layout the 
>>>> hot BBs got a different suboptimal alignment that caused the overall 
>>>> slowdown. So I don’t think we have anything split-paths-related to do here.
>>>> Alignment for big AArch64 CPUs is something that has given me headaches 
>>>> recently and I think we should be looking at things like stricter 
>>>> alignment on small hot loops and better function alignment for hot 
>>>> functions (ignoring skip in those cases), but those are separate 
>>>> discussions.
>>>> Thanks,
>>>> Kyrill
>>> Hi Kyrill,
>>> 
>>> Our analysis showed a larger code generation difference.
>>> 
>>> Speicifcally in quantum_toffoli near the end of the loop we see the 
>>> following change:
>>> 
>>> NEW:
>>> 
>>> //.L1
>>> cmp w2, w1
>>> b.le 404a40 <quantum_toffoli+0x174> // Out of loop
>>> 
>>> //.L2
>>> ldr x0, [x19, #16]
>>> add x0, x0, w1, uxtw #4
>>> add x1, x1, #0x1
>>> ldr x3, [x0, #8]
>>> bics xzr, x5, x3
>>> b.ne 404a0c <quantum_toffoli+0x140>  // b.any // .L1
>>> //.L3
>>> eor x3, x4, x3
>>> str x3, [x0, #8]
>>> ldr w2, [x19, #4]
>>> cmp w2, w1 b.gt 404a14 <quantum_toffoli+0x148> // .L2
>>> 
>>> 
>>> 
>>> OLD:
>>> 
>>> //.L1
>>> add x3, x5, w0, uxtw #4
>>> ldr x1, [x3, #8]
>>> bics xzr, x19, x1
>>> b.eq 404a9c <quantum_toffoli+0x17c>  // b.none // Out of loop
>>> //.L2
>>> add x0, x0, #0x1
>>> cmp w2, w0
>>> b.gt 404a68 <quantum_toffoli+0x148> // .L1
>>> 
>>> (Roughly, comments are my own)
>>> 
>>> We think this is the primary change causing this regression due to the 
>>> extra load in the loop?
>>> 
>>> (credit to Tamar for most of the analysis here)
>> Thanks for this. Can you also share the options used?
>> Kyrill
> 
> I believe they are:
> 
> -mcpu=native -Ofast -fomit-frame-pointer -flto=auto --param 
> ipa-cp-eval-threshold=1 --param ipa-cp-unit-growth=80

Ok, I think the key option here is -fno-strict-aliasing and the reduced test 
case would be:
https://godbolt.org/z/x8dP1P75e
-fsplit-paths legitimately removes a pointer-chasing load outside the hot loop 
(the x4 load).

Any opinions from others on how to approach this? Shall I file this case as a 
missed-optimization PR so we can discuss how else we may handle this case? Or 
we can reinstate -fsplit-paths as off-by-default and tell people who want 
libquantum performance to use -fsplit-paths explicitly in their options?

Thanks,
Kyrill  

> 
>>> 
>>> KR,
>>> Alfie
>>> 
>>>>> Thanks,
>>>>> Kyrill
>>>>> 
>>>>>> 
>>>>>> Thanks,
>>>>>> Alfie
>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> Thanks,
>>>>>>>>> Richard.
>>>>>>>>> 
>>>>>>>>>> gcc/ChangeLog:
>>>>>>>>>> 
>>>>>>>>>> PR tree-optimization/120892
>>>>>>>>>> * gimple-ssa-split-paths.cc: Remove.
>>>>>>>>>> * passes.def (pass_split_paths): Remove.
>>>>>>>>>> * tree-pass.h (make_pass_split_paths): Remove.
>>>>>>>>>> * Makefile.in (OBJS): Remove gimple-ssa-split-paths.o.
>>>>>>>>>> * timevar.def (TV_SPLIT_PATHS): Remove.
>>>>>>>>>> * opts.cc (default_options_table): Remove the OPT_LEVELS_3_PLUS entry
>>>>>>>>>> for OPT_fsplit_paths.
>>>>>>>>>> * common.opt (fsplit-paths): Make it a deprecated no-op using Ignore.
>>>>>>>>>> * doc/invoke.texi (-fsplit-paths): Document as deprecated and remove 
>>>>>>>>>> it
>>>>>>>>>> from the option summary and the -O3 list.
>>>>>>>>>> 
>>>>>>>>>> gcc/testsuite/ChangeLog:
>>>>>>>>>> 
>>>>>>>>>> PR tree-optimization/120892
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-1.c: Move to...
>>>>>>>>>> * gcc.c-torture/execute/split-path-1.c: ...here.  Adjust to a plain
>>>>>>>>>> compile and run test without the split-paths dump scan.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-2.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-2.c: ...here.  Adjust to a
>>>>>>>>>> compile-only test without the split-paths dump scan.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-3.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-3.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-4.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-4.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-5.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-5.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-6.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-6.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-7.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-7.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-8.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-8.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-9.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-9.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-10.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-10.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-11.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-11.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-12.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-12.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/split-path-13.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/split-path-13.c: ...here.  Likewise.
>>>>>>>>>> * gcc.dg/tree-ssa/pr69270.c: Move to...
>>>>>>>>>> * gcc.c-torture/compile/pr69270.c: ...here.  Likewise.
>>>>>>>>>> * g++.dg/tree-ssa/pr88797.C: Move to...
>>>>>>>>>> * g++.dg/torture/pr88797.C: ...here.  Adjust to a compile-only test.
>>>>>>>>>> * gcc.target/i386/pr106450.c: Remove -fsplit-paths from dg-options.
>>>>>>>>>> 
>>>>>>>>>> Signed-off-by: Kyrylo Tkachov <[email protected]>
>>>>>>>>>> ---
>>>>>>>>>> gcc/Makefile.in                               |   1 -
>>>>>>>>>> gcc/common.opt                                |   4 +-
>>>>>>>>>> gcc/doc/invoke.texi                           |   8 +-
>>>>>>>>>> gcc/gimple-ssa-split-paths.cc                 | 545 
>>>>>>>>>> ------------------
>>>>>>>>>> gcc/opts.cc                                   |   1 -
>>>>>>>>>> gcc/passes.def                                |   1 -
>>>>>>>>>> gcc/testsuite/g++.dg/torture/pr88797.C        |  13 +
>>>>>>>>>> gcc/testsuite/g++.dg/tree-ssa/pr88797.C       |  16 -
>>>>>>>>>> .../compile}/pr69270.c                        |   7 -
>>>>>>>>>> .../compile}/split-path-10.c                  |   3 -
>>>>>>>>>> .../compile}/split-path-11.c                  |   3 -
>>>>>>>>>> .../compile}/split-path-12.c                  |   3 -
>>>>>>>>>> .../compile}/split-path-13.c                  |   3 -
>>>>>>>>>> .../gcc.c-torture/compile/split-path-2.c      |  18 +
>>>>>>>>>> .../compile}/split-path-3.c                   |   3 -
>>>>>>>>>> .../compile}/split-path-4.c                   |   3 -
>>>>>>>>>> .../compile}/split-path-5.c                   |   3 -
>>>>>>>>>> .../compile}/split-path-6.c                   |   3 -
>>>>>>>>>> .../compile}/split-path-7.c                   |   3 -
>>>>>>>>>> .../gcc.c-torture/compile/split-path-8.c      |  11 +
>>>>>>>>>> .../gcc.c-torture/compile/split-path-9.c      |  14 +
>>>>>>>>>> .../execute}/split-path-1.c                   |   3 -
>>>>>>>>>> gcc/testsuite/gcc.dg/tree-ssa/split-path-2.c  |  21 -
>>>>>>>>>> gcc/testsuite/gcc.dg/tree-ssa/split-path-8.c  |  14 -
>>>>>>>>>> gcc/testsuite/gcc.dg/tree-ssa/split-path-9.c  |  17 -
>>>>>>>>>> gcc/testsuite/gcc.target/i386/pr106450.c      |   2 +-
>>>>>>>>>> gcc/timevar.def                               |   1 -
>>>>>>>>>> gcc/tree-pass.h                               |   1 -
>>>>>>>>>> 28 files changed, 62 insertions(+), 663 deletions(-)
>>>>>>>>>> delete mode 100644 gcc/gimple-ssa-split-paths.cc
>>>>>>>>>> create mode 100644 gcc/testsuite/g++.dg/torture/pr88797.C
>>>>>>>>>> delete mode 100644 gcc/testsuite/g++.dg/tree-ssa/pr88797.C
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/pr69270.c (57%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-10.c (81%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-11.c (55%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-12.c (70%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-13.c (62%)
>>>>>>>>>> create mode 100644 gcc/testsuite/gcc.c-torture/compile/split-path-2.c
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-3.c (93%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-4.c (62%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-5.c (84%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-6.c (82%)
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/compile}/split-path-7.c (87%)
>>>>>>>>>> create mode 100644 gcc/testsuite/gcc.c-torture/compile/split-path-8.c
>>>>>>>>>> create mode 100644 gcc/testsuite/gcc.c-torture/compile/split-path-9.c
>>>>>>>>>> rename gcc/testsuite/{gcc.dg/tree-ssa => 
>>>>>>>>>> gcc.c-torture/execute}/split-path-1.c (83%)
>>>>>>>>>> delete mode 100644 gcc/testsuite/gcc.dg/tree-ssa/split-path-2.c
>>>>>>>>>> delete mode 100644 gcc/testsuite/gcc.dg/tree-ssa/split-path-8.c
>>>>>>>>>> delete mode 100644 gcc/testsuite/gcc.dg/tree-ssa/split-path-9.c
>>>>>>>>>> 
>>>>>>>>>> diff --git a/gcc/Makefile.in b/gcc/Makefile.in
>>>>>>>>>> index 536e2faaaf8..06e711094ef 100644
>>>>>>>>>> --- a/gcc/Makefile.in
>>>>>>>>>> +++ b/gcc/Makefile.in
>>>>>>>>>> @@ -1542,7 +1542,6 @@ OBJS = \
>>>>>>>>>> gimple-ssa-isolate-paths.o \
>>>>>>>>>> gimple-ssa-nonnull-compare.o \
>>>>>>>>>> gimple-ssa-sccopy.o \
>>>>>>>>>> - gimple-ssa-split-paths.o \
>>>>>>>>>> gimple-ssa-store-merging.o \
>>>>>>>>>> gimple-ssa-strength-reduction.o \
>>>>>>>>>> gimple-ssa-sprintf.o \
>>>>>>>>>> diff --git a/gcc/common.opt b/gcc/common.opt
>>>>>>>>>> index 218dddf5dfe..0055578ccf6 100644
>>>>>>>>>> --- a/gcc/common.opt
>>>>>>>>>> +++ b/gcc/common.opt
>>>>>>>>>> @@ -3388,8 +3388,8 @@ Common Var(flag_tree_vrp) Init(0) Optimization
>>>>>>>>>> Perform Value Range Propagation on trees.
>>>>>>>>>> 
>>>>>>>>>> fsplit-paths
>>>>>>>>>> -Common Var(flag_split_paths) Init(0) Optimization
>>>>>>>>>> -Split paths leading to loop backedges.
>>>>>>>>>> +Common Ignore
>>>>>>>>>> +Does nothing.  Preserved for backward compatibility.
>>>>>>>>>> 
>>>>>>>>>> funconstrained-commons
>>>>>>>>>> Common Var(flag_unconstrained_commons) Optimization
>>>>>>>>>> diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
>>>>>>>>>> index 8da5f03ccbd..932def96d8f 100644
>>>>>>>>>> --- a/gcc/doc/invoke.texi
>>>>>>>>>> +++ b/gcc/doc/invoke.texi
>>>>>>>>>> @@ -676,7 +676,7 @@ Objective-C and Objective-C++ Dialects}.
>>>>>>>>>> -fsemantic-interposition  -fshrink-wrap  -fshrink-wrap-separate
>>>>>>>>>> -fsignaling-nans
>>>>>>>>>> -fsingle-precision-constant  -fsplit-ivs-in-unroller  -fsplit-loops
>>>>>>>>>> --fspeculatively-call-stored-functions  -fsplit-paths
>>>>>>>>>> +-fspeculatively-call-stored-functions
>>>>>>>>>> -fsplit-wide-types  -fsplit-wide-types-early  -fssa-backprop  
>>>>>>>>>> -fssa-phiopt
>>>>>>>>>> -fstdarg-opt  -fstore-merging  -fstrict-aliasing 
>>>>>>>>>> -fipa-strict-aliasing
>>>>>>>>>> -fthread-jumps  -ftracer  -ftree-bit-ccp
>>>>>>>>>> @@ -13619,7 +13619,6 @@ by @option{-O2} and also turns on the 
>>>>>>>>>> following optimization flags:
>>>>>>>>>> -fpeel-loops
>>>>>>>>>> -fpredictive-commoning
>>>>>>>>>> -fsplit-loops
>>>>>>>>>> --fsplit-paths
>>>>>>>>>> -ftree-loop-distribution
>>>>>>>>>> -ftree-partial-pre
>>>>>>>>>> -funswitch-loops
>>>>>>>>>> @@ -15462,9 +15461,8 @@ enabled.
>>>>>>>>>> @opindex fsplit-paths
>>>>>>>>>> @opindex fno-split-paths
>>>>>>>>>> @item -fsplit-paths
>>>>>>>>>> -Split paths leading to loop backedges.  This can improve dead code
>>>>>>>>>> -elimination and common subexpression elimination.  This is enabled 
>>>>>>>>>> by
>>>>>>>>>> -default at @option{-O3} and above.
>>>>>>>>>> +This option is deprecated and does nothing.  It is accepted only for
>>>>>>>>>> +backward compatibility.
>>>>>>>>>> 
>>>>>>>>>> @opindex fsplit-ivs-in-unroller
>>>>>>>>>> @opindex fno-split-ivs-in-unroller
>>>>>>>>>> diff --git a/gcc/gimple-ssa-split-paths.cc 
>>>>>>>>>> b/gcc/gimple-ssa-split-paths.cc
>>>>>>>>>> deleted file mode 100644
>>>>>>>>>> index 171a581e8f8..00000000000
>>>>>>>>>> --- a/gcc/gimple-ssa-split-paths.cc
>>>>>>>>>> +++ /dev/null
>>>>>>>>>> @@ -1,545 +0,0 @@
>>>>>>>>>> -/* Support routines for Splitting Paths to loop backedges
>>>>>>>>>> -   Copyright (C) 2015-2026 Free Software Foundation, Inc.
>>>>>>>>>> -   Contributed by Ajit Kumar Agarwal <[email protected]>.
>>>>>>>>>> -
>>>>>>>>>> - This file is part of GCC.
>>>>>>>>>> -
>>>>>>>>>> - GCC is free software; you can redistribute it and/or modify
>>>>>>>>>> - it under the terms of the GNU General Public License as published 
>>>>>>>>>> by
>>>>>>>>>> - the Free Software Foundation; either version 3, or (at your option)
>>>>>>>>>> - any later version.
>>>>>>>>>> -
>>>>>>>>>> -GCC is distributed in the hope that it will be useful,
>>>>>>>>>> -but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>>>>>>>> -MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>>>>>>>>>> -GNU General Public License for more details.
>>>>>>>>>> -
>>>>>>>>>> -You should have received a copy of the GNU General Public License
>>>>>>>>>> -along with GCC; see the file COPYING3.  If not see
>>>>>>>>>> -<http://www.gnu.org/licenses/>.  */
>>>>>>>>>> -
>>>>>>>>>> -#include "config.h"
>>>>>>>>>> -#include "system.h"
>>>>>>>>>> -#include "coretypes.h"
>>>>>>>>>> -#include "backend.h"
>>>>>>>>>> -#include "tree.h"
>>>>>>>>>> -#include "gimple.h"
>>>>>>>>>> -#include "tree-pass.h"
>>>>>>>>>> -#include "tree-cfg.h"
>>>>>>>>>> -#include "cfganal.h"
>>>>>>>>>> -#include "cfgloop.h"
>>>>>>>>>> -#include "gimple-iterator.h"
>>>>>>>>>> -#include "tracer.h"
>>>>>>>>>> -#include "predict.h"
>>>>>>>>>> -#include "gimple-ssa.h"
>>>>>>>>>> -#include "tree-phinodes.h"
>>>>>>>>>> -#include "ssa-iterators.h"
>>>>>>>>>> -#include "fold-const.h"
>>>>>>>>>> -#include "cfghooks.h"
>>>>>>>>>> -
>>>>>>>>>> -/* Given LATCH, the latch block in a loop, see if the shape of the
>>>>>>>>>> -   path reaching LATCH is suitable for being split by duplication.
>>>>>>>>>> -   If so, return the block that will be duplicated into its 
>>>>>>>>>> predecessor
>>>>>>>>>> -   paths.  Else return NULL.  */
>>>>>>>>>> -
>>>>>>>>>> -static basic_block
>>>>>>>>>> -find_block_to_duplicate_for_splitting_paths (basic_block latch)
>>>>>>>>>> -{
>>>>>>>>>> -  /* We should have simple latches at this point.  So the latch 
>>>>>>>>>> should
>>>>>>>>>> -     have a single successor.  This implies the predecessor of the 
>>>>>>>>>> latch
>>>>>>>>>> -     likely has the loop exit.  And it's that predecessor we're most
>>>>>>>>>> -     interested in. To keep things simple, we're going to require 
>>>>>>>>>> that
>>>>>>>>>> -     the latch have a single predecessor too.  */
>>>>>>>>>> -  if (single_succ_p (latch) && single_pred_p (latch))
>>>>>>>>>> -    {
>>>>>>>>>> -      basic_block bb = get_immediate_dominator (CDI_DOMINATORS, 
>>>>>>>>>> latch);
>>>>>>>>>> -      gcc_assert (single_pred_edge (latch)->src == bb);
>>>>>>>>>> -
>>>>>>>>>> -      /* If BB has been marked as not to be duplicated, then honor 
>>>>>>>>>> that
>>>>>>>>>> -  request.  */
>>>>>>>>>> -      if (ignore_bb_p (bb))
>>>>>>>>>> - return NULL;
>>>>>>>>>> -
>>>>>>>>>> -      gimple *last = gsi_stmt (gsi_last_nondebug_bb (bb));
>>>>>>>>>> -      /* The immediate dominator of the latch must end in a 
>>>>>>>>>> conditional.  */
>>>>>>>>>> -      if (!last || gimple_code (last) != GIMPLE_COND)
>>>>>>>>>> - return NULL;
>>>>>>>>>> -
>>>>>>>>>> -      /* We're hoping that BB is a join point for an IF-THEN-ELSE 
>>>>>>>>>> diamond
>>>>>>>>>> -  region.  Verify that it is.
>>>>>>>>>> -
>>>>>>>>>> -  First, verify that BB has two predecessors (each arm of the
>>>>>>>>>> -  IF-THEN-ELSE) and two successors (the latch and exit) and that
>>>>>>>>>> -  all edges are normal.  */
>>>>>>>>>> -      if (EDGE_COUNT (bb->preds) == 2
>>>>>>>>>> -   && !(EDGE_PRED (bb, 0)->flags & EDGE_COMPLEX)
>>>>>>>>>> -   && !(EDGE_PRED (bb, 1)->flags & EDGE_COMPLEX)
>>>>>>>>>> -   && EDGE_COUNT (bb->succs) == 2
>>>>>>>>>> -   && !(EDGE_SUCC (bb, 0)->flags & EDGE_COMPLEX)
>>>>>>>>>> -   && !(EDGE_SUCC (bb, 1)->flags & EDGE_COMPLEX))
>>>>>>>>>> - {
>>>>>>>>>> -   /* Now verify that BB's immediate dominator ends in a
>>>>>>>>>> -      conditional as well.  */
>>>>>>>>>> -   basic_block bb_idom = get_immediate_dominator (CDI_DOMINATORS, 
>>>>>>>>>> bb);
>>>>>>>>>> -   gimple *last = gsi_stmt (gsi_last_nondebug_bb (bb_idom));
>>>>>>>>>> -   if (!last || gimple_code (last) != GIMPLE_COND)
>>>>>>>>>> -     return NULL;
>>>>>>>>>> -
>>>>>>>>>> -   /* And that BB's immediate dominator's successors are the
>>>>>>>>>> -      predecessors of BB or BB itself.  */
>>>>>>>>>> -   if (!(EDGE_PRED (bb, 0)->src == bb_idom
>>>>>>>>>> - || find_edge (bb_idom, EDGE_PRED (bb, 0)->src))
>>>>>>>>>> -       || !(EDGE_PRED (bb, 1)->src == bb_idom
>>>>>>>>>> -    || find_edge (bb_idom, EDGE_PRED (bb, 1)->src)))
>>>>>>>>>> -     return NULL;
>>>>>>>>>> -
>>>>>>>>>> -   /* And that the predecessors of BB each have a single successor
>>>>>>>>>> -      or are BB's immediate domiator itself.  */
>>>>>>>>>> -   if (!(EDGE_PRED (bb, 0)->src == bb_idom
>>>>>>>>>> - || single_succ_p (EDGE_PRED (bb, 0)->src))
>>>>>>>>>> -       || !(EDGE_PRED (bb, 1)->src == bb_idom
>>>>>>>>>> -    || single_succ_p (EDGE_PRED (bb, 1)->src)))
>>>>>>>>>> -     return NULL;
>>>>>>>>>> -
>>>>>>>>>> -   /* So at this point we have a simple diamond for an IF-THEN-ELSE
>>>>>>>>>> -      construct starting at BB_IDOM, with a join point at BB.  BB
>>>>>>>>>> -      pass control outside the loop or to the loop latch.
>>>>>>>>>> -
>>>>>>>>>> -      We're going to want to create two duplicates of BB, one for
>>>>>>>>>> -      each successor of BB_IDOM.  */
>>>>>>>>>> -   return bb;
>>>>>>>>>> - }
>>>>>>>>>> -    }
>>>>>>>>>> -  return NULL;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* Return the number of non-debug statements in a block.  */
>>>>>>>>>> -static unsigned int
>>>>>>>>>> -count_stmts_in_block (basic_block bb)
>>>>>>>>>> -{
>>>>>>>>>> -  gimple_stmt_iterator gsi;
>>>>>>>>>> -  unsigned int num_stmts = 0;
>>>>>>>>>> -
>>>>>>>>>> -  for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
>>>>>>>>>> -    {
>>>>>>>>>> -      gimple *stmt = gsi_stmt (gsi);
>>>>>>>>>> -      if (!is_gimple_debug (stmt))
>>>>>>>>>> - num_stmts++;
>>>>>>>>>> -    }
>>>>>>>>>> -  return num_stmts;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* Return TRUE if CODE represents a tree code that is not likely to
>>>>>>>>>> -   be easily if-convertible because it likely expands into multiple
>>>>>>>>>> -   insns, FALSE otherwise.  */
>>>>>>>>>> -static bool
>>>>>>>>>> -poor_ifcvt_candidate_code (enum tree_code code)
>>>>>>>>>> -{
>>>>>>>>>> -  return (code == MIN_EXPR
>>>>>>>>>> -   || code == MAX_EXPR
>>>>>>>>>> -   || code == ABS_EXPR
>>>>>>>>>> -   || code == COND_EXPR);
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* Return TRUE if PRED of BB is an poor ifcvt candidate. */
>>>>>>>>>> -static bool
>>>>>>>>>> -poor_ifcvt_pred (basic_block pred, basic_block bb)
>>>>>>>>>> -{
>>>>>>>>>> -  /* If the edge count of the pred is not 1, then
>>>>>>>>>> -     this is the predecessor from the if rather
>>>>>>>>>> -     than middle one. */
>>>>>>>>>> -  if (EDGE_COUNT (pred->succs) != 1)
>>>>>>>>>> -    return false;
>>>>>>>>>> -
>>>>>>>>>> -  /* Empty middle bb are never a poor ifcvt candidate. */
>>>>>>>>>> -  if (empty_block_p (pred))
>>>>>>>>>> -    return false;
>>>>>>>>>> -  /* If BB's predecessors are single statement blocks where
>>>>>>>>>> -     the output of that statement feed the same PHI in BB,
>>>>>>>>>> -     it an ifcvt candidate. */
>>>>>>>>>> -  gimple *stmt = last_and_only_stmt (pred);
>>>>>>>>>> -  if (!stmt || gimple_code (stmt) != GIMPLE_ASSIGN)
>>>>>>>>>> -    return true;
>>>>>>>>>> -
>>>>>>>>>> -  /* If the statement could trap, then this is a poor ifcvt 
>>>>>>>>>> candidate. */
>>>>>>>>>> -  if (gimple_could_trap_p (stmt))
>>>>>>>>>> -    return true;
>>>>>>>>>> -
>>>>>>>>>> -  tree_code code = gimple_assign_rhs_code (stmt);
>>>>>>>>>> -  if (poor_ifcvt_candidate_code (code))
>>>>>>>>>> -    return true;
>>>>>>>>>> -  tree lhs = gimple_assign_lhs (stmt);
>>>>>>>>>> -  gimple_stmt_iterator gsi;
>>>>>>>>>> -  for (gsi = gsi_start_phis (bb); !gsi_end_p (gsi); gsi_next (&gsi))
>>>>>>>>>> -    {
>>>>>>>>>> -      gimple *phi = gsi_stmt (gsi);
>>>>>>>>>> -      if (gimple_phi_arg_def (phi, 0) == lhs
>>>>>>>>>> -   || gimple_phi_arg_def (phi, 1) == lhs)
>>>>>>>>>> - return false;
>>>>>>>>>> -    }
>>>>>>>>>> -  return true;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* Return TRUE if BB is a reasonable block to duplicate by examining
>>>>>>>>>> -   its size, false otherwise.  BB will always be a loop latch block.
>>>>>>>>>> -
>>>>>>>>>> -   Things to consider:
>>>>>>>>>> -
>>>>>>>>>> -     We do not want to spoil if-conversion if at all possible.
>>>>>>>>>> -
>>>>>>>>>> -     Most of the benefit seems to be from eliminating the 
>>>>>>>>>> unconditional
>>>>>>>>>> -     jump rather than CSE/DCE opportunities.  So favor duplicating
>>>>>>>>>> -     small latches.  A latch with just a conditional branch is 
>>>>>>>>>> ideal.
>>>>>>>>>> -
>>>>>>>>>> -     CSE/DCE opportunities crop up when statements from the 
>>>>>>>>>> predecessors
>>>>>>>>>> -     feed statements in the latch and allow statements in the latch 
>>>>>>>>>> to
>>>>>>>>>> -     simplify.  */
>>>>>>>>>> -
>>>>>>>>>> -static bool
>>>>>>>>>> -is_feasible_trace (basic_block bb)
>>>>>>>>>> -{
>>>>>>>>>> -  basic_block pred1 = EDGE_PRED (bb, 0)->src;
>>>>>>>>>> -  basic_block pred2 = EDGE_PRED (bb, 1)->src;
>>>>>>>>>> -  int num_stmts_in_join = count_stmts_in_block (bb);
>>>>>>>>>> -  int num_stmts_in_pred1
>>>>>>>>>> -    = EDGE_COUNT (pred1->succs) == 1 ? count_stmts_in_block (pred1) 
>>>>>>>>>> : 0;
>>>>>>>>>> -  int num_stmts_in_pred2
>>>>>>>>>> -    = EDGE_COUNT (pred2->succs) == 1 ? count_stmts_in_block (pred2) 
>>>>>>>>>> : 0;
>>>>>>>>>> -
>>>>>>>>>> -  /* Upper Hard limit on the number statements to copy.  */
>>>>>>>>>> -  if (num_stmts_in_join
>>>>>>>>>> -      >= param_max_jump_thread_duplication_stmts)
>>>>>>>>>> -    {
>>>>>>>>>> -      if (dump_file && (dump_flags & TDF_DETAILS))
>>>>>>>>>> - fprintf (dump_file,
>>>>>>>>>> -  "Duplicating block %d would duplicate "
>>>>>>>>>> -  "too many statements: %d >= %d\n",
>>>>>>>>>> -  bb->index, num_stmts_in_join,
>>>>>>>>>> -  param_max_jump_thread_duplication_stmts);
>>>>>>>>>> -      return false;
>>>>>>>>>> -    }
>>>>>>>>>> -
>>>>>>>>>> -  /* This is meant to catch cases that are likely opportunities for
>>>>>>>>>> -     if-conversion.  */
>>>>>>>>>> -  if (num_stmts_in_pred1 <= 1 && num_stmts_in_pred2 <= 1)
>>>>>>>>>> -    {
>>>>>>>>>> -      int num_phis = 0;
>>>>>>>>>> -      /* The max number of PHIs that should be considered for an 
>>>>>>>>>> ifcvt
>>>>>>>>>> -  candidate.  */
>>>>>>>>>> -      const int max_num_phis = 3;
>>>>>>>>>> -      for (gphi_iterator si = gsi_start_phis (bb); ! gsi_end_p (si);
>>>>>>>>>> -   gsi_next (&si))
>>>>>>>>>> - {
>>>>>>>>>> -   num_phis++;
>>>>>>>>>> -   if (num_phis > max_num_phis)
>>>>>>>>>> -     break;
>>>>>>>>>> - }
>>>>>>>>>> -      if (num_phis <= max_num_phis
>>>>>>>>>> -   && !poor_ifcvt_pred (pred1, bb)
>>>>>>>>>> -   && !poor_ifcvt_pred (pred2, bb))
>>>>>>>>>> - {
>>>>>>>>>> -   if (dump_file && (dump_flags & TDF_DETAILS))
>>>>>>>>>> -     fprintf (dump_file,
>>>>>>>>>> -      "Block %d appears to be a join point for "
>>>>>>>>>> -      "if-convertible bbs.\n",
>>>>>>>>>> -      bb->index);
>>>>>>>>>> -   return false;
>>>>>>>>>> - }
>>>>>>>>>> -    }
>>>>>>>>>> -
>>>>>>>>>> -  /* If the joiner has no PHIs with useful uses there is zero chance
>>>>>>>>>> -     of CSE/DCE/jump-threading possibilities exposed by duplicating 
>>>>>>>>>> it.  */
>>>>>>>>>> -  bool found_useful_phi = false;
>>>>>>>>>> -  for (gphi_iterator si = gsi_start_phis (bb); ! gsi_end_p (si);
>>>>>>>>>> -       gsi_next (&si))
>>>>>>>>>> -    {
>>>>>>>>>> -      gphi *phi = si.phi ();
>>>>>>>>>> -      use_operand_p use_p;
>>>>>>>>>> -      imm_use_iterator iter;
>>>>>>>>>> -      FOR_EACH_IMM_USE_FAST (use_p, iter, gimple_phi_result (phi))
>>>>>>>>>> - {
>>>>>>>>>> -   gimple *stmt = USE_STMT (use_p);
>>>>>>>>>> -   if (is_gimple_debug (stmt))
>>>>>>>>>> -     continue;
>>>>>>>>>> -   /* If there's a use in the joiner this might be a CSE/DCE
>>>>>>>>>> -      opportunity, but not if the use is in a conditional
>>>>>>>>>> -      which makes this a likely if-conversion candidate.  */
>>>>>>>>>> -   if (gimple_bb (stmt) == bb
>>>>>>>>>> -       && (!is_gimple_assign (stmt)
>>>>>>>>>> -   || (TREE_CODE_CLASS (gimple_assign_rhs_code (stmt))
>>>>>>>>>> -       != tcc_comparison)))
>>>>>>>>>> -     {
>>>>>>>>>> -       found_useful_phi = true;
>>>>>>>>>> -       break;
>>>>>>>>>> -     }
>>>>>>>>>> -   /* If the use is on a loop header PHI and on one path the
>>>>>>>>>> -      value is unchanged this might expose a jump threading
>>>>>>>>>> -      opportunity.  */
>>>>>>>>>> -   if (gimple_code (stmt) == GIMPLE_PHI
>>>>>>>>>> -       && gimple_bb (stmt) == bb->loop_father->header
>>>>>>>>>> -       /* But for memory the PHI alone isn't good enough.  */
>>>>>>>>>> -       && ! virtual_operand_p (gimple_phi_result (stmt)))
>>>>>>>>>> -     {
>>>>>>>>>> -       bool found_unchanged_path = false;
>>>>>>>>>> -       for (unsigned i = 0; i < gimple_phi_num_args (phi); ++i)
>>>>>>>>>> - if (gimple_phi_arg_def (phi, i) == gimple_phi_result (stmt))
>>>>>>>>>> -   {
>>>>>>>>>> -     found_unchanged_path = true;
>>>>>>>>>> -     break;
>>>>>>>>>> -   }
>>>>>>>>>> -       /* If we found an unchanged path this can only be a threading
>>>>>>>>>> -          opportunity if we have uses of the loop header PHI result
>>>>>>>>>> -  in a stmt dominating the merge block.  Otherwise the
>>>>>>>>>> -  splitting may prevent if-conversion.  */
>>>>>>>>>> -       if (found_unchanged_path)
>>>>>>>>>> - {
>>>>>>>>>> -   use_operand_p use2_p;
>>>>>>>>>> -   imm_use_iterator iter2;
>>>>>>>>>> -   FOR_EACH_IMM_USE_FAST (use2_p, iter2, gimple_phi_result (stmt))
>>>>>>>>>> -     {
>>>>>>>>>> -       gimple *use_stmt = USE_STMT (use2_p);
>>>>>>>>>> -       if (is_gimple_debug (use_stmt))
>>>>>>>>>> - continue;
>>>>>>>>>> -       basic_block use_bb = gimple_bb (use_stmt);
>>>>>>>>>> -       if (use_bb != bb
>>>>>>>>>> -   && dominated_by_p (CDI_DOMINATORS, bb, use_bb))
>>>>>>>>>> - {
>>>>>>>>>> -   if (gcond *cond = dyn_cast <gcond *> (use_stmt))
>>>>>>>>>> -     if (gimple_cond_code (cond) == EQ_EXPR
>>>>>>>>>> - || gimple_cond_code (cond) == NE_EXPR)
>>>>>>>>>> -       found_useful_phi = true;
>>>>>>>>>> -   break;
>>>>>>>>>> - }
>>>>>>>>>> -     }
>>>>>>>>>> - }
>>>>>>>>>> -       if (found_useful_phi)
>>>>>>>>>> - break;
>>>>>>>>>> -     }
>>>>>>>>>> - }
>>>>>>>>>> -      if (found_useful_phi)
>>>>>>>>>> - break;
>>>>>>>>>> -    }
>>>>>>>>>> -  /* There is one exception namely a controlling condition we can 
>>>>>>>>>> propagate
>>>>>>>>>> -     an equivalence from to the joiner.  */
>>>>>>>>>> -  bool found_cprop_opportunity = false;
>>>>>>>>>> -  basic_block dom = get_immediate_dominator (CDI_DOMINATORS, bb);
>>>>>>>>>> -  gcond *cond = as_a <gcond *> (*gsi_last_bb (dom));
>>>>>>>>>> -  if (gimple_cond_code (cond) == EQ_EXPR
>>>>>>>>>> -      || gimple_cond_code (cond) == NE_EXPR)
>>>>>>>>>> -    for (unsigned i = 0; i < 2; ++i)
>>>>>>>>>> -      {
>>>>>>>>>> - tree op = gimple_op (cond, i);
>>>>>>>>>> - if (TREE_CODE (op) == SSA_NAME)
>>>>>>>>>> -   {
>>>>>>>>>> -     use_operand_p use_p;
>>>>>>>>>> -     imm_use_iterator iter;
>>>>>>>>>> -     FOR_EACH_IMM_USE_FAST (use_p, iter, op)
>>>>>>>>>> -       {
>>>>>>>>>> - if (is_gimple_debug (USE_STMT (use_p)))
>>>>>>>>>> -   continue;
>>>>>>>>>> - if (gimple_bb (USE_STMT (use_p)) == bb)
>>>>>>>>>> -   {
>>>>>>>>>> -     found_cprop_opportunity = true;
>>>>>>>>>> -     break;
>>>>>>>>>> -   }
>>>>>>>>>> -       }
>>>>>>>>>> -   }
>>>>>>>>>> - if (found_cprop_opportunity)
>>>>>>>>>> -   break;
>>>>>>>>>> -      }
>>>>>>>>>> -
>>>>>>>>>> -  if (! found_useful_phi && ! found_cprop_opportunity)
>>>>>>>>>> -    {
>>>>>>>>>> -      if (dump_file && (dump_flags & TDF_DETAILS))
>>>>>>>>>> - fprintf (dump_file,
>>>>>>>>>> -  "Block %d is a join that does not expose CSE/DCE/jump-thread "
>>>>>>>>>> -  "opportunities when duplicated.\n",
>>>>>>>>>> -  bb->index);
>>>>>>>>>> -      return false;
>>>>>>>>>> -    }
>>>>>>>>>> -
>>>>>>>>>> -  /* We may want something here which looks at dataflow and tries
>>>>>>>>>> -     to guess if duplication of BB is likely to result in 
>>>>>>>>>> simplification
>>>>>>>>>> -     of instructions in BB in either the original or the duplicate. 
>>>>>>>>>>  */
>>>>>>>>>> -  return true;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* If the immediate dominator of the latch of the loop is
>>>>>>>>>> -   block with conditional branch, then the loop latch  is
>>>>>>>>>> -   duplicated to its predecessors path preserving the SSA
>>>>>>>>>> -   semantics.
>>>>>>>>>> -
>>>>>>>>>> -   CFG before transformation.
>>>>>>>>>> -
>>>>>>>>>> -              2
>>>>>>>>>> -              |
>>>>>>>>>> -              |
>>>>>>>>>> -        +---->3
>>>>>>>>>> -        |    / \
>>>>>>>>>> -        |   /   \
>>>>>>>>>> -        |  4     5
>>>>>>>>>> -        |   \   /
>>>>>>>>>> -        |    \ /
>>>>>>>>>> -        |     6
>>>>>>>>>> -        |    / \
>>>>>>>>>> -        |   /   \
>>>>>>>>>> -        |  8     7
>>>>>>>>>> -        |  |     |
>>>>>>>>>> -        ---+     E
>>>>>>>>>> -
>>>>>>>>>> -
>>>>>>>>>> -
>>>>>>>>>> -    Block 8 is the latch.  We're going to make copies of block 6 (9 
>>>>>>>>>> & 10)
>>>>>>>>>> -    and wire things up so they look like this:
>>>>>>>>>> -
>>>>>>>>>> -              2
>>>>>>>>>> -              |
>>>>>>>>>> -              |
>>>>>>>>>> -        +---->3
>>>>>>>>>> -        |    / \
>>>>>>>>>> -        |   /   \
>>>>>>>>>> -        |  4     5
>>>>>>>>>> -        |  |     |
>>>>>>>>>> -        |  |     |
>>>>>>>>>> -        |  9    10
>>>>>>>>>> -        |  |\   /|
>>>>>>>>>> -        |  | \ / |
>>>>>>>>>> -        |  |  7  |
>>>>>>>>>> -        |  |  |  |
>>>>>>>>>> -        |  |  E  |
>>>>>>>>>> -        |  |     |
>>>>>>>>>> -        |   \   /
>>>>>>>>>> -        |    \ /
>>>>>>>>>> -        +-----8
>>>>>>>>>> -
>>>>>>>>>> -
>>>>>>>>>> -    Blocks 9 and 10 will get merged into blocks 4 & 5 respectively 
>>>>>>>>>> which
>>>>>>>>>> -    enables CSE, DCE and other optimizations to occur on a larger 
>>>>>>>>>> block
>>>>>>>>>> -    of code.   */
>>>>>>>>>> -
>>>>>>>>>> -static bool
>>>>>>>>>> -split_paths ()
>>>>>>>>>> -{
>>>>>>>>>> -  bool changed = false;
>>>>>>>>>> -
>>>>>>>>>> -  loop_optimizer_init (LOOPS_NORMAL | LOOPS_HAVE_RECORDED_EXITS);
>>>>>>>>>> -  initialize_original_copy_tables ();
>>>>>>>>>> -  calculate_dominance_info (CDI_DOMINATORS);
>>>>>>>>>> -
>>>>>>>>>> -  for (auto loop : loops_list (cfun, LI_FROM_INNERMOST))
>>>>>>>>>> -    {
>>>>>>>>>> -      /* Only split paths if we are optimizing this loop for speed. 
>>>>>>>>>>  */
>>>>>>>>>> -      if (!optimize_loop_for_speed_p (loop))
>>>>>>>>>> - continue;
>>>>>>>>>> -
>>>>>>>>>> -      /* See if there is a block that we can duplicate to split the
>>>>>>>>>> -  path to the loop latch.  */
>>>>>>>>>> -      basic_block bb
>>>>>>>>>> - = find_block_to_duplicate_for_splitting_paths (loop->latch);
>>>>>>>>>> -
>>>>>>>>>> -      /* BB is the merge point for an IF-THEN-ELSE we want to 
>>>>>>>>>> transform.
>>>>>>>>>> -
>>>>>>>>>> -  Essentially we want to create a duplicate of bb and redirect the
>>>>>>>>>> -  first predecessor of BB to the duplicate (leaving the second
>>>>>>>>>> -  predecessor as is.  This will split the path leading to the latch
>>>>>>>>>> -  re-using BB to avoid useless copying.  */
>>>>>>>>>> -      if (bb && is_feasible_trace (bb))
>>>>>>>>>> - {
>>>>>>>>>> -   if (dump_file && (dump_flags & TDF_DETAILS))
>>>>>>>>>> -     fprintf (dump_file,
>>>>>>>>>> -      "Duplicating join block %d into predecessor paths\n",
>>>>>>>>>> -      bb->index);
>>>>>>>>>> -   basic_block pred0 = EDGE_PRED (bb, 0)->src;
>>>>>>>>>> -   if (EDGE_COUNT (pred0->succs) != 1)
>>>>>>>>>> -     pred0 = EDGE_PRED (bb, 1)->src;
>>>>>>>>>> -   transform_duplicate (pred0, bb);
>>>>>>>>>> -   changed = true;
>>>>>>>>>> -
>>>>>>>>>> -   /* If BB has an outgoing edge marked as IRREDUCIBLE, then
>>>>>>>>>> -      duplicating BB may result in an irreducible region turning
>>>>>>>>>> -      into a natural loop.
>>>>>>>>>> -
>>>>>>>>>> -      Long term we might want to hook this into the block
>>>>>>>>>> -      duplication code, but as we've seen with similar changes
>>>>>>>>>> -      for edge removal, that can be somewhat risky.  */
>>>>>>>>>> -   if (EDGE_SUCC (bb, 0)->flags & EDGE_IRREDUCIBLE_LOOP
>>>>>>>>>> -       || EDGE_SUCC (bb, 1)->flags & EDGE_IRREDUCIBLE_LOOP)
>>>>>>>>>> -     {
>>>>>>>>>> -       if (dump_file && (dump_flags & TDF_DETAILS))
>>>>>>>>>> -   fprintf (dump_file,
>>>>>>>>>> -    "Join block %d has EDGE_IRREDUCIBLE_LOOP set.  "
>>>>>>>>>> -    "Scheduling loop fixups.\n",
>>>>>>>>>> -    bb->index);
>>>>>>>>>> -       loops_state_set (LOOPS_NEED_FIXUP);
>>>>>>>>>> -     }
>>>>>>>>>> - }
>>>>>>>>>> -    }
>>>>>>>>>> -
>>>>>>>>>> -  loop_optimizer_finalize ();
>>>>>>>>>> -  free_original_copy_tables ();
>>>>>>>>>> -  return changed;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* Main entry point for splitting paths.  Returns TODO_cleanup_cfg 
>>>>>>>>>> if any
>>>>>>>>>> -   paths where split, otherwise return zero.  */
>>>>>>>>>> -
>>>>>>>>>> -static unsigned int
>>>>>>>>>> -execute_split_paths ()
>>>>>>>>>> -{
>>>>>>>>>> -  /* If we don't have at least 2 real blocks and backedges in the
>>>>>>>>>> -     CFG, then there's no point in trying to perform path 
>>>>>>>>>> splitting.  */
>>>>>>>>>> -  if (n_basic_blocks_for_fn (cfun) <= NUM_FIXED_BLOCKS + 1
>>>>>>>>>> -      || !mark_dfs_back_edges ())
>>>>>>>>>> -    return 0;
>>>>>>>>>> -
>>>>>>>>>> -  bool changed = split_paths();
>>>>>>>>>> -  if (changed)
>>>>>>>>>> -    free_dominance_info (CDI_DOMINATORS);
>>>>>>>>>> -
>>>>>>>>>> -  return changed ? TODO_cleanup_cfg : 0;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -static bool
>>>>>>>>>> -gate_split_paths ()
>>>>>>>>>> -{
>>>>>>>>>> -  return flag_split_paths;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -namespace {
>>>>>>>>>> -
>>>>>>>>>> -const pass_data pass_data_split_paths =
>>>>>>>>>> -{
>>>>>>>>>> -  GIMPLE_PASS, /* type */
>>>>>>>>>> -  "split-paths", /* name */
>>>>>>>>>> -  OPTGROUP_NONE, /* optinfo_flags */
>>>>>>>>>> -  TV_SPLIT_PATHS, /* tv_id */
>>>>>>>>>> -  PROP_ssa, /* properties_required */
>>>>>>>>>> -  0, /* properties_provided */
>>>>>>>>>> -  0, /* properties_destroyed */
>>>>>>>>>> -  0, /* todo_flags_start */
>>>>>>>>>> -  TODO_update_ssa, /* todo_flags_finish */
>>>>>>>>>> -};
>>>>>>>>>> -
>>>>>>>>>> -class pass_split_paths : public gimple_opt_pass
>>>>>>>>>> -{
>>>>>>>>>> -   public:
>>>>>>>>>> -    pass_split_paths (gcc::context *ctxt)
>>>>>>>>>> -      : gimple_opt_pass (pass_data_split_paths, ctxt)
>>>>>>>>>> -    {}
>>>>>>>>>> -   /* opt_pass methods: */
>>>>>>>>>> -  opt_pass * clone () final override { return new pass_split_paths 
>>>>>>>>>> (m_ctxt); }
>>>>>>>>>> -  bool gate (function *) final override { return gate_split_paths 
>>>>>>>>>> (); }
>>>>>>>>>> -  unsigned int execute (function *) final override
>>>>>>>>>> -  {
>>>>>>>>>> -    return execute_split_paths ();
>>>>>>>>>> -  }
>>>>>>>>>> -
>>>>>>>>>> -}; // class pass_split_paths
>>>>>>>>>> -
>>>>>>>>>> -} // anon namespace
>>>>>>>>>> -
>>>>>>>>>> -gimple_opt_pass *
>>>>>>>>>> -make_pass_split_paths (gcc::context *ctxt)
>>>>>>>>>> -{
>>>>>>>>>> -  return new pass_split_paths (ctxt);
>>>>>>>>>> -}
>>>>>>>>>> diff --git a/gcc/opts.cc b/gcc/opts.cc
>>>>>>>>>> index 342517528e8..5522f41f906 100644
>>>>>>>>>> --- a/gcc/opts.cc
>>>>>>>>>> +++ b/gcc/opts.cc
>>>>>>>>>> @@ -707,7 +707,6 @@ static const struct default_options 
>>>>>>>>>> default_options_table[] =
>>>>>>>>>>   { OPT_LEVELS_3_PLUS, OPT_fpeel_loops, NULL, 1 },
>>>>>>>>>>   { OPT_LEVELS_3_PLUS, OPT_fpredictive_commoning, NULL, 1 },
>>>>>>>>>>   { OPT_LEVELS_3_PLUS, OPT_fsplit_loops, NULL, 1 },
>>>>>>>>>> -    { OPT_LEVELS_3_PLUS, OPT_fsplit_paths, NULL, 1 },
>>>>>>>>>>   { OPT_LEVELS_3_PLUS, OPT_ftree_loop_distribution, NULL, 1 },
>>>>>>>>>>   { OPT_LEVELS_3_PLUS, OPT_ftree_partial_pre, NULL, 1 },
>>>>>>>>>>   { OPT_LEVELS_3_PLUS, OPT_funswitch_loops, NULL, 1 },
>>>>>>>>>> diff --git a/gcc/passes.def b/gcc/passes.def
>>>>>>>>>> index 1fc867fae51..9095c134f49 100644
>>>>>>>>>> --- a/gcc/passes.def
>>>>>>>>>> +++ b/gcc/passes.def
>>>>>>>>>> @@ -351,7 +351,6 @@ along with GCC; see the file COPYING3.  If not 
>>>>>>>>>> see
>>>>>>>>>>     NEXT_PASS (pass_cse_reciprocals);
>>>>>>>>>>     NEXT_PASS (pass_reassoc, false /* early_p */);
>>>>>>>>>>     NEXT_PASS (pass_strength_reduction);
>>>>>>>>>> -      NEXT_PASS (pass_split_paths);
>>>>>>>>>>     NEXT_PASS (pass_tracer);
>>>>>>>>>>     NEXT_PASS (pass_fre, false /* may_iterate */);
>>>>>>>>>>     /* After late FRE we rewrite no longer addressed locals into SSA
>>>>>>>>>> diff --git a/gcc/testsuite/g++.dg/torture/pr88797.C 
>>>>>>>>>> b/gcc/testsuite/g++.dg/torture/pr88797.C
>>>>>>>>>> new file mode 100644
>>>>>>>>>> index 00000000000..3dd00fcb7a9
>>>>>>>>>> --- /dev/null
>>>>>>>>>> +++ b/gcc/testsuite/g++.dg/torture/pr88797.C
>>>>>>>>>> @@ -0,0 +1,13 @@
>>>>>>>>>> +/* { dg-do compile } */
>>>>>>>>>> +
>>>>>>>>>> +
>>>>>>>>>> +void use(unsigned);
>>>>>>>>>> +bool f(unsigned x, unsigned y) {
>>>>>>>>>> +    return x < 1111 + (y <= 2222);
>>>>>>>>>> +}
>>>>>>>>>> +void test_f(unsigned x, unsigned y) {
>>>>>>>>>> +    for (unsigned i = 0; i < 3333; ++i)
>>>>>>>>>> +        use(f(x++, y++));
>>>>>>>>>> +}
>>>>>>>>>> +
>>>>>>>>>> +
>>>>>>>>>> diff --git a/gcc/testsuite/g++.dg/tree-ssa/pr88797.C 
>>>>>>>>>> b/gcc/testsuite/g++.dg/tree-ssa/pr88797.C
>>>>>>>>>> deleted file mode 100644
>>>>>>>>>> index 541ae8e71e1..00000000000
>>>>>>>>>> --- a/gcc/testsuite/g++.dg/tree-ssa/pr88797.C
>>>>>>>>>> +++ /dev/null
>>>>>>>>>> @@ -1,16 +0,0 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O3 -fdump-tree-split-paths-details" } */
>>>>>>>>>> -
>>>>>>>>>> -
>>>>>>>>>> -void use(unsigned);
>>>>>>>>>> -bool f(unsigned x, unsigned y) {
>>>>>>>>>> -    return x < 1111 + (y <= 2222);
>>>>>>>>>> -}
>>>>>>>>>> -void test_f(unsigned x, unsigned y) {
>>>>>>>>>> -    for (unsigned i = 0; i < 3333; ++i)
>>>>>>>>>> -        use(f(x++, y++));
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* { dg-final { scan-tree-dump-not "Duplicating join block" 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "appears to be a join point 
>>>>>>>>>> for if-convertible bbs." 1 "split-paths" } } */
>>>>>>>>>> -
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/pr69270.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/pr69270.c
>>>>>>>>>> similarity index 57%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/pr69270.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/pr69270.c
>>>>>>>>>> index b08ec9d6ddb..daed456ec60 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/pr69270.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/pr69270.c
>>>>>>>>>> @@ -1,17 +1,10 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-dom3-details" } */
>>>>>>>>>> 
>>>>>>>>>> /* There should be two references to bufferstep that turn into
>>>>>>>>>>  constants.  */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Replaced 
>>>>>>>>>> .bufferstep_\[0-9\]+. with constant .0." 1 "dom3"} } */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Replaced 
>>>>>>>>>> .bufferstep_\[0-9\]+. with constant .1." 1 "dom3"} } */
>>>>>>>>>> 
>>>>>>>>>> /* And some assignments ought to fold down to constants.  */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Folded to: 
>>>>>>>>>> (?:bufferstep)?_\[0-9\]+ = 1;" 1 "dom3"} } */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Folded to: 
>>>>>>>>>> (?:bufferstep)?_\[0-9\]+ = 0;" 1 "dom3"} } */
>>>>>>>>>> 
>>>>>>>>>> /* The XOR operations should have been optimized to constants.  */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-not "bit_xor" "dom3"} } */
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> extern int *stepsizeTable;
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-10.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-10.c
>>>>>>>>>> similarity index 81%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-10.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-10.c
>>>>>>>>>> index ed208795488..336b11893b6 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-10.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-10.c
>>>>>>>>>> @@ -1,6 +1,4 @@
>>>>>>>>>> /* PR tree-optimization/79389  */
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O3 -fdump-tree-split-paths-details" } */
>>>>>>>>>> 
>>>>>>>>>> typedef struct
>>>>>>>>>> {
>>>>>>>>>> @@ -46,4 +44,3 @@ double MonteCarlo_integrate(int Num_samples)
>>>>>>>>>> return ((double) under_curve / Num_samples) * 4.0;
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Duplicating join block" 0 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-11.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-11.c
>>>>>>>>>> similarity index 55%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-11.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-11.c
>>>>>>>>>> index fa33367feaa..44b8b059481 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-11.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-11.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fno-tree-vectorize -fsplit-paths 
>>>>>>>>>> -fdump-tree-split-paths-details -w" } */
>>>>>>>>>> 
>>>>>>>>>> void foo(unsigned long long *M)
>>>>>>>>>> {
>>>>>>>>>> @@ -11,4 +9,3 @@ void foo(unsigned long long *M)
>>>>>>>>>>   }
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "join point for 
>>>>>>>>>> if-convertible" 1 "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-12.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-12.c
>>>>>>>>>> similarity index 70%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-12.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-12.c
>>>>>>>>>> index da00f795ef0..dc9da4e90ef 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-12.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-12.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> " } */
>>>>>>>>>> 
>>>>>>>>>> double
>>>>>>>>>> foo(double *d1, double *d2, double *d3, int num, double *ip)
>>>>>>>>>> @@ -18,4 +16,3 @@ foo(double *d1, double *d2, double *d3, int num, 
>>>>>>>>>> double *ip)
>>>>>>>>>> 
>>>>>>>>>> /* Split-paths shouldn't do anything here, if there's a diamond it 
>>>>>>>>>> would
>>>>>>>>>>  be if-convertible.  */
>>>>>>>>>> -/* { dg-final { scan-tree-dump-not "Duplicating join block" 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-13.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-13.c
>>>>>>>>>> similarity index 62%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-13.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-13.c
>>>>>>>>>> index d24e23bc59d..554ae4d8f39 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-13.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-13.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> " } */
>>>>>>>>>> /* PR tree-optimization/112402 */
>>>>>>>>>> /* This is similar to split-path-2.c but instead of the add
>>>>>>>>>>  being inside both sides, we have a constant. */
>>>>>>>>>> @@ -22,5 +20,4 @@ foo(signed char *p, int n)
>>>>>>>>>> return s;
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump "appears to be a join point for 
>>>>>>>>>> if-convertible" "split-paths" } } */
>>>>>>>>>> 
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.c-torture/compile/split-path-2.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-2.c
>>>>>>>>>> new file mode 100644
>>>>>>>>>> index 00000000000..9418c9488d4
>>>>>>>>>> --- /dev/null
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-2.c
>>>>>>>>>> @@ -0,0 +1,18 @@
>>>>>>>>>> +
>>>>>>>>>> +int
>>>>>>>>>> +foo(signed char *p, int n)
>>>>>>>>>> +{
>>>>>>>>>> +  int s = 0;
>>>>>>>>>> +  int i;
>>>>>>>>>> +
>>>>>>>>>> +  for (i = 0; i < n; i++) {
>>>>>>>>>> +    if (p[i] >= 0)
>>>>>>>>>> +      s++;
>>>>>>>>>> +    else
>>>>>>>>>> +      s--;
>>>>>>>>>> +  }
>>>>>>>>>> +
>>>>>>>>>> +  return s;
>>>>>>>>>> +}
>>>>>>>>>> +
>>>>>>>>>> +
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-3.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-3.c
>>>>>>>>>> similarity index 93%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-3.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-3.c
>>>>>>>>>> index f6a29f628a2..9ea9fb2fed1 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-3.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-3.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> -w" } */
>>>>>>>>>> 
>>>>>>>>>> typedef struct bitmap_head_def *bitmap;
>>>>>>>>>> extern void vec_assert_fail (const char *, const char *, const char 
>>>>>>>>>> *file_,
>>>>>>>>>> @@ -94,4 +92,3 @@ compute_idf (bitmap_head * dfs)
>>>>>>>>>> (VEC_int_heap_free (&work_stack));
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump-not "Duplicating join block" 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-4.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-4.c
>>>>>>>>>> similarity index 62%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-4.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-4.c
>>>>>>>>>> index cbee6fce4b7..6705f9e8426 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-4.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-4.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> -w -fdisable-tree-thread1 -fdisable-tree-thread2" } */
>>>>>>>>>> 
>>>>>>>>>> /* Note: Threader causes the infinite loop in val & 1 sooner.  */
>>>>>>>>>> 
>>>>>>>>>> @@ -25,5 +23,4 @@ powi_cost (long n)
>>>>>>>>>>   }
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Duplicating join block" 1 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> 
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-5.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-5.c
>>>>>>>>>> similarity index 84%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-5.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-5.c
>>>>>>>>>> index 2f5e7edcfa8..aff1f038195 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-5.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-5.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> -w" } */
>>>>>>>>>> 
>>>>>>>>>> const extern signed char *__ctype_ptr__;
>>>>>>>>>> typedef unsigned char uchar;
>>>>>>>>>> @@ -41,4 +39,3 @@ bmhi_init (const signed char *pattern)
>>>>>>>>>>   }
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "join point for 
>>>>>>>>>> if-convertible" 1 "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-6.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-6.c
>>>>>>>>>> similarity index 82%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-6.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-6.c
>>>>>>>>>> index e2b0a9571f0..9c32c813b60 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-6.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-6.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fno-tree-cselim -fno-ssa-phiopt 
>>>>>>>>>> -fdump-tree-split-paths-details -fno-finite-loops 
>>>>>>>>>> -fno-tree-dominator-opts -fno-tree-vrp -w" } */
>>>>>>>>>> 
>>>>>>>>>> struct __sFILE
>>>>>>>>>> {
>>>>>>>>>> @@ -75,4 +73,3 @@ lookharder (char *string)
>>>>>>>>>>   }
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Duplicating join block" 2 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-7.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-7.c
>>>>>>>>>> similarity index 87%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-7.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/compile/split-path-7.c
>>>>>>>>>> index 35634ab3bd8..33d04b4ff7d 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-7.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-7.c
>>>>>>>>>> @@ -1,5 +1,3 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fno-tree-cselim -fno-ssa-phiopt 
>>>>>>>>>> -fno-tree-sink -fdump-tree-split-paths-details -w" } */
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> struct _reent
>>>>>>>>>> @@ -91,4 +89,3 @@ linit ()
>>>>>>>>>> }
>>>>>>>>>>   }
>>>>>>>>>> }
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Duplicating join block" 0 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.c-torture/compile/split-path-8.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-8.c
>>>>>>>>>> new file mode 100644
>>>>>>>>>> index 00000000000..cccbf0ca0a3
>>>>>>>>>> --- /dev/null
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-8.c
>>>>>>>>>> @@ -0,0 +1,11 @@
>>>>>>>>>> +/* PR77283 */
>>>>>>>>>> +
>>>>>>>>>> +void
>>>>>>>>>> +foo (double *x, double *a, double *b, long n, double limit)
>>>>>>>>>> +{
>>>>>>>>>> +  long i;
>>>>>>>>>> +  for (i=0; i < n; i++)
>>>>>>>>>> +    if (a[i] < limit)
>>>>>>>>>> +      x[i] = b[i];
>>>>>>>>>> +}
>>>>>>>>>> +
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.c-torture/compile/split-path-9.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/compile/split-path-9.c
>>>>>>>>>> new file mode 100644
>>>>>>>>>> index 00000000000..4eacad2978a
>>>>>>>>>> --- /dev/null
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/compile/split-path-9.c
>>>>>>>>>> @@ -0,0 +1,14 @@
>>>>>>>>>> +/* PR77366 */
>>>>>>>>>> +
>>>>>>>>>> +void
>>>>>>>>>> +foo(unsigned int size, unsigned int *state)
>>>>>>>>>> +{
>>>>>>>>>> +  unsigned int i;
>>>>>>>>>> +
>>>>>>>>>> +  for(i = 0; i < size; i++)
>>>>>>>>>> +    {
>>>>>>>>>> +      if(state[i] & 1)
>>>>>>>>>> + state[i] ^= 1;
>>>>>>>>>> +    }
>>>>>>>>>> +}
>>>>>>>>>> +
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-1.c 
>>>>>>>>>> b/gcc/testsuite/gcc.c-torture/execute/split-path-1.c
>>>>>>>>>> similarity index 83%
>>>>>>>>>> rename from gcc/testsuite/gcc.dg/tree-ssa/split-path-1.c
>>>>>>>>>> rename to gcc/testsuite/gcc.c-torture/execute/split-path-1.c
>>>>>>>>>> index b670dee8d10..e0ca386b4ea 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-1.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.c-torture/execute/split-path-1.c
>>>>>>>>>> @@ -1,6 +1,4 @@
>>>>>>>>>> -/* { dg-do run } */
>>>>>>>>>> /* Note both PHI-OPT and the loop if conversion pass converts the 
>>>>>>>>>> inner if to be branchless using min/max. */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> --param max-jump-thread-duplication-stmts=20 -fno-ssa-phiopt 
>>>>>>>>>> -fno-tree-loop-if-convert" } */
>>>>>>>>>> 
>>>>>>>>>> #include <stdio.h>
>>>>>>>>>> #include <stdlib.h>
>>>>>>>>>> @@ -67,4 +65,3 @@ main()
>>>>>>>>>> return 0;
>>>>>>>>>> }
>>>>>>>>>> 
>>>>>>>>>> -/* { dg-final { scan-tree-dump "Duplicating join block" 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-2.c 
>>>>>>>>>> b/gcc/testsuite/gcc.dg/tree-ssa/split-path-2.c
>>>>>>>>>> deleted file mode 100644
>>>>>>>>>> index 61697a35e56..00000000000
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-2.c
>>>>>>>>>> +++ /dev/null
>>>>>>>>>> @@ -1,21 +0,0 @@
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths -fdump-tree-split-paths-details 
>>>>>>>>>> " } */
>>>>>>>>>> -
>>>>>>>>>> -int
>>>>>>>>>> -foo(signed char *p, int n)
>>>>>>>>>> -{
>>>>>>>>>> -  int s = 0;
>>>>>>>>>> -  int i;
>>>>>>>>>> -
>>>>>>>>>> -  for (i = 0; i < n; i++) {
>>>>>>>>>> -    if (p[i] >= 0)
>>>>>>>>>> -      s++;
>>>>>>>>>> -    else
>>>>>>>>>> -      s--;
>>>>>>>>>> -  }
>>>>>>>>>> -
>>>>>>>>>> -  return s;
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* { dg-final { scan-tree-dump "appears to be a join point for 
>>>>>>>>>> if-convertible" "split-paths" } } */
>>>>>>>>>> -
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-8.c 
>>>>>>>>>> b/gcc/testsuite/gcc.dg/tree-ssa/split-path-8.c
>>>>>>>>>> deleted file mode 100644
>>>>>>>>>> index fb54f5dc512..00000000000
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-8.c
>>>>>>>>>> +++ /dev/null
>>>>>>>>>> @@ -1,14 +0,0 @@
>>>>>>>>>> -/* PR77283 */
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O3 -fdump-tree-split-paths-details" } */
>>>>>>>>>> -
>>>>>>>>>> -void
>>>>>>>>>> -foo (double *x, double *a, double *b, long n, double limit)
>>>>>>>>>> -{
>>>>>>>>>> -  long i;
>>>>>>>>>> -  for (i=0; i < n; i++)
>>>>>>>>>> -    if (a[i] < limit)
>>>>>>>>>> -      x[i] = b[i];
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Duplicating join block" 0 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/split-path-9.c 
>>>>>>>>>> b/gcc/testsuite/gcc.dg/tree-ssa/split-path-9.c
>>>>>>>>>> deleted file mode 100644
>>>>>>>>>> index 8be0cd01201..00000000000
>>>>>>>>>> --- a/gcc/testsuite/gcc.dg/tree-ssa/split-path-9.c
>>>>>>>>>> +++ /dev/null
>>>>>>>>>> @@ -1,17 +0,0 @@
>>>>>>>>>> -/* PR77366 */
>>>>>>>>>> -/* { dg-do compile } */
>>>>>>>>>> -/* { dg-options "-O3 -fdump-tree-split-paths-details" } */
>>>>>>>>>> -
>>>>>>>>>> -void
>>>>>>>>>> -foo(unsigned int size, unsigned int *state)
>>>>>>>>>> -{
>>>>>>>>>> -  unsigned int i;
>>>>>>>>>> -
>>>>>>>>>> -  for(i = 0; i < size; i++)
>>>>>>>>>> -    {
>>>>>>>>>> -      if(state[i] & 1)
>>>>>>>>>> - state[i] ^= 1;
>>>>>>>>>> -    }
>>>>>>>>>> -}
>>>>>>>>>> -
>>>>>>>>>> -/* { dg-final { scan-tree-dump-times "Duplicating join block" 0 
>>>>>>>>>> "split-paths" } } */
>>>>>>>>>> diff --git a/gcc/testsuite/gcc.target/i386/pr106450.c 
>>>>>>>>>> b/gcc/testsuite/gcc.target/i386/pr106450.c
>>>>>>>>>> index d16231f6abd..009e3eb8c74 100644
>>>>>>>>>> --- a/gcc/testsuite/gcc.target/i386/pr106450.c
>>>>>>>>>> +++ b/gcc/testsuite/gcc.target/i386/pr106450.c
>>>>>>>>>> @@ -1,5 +1,5 @@
>>>>>>>>>> /* { dg-do compile { target int128 } } */
>>>>>>>>>> -/* { dg-options "-O2 -fsplit-paths" } */
>>>>>>>>>> +/* { dg-options "-O2" } */
>>>>>>>>>> 
>>>>>>>>>> __int128 n;
>>>>>>>>>> 
>>>>>>>>>> diff --git a/gcc/timevar.def b/gcc/timevar.def
>>>>>>>>>> index 588d987ec35..c7020997609 100644
>>>>>>>>>> --- a/gcc/timevar.def
>>>>>>>>>> +++ b/gcc/timevar.def
>>>>>>>>>> @@ -287,7 +287,6 @@ DEFTIMEVAR (TV_GCSE_AFTER_RELOAD     , "load CSE 
>>>>>>>>>> after reload")
>>>>>>>>>> DEFTIMEVAR (TV_REE      , "ree")
>>>>>>>>>> DEFTIMEVAR (TV_THREAD_PROLOGUE_AND_EPILOGUE, "thread pro- & 
>>>>>>>>>> epilogue")
>>>>>>>>>> DEFTIMEVAR (TV_IFCVT2      , "if-conversion 2")
>>>>>>>>>> -DEFTIMEVAR (TV_SPLIT_PATHS      , "split paths")
>>>>>>>>>> DEFTIMEVAR (TV_COMBINE_STACK_ADJUST  , "combine stack adjustments")
>>>>>>>>>> DEFTIMEVAR (TV_PEEPHOLE2             , "peephole 2")
>>>>>>>>>> DEFTIMEVAR (TV_RENAME_REGISTERS      , "rename registers")
>>>>>>>>>> diff --git a/gcc/tree-pass.h b/gcc/tree-pass.h
>>>>>>>>>> index b3c97658a8f..a3b35e009e0 100644
>>>>>>>>>> --- a/gcc/tree-pass.h
>>>>>>>>>> +++ b/gcc/tree-pass.h
>>>>>>>>>> @@ -404,7 +404,6 @@ extern gimple_opt_pass *make_pass_ch 
>>>>>>>>>> (gcc::context *ctxt);
>>>>>>>>>> extern gimple_opt_pass *make_pass_ch_vect (gcc::context *ctxt);
>>>>>>>>>> extern gimple_opt_pass *make_pass_sccopy (gcc::context *ctxt);
>>>>>>>>>> extern gimple_opt_pass *make_pass_ccp (gcc::context *ctxt);
>>>>>>>>>> -extern gimple_opt_pass *make_pass_split_paths (gcc::context *ctxt);
>>>>>>>>>> extern gimple_opt_pass *make_pass_build_ssa (gcc::context *ctxt);
>>>>>>>>>> extern gimple_opt_pass *make_pass_build_alias (gcc::context *ctxt);
>>>>>>>>>> extern gimple_opt_pass *make_pass_build_ealias (gcc::context *ctxt);
>>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> -- 
>>>>>>>>> Richard Biener <[email protected]>
>>>>>>>>> SUSE Software Solutions Germany GmbH,
>>>>>>>>> Frankenstrasse 146, 90461 Nuernberg, Germany;
>>>>>>>>> GF: Jochen Jaser, Andrew McDonald; (HRB 36809, AG Nuernberg)
> 

Reply via email to