[Bug tree-optimization/113385] New: [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[Bug tree-optimization/113385] New: [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

            Bug ID: 113385
           Summary: [14 regression] ICE when building opencv
                    (dfs_enumerate_from, at cfganal.cc:1590)
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: sjames at gcc dot gnu.org
                CC: tnfchris at gcc dot gnu.org
  Target Milestone: ---

Created attachment 57077
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=57077&action=edit
essential_solver.cpp.ii

(Note that this is with https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113364#c9
applied so I could bootstrap.)

```
$ g++ -c essential_solver.cpp.ii  -march=znver2 -O3 -std=c++14
during GIMPLE pass: vect
In file included from /usr/include/eigen3/Eigen/Householder:23,
                 from /usr/include/eigen3/Eigen/QR:15,
                 from /usr/include/eigen3/Eigen/Dense:4,
                 from /usr/include/eigen3/Eigen/Eigen:1,
                 from
/var/tmp/portage/media-libs/opencv-4.8.1-r1/work/opencv-4.8.1/modules/calib3d/src/usac/essential_solver.cpp:8:
/usr/include/eigen3/Eigen/src/Householder/Householder.h: In member function
‘void Eigen::MatrixBase<Derived>::applyHouseholderOnTheRight(const
EssentialPart&, const Scalar&, Scalar*) [with EssentialPart =
Eigen::Matrix<double, 2, 1>; Derived = Eigen::Block<Eigen::Matrix<double, 10,
10>, -1, -1, false>]’:
/usr/include/eigen3/Eigen/src/Householder/Householder.h:154:6: internal
compiler error: in dfs_enumerate_from, at cfganal.cc:1590
  154 | void MatrixBase<Derived>::applyHouseholderOnTheRight(
      |      ^~~~~~~~~~~~~~~~~~~
0x56284e95dedb dfs_enumerate_from(basic_block_def*, int, bool
(*)(basic_block_def const*, void const*), basic_block_def**, int, void const*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/cfganal.cc:1590
0x562850340e8b get_loop_body_with_size(loop const*, basic_block_def**, unsigned
int)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/cfgloop.cc:872
0x562850340e8b get_loop_body(loop const*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/cfgloop.cc:901
0x56285033ca5c estimate_numbers_of_iterations(loop*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4833
0x5628508c5072 loop_exits_before_overflow
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:5259
0x5628504227b2 scev_probably_wraps_p(tree_node*, tree_node*, tree_node*,
gimple*, loop*, bool)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:5511
0x5628503d64cd convert_affine_scev(loop*, tree_node*, tree_node**, tree_node**,
gimple*, bool, tree_node*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-chrec.cc:1425
0x562850354ba7 chrec_convert_1
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-chrec.cc:1483
0x562850350aea interpret_gimple_assign
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-scalar-evolution.cc:1919
0x562850350aea analyze_scalar_evolution_1
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-scalar-evolution.cc:1970
0x562850344f7d analyze_scalar_evolution(loop*, tree_node*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-scalar-evolution.cc:2039
0x562850343721 analyze_scalar_evolution_in_loop
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-scalar-evolution.cc:2165
0x562850343721 simple_iv_with_niters(loop*, loop*, tree_node*, affine_iv*,
tree_node**, bool)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-scalar-evolution.cc:3237
0x562850904a0f get_cst_init_from_scev
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4015
0x5628508ab276 record_nonwrapping_iv(loop*, tree_node*, tree_node*, gimple*,
tree_node*, tree_node*, bool, bool) [clone .constprop.0]
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4101
0x56285033d648 infer_loop_bounds_from_signedness
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4380
0x56285033d648 infer_loop_bounds_from_undefined
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4418
0x56285033d648 estimate_numbers_of_iterations(loop*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4872
0x56285066ec47 estimated_loop_iterations(loop*,
generic_wide_int<widest_int_storage<131072> >*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4904
0x56285066ebe1 estimated_loop_iterations_int(loop*)
       
/usr/src/debug/sys-devel/gcc-14.0.1.9999/gcc-14.0.1.9999/gcc/tree-ssa-loop-niter.cc:4919
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://bugs.gentoo.org/> for instructions.
```

```
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/usr/libexec/gcc/x86_64-pc-linux-gnu/14/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with:
/var/tmp/portage/sys-devel/gcc-14.0.1.9999/work/gcc-14.0.1.9999/configure
--host=x86_64-pc-linux-gnu --build=x86_64-pc-linux-gnu --prefix=/usr
--bindir=/usr/x86_64-pc-linux-gnu/gcc-bin/14
--includedir=/usr/lib/gcc/x86_64-pc-linux-gnu/14/include
--datadir=/usr/share/gcc-data/x86_64-pc-linux-gnu/14
--mandir=/usr/share/gcc-data/x86_64-pc-linux-gnu/14/man
--infodir=/usr/share/gcc-data/x86_64-pc-linux-gnu/14/info
--with-gxx-include-dir=/usr/lib/gcc/x86_64-pc-linux-gnu/14/include/g++-v14
--disable-silent-rules --disable-dependency-tracking
--with-python-dir=/share/gcc-data/x86_64-pc-linux-gnu/14/python
--enable-languages=c,c++,fortran --enable-obsolete --enable-secureplt
--disable-werror --with-system-zlib --enable-nls --without-included-gettext
--disable-libunwind-exceptions --enable-checking=yes,extra,rtl,df
--with-bugurl=https://bugs.gentoo.org/ --with-pkgversion='Gentoo 14.0.1 p,
commit 9d69e54a3b402b0fad067464bd402e92c14504a9' --with-gcc-major-version-only
--enable-libstdcxx-time --enable-lto --disable-libstdcxx-pch --enable-shared
--enable-threads=posix --enable-__cxa_atexit --enable-clocale=gnu
--enable-multilib --with-multilib-list=m32,m64 --disable-fixed-point
--enable-targets=all --enable-libgomp --disable-libssp --disable-libada
--disable-cet --disable-systemtap --disable-valgrind-annotations
--disable-vtable-verify --disable-libvtv --with-zstd --without-isl
--enable-default-pie --enable-host-pie --disable-host-bind-now
--enable-default-ssp --with-build-config='bootstrap-O3 bootstrap-lto'
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 14.0.1 20240113 (experimental)
9935667a69896865b848dfa690f94c9c693a48a3 (Gentoo 14.0.1 p, commit
9d69e54a3b402b0fad067464bd402e92c14504a9)
```

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Reducing ...

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |14.0
           Keywords|                            |ice-on-valid-code

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fuller backtrace:
```
#2  0x000000000083f554 in dfs_enumerate_from (bb=<optimized out>,
reverse=reverse@entry=1, predicate=predicate@entry=0xe00620
<glb_enum_p(const_basic_block, void const*)>, rslt=<optimized out>,
rslt@entry=0x4604c70, rslt_max=36, data=data@entry=0x7fffddb02960)
    at /home/apinski/src/upstream-gcc-match/gcc/gcc/cfganal.cc:1590
#3  0x0000000000e01272 in get_loop_body_with_size (max_size=<optimized out>,
body=0x4604c70, loop=0x7fffddb02960) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/cfgloop.cc:872
#4  get_loop_body (loop=loop@entry=0x7fffddb02960) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/cfgloop.cc:901
#5  0x00000000014d0bea in estimate_numbers_of_iterations (loop=0x7fffddb02960)
at /home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4833
#6  0x00000000014d4616 in loop_exits_before_overflow (loop=0x7fffddb02960,
at_stmt=0x7fffd7d63d20, step=0x7ffff4314f48, base=0x7ffff76603a8) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:5259
#7  scev_probably_wraps_p (var=var@entry=0x0, base=base@entry=0x7ffff76603a8,
step=step@entry=0x7ffff4314f48, at_stmt=0x7fffd7d63d20,
loop=loop@entry=0x7fffddb02960,
use_overflow_semantics=use_overflow_semantics@entry=false) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:5511
#8  0x0000000002462fff in convert_affine_scev (loop=0x7fffddb02960,
type=<optimized out>, base=0x7fffffffc1d0, step=0x7fffffffc1d8,
at_stmt=<optimized out>, use_overflow_semantics=<optimized out>,
from=0x7fffd814b708) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-chrec.cc:1425
#9  0x0000000002462a59 in chrec_convert_1 (type=0x7ffff7826738,
chrec=0x7fffd7d819b0, at_stmt=0x7fffd7d63d20, use_overflow_semantics=<optimized
out>, from=0x7fffd814b708) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-chrec.cc:1483
#10 0x000000000141b887 in interpret_gimple_assign (stmt=0x7fffd7d63d20,
loop=0x7fffddb02960) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-scalar-evolution.cc:1919
#11 analyze_scalar_evolution_1 (loop=0x7fffddb02960, var=0x7fffd8130828) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-scalar-evolution.cc:1970
#12 0x000000000141c193 in analyze_scalar_evolution (loop=0x7fffddb02960,
var=0x7fffd8130828) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-scalar-evolution.cc:2039
#13 0x000000000141eceb in analyze_scalar_evolution_in_loop
(wrto_loop=wrto_loop@entry=0x7fffddb02960,
use_loop=use_loop@entry=0x7fffddb02960, version=<optimized out>,
folded_casts=folded_casts@entry=0x7fffffffc37b) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-scalar-evolution.cc:2165
#14 0x0000000001420696 in simple_iv_with_niters (wrto_loop=0x7fffddb02960,
use_loop=0x7fffddb02960, op=<optimized out>, iv=0x7fffffffc480, iv_niters=0x0,
allow_nonconstant_step=false) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-scalar-evolution.cc:3237
#15 0x00000000014c2c02 in get_cst_init_from_scev (var=var@entry=0x7fffd8130828,
init=init@entry=0x7fffffffc580, is_min=is_min@entry=true) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4015
#16 0x00000000014c64f0 in record_nonwrapping_iv (loop=0x7fffddaf0320,
base=0x7fffe6031d00, step=0x7ffff692fbb8, stmt=0x7fffd8121cb8,
low=0x7ffff76603a8, high=<optimized out>, upper=true, realistic=false) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4101
#17 0x00000000014c6bb9 in infer_loop_bounds_from_signedness
(loop=0x7fffddaf0320, stmt=0x7fffd8121cb8) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4380
#18 0x00000000014d21c3 in infer_loop_bounds_from_undefined (bbs=<optimized
out>, loop=<optimized out>) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4418
#19 estimate_numbers_of_iterations (loop=0x7fffddaf0320) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4872
#20 0x00000000014d3f38 in estimated_loop_iterations (nit=0x7fffffffd340,
loop=0x7fffddaf0320) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4904
#21 estimated_loop_iterations_int (loop=loop@entry=0x7fffddaf0320) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:4919
#22 0x00000000014d3f69 in estimated_stmt_executions_int
(loop=loop@entry=0x7fffddaf0320) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-ssa-loop-niter.cc:5007
#23 0x00000000015e1c58 in vect_analyze_loop_costing (loop_vinfo=<optimized
out>, suggested_unroll_factor=<optimized out>) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-vect-loop.cc:2450
#24 0x00000000015ee8df in vect_analyze_loop_2 (loop_vinfo=<optimized out>,
fatal=<optimized out>, suggested_unroll_factor=0x7fffffffd83c,
slp_done_for_suggested_uf=@0x7fffffffd83b: false) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-vect-loop.cc:3187
#25 0x00000000015f050f in vect_analyze_loop_1 (loop=0x7fffddaf0320,
shared=0x7fffffffda70, loop_form_info=0x7fffffffd900, main_loop_vinfo=0x0,
vector_modes=..., mode_i=@0x7fffffffd8dc: 0,
autodetected_vector_mode=@0x7fffffffd8d8: E_VOIDmode, fatal=@0x7fffffffd8d7:
false)
    at /home/apinski/src/upstream-gcc-match/gcc/gcc/tree-vect-loop.cc:3482
#26 0x00000000015f0d38 in vect_analyze_loop (loop=loop@entry=0x7fffddaf0320,
shared=shared@entry=0x7fffffffda70) at
/home/apinski/src/upstream-gcc-match/gcc/gcc/tree-vect-loop.cc:3639

```

I am suspecting it was related to Tamar's patches but more related to the patch
which fixed PR113210 .

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

--- Comment #3 from Sam James <sjames at gcc dot gnu.org> ---
Created attachment 57078
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=57078&action=edit
reduced.ii

Attached something a bit smaller but it's not great (not very elegant and too
many warnings).

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2024-01-15
           Priority|P3                          |P1
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.  The ICE means that loop->num_nodes is out-of-date.

I see an early exit created BB which is in the wrong loop.  This:

      for (auto exit : loop_exits)
        {                         
          basic_block dest = main_loop_exit_block;
          if (exit != loop_exit)
            {
              if (!alt_loop_exit_block)
                {
                  alt_loop_exit_block = split_edge (exit);

splits an edge to __assert_fail but is then re-wried

                  edge res = redirect_edge_and_branch (
                                single_succ_edge (alt_loop_exit_block),
                                new_preheader);
                  flush_pending_stmts (res);
                  continue;

not sure why we split the edge before redirecting, swapping fixes the
testcase.  remove_bb_from_loops/add_bb_to_loop could also fixup with
the original order.

I'm testing swapping.

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Created attachment 57080
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=57080&action=edit
Further reduced

Reduced further, removes the warnings too.

Still has so many C++ layers in it though.

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

--- Comment #6 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:afac1bd33657a5054f5e6ea6746c25bbb70b82f2

commit r14-7243-gafac1bd33657a5054f5e6ea6746c25bbb70b82f2
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Jan 15 10:03:59 2024 +0100

    tree-optimization/113385 - wrong loop father with early exit vectorization

    The following avoids splitting an edge before redirecting it.  This
    allows the loop father of the new block to be correct in the first
    place.

            PR tree-optimization/113385
            * tree-vect-loop-manip.cc (slpeel_tree_duplicate_loop_to_edge_cfg):
            First redirect, then split the exit edge.

[Bug tree-optimization/113385] [14 regression] ICE when building opencv (dfs_enumerate_from, at cfganal.cc:1590)

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113385

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> ---
Should be fixed.