[Bug debug/100148] New: [10/11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[Bug debug/100148] New: [10/11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

            Bug ID: 100148
           Summary: [10/11 Regression] -fcompare-debug failure (length)
                    with -O2 -fno-dce -fno-tree-dce
                    -fno-tree-dominator-opts -fno-tree-sink
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
          Assignee: unassigned at gcc dot gnu.org
          Reporter: zsojka at seznam dot cz
                CC: aoliva at gcc dot gnu.org
  Target Milestone: ---
              Host: x86_64-pc-linux-gnu

Created attachment 50630
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50630&action=edit
reduced testcase

Compiler output:
$ x86_64-pc-linux-gnu-gcc -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts
-fno-tree-sink -fcompare-debug testcase.C 
x86_64-pc-linux-gnu-gcc: error: testcase.C: '-fcompare-debug' failure (length)

$ x86_64-pc-linux-gnu-gcc -v
Using built-in specs.
COLLECT_GCC=/repo/gcc-trunk/binary-latest-amd64/bin/x86_64-pc-linux-gnu-gcc
COLLECT_LTO_WRAPPER=/repo/gcc-trunk/binary-trunk-r11-8244-20210419132718-g714bdc31b69-checking-yes-rtl-df-extra-amd64/bin/../libexec/gcc/x86_64-pc-linux-gnu/11.0.1/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: /repo/gcc-trunk//configure --enable-languages=c,c++
--enable-valgrind-annotations --disable-nls --enable-checking=yes,rtl,df,extra
--with-cloog --with-ppl --with-isl --build=x86_64-pc-linux-gnu
--host=x86_64-pc-linux-gnu --target=x86_64-pc-linux-gnu
--with-ld=/usr/bin/x86_64-pc-linux-gnu-ld
--with-as=/usr/bin/x86_64-pc-linux-gnu-as --disable-libstdcxx-pch
--prefix=/repo/gcc-trunk//binary-trunk-r11-8244-20210419132718-g714bdc31b69-checking-yes-rtl-df-extra-amd64
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 11.0.1 20210419 (experimental) (GCC)

[Bug debug/100148] [10/11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
             Status|UNCONFIRMED                 |NEW
   Target Milestone|---                         |10.4
   Last reconfirmed|                            |2021-04-19
                 CC|                            |jakub at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Started with r10-7268-g529ea7d9596b26ba103578eeab448e9862a2d2c5

[Bug debug/100148] [10/11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I bet it goes wrong during cprop1, at least fwprop1 dumps with --param
min-nondebug-insn-uid=10000 look good to me and cprop1 contains
-rescanning insn with uid = 10033.
-GLOBAL CONST-PROP: Replacing reg 82 in jump_insn 10033 with constant
(const_int 0 [0])
-Purged edges from bb 8
-deleting insn with uid = 10033.
-GLOBAL CONST-PROP: Replacing reg 82 in insn 10029 with constant (const_int 0
[0])
-CPROP of foo, 14 basic blocks, 592 bytes needed, 0 local const props, 0 local
copy props, 2 global const props, 0 global copy props
+CPROP of foo, 14 basic blocks, 592 bytes needed, 0 local const props, 0 local
copy props, 0 global const props, 0 global copy props

as the first major difference.

[Bug rtl-optimization/100148] [10/11/12 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

[Bug rtl-optimization/100148] [10/11/12 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 50637
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50637&action=edit
gcc11-pr100148.patch

Untested fix.

[Bug rtl-optimization/100148] [10/11/12 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:022f6ee3ad67ee30f62c8c2aeeb4156494f3284e

commit r12-24-g022f6ee3ad67ee30f62c8c2aeeb4156494f3284e
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Apr 21 12:31:45 2021 +0200

    cprop: Fix -fcompare-debug bug in constprop_register [PR100148]

    The following testcase shows different behavior between -g and -g0
    in constprop_register, if a flags register setter is separated
    from a conditional jump using those flags with -g by a DEBUG_INSN.
    As it uses just NEXT_INSN, for -g it will look at the DEBUG_INSN which is
    not a conditional jump, while otherwise it would look at the conditional
    jump and call cprop_jump.

    2021-04-21  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/100148
            * cprop.c (constprop_register): Use next_nondebug_insn instead of
            NEXT_INSN.

            * g++.dg/opt/pr100148.C: New test.

[Bug rtl-optimization/100148] [10/11/12 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:63353b2aebd2cf171ee8aa747b2dc84f5534de6b

commit r10-9741-g63353b2aebd2cf171ee8aa747b2dc84f5534de6b
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Apr 21 12:31:45 2021 +0200

    cprop: Fix -fcompare-debug bug in constprop_register [PR100148]

    The following testcase shows different behavior between -g and -g0
    in constprop_register, if a flags register setter is separated
    from a conditional jump using those flags with -g by a DEBUG_INSN.
    As it uses just NEXT_INSN, for -g it will look at the DEBUG_INSN which is
    not a conditional jump, while otherwise it would look at the conditional
    jump and call cprop_jump.

    2021-04-21  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/100148
            * cprop.c (constprop_register): Use next_nondebug_insn instead of
            NEXT_INSN.

            * g++.dg/opt/pr100148.C: New test.

    (cherry picked from commit 022f6ee3ad67ee30f62c8c2aeeb4156494f3284e)

[Bug rtl-optimization/100148] [11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[10/11/12 Regression]       |[11 Regression]
                   |-fcompare-debug failure     |-fcompare-debug failure
                   |(length) with -O2 -fno-dce  |(length) with -O2 -fno-dce
                   |-fno-tree-dce               |-fno-tree-dce
                   |-fno-tree-dominator-opts    |-fno-tree-dominator-opts
                   |-fno-tree-sink              |-fno-tree-sink

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed on trunk and for 10.4 so far, queued for 11.2.

[Bug rtl-optimization/100148] [11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:12333b7e7636d7337054b85d37cb2def3b2e4ca4

commit r11-8296-g12333b7e7636d7337054b85d37cb2def3b2e4ca4
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Apr 21 12:31:45 2021 +0200

    cprop: Fix -fcompare-debug bug in constprop_register [PR100148]

    The following testcase shows different behavior between -g and -g0
    in constprop_register, if a flags register setter is separated
    from a conditional jump using those flags with -g by a DEBUG_INSN.
    As it uses just NEXT_INSN, for -g it will look at the DEBUG_INSN which is
    not a conditional jump, while otherwise it would look at the conditional
    jump and call cprop_jump.

    2021-04-21  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/100148
            * cprop.c (constprop_register): Use next_nondebug_insn instead of
            NEXT_INSN.

            * g++.dg/opt/pr100148.C: New test.

    (cherry picked from commit 022f6ee3ad67ee30f62c8c2aeeb4156494f3284e)

[Bug rtl-optimization/100148] [11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed for 11.2 too.

[Bug rtl-optimization/100148] [11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #9 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-9 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:959719061ab3e9cfa11a2551e20cc0d8c5570a4c

commit r9-9503-g959719061ab3e9cfa11a2551e20cc0d8c5570a4c
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Apr 21 12:31:45 2021 +0200

    cprop: Fix -fcompare-debug bug in constprop_register [PR100148]

    The following testcase shows different behavior between -g and -g0
    in constprop_register, if a flags register setter is separated
    from a conditional jump using those flags with -g by a DEBUG_INSN.
    As it uses just NEXT_INSN, for -g it will look at the DEBUG_INSN which is
    not a conditional jump, while otherwise it would look at the conditional
    jump and call cprop_jump.

    2021-04-21  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/100148
            * cprop.c (constprop_register): Use next_nondebug_insn instead of
            NEXT_INSN.

            * g++.dg/opt/pr100148.C: New test.

    (cherry picked from commit 022f6ee3ad67ee30f62c8c2aeeb4156494f3284e)

[Bug rtl-optimization/100148] [11 Regression] -fcompare-debug failure (length) with -O2 -fno-dce -fno-tree-dce -fno-tree-dominator-opts -fno-tree-sink

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100148

--- Comment #10 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-8 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:bef3ed832a97b0cdfb540fc5c5dee923ccdc540a

commit r8-10944-gbef3ed832a97b0cdfb540fc5c5dee923ccdc540a
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Apr 21 12:31:45 2021 +0200

    cprop: Fix -fcompare-debug bug in constprop_register [PR100148]

    The following testcase shows different behavior between -g and -g0
    in constprop_register, if a flags register setter is separated
    from a conditional jump using those flags with -g by a DEBUG_INSN.
    As it uses just NEXT_INSN, for -g it will look at the DEBUG_INSN which is
    not a conditional jump, while otherwise it would look at the conditional
    jump and call cprop_jump.

    2021-04-21  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/100148
            * cprop.c (constprop_register): Use next_nondebug_insn instead of
            NEXT_INSN.

            * g++.dg/opt/pr100148.C: New test.

    (cherry picked from commit 022f6ee3ad67ee30f62c8c2aeeb4156494f3284e)