[Bug gcov-profile/95494] New: [11 regression] Several -fprofile-use tests FAIL

[Bug gcov-profile/95494] New: [11 regression] Several -fprofile-use tests FAIL

[ro at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

            Bug ID: 95494
           Summary: [11 regression] Several -fprofile-use tests FAIL
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: gcov-profile
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ro at gcc dot gnu.org
                CC: marxin at gcc dot gnu.org
  Target Milestone: ---
            Target: sparc-sun-solaris2.11

Between 20200601 (e41b988cc5af34e9c1a3d37b717fbfcc52d7ff90) and 20200602
(d3b6767dce45a7100e4cc32d2986a55f09a2cce2),
several -fprofile-use tests started to FAIL on 32-bit Solaris/SPARC only:

+FAIL: g++.dg/tree-prof/pr35545.C compilation,  -fprofile-use -D_PROFILE_USE
+UNRESOLVED: g++.dg/tree-prof/pr35545.C execution,    -fprofile-use
-D_PROFILE_USE

/vol/gcc/src/hg/master/local/gcc/testsuite/g++.dg/tree-prof/pr35545.C: In
member function 'virtual int B::foo()':
/vol/gcc/src/hg/master/local/gcc/testsuite/g++.dg/tree-prof/pr35545.C:16:15:
warning: profile for function 'virtual int B::foo()' not found in profile data
[-Wmissing-profile]
/vol/gcc/src/hg/master/local/gcc/testsuite/g++.dg/tree-prof/pr35545.C: In
member function 'virtual int A::foo()':
/vol/gcc/src/hg/master/local/gcc/testsuite/g++.dg/tree-prof/pr35545.C:6:15:
warning: profile for function 'virtual int A::foo()' not found in profile data
[-Wmissing-profile]

+FAIL: gcc.dg/tree-prof/indir-call-prof-2.c compilation,  -fprofile-use
-D_PROFILE_USE
+UNRESOLVED: gcc.dg/tree-prof/indir-call-prof-2.c execution,    -fprofile-use
-D_PROFILE_USE

/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-2.c:
In function 'sub1':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-2.c:10:1:
warning: profile for function 'sub1' not found in profile data
[-Wmissing-profile]
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-2.c:
In function 'add1':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-2.c:4:1:
warning: profile for function 'add1' not found in profile data
[-Wmissing-profile]

+FAIL: gcc.dg/tree-prof/indir-call-prof-topn.c compilation,  -fprofile-use
-D_PROFILE_USE
+UNRESOLVED: gcc.dg/tree-prof/indir-call-prof-topn.c execution,   
-fprofile-use -D_PROFILE_USE

/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-topn.c:
In function 'two':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-topn.c:14:1:
warning: profile for function 'two' not found in profile data
[-Wmissing-profile]
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-topn.c:
In function 'one':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof-topn.c:8:1:
warning: profile for function 'one' not found in profile data
[-Wmissing-profile]

+FAIL: gcc.dg/tree-prof/indir-call-prof.c compilation,  -fprofile-use
-D_PROFILE_USE
+UNRESOLVED: gcc.dg/tree-prof/indir-call-prof.c execution,    -fprofile-use
-D_PROFILE_USE

/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof.c:
In function 'setp':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof.c:17:6:
warning: profile for function 'setp' not found in profile data
[-Wmissing-profile]
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof.c:
In function 'a2':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof.c:8:12:
warning: profile for function 'a2' not found in profile data
[-Wmissing-profile]
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof.c:
In function 'a1':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/indir-call-prof.c:3:12:
warning: profile for function 'a1' not found in profile data
[-Wmissing-profile]

+FAIL: gcc.dg/tree-prof/pr59003.c compilation,  -fprofile-use -D_PROFILE_USE
+UNRESOLVED: gcc.dg/tree-prof/pr59003.c execution,    -fprofile-use
-D_PROFILE_USE

/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/pr59003.c: In
function 'foo':
/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/pr59003.c:9:10:
error: corrupted value profile: stringops profile counter (28246016 out of
100001) inconsistent with basic-block count (100001)

+FAIL: gcc.dg/tree-prof/val-prof-2.c scan-ipa-dump profile "Transformation
done: div/mod by constant 256"

+FAIL: gcc.dg/tree-prof/val-prof-6.c compilation,  -fprofile-use -D_PROFILE_USE
+UNRESOLVED: gcc.dg/tree-prof/val-prof-6.c execution,    -fprofile-use
-D_PROFILE_USE

/vol/gcc/src/hg/master/local/gcc/testsuite/gcc.dg/tree-prof/val-prof-6.c:8:3:
error: corrupted value profile: stringops profile counter (28246016 out of
1000) inconsistent with basic-block count (1000)

All of 64-bit Solaris/SPARC, 32 and 64-bit Solaris/x86 are still ok.

It seems highly probably that this was caused by

changeset:   233568:8312a62869f6
user:        Martin Liska <mliska@suse.cz>
date:        Fri Jan 31 13:10:14 2020 +0100
files:       gcc/coverage.c gcc/doc/gcov-dump.texi gcc/gcov-dump.c
gcc/gcov-io.h gcc/ipa-profile.c gcc/profile.c gcc/value-prof.c
libgcc/libgcov-driver.c libgcc/libgcov-merge.c libgcc/libgcov-profiler.c
libgcc/libgcov.h
description:
Make TOPN counter dynamically allocated.

the only pertinent patch in that revision range.

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[ro at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

Rainer Orth <ro at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |11.0

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2020-06-03
             Status|UNCONFIRMED                 |WAITING
     Ever confirmed|0                           |1

--- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> ---
Can you please test the current master?
This patch could fix it: a04b7410d305800b747963ab940d2b1a602b5ddf

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

--- Comment #2 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> ---
> Can you please test the current master?
> This patch could fix it: a04b7410d305800b747963ab940d2b1a602b5ddf

Unfortunately, it doesn't make a difference.

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Is there a compile farm machine I can test it on?
Btw. can you please make a brief analysis why it fails (valgrind)?

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

--- Comment #4 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
> Is there a compile farm machine I can test it on?

Sure: gcc211 should do the trick.

> Btw. can you please make a brief analysis why it fails (valgrind)?

Unfortunately not: the Solaris/x86 valgrind port has been abandoned and
doesn't do anything useful, and there never was a sparc port.

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

--- Comment #5 from Martin Liška <marxin at gcc dot gnu.org> ---
(In reply to ro@CeBiTec.Uni-Bielefeld.DE from comment #4)
> > --- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
> > Is there a compile farm machine I can test it on?
> 
> Sure: gcc211 should do the trick.

Thanks.
Unfortunately, I can't build the current master due to switch to C++11:

../../gcc/expr.c: In function ‘rtx_insn* emit_move_insn(rtx, rtx)’:
../../gcc/expr.c:3830:3: warning: ‘auto’ changes meaning in C++11; please
remove it [-Wc++0x-compat]
   auto candidate_subreg_p = [&](rtx subreg) {
   ^
../../gcc/expr.c:3830:8: error: ‘candidate_subreg_p’ does not name a type
   auto candidate_subreg_p = [&](rtx subreg) {
        ^
../../gcc/expr.c:3838:3: warning: ‘auto’ changes meaning in C++11; please
remove it [-Wc++0x-compat]
   auto candidate_mem_p = [&](machine_mode innermode, rtx mem) {
   ^
../../gcc/expr.c:3838:8: error: ‘candidate_mem_p’ does not name a type
   auto candidate_mem_p = [&](machine_mode innermode, rtx mem) {
        ^
../../gcc/expr.c:3849:44: error: ‘candidate_subreg_p’ was not declared in this
scope
   if (SUBREG_P (x) && candidate_subreg_p (x))
                                            ^
../../gcc/expr.c:3852:44: error: ‘candidate_subreg_p’ was not declared in this
scope
   if (SUBREG_P (y) && candidate_subreg_p (y))
                                            ^
../../gcc/expr.c:3866:46: error: ‘candidate_mem_p’ was not declared in this
scope
     && candidate_mem_p (GET_MODE (x_inner), y))

Is there a compiler in /opt that can build current GCC?

> 
> > Btw. can you please make a brief analysis why it fails (valgrind)?
> 
> Unfortunately not: the Solaris/x86 valgrind port has been abandoned and
> doesn't do anything useful, and there never was a sparc port.

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

--- Comment #6 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #5 from Martin Liška <marxin at gcc dot gnu.org> ---
> (In reply to ro@CeBiTec.Uni-Bielefeld.DE from comment #4)
>> > --- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
>> > Is there a compile farm machine I can test it on?
>> 
>> Sure: gcc211 should do the trick.
>
> Thanks.
> Unfortunately, I can't build the current master due to switch to C++11:
[...]
> Is there a compiler in /opt that can build current GCC?

Drats, this system (like several others in the cfarm) urgently needs
some love ;-)  I've placed the self-compiled gcc 8.1.0 which I'm using
all the time on my systems to bootstrap master into ~ro/gcc/8.

You will probably have to point configure at /opt/csw for sufficiently
recent gmp, mpfr, and mpc, though.

All this should live in /opt/cfarm/{gcc,gmp,mpfr,mpc}-latest, following
cfarm standards...

        Rainer

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |ASSIGNED

--- Comment #7 from Martin Liška <marxin at gcc dot gnu.org> ---
Confirmed, I've got hopefully a patch candidate for it.

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |clyon at gcc dot gnu.org

--- Comment #8 from Martin Liška <marxin at gcc dot gnu.org> ---
*** Bug 95600 has been marked as a duplicate of this bug. ***

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

--- Comment #9 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Martin Liska <marxin@gcc.gnu.org>:

https://gcc.gnu.org/g:862b9b225fba6cf3c63234206f2dbc47f1ab5350

commit r11-1114-g862b9b225fba6cf3c63234206f2dbc47f1ab5350
Author: Martin Liska <mliska@suse.cz>
Date:   Mon Jun 8 20:07:08 2020 +0200

    libgcov: fix TOPN type casting

    The patch fixes tree-prof.exp tests on solaris11 and i686-linux-gnu,
    problem was that sizeof of a pointer is different from sizeof gcov_type.

    I'm going to install it if there are no objections.
    Thanks,
    Martin

    libgcc/ChangeLog:

            PR gcov-profile/95494
            * libgcov-driver.c (write_top_counters): Cast first to
            intptr_t as sizeof(*) != sizeof(gcov_type).
            * libgcov.h (gcov_counter_set_if_null): Remove.
            (gcov_topn_add_value): Cast first to intptr_t and update
            linked list directly.

[Bug gcov-profile/95494] [11 regression] Several -fprofile-use tests FAIL

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95494

Martin Liška <marxin at gcc dot gnu.org> changed:

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

--- Comment #10 from Martin Liška <marxin at gcc dot gnu.org> ---
Fixed now.