[Bug fortran/64927] New: [4.7/4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[Bug fortran/64927] New: [4.7/4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[anlauf at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

            Bug ID: 64927
           Summary: [4.7/4.8 Regression] Surprising error with
                    -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: anlauf at gmx dot de

The following error occurs with 4.7 and 4.8, but not with 4.6 or 4.9.
The error is slightly annoying since -Wsurprising is automatically
enabled by -Wall, and it was a try-and-error by a colleague to trace
the problem to -Wsurprising.

The error shows up only if both indicated code lines with TRANSFER
and C_ASSOCIATED are active, commenting out any of them makes the
error disappear.  The dump-tree for 4.8 and 4.9 does not show
any difference if -Wno-surprising is set.  The error is bogus.

> cat gfcbug130.f90
! gfortran -c gfcbug130.f90 -Wall
! false error, compare -Wsurprising vs. -Wno-surprising

module surprise
  use iso_c_binding
  implicit none
  TYPE t_pointer
     TYPE(c_ptr) :: cptr
  END TYPE t_pointer
contains
  SUBROUTINE get_pointer (ibuf)
    INTEGER, INTENT(IN) :: ibuf(:)
    TYPE(t_pointer)     :: zp
! Error disappears if any line is commented out:
    zp = TRANSFER (ibuf, zp)
    IF (C_ASSOCIATED (zp% cptr)) Print *, "OK"
  END SUBROUTINE get_pointer
end module surprise

> gfc-48 -c gfcbug130.f90 -Wall       
gfcbug130.f90:16.22:

    IF (C_ASSOCIATED (zp% cptr)) Print *, "OK"
                      1
Error: Type mismatch in argument 'c_ptr_1' at (1); passed INTEGER(4) to
TYPE(c_ptr)

> gfc-48 -c gfcbug130.f90 -Wall -Wno-surprising

The problem does not occur in 4.9.  If there is an
easy fix for 4.8, could it be backported?

(I understand that the 4.7 branch is already closed.)

Thanks,
Harald

[Bug fortran/64927] [4.7/4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[anlauf at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #1 from Harald Anlauf <anlauf at gmx dot de> ---
There's also a typo in the online documentation on C_ASSOCIATED:

https://gcc.gnu.org/onlinedocs/gfortran/C_005fASSOCIATED.html

"c_prt_1" should read "c_ptr_1" in two places.

[Bug fortran/64927] [4.7/4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

kargl at gcc dot gnu.org changed:

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

--- Comment #2 from kargl at gcc dot gnu.org ---
As the error does not appear in either 4.9 or 5.0, I
suspect that this should be closest as WONTFIX.  But,
I'll let other, more active, gfortran developers make
that call.

I'll fix the typos in the manual in few minutes.

[Bug fortran/64927] [4.7/4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #3 from Steve Kargl <sgk at troutmask dot apl.washington.edu> ---
On Tue, Feb 03, 2015 at 09:49:16PM +0000, kargl at gcc dot gnu.org wrote:
> 
> I'll fix the typos in the manual in few minutes.
>

Fixed by r220381 in 5.0 and by r220382 in 4.9.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4
   Target Milestone|---                         |4.8.5
            Summary|[4.7/4.8 Regression]        |[4.8 Regression] Surprising
                   |Surprising error with       |error with -Wsurprising
                   |-Wsurprising (-Wall) and    |(-Wall) and TRANSFER +
                   |TRANSFER + C_ASSOCIATED     |C_ASSOCIATED

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2015-02-11
     Ever confirmed|0                           |1

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
This PR has been fixed between r197969 (2013-04-15, warning) and r198189
(2013-04-23, no warning). I did not find any obvious commit for the change. I
share with Steve the feeling that this PR can be closed as FIXED.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[anlauf at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #5 from Harald Anlauf <anlauf at gmx dot de> ---
(In reply to Dominique d'Humieres from comment #4)
> This PR has been fixed between r197969 (2013-04-15, warning) and r198189
> (2013-04-23, no warning). I did not find any obvious commit for the change.
> I share with Steve the feeling that this PR can be closed as FIXED.

The revision numbers you refer to belong to the 4.9-branch.  The bug
is still present on the 4.8 branch; I see it with gfortran 4.8.2
on a Debian system and on an OpenSuse 13.1.  I think I have also
verified against some pre-4.8.5 snapshot (need to check the exact
date when I pulled and built that one).

IMO it is not fixed on 4.8.  If there is no easy solution, I'd rather
prefer to mark it in an appropriate way (wontfix?), so that others
can see that this problem is known for particular gcc versions.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #6 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> The revision numbers you refer to belong to the 4.9-branch.

Indeed -> I was trying to find the commit that fixed the problem (without
success).

> IMO it is not fixed on 4.8. If there is no easy solution, I'd rather
> prefer to mark it in an appropriate way (wontfix?), so that others
> can see that this problem is known for particular gcc versions.

I think the only "appropriate way" is FIXED as it is the case for 4.9 and
trunk. Closing as WONTFIX will be misleading for these revisions. Note that 4.8
will be closed pretty soon.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #7 from Steve Kargl <sgk at troutmask dot apl.washington.edu> ---
On Wed, Feb 11, 2015 at 07:31:50PM +0000, anlauf at gmx dot de wrote:
> 
> IMO it is not fixed on 4.8.  If there is no easy solution, I'd rather
> prefer to mark it in an appropriate way (wontfix?), so that others
> can see that this problem is known for particular gcc versions.
> 

You're more than welcomed to backport whatever patch(es) fixes
the issue to 4.8 branch.  The reality of the matter is that
there are too few gfortran contributors to support 3 active
branches.  With the encroachment of C++ into the gfortran
frontend, I know at least one long time gfortran contributor
who is essentially inactive because he does not know C++ and
has no time or desire to learn it.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[anlauf at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #9 from Harald Anlauf <anlauf at gmx dot de> ---
(In reply to Steve Kargl from comment #7)
> On Wed, Feb 11, 2015 at 07:31:50PM +0000, anlauf at gmx dot de wrote:
> > 
> > IMO it is not fixed on 4.8.  If there is no easy solution, I'd rather
> > prefer to mark it in an appropriate way (wontfix?), so that others
> > can see that this problem is known for particular gcc versions.
> > 
> 
> You're more than welcomed to backport whatever patch(es) fixes
> the issue to 4.8 branch.

After Dominique pointed out a range of revisions where the
bug disappeared in the 4.9 branch, I browsed through the list
of svn log messages.  However, all fortran-related commits that
looked interesting to me (some of them TRANSFER related) appeared
to have backports to the 4.8 branch.  Maybe I was not careful enough.

If I'd knew which one fixed it for 4.9, I could at least try.
And, since you mentioned that, I'm not thrown off by C++.
(Although I personally don't use it if I can avoid it).

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #10 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> After Dominique pointed out a range of revisions where the
> bug disappeared in the 4.9 branch, I browsed through the list
> of svn log messages.  However, all fortran-related commits that
> looked interesting to me (some of them TRANSFER related) appeared
> to have backports to the 4.8 branch.  Maybe I was not careful enough.

AFAICT r198000 (pr56969) has not been back ported. The patch is almost trivial.
If you don't beat me, I may find the motivation to do some testing during the
weekend.

> My reasoning is based on the observation that Tobias Burnus
> just closed PR64474 as WONTFIX for the similar reason:
> 4.9 and 5 don't have the resp. bug, all 4.8.x have.

Well, at the same time Bernd Edlinger closed pr60718 as FIXED while the problem
is still present in versions 4.8 and 4.9. The problem is that 'WONTFIX for
version m, FIXED for version n>m' cannot be encoded in the Status field. My
decision rule is the following:

WONTFIX: the issue is present in version m and will be present in any version
n>m.
This typically used for extension or diagnostic requests.

FIXED: the issue is present in version m, has been fixed in version n, and will
be absent in any version p>=n (otherwise it will be a regression).
This is the default rule when a PR is not a regression.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[anlauf at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #11 from Harald Anlauf <anlauf at gmx dot de> ---
(In reply to Dominique d'Humieres from comment #10)
> > After Dominique pointed out a range of revisions where the
> > bug disappeared in the 4.9 branch, I browsed through the list
> > of svn log messages.  However, all fortran-related commits that
> > looked interesting to me (some of them TRANSFER related) appeared
> > to have backports to the 4.8 branch.  Maybe I was not careful enough.
> 
> AFAICT r198000 (pr56969) has not been back ported. The patch is almost
> trivial. If you don't beat me, I may find the motivation to do some testing
> during the weekend.

I pulled yesterday's 4.8 snapshot, bootstrapped it to verify that
the problem still existed, and applied the 1-liner from r198000.
It does not fix the issue.

Anyway, thanks for suggesting to try this one.  Don't waste your time.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #12 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> I pulled yesterday's 4.8 snapshot, bootstrapped it to verify that
> the problem still existed, and applied the 1-liner from r198000.
> It does not fix the issue.

Confirmed, for the record, testing results at

https://gcc.gnu.org/ml/gcc-testresults/2015-02/msg01669.html

> Anyway, thanks for suggesting to try this one.  Don't waste your time.

I am leaving this PR in the WAITING state since the issue may be latent in 4.9
and above.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #13 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Bisecting the range given in comment 4 shows that this PR is fixed by r198155
for 4.9. The fix was supposed to be back ported to 4.8 by r198345. However
comparing the logs shows no entry for gcc/fortran/simplify.c in r198345.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

--- Comment #14 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
The reason why the changes for gcc/fortran/simplify.c on 4.9 have not been
applied to 4.8 is that these changes apply to procs introduced on 4.9 at
r197159, but not on 4.8. AFAICT these changes are needed to fix this PR and I
did not find a simple way to fix this PR for 4.8.

Since 4.8 will be history in a couple of months, I don't think fixing this PR
for 4.8 is worth the work.

[Bug fortran/64927] [4.8 Regression] Surprising error with -Wsurprising (-Wall) and TRANSFER + C_ASSOCIATED

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64927

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|---                         |FIXED
   Target Milestone|4.8.5                       |4.9.0
      Known to fail|                            |4.8.5

--- Comment #15 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed for 4.9.0.