[Bug testsuite/108192] New: g++.dg/cet-notrack-1.C searching for wrong function on mingw

[Bug testsuite/108192] New: g++.dg/cet-notrack-1.C searching for wrong function on mingw

[nightstrike at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

            Bug ID: 108192
           Summary: g++.dg/cet-notrack-1.C searching for wrong function on
                    mingw
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: testsuite
          Assignee: unassigned at gcc dot gnu.org
          Reporter: nightstrike at gmail dot com
  Target Milestone: ---

Created attachment 54139
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54139&action=edit
cet-notrack-1.s

g++.dg/cet-notrack-1.C Tries to find 'call puts()' in the asm after calling
printf() on the c++ side.  On Mingw, we do magic things with converting
printf() into __mingw_ variants to support MS style printf and other things. 
So the assembler dump on mingw results in a call to __mingw_vfprintf in this
particular case.

It would likely be possible to fix the testcase by searching for "call
__mingw_*printf" or something, but I'm curious if this is really a case of a
missed optimization.  The strings in question are simple string literals, so we
don't care about mingw or ms special versions of printf, and a call to puts
should be fine.  I don't know if this testcase is specifically trying to
validate that optimization, but it at least indirectly revealed it.

Asm attached.

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> ---
Since Windows doesn't support IBT, this test can be limited to Linux.

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[nightstrike at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

--- Comment #2 from nightstrike <nightstrike at gmail dot com> ---
(In reply to H.J. Lu from comment #1)
> Since Windows doesn't support IBT, this test can be limited to Linux.

I don't know what IBT is, but if I change the two printf's to puts(), the tests
pass.  So, maybe they don't have to be skipped?

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[nightstrike at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

--- Comment #3 from nightstrike <nightstrike at gmail dot com> ---
Created attachment 54209
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54209&action=edit
Patch to change printf to puts so it works everywhere

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[10walls at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

jon_y <10walls at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |10walls at gmail dot com

--- Comment #4 from jon_y <10walls at gmail dot com> ---
Is Nightstrike's patch OK?

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to jon_y from comment #4)
> Is Nightstrike's patch OK?

I think it is correct but patches should goto gcc-patches@ really.

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jonathan Yong <jyong@gcc.gnu.org>:

https://gcc.gnu.org/g:6464409b0c7077e36858989b5dae302e2d26f45a

commit r13-5475-g6464409b0c7077e36858989b5dae302e2d26f45a
Author: Jonathan Yong <10walls@gmail.com>
Date:   Wed Jan 11 09:51:02 2023 +0000

    PR c/108192 - Fix test for mingw

    gcc/testsuite/ChangeLog:

            PR c/108192
            * g++.dg/cet-notrack-1.C: Use puts instead of printf,
            so function call is not mangled by __mingw_printf when
            doing assembly symbol inspection.

    Signed-off-by: Jonathan Yong <10walls@gmail.com>

[Bug testsuite/108192] g++.dg/cet-notrack-1.C searching for wrong function on mingw

[jyong at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108192

jyong at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jyong at gcc dot gnu.org
         Resolution|---                         |FIXED
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #7 from jyong at gcc dot gnu.org ---
Changed to use __mingw_*printf.