[Bug c++/56652] New: Wrong code in O2 - if in inline function

[Bug c++/56652] New: Wrong code in O2 - if in inline function

[dominik.siatkowski at aldec dot com.pl]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

             Bug #: 56652
           Summary: Wrong code in O2 - if in inline function
    Classification: Unclassified
           Product: gcc
           Version: 4.7.2
            Status: UNCONFIRMED
          Severity: major
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dominik.siatkowski@aldec.com.pl


Bug detected with gcc-4.7.2 on i386-linux-gnu

Attached example segfalts compiled with -O2 (or even with '-O1
-finline-small-functions -fstrict-overflow -ftree-vrp')
Works with -O1. Works also if compiled in one compilation unit.

Stepping thru code shows it enters
    if (count >= 0 && count < size)
in shift_right, where count is INT_MIN and size is 2.

[Bug c++/56652] Wrong code in O2 - if in inline function

[dominik.siatkowski at aldec dot com.pl]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

--- Comment #1 from dominik.siatkowski at aldec dot com.pl 2013-03-18 12:05:05 UTC ---
Created attachment 29691
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29691
example

[Bug c++/56652] Wrong code in O2 - if in inline function

[markus at trippelsdorf dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

Markus Trippelsdorf <markus at trippelsdorf dot de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |markus at trippelsdorf dot
                   |                            |de

--- Comment #2 from Markus Trippelsdorf <markus at trippelsdorf dot de> 2013-03-18 12:15:21 UTC ---
Quick check with clang -fsanitize=undefined shows:
...
numeric.cc:47:30: fatal error: negation of -2147483648 cannot be represented in
type 'int'; cast to an unsigned type to negate this value to itself

Thus invalid.

[Bug c++/56652] Wrong code in O2 - if in inline function

[dominik.siatkowski at aldec dot com.pl]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

--- Comment #3 from dominik.siatkowski at aldec dot com.pl 2013-03-18 12:16:35 UTC ---
Is seems its 4.7 reggresion.
Example works if compiled with 4.6.2

[Bug c++/56652] Wrong code in O2 - if in inline function

[dominik.siatkowski at aldec dot com.pl]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

--- Comment #4 from dominik.siatkowski at aldec dot com.pl 2013-03-18 12:24:17 UTC ---
(In reply to comment #2)
> Quick check with clang -fsanitize=undefined shows:
> ...
> numeric.cc:47:30: fatal error: negation of -2147483648 cannot be represented in
> type 'int'; cast to an unsigned type to negate this value to itself
> 
> Thus invalid.

Thank you. OK, works if I change type of count and size to unsigned.
So INT_MIN has wrong value?

[Bug c++/56652] Wrong code in O2 - if in inline function

[markus at trippelsdorf dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

--- Comment #5 from Markus Trippelsdorf <markus at trippelsdorf dot de> 2013-03-18 12:33:50 UTC ---
(In reply to comment #4)
> (In reply to comment #2)
> > Quick check with clang -fsanitize=undefined shows:
> > ...
> > numeric.cc:47:30: fatal error: negation of -2147483648 cannot be represented in
> > type 'int'; cast to an unsigned type to negate this value to itself
> > 
> > Thus invalid.
> 
> Thank you. OK, works if I change type of count and size to unsigned.
> So INT_MIN has wrong value?

-INT_MIN is an integer overflow and is undefined behavior in C.
http://www.google.com/search?hl=en&q=+INT_MIN

[Bug c++/56652] Wrong code in O2 - if in inline function

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56652

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID
           Severity|major                       |normal

--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> 2013-03-20 00:49:45 UTC ---
Invalid, thanks, Markus