Re: [PING 2] [PATCH] c++/67942 - diagnose placement new buffer overflow

[Martin Sebor <msebor@gmail.com>]

On 11/06/2015 05:50 AM, Andreas Schwab wrote:
> I see this failure on m68k:
>
> FAIL: g++.dg/warn/Wplacement-new-size.C  -std=gnu++11 (test for excess errors)
> Excess errors:
> /daten/aranym/gcc/gcc-20151106/gcc/testsuite/g++.dg/warn/Wplacement-new-size.C:189:19: warning: placement new constructing an object of type 'int' and size '4' in a region of type 'char [4]' and size '0' [-Wplacement-new]
> /daten/aranym/gcc/gcc-20151106/gcc/testsuite/g++.dg/warn/Wplacement-new-size.C:191:19: warning: placement new constructing an object of type 'int' and size '4' in a region of type 'char [4]' and size '0' [-Wplacement-new]
> /daten/aranym/gcc/gcc-20151106/gcc/testsuite/g++.dg/warn/Wplacement-new-size.C:194:19: warning: placement new constructing an object of type 'int' and size '4' in a region of type 'char [4]' and size '0' [-Wplacement-new]
> /daten/aranym/gcc/gcc-20151106/gcc/testsuite/g++.dg/warn/Wplacement-new-size.C:198:19: warning: placement new constructing an object of type 'int' and size '4' in a region of type 'char [4]' and size '0' [-Wplacement-new]
>
> That appears to be a 32-bit problem, the test also fails here on x86-64
> with -m32 <http://gcc.gnu.org/ml/gcc-testresults/2015-11/msg00522.html>
> or here on powerpc
> <http://gcc.gnu.org/ml/gcc-testresults/2015-11/msg00520.html>

This should be fixed now via r229959 (tested on x86_64 with -m32).

The problem was caused by assuming that the POINTER_PLUS_EXPR offset
which is stored as sizetype, an unsigned 32-bit type in ILP32, can
be "extracted" as an unsigned HOST_WIDE_INT (a 64-bit type when
the host compiler is LP64), and converted to signed to obtain the
original negative offset.

Martin