Re: type punning error in packet-gtpv2.c

[Martin Mathieson <martin.r.mathieson () googlemail com>]

On Tue, Jul 6, 2010 at 3:33 PM, Ed Beroset <beroset () mindspring com> wrote:

> I just updated and attempted to build Wireshark on a 32-bit Linux box using
> gcc 4.1.2 and I find that I get an error in the
> epan/dissectors/packet-gtpv2.c file:
>
> packet-gtpv2.c: In function 'proto_register_gtpv2':
> packet-gtpv2.c:2136: warning: dereferencing type-punned pointer will break
> strict-aliasing rules
> packet-gtpv2.c:2161: warning: dereferencing type-punned pointer will break
> strict-aliasing rules
>
> Line 2136 contains the VALS() macro invocation:
>        { &hf_gtpv2_cause_type,
>        {"Cause Type","gtpv2.cause_type",
>        FT_UINT8, BASE_DEC|BASE_EXT_STRING,
> VALS(&gtpv2_cause_type_vals_ext),0x0f,
>        NULL, HFILL}
>        },
>
> The problem is with the VALS() invocation, but before I dive into it, I
> thought I'd see if anybody else is seeing this or is already working on it.
>
> If I back up to version 33350 of packet-gtpv2.c, I don't get any errors and
> it builds just fine.  As an interesting point of reference, I don't get this
> error on my 64-bit Linux box (gcc 4.4.3), nor on the Windows machine (MSVC
> Express 2008).  I wonder if this might be a compiler version issue.  This
> box is running Fedora Core 8 (yes, I know it's old).
>
> Ed

Hi Ed,
I had the same warnings under gcc 3.4.6.  Until now I'd just removed it from
my local copy of epan/dissectors/Makefile.common, but I might actually want
to use this dissector now.

value_type_ext hasn't been used  much, but I followed the same advice that
arose from a discussion when packet-q708.c was broken in a similar way.

Martin

___________________________________________________________________________
Sent via:    Wireshark-dev mailing list <wireshark-dev () wireshark org>
Archives:    http://www.wireshark.org/lists/wireshark-dev
Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev
             mailto:wireshark-dev-request () wireshark org?subject=unsubscribe