Re: iphlapi: Fix some leaks (coverity)

[Alexandre Julliard]

Frédéric Delanoy <frederic.delanoy(a)gmail.com> writes:
> @@ -1520,8 +1521,8 @@ DWORD WINAPI GetIpAddrTable(PMIB_IPADDRTABLE pIpAddrTable, PULONG pdwSize, BOOL
>             sizeof(MIB_IPADDRROW), IpAddrTableSorter);
>          ret = NO_ERROR;
>        }
> -      HeapFree(GetProcessHeap(), 0, table);
>      }
> +    HeapFree(GetProcessHeap(), 0, table);
>    }

That's not the correct fix.

-- 
Alexandre Julliard
julliard(a)winehq.org

iphlapi: Fix some leaks (coverity)

[Juan Lang]

Hi Frédéric,

thanks for the patch. Allow me to elaborate:

On Tue, Nov 6, 2012 at 6:05 AM, Alexandre Julliard <julliard(a)winehq.org>wrote:
> Frédéric Delanoy <frederic.delanoy(a)gmail.com> writes:
> > @@ -1520,8 +1521,8 @@ DWORD WINAPI GetIpAddrTable(PMIB_IPADDRTABLE
> > pIpAddrTable, PULONG pdwSize, BOOL
> >             sizeof(MIB_IPADDRROW), IpAddrTableSorter);
> >          ret = NO_ERROR;
> >        }
> > -      HeapFree(GetProcessHeap(), 0, table);
> >      }
> > +    HeapFree(GetProcessHeap(), 0, table);
> >    }
>
> That's not the correct fix.

The correct place to fix is getIpAddrTable in ifenum.c, so that should it
fail, it frees the memory it allocated. This was obviously an oversight on
my part. See the for loop here:
http://source.winehq.org/source/dlls/iphlpapi/ifenum.c#L735

This loop terminates if something fails, so after the for loop is an ideal
place to add a HeapFree. In this way, the callers (GetIpAddrTable is but
one) never have to cope with freeing memory when getIpAddrTable fails.
--Juan