Date: Tue, 19 Aug 2003 15:02:48 -0700 From: Marcel Moolenaar <marcel@xcllnt.net> To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> Cc: cvs-all@FreeBSD.org Subject: Re: cvs commit: src/include unistd.h Message-ID: <20030819220248.GA91232@athlon.pn.xcllnt.net> In-Reply-To: <200308192132.h7JLWnZk035752@khavrinen.lcs.mit.edu> References: <200308192039.h7JKdni9081097@repoman.freebsd.org> <20030819210421.GA28597@athlon.pn.xcllnt.net> <200308192132.h7JLWnZk035752@khavrinen.lcs.mit.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Aug 19, 2003 at 05:32:49PM -0400, Garrett Wollman wrote:
> <<On Tue, 19 Aug 2003 14:04:21 -0700, Marcel Moolenaar <marcel@xcllnt.net> said:
>
> > Changing the type of the namelen argument from int to size_t in
> > at least the implementation of the function breaks the ABI on
> > 64-bit platforms. The problem is that sign-extension is moved
> > from the callee to the caller and previously compiled code will
> > not have said sign-extension.
>
> Hmmm. I hadn't thought of that, but you're absolutely right.
> Suggestions as to the best way to deal? (I want the type to be
> correct going forward.)
If we care about it (we may not, because I don't think it
imposes a security problem) we could force the sign-extension
in gethostname() by assigning namelen before we do anything
else. Something like:
int
gethostname(name, namelen)
char *name;
size_t namelen;
{
int mib[2];
size_t size;
/*
* Preserve behaviour for when namelen was an int.
* This avoids ABI breakages.
*/
namelen = (int)namelen;
:
:
}
I think that should do it.
--
Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030819220248.GA91232>