Re: Invalid compilation without -fno-strict-aliasing

[Daniel Jacobowitz <dan@debian.org>]

On Wed, Feb 26, 2003 at 05:47:48PM -0300, Horst von Brand wrote:
> Falk Hueffner <falk.hueffner@student.uni-tuebingen.de> said:
> > Horst von Brand <vonbrand@inf.utfsm.cl> writes:
> > > Jean Tourrilhes <jt@bougret.hpl.hp.com> said:
> > > > 	if((stream + event_len) < ends) {
> > > > 		iwe->len = event_len;
> > > > 		memcpy(stream, (char *) iwe, event_len);
> > > > 		stream += event_len;
> > > > 	}
> > > > 	return stream;
> > > > }
> > > 
> > > The compiler is free to assume char *stream and struct iw_event *iwe
> > > point to separate areas of memory, due to strict aliasing.
> > 
> > The relevant paragraph of the C99 standard is:
> > 
> > An object shall have its stored value accessed only by an lvalue
> > expression that has one of the following types:
> [...]
> > -- a character type.
> 
> (char *) gives you a (pointer to) a character type.
> 
> > I can't really spot any lvalue here that might violate this rule.  It
> > would be nice if somebody could report a bug with a testcase.
> 
> stream and (char *) iwe

Stream is not the same storage as iwe, so this is hardly the issue. 
Writes to stream don't affect iwe.  The problem was the assignment to
iwe->len being moved after the access, according to the report.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer