This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH, PR43814] Assume function arguments of pointer type are aligned.


On Wed, Sep 28, 2011 at 06:43:04PM -0400, David Miller wrote:
> From: Mike Stump <mikestump@comcast.net>
> Date: Wed, 28 Sep 2011 15:19:10 -0700
> 
> > If Android is safe in this respect, then, they can just turn it on,
> > and then force anyone porting software to their platform to `fix'
> > their code.
> 
> They'd have to then know to turn this option off when building the kernel,
> which does use such constructs extensively.
> 
> I think this whole idea has too many downsides to be considered seriously.
> 
> People write problems like this, lots of people.  It's a pervasive
> technique to encode boolean state into the low bits of a pointer or to
> represent special "token" pointers using integers such as "-1".

Or "1".  Just do
grep '\*)[[:blank:]]*1' *.[chS]
to see how often an integer value is stored into a pointer.
And it is not just void * pointers, it is struct cgraph_node *
or struct varpool_node * too and the pointed types there certainly
have higher alignment than 1.
Now repeat the same with google codesearch.

	Jakub


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]