Re: [PATCH 1/6] mm/gup_benchmark: Time put_page

[Keith Busch <keith.busch@intel.com>]

On Wed, Oct 10, 2018 at 03:41:11PM -0700, Andrew Morton wrote:
> On Wed, 10 Oct 2018 16:28:43 -0600 Keith Busch <keith.busch@intel.com> wrote:
> 
> > > >  struct gup_benchmark {
> > > > -	__u64 delta_usec;
> > > > +	__u64 get_delta_usec;
> > > > +	__u64 put_delta_usec;
> > > >  	__u64 addr;
> > > >  	__u64 size;
> > > >  	__u32 nr_pages_per_call;
> > > 
> > > If we move put_delta_usec to the end of this struct, the ABI remains
> > > back-compatible?
> > 
> > If the kernel writes to a new value appended to the end of the struct,
> > and the application allocated the older sized struct, wouldn't that
> > corrupt the user memory?
> 
> Looks like it.  How about we do this while we're breaking it?

Yep, that sounds good to me!
 
> --- a/mm/gup_benchmark.c~mm-gup_benchmark-time-put_page-fix
> +++ a/mm/gup_benchmark.c
> @@ -14,6 +14,7 @@ struct gup_benchmark {
>  	__u64 size;
>  	__u32 nr_pages_per_call;
>  	__u32 flags;
> +	__u64 expansion[10];	/* For future use */
>  };
>  
>  static int __gup_benchmark_ioctl(unsigned int cmd,
>