Re: [ 03/34] lib: proportion: lower PROP_MAX_SHIFT to 32 on 64-bit kernel

[Fengguang Wu <fengguang.wu@intel.com>]

On Mon, Mar 05, 2012 at 04:06:57PM +0100, Jan Kara wrote:
> On Thu 01-03-12 13:39:25, Greg KH wrote:
> > 2.6.32-longterm review patch.  If anyone has any objections, please let me know.
>   Not that I'd see anything wrong with this patch for 2.6.32. But it is
> also unnecessary since the code which was triggering the overflow does not
> exist in 2.6.32. So maybe just on the grounds of not applying unneeded
> patchs I'd skip this one.

FYI I never see this divide error for pre-3.2 kernels. However I've
run into problem (2) before 3.2 which makes bdi dirty threshold go
wild. So it seems safer to go with this patch.

To be frank the boxes that run into bugs (1) or (2) do not have
Terabytes of memory to create the big shift value in
calc_period_shift() which is the sufficient condition for triggering
the bugs as described in the below changelog. However the bugs do
magically go away with the patch applied. Perhaps this patch breaks
one necessary condition for triggering the bugs in a small memory box.

Thanks,
Fengguang

> > ------------------
> > 
> > From: Wu Fengguang <fengguang.wu@intel.com>
> > 
> > commit 3310225dfc71a35a2cc9340c15c0e08b14b3c754 upstream.
> > 
> > PROP_MAX_SHIFT should be set to <=32 on 64-bit box. This fixes two bugs
> > in the below lines of bdi_dirty_limit():
> > 
> > 	bdi_dirty *= numerator;
> > 	do_div(bdi_dirty, denominator);
> > 
> > 1) divide error: do_div() only uses the lower 32 bit of the denominator,
> >    which may trimmed to be 0 when PROP_MAX_SHIFT > 32.
> > 
> > 2) overflow: (bdi_dirty * numerator) could easily overflow if numerator
> >    used up to 48 bits, leaving only 16 bits to bdi_dirty
> > 
> > Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
> > Reported-by: Ilya Tumaykin <librarian_rus@yahoo.com>
> > Tested-by: Ilya Tumaykin <librarian_rus@yahoo.com>
> > Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
> > Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > 
> > ---
> >  include/linux/proportions.h |    4 ++++
> >  1 file changed, 4 insertions(+)
> > 
> > --- a/include/linux/proportions.h
> > +++ b/include/linux/proportions.h
> > @@ -81,7 +81,11 @@ void prop_inc_percpu(struct prop_descrip
> >   * Limit the time part in order to ensure there are some bits left for the
> >   * cycle counter and fraction multiply.
> >   */
> > +#if BITS_PER_LONG == 32
> >  #define PROP_MAX_SHIFT (3*BITS_PER_LONG/4)
> > +#else
> > +#define PROP_MAX_SHIFT (BITS_PER_LONG/2)
> > +#endif
> >  
> >  #define PROP_FRAC_SHIFT		(BITS_PER_LONG - PROP_MAX_SHIFT - 1)
> >  #define PROP_FRAC_BASE		(1UL << PROP_FRAC_SHIFT)
> > 
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
> -- 
> Jan Kara <jack@suse.cz>
> SUSE Labs, CR