Re: [PATCH] find: Do not read beyond variable boundaries on small sizes

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Tue, Dec 07, 2021 at 03:39:30PM -0800, Yury Norov wrote:
> On Fri, Dec 03, 2021 at 03:01:30PM -0800, Kees Cook wrote:
> > On Fri, Dec 03, 2021 at 10:26:38AM -0800, Yury Norov wrote:
> > > On Fri, Dec 03, 2021 at 02:30:35PM +0200, Andy Shevchenko wrote:
> > > > On Fri, Dec 03, 2021 at 02:08:46AM -0800, Kees Cook wrote:
> > > > > It's common practice to cast small variable arguments to the find_*_bit()
> > > 
> > > Not that common - I found 19 examples of this cast, and most of them
> > > are in drivers.
> > 
> > I find 51 (most are in the for_each_* wrappers):
> > 
> > $ RE=$(echo '\b('$(echo $(grep -E '^(unsigned long find|#define for_each)_' include/linux/find.h | cut -d'(' -f1 | awk '{print $NF}') | tr ' ' '|')')\(.*\(unsigned long \*\)')
> > $ git grep -E "$RE" | wc -l
> > 51
> > 
> > > > > This leads to the find helper dereferencing a full unsigned long,
> > > > > regardless of the size of the actual variable. The unwanted bits
> > > > > get masked away, but strictly speaking, a read beyond the end of
> > > > > the target variable happens. Builds under -Warray-bounds complain
> > > > > about this situation, for example:
> > > > > 
> > > > > In file included from ./include/linux/bitmap.h:9,
> > > > >                  from drivers/iommu/intel/iommu.c:17:
> > > > > drivers/iommu/intel/iommu.c: In function 'domain_context_mapping_one':
> > > > > ./include/linux/find.h:119:37: error: array subscript 'long unsigned int[0]' is partly outside array bounds of 'int[1]' [-Werror=array-bounds]
> > > > >   119 |                 unsigned long val = *addr & GENMASK(size - 1, 0);
> > > > >       |                                     ^~~~~
> > > > > drivers/iommu/intel/iommu.c:2115:18: note: while referencing 'max_pde'
> > > > >  2115 |         int pds, max_pde;
> > > > >       |                  ^~~~~~~
> > > 
> > > The driver should be fixed. I would suggest using one of ffs/fls/ffz from
> > > include/asm/bitops.h
> > 
> > I don't think it's a good API design to make developers choose between
> > functions based on the size of their target.
> 
> Bitmap functions work identically for all sizes from 0 to INT_MAX - 1. 
> Users don't 'choose between functions based on the size of their target'.
> 
> Can you explain more what you mean?

I believe it was a reaction to your suggestion about ffs/ffz/etc.

Kees, if we _know_ that the size of the value in question will always
fit 32/64-bit, then ffs/ffz/etc is okay to use. OTOH, if the type of
that value is unsigned long [] and bitmap APIs() is used, then of
course the consistent use of bitmap APIs is preferable.

I.o.w.
 uXX: ffX()/etc is fine.
 unsigned long: bitmap API.

I believe that's what Yury meant.

-- 
With Best Regards,
Andy Shevchenko