Re: usb/gadget: warning in ep_write_iter/__alloc_pages_nodemask

[Michal Hocko <mhocko@kernel.org>]

On Mon 12-12-16 16:12:16, Alan Stern wrote:
> On Mon, 12 Dec 2016, Michal Hocko wrote:
> 
> > On Mon 12-12-16 21:32:35, Andrey Konovalov wrote:
> > > On Mon, Dec 12, 2016 at 9:31 PM, Andrey Konovalov <andreyknvl@google.com> wrote:
> > > > Hi!
> > > >
> > > > While running the syzkaller fuzzer I've got the following error report.
> > > >
> > > > The issue is that the len argument is not checked for being too big.
> > 
> > Well, the value is checked in kmalloc_slab. There is a discrepancy
> > though. While the page allocator enforces order < MAX_ORDER, slab
> > allocators enforce size <= KMALLOC_MAX_SIZE and
> > 
> > KMALLOC_MAX_SIZE is 1UL << (MAX_ORDER + PAGE_SHIFT) for SLUB which is
> > what I expect you have or 1UL << (MAX_ORDER + PAGE_SHIFT - 1) for SLAB
> > on most archs.  This means that KMALLOC_MAX_SIZE might be MAX_ORDER for
> > SLUB which would explain your warning.
> > 
> > Let's CC Christoph, this is nothing really new, but I suspect that SLUB
> > should change the maximum size to something like SLAB does.
> > 
> > That being said, what ep_write_iter does sounds quite stupit. It just
> > allocates a large continuous buffer which seems to be under user
> > control...  Aka no good! It should do that per pages or something like
> > that. Something worth fixing
> 
> It's not important enough to make the driver do all this work.  If
> users want to send large amounts of data, they can send it a page at a
> time (or something like that).

Is it really necessary to allocate the full iov_iter_count? Why cannot
we process the from buffer one page at a time?

> If you really want to prevent the driver from attempting to allocate a
> large buffer, all that's needed is an upper limit on the total size.  
> For example, 64 KB.

Well, my point was that it is not really hard to imagine to deplete
larger contiguous memory blocks (say PAGE_ALLOC_COSTLY_ORDER). Those are
still causing the OOM killer and chances are that a controlled flood of
these requests could completely DoS the system.
-- 
Michal Hocko
SUSE Labs