Re: WARNING in kmalloc_slab (3)

[Eric Biggers]

On Mon, Dec 04, 2017 at 12:26:32PM +0300, Dan Carpenter wrote:
> On Mon, Dec 04, 2017 at 09:18:05AM +0100, Dmitry Vyukov wrote:
> > On Mon, Dec 4, 2017 at 9:14 AM, Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
> > > On Sun, Dec 03, 2017 at 12:16:08PM -0800, Eric Biggers wrote:
> > >> Looks like BLKTRACESETUP doesn't limit the '.buf_nr' parameter, allowing anyone
> > >> who can open a block device to cause an extremely large kmalloc.  Here's a
> > >> simplified reproducer:
> > >>
> > >
> > > There are lots of places which allow people to allocate as much as they
> > > want.  With Syzcaller, you might want to just hard code a __GFP_NOWARN
> > > in to disable it.
> > 
> > Hi,
> > 
> > Hard code it where?
> 
> My idea was to just make warn_alloc() a no-op.
> 
> > 
> > User-controllable allocation are supposed to use __GFP_NOWARN.
> 
> No that's not right.  What we don't want is unprivileged users to use
> all the memory and we don't want unprivileged users to spam
> /var/log/messages.  But you have to have slightly elevated permissions
> to open block devices right?  The warning is helpful.  Admins should
> "don't do that" if they don't want the warning.

WARN_ON() should only be used for kernel bugs.  printk can be a different story.
If it's a "userspace shouldn't do this" kind of thing, then if there is any
message at all it should be a rate-limited printk that actually explains what
the problem is, not a random WARN_ON() that can only be interpreted by kernel
developers.

And yes, the fact that anyone with read access to any block device, even e.g. a
loop device, can cause the kernel to do an unbounded kmalloc *is* a bug.  It
needs to have a reasonable limit.  It is not a problem on all systems, but on
some systems "the admin" might give users read access to some block devices.

Eric