Re: [PATCH -next] mm: delete oversized WARN_ON() in kvmalloc() calls

[Matthew Wilcox]

On Thu, Dec 02, 2021 at 02:03:43PM -0800, Andrew Morton wrote:
> On Thu, 2 Dec 2021 13:23:13 -0800 Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> 
> > > > I think we have two cases:
> > > > 
> > > > - limiting kvmalloc allocations to INT_MAX
> > > > - issuing a WARN when that limit is exceeded
> > > > 
> > > > The argument for the having the WARN is "that amount should never be
> > > > allocated so we want to find the pathological callers".
> > > > 
> > > > But if the actual issue is that >INT_MAX is _acceptable_, then we have
> > > > to do away with the entire check, not just the WARN.
> > > 
> > > First we need to get rid from WARN_ON(), which is completely safe thing to do.
> > > 
> > > Removal of the check can be done in second step as it will require audit
> > > of whole kvmalloc* path.
> > 
> > If those are legit sizes, I'm fine with dropping the WARN. (But I still
> > think if they're legit sizes, we must also drop the INT_MAX limit.)
> 
> Can we suppress the WARN if the caller passed __GFP_NOWARN?

I don't think that's a good idea.  NOWARN is for allocation failure
messages whereas this warning is more of a "You're doing something
wrong" -- ENOMEM vs EINVAL.

I'm still agnostic on whether this should be a check at all, or whether
we should let people kvmalloc(20GB).  But I don't like conditioning the
warning on GFP_NOWARN.