Re: [PATCH] sys_getppid oopses on debug kernel

[Martin Schwidefsky <schwidefsky@de.ibm.com>]

On Tue, 2006-08-08 at 08:49 -0700, Dave Hansen wrote:
> On Tue, 2006-08-08 at 19:43 +0400, Kirill Korotaev wrote:
> > > Accessing freed memory is a bug, always, not just *only* when slab
> > > debugging is on, right?  Doesn't this mean we could get junk, or that
> > > the reader could potentially run off a bad pointer?
> > no, read the comment in sys_getppid.
> > It is a valid optimization. _safe_ and alowing to bypass taking the lock.
> > BUT! This optimization relies on the fact that kernel memory (DMA + normal zone)
> > is always mapped into virtual address space.
> > Which is invalid for debug kernels only.
> 
> Actually, it might also be invalid in hypervisor environments.  s390 and
> Xen use ballooning drivers to tell the hypervisor which pages are not
> currently in use by the OS so that they may be used in virtual machines
> elsewhere.
> 
> I'm cc'ing the s390 guys.  Will the s390 kernel oops if it accesses a
> page which was ballooned back to the hypervisor?

Not with the ballooner, that just tells the hypervisor that it can
forget the current content. On the next access the hypervisor hands out
a zeroed page so the access will succeed. But with my guest page hinting
code the kernel will oops if a free page is accessed.

-- 
blue skies,
  Martin.

Martin Schwidefsky
Linux for zSeries Development & Services
IBM Deutschland Entwicklung GmbH

"Reality continues to ruin my life." - Calvin.