Re: [uml-devel] Re: [PATCH 07/10] uml: avoid fixing faults while atomic

[Andrew Morton <akpm@osdl.org>]

Blaisorblade <blaisorblade@yahoo.it> wrote:
>
> On Wednesday 21 September 2005 22:47, Andrew Morton wrote:
> > Blaisorblade <blaisorblade@yahoo.it> wrote:
> > > On Wednesday 21 September 2005 21:49, Andrew Morton wrote:
> > > > "Paolo 'Blaisorblade' Giarrusso" <blaisorblade@yahoo.it> wrote:
> > > > > From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
> 
> > > > It has accidental side-effects,
> > > > such as making copy_to_user() fail if inside spinlocks when
> > > > CONFIG_PREEMPT=y.
> 
> > > Sorry, but should it ever succeed inside spinlocks? I mean, should it
> > > ever call down() inside spinlocks? (We never do down_trylock, and ever if
> > > we did the x86 trick, that wouldn't make the whole thing safe at all -
> > > they still take the spinlock and potentially sleep. And it's legal only
> > > if no spinlock is held).
> 
> > Not sure what you're asking here.
> 
> > copy_to/from_user() will fail inside spinlock if CONFIG_PREMPT=y and if the
> > copy happens to cause a fault.
> 
> > Otherwise it will succeed inside spinlock, 
> > and it won't spew a sleeping-while-atomic warning, because that uses
> > in_atomic() too.
> 
> > It might deadlock if we schedule away and try to retake 
> > the same lock.
> Exactly - the point is: is it legal to call copy_from_user() while holding a 
> spinlock (which is my original question)? Or should copy_from_user try to 
> satisfy the fault, instead of seeing in_atomic() or something similar and 
> fail?

No, it is not legal to call copy_*_user() while holding a spinlock.

If CONFIG_PREEMPT=n, do_page_fault() has no way of knowing that the caller
holds a spinlock.