Re: [linux-safety] [RFC PATCH 2/2] bust_spinlocks: do not decrement oops_in_progress unconditionally

Hi Lukas

> -----Original Message-----
> From: Lukas Bulwahn <[email protected]>
> Sent: Wednesday, October 14, 2020 7:53 AM
> To: Paoloni, Gabriele <[email protected]>
> Cc: [email protected]
> Subject: Re: [linux-safety] [RFC PATCH 2/2] bust_spinlocks: do not
> decrement oops_in_progress unconditionally
> 
> 
> 
> On Tue, 13 Oct 2020, Paoloni, Gabriele wrote:
> 
> > In the current implementation if the input flag is 0
> > oops_in_progress is unconditionally decremented, thus allowing
> > to become a negative number. Since right now oops_in_progress
> > is a global variable used in the kernel as a conditional flag
> > to check if oops, panic(), BUG() or die() is in progress the
> > current unconditional decrement may lead to unexpected behavior
> > in the Kernel paths conditionally executing over this flag.
> >
> > This patch only decrement oops_in_progress if it is non zero
> >
> > Signed-off-by: Gabriele Paoloni <[email protected]>
> > ---
> >  lib/bust_spinlocks.c | 7 ++++++-
> >  1 file changed, 6 insertions(+), 1 deletion(-)
> >
> > diff --git a/lib/bust_spinlocks.c b/lib/bust_spinlocks.c
> > index 594b270161d9..842633ac9130 100644
> > --- a/lib/bust_spinlocks.c
> > +++ b/lib/bust_spinlocks.c
> > @@ -23,6 +23,9 @@
> >   * @yes: input flag; if zero decreases oops_in_progress,
> >   * otherwise increases it.
> >   *
> > + * Note: if oops_in_progress is already 0 it will not
> > + * be decreased
> > + *
> >   */
> >  void bust_spinlocks(int yes)
> >  {
> > @@ -33,7 +36,9 @@ void bust_spinlocks(int yes)
> >  		unblank_screen();
> >  #endif
> >  		console_unblank();
> > -		if (--oops_in_progress == 0)
> > +		if (oops_in_progress)
> > +			oops_in_progress--;
> > +		if (!oops_in_progress)
> >  			wake_up_klogd();
> 
> I did not get the original motivation stated above.
> 
> But I believe you meant:
> 
> 'I think there is race condition here (before this patch).'
> 
> So do something in this patch: 'And now the race condition is gone?'
> 
> 
> I think:
> 
> 'The could be a race condition before, and probably the race condition is
> still there after this patch.'
> 
> But maybe I did even get the intent of this patch in the first place...

What I meant is the following scenario:
Let's assume oops_in_progress = 0, then we have

func_a()
{
	bust_spinlocks(0);
}

In this case after the call, with the current implementation oops_in_progress = -1; that is not acceptable...

Thanks
Gab 

> 
> Lukas
---------------------------------------------------------------------
INTEL CORPORATION ITALIA S.p.A. con unico socio
Sede: Milanofiori Palazzo E 4 
CAP 20094 Assago (MI)
Capitale Sociale Euro 104.000,00 interamente versato
Partita I.V.A. e Codice Fiscale  04236760155
Repertorio Economico Amministrativo n. 997124 
Registro delle Imprese di Milano nr. 183983/5281/33
Soggetta ad attivita' di direzione e coordinamento di 
INTEL CORPORATION, USA

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.