Re: [PATCH 20/21] context_tracking: Convert state to atomic_t

[Frederic Weisbecker <frederic@kernel.org>]

On Wed, May 18, 2022 at 05:09:55PM +0200, nicolas saenz julienne wrote:
> On Tue, 2022-05-03 at 12:00 +0200, Frederic Weisbecker wrote:
> 
> [...]
> 
> > +/**
> > + * ct_state() - return the current context tracking state if known
> > + *
> > + * Returns the current cpu's context tracking state if context tracking
> > + * is enabled.  If context tracking is disabled, returns
> > + * CONTEXT_DISABLED.  This should be used primarily for debugging.
> > + */
> > +static __always_inline int ct_state(void)
> > +{
> > +	int ret;
> > +
> > +	if (!context_tracking_enabled())
> > +		return CONTEXT_DISABLED;
> > +
> > +	preempt_disable();
> > +	ret = __ct_state();
> > +	preempt_enable();
> > +
> > +	return ret;
> > +}
> > +
> 
> I can't see any use for this function with preemption enabled. You can't trust
> the data due to CPU migration and it could be a source of bugs in the future.
> Wouldn't it make more sense to move the burden into the callers? They all DTRT,
> plus, this_cpu_ptr() will spew warnings if someone shows up and doesn't comply.

I believe syscall_exit_to_user_mode_prepare() has preemption enabled.
Then it's ok if we are scheduled away right before the check, it still applies
on any CPU.