Re: [patch 11/12] hrtimer: Prepare support for PREEMPT_RT

[Juergen Gross <jgross@suse.com>]

On 26.07.19 20:30, Thomas Gleixner wrote:
> From: Anna-Maria Gleixner <anna-maria@linutronix.de>
> 
> When PREEMPT_RT is enabled, the soft interrupt thread can be preempted.  If
> the soft interrupt thread is preempted in the middle of a timer callback,
> then calling hrtimer_cancel() can lead to two issues:
> 
>    - If the caller is on a remote CPU then it has to spin wait for the timer
>      handler to complete. This can result in unbound priority inversion.
> 
>    - If the caller originates from the task which preempted the timer
>      handler on the same CPU, then spin waiting for the timer handler to
>      complete is never going to end.
> 
> To avoid these issues, add a new lock to the timer base which is held
> around the execution of the timer callbacks. If hrtimer_cancel() detects
> that the timer callback is currently running, it blocks on the expiry
> lock. When the callback is finished, the expiry lock is dropped by the
> softirq thread which wakes up the waiter and the system makes progress.
> 
> This addresses both the priority inversion and the life lock issues.
> 
> The same issue can happen in virtual machines when the vCPU which runs a
> timer callback is scheduled out. If a second vCPU of the same guest calls
> hrtimer_cancel() it will spin wait for the other vCPU to be scheduled back
> in. The expiry lock mechanism would avoid that. It'd be trivial to enable
> this when paravirt spinlocks are enabled in a guest, but it's not clear
> whether this is an actual problem in the wild, so for now it's an RT only
> mechanism.

As in virtual machines the soft interrupt thread preemption should not
be an issue, I guess the spinning is "just" sub-optimal (similar to not
using paravirt spinlocks).

In case we'd want to change that I'd rather not special case timers, but
apply a more general solution to the quite large amount of similar
cases: I assume the majority of cpu_relax() uses are affected, so adding
a paravirt op cpu_relax() might be appropriate.

That could be put under CONFIG_PARAVIRT_SPINLOCK. If called in a guest
it could ask the hypervisor to give up the physical cpu voluntarily
(in Xen this would be a "yield" hypercall).


Juergen