2024-01-16 07:33:38

by Leonardo Bras

Subject: [RFC PATCH v1 2/2] serial/8250: Avoid getting lock in RT atomic context

With PREEMPT_RT enabled, a spin_lock_irqsave() becomes a possibly sleeping
spin_lock(), without preempt_disable() or irq_disable().

This allows a task T1 to get preempted or interrupted while holding the
port->lock. If the preempting task T2 need the lock, spin_lock() code
will schedule T1 back until it finishes using the lock, and then go back to
T2.

There is an issue if a T1 holding port->lock is interrupted by an
IRQ, and this IRQ handler needs to get port->lock for writting (printk):
spin_lock() code will try to reschedule the interrupt handler, which is in
atomic context, causing a BUG() for trying to reschedule/sleep in atomic
context.

So for the case (PREEMPT_RT && in_atomic()) try to get the lock, and if it
fails proceed anyway, just like it's done in oops_in_progress case.

Signed-off-by: Leonardo Bras <[email protected]>
---
drivers/tty/serial/8250/8250_port.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/tty/serial/8250/8250_port.c b/drivers/tty/serial/8250/8250_port.c
index 8ca061d3bbb92..8480832846319 100644
--- a/drivers/tty/serial/8250/8250_port.c
+++ b/drivers/tty/serial/8250/8250_port.c
@@ -3397,7 +3397,7 @@ void serial8250_console_write(struct uart_8250_port *up, const char *s,

touch_nmi_watchdog();

- if (oops_in_progress)
+ if (oops_in_progress || (IS_ENABLED(CONFIG_PREEMPT_RT) && in_atomic())
locked = uart_port_trylock_irqsave(port, &flags);
else
uart_port_lock_irqsave(port, &flags);
--
2.43.0

2024-01-16 07:51:25

by Jiri Slaby

Subject: Re: [RFC PATCH v1 2/2] serial/8250: Avoid getting lock in RT atomic context

On 16. 01. 24, 8:32, Leonardo Bras wrote:
> With PREEMPT_RT enabled, a spin_lock_irqsave() becomes a possibly sleeping
> spin_lock(), without preempt_disable() or irq_disable().
>
> This allows a task T1 to get preempted or interrupted while holding the
> port->lock. If the preempting task T2 need the lock, spin_lock() code
> will schedule T1 back until it finishes using the lock, and then go back to
> T2.
>
> There is an issue if a T1 holding port->lock is interrupted by an
> IRQ, and this IRQ handler needs to get port->lock for writting (printk):
> spin_lock() code will try to reschedule the interrupt handler, which is in
> atomic context, causing a BUG() for trying to reschedule/sleep in atomic
> context.
>
> So for the case (PREEMPT_RT && in_atomic()) try to get the lock, and if it
> fails proceed anyway, just like it's done in oops_in_progress case.

Hmm, that appears incorrect to me.

Perhaps we need a raw spin lock? Or maybe I am totally off, as my RT
knowledge is close to zero.

This needs advices from RT folks...

> Signed-off-by: Leonardo Bras <[email protected]>
> ---
> drivers/tty/serial/8250/8250_port.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/tty/serial/8250/8250_port.c b/drivers/tty/serial/8250/8250_port.c
> index 8ca061d3bbb92..8480832846319 100644
> --- a/drivers/tty/serial/8250/8250_port.c
> +++ b/drivers/tty/serial/8250/8250_port.c
> @@ -3397,7 +3397,7 @@ void serial8250_console_write(struct uart_8250_port *up, const char *s,
>
> touch_nmi_watchdog();
>
> - if (oops_in_progress)
> + if (oops_in_progress || (IS_ENABLED(CONFIG_PREEMPT_RT) && in_atomic())
> locked = uart_port_trylock_irqsave(port, &flags);
> else
> uart_port_lock_irqsave(port, &flags);

--
js
suse labs

2024-01-16 18:24:32

by Leonardo Bras

Subject: Re: [RFC PATCH v1 2/2] serial/8250: Avoid getting lock in RT atomic context

On Tue, Jan 16, 2024 at 08:49:14AM +0100, Jiri Slaby wrote:
> On 16. 01. 24, 8:32, Leonardo Bras wrote:
> > With PREEMPT_RT enabled, a spin_lock_irqsave() becomes a possibly sleeping
> > spin_lock(), without preempt_disable() or irq_disable().
> >
> > This allows a task T1 to get preempted or interrupted while holding the
> > port->lock. If the preempting task T2 need the lock, spin_lock() code
> > will schedule T1 back until it finishes using the lock, and then go back to
> > T2.
> >
> > There is an issue if a T1 holding port->lock is interrupted by an
> > IRQ, and this IRQ handler needs to get port->lock for writting (printk):
> > spin_lock() code will try to reschedule the interrupt handler, which is in
> > atomic context, causing a BUG() for trying to reschedule/sleep in atomic
> > context.
> >
> > So for the case (PREEMPT_RT && in_atomic()) try to get the lock, and if it
> > fails proceed anyway, just like it's done in oops_in_progress case.
>
> Hmm, that appears incorrect to me.
>
> Perhaps we need a raw spin lock? Or maybe I am totally off, as my RT
> knowledge is close to zero.

If we have a raw_spin_lock_irqsave() here, it would hurt RT by a lot since
disabling interrupts is usually bad at the RT kernel, and serial console
can be used a lot.

>
> This needs advices from RT folks...

Agree. All help is welcome in this case!

Thanks!
Leo

>
> > Signed-off-by: Leonardo Bras <[email protected]>
> > ---
> > drivers/tty/serial/8250/8250_port.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/tty/serial/8250/8250_port.c b/drivers/tty/serial/8250/8250_port.c
> > index 8ca061d3bbb92..8480832846319 100644
> > --- a/drivers/tty/serial/8250/8250_port.c
> > +++ b/drivers/tty/serial/8250/8250_port.c
> > @@ -3397,7 +3397,7 @@ void serial8250_console_write(struct uart_8250_port *up, const char *s,
> > touch_nmi_watchdog();
> > - if (oops_in_progress)
> > + if (oops_in_progress || (IS_ENABLED(CONFIG_PREEMPT_RT) && in_atomic())
> > locked = uart_port_trylock_irqsave(port, &flags);
> > else
> > uart_port_lock_irqsave(port, &flags);
>
> --
> js
> suse labs
>