On Thu, Jan 19, 2012 at 11:51:47AM -0800, Simon Glass wrote:
> Hi Paul,
>
> On Thu, Jan 19, 2012 at 11:42 AM, Paul E. McKenney
> <paulmck@xxxxxxxxxxxxxxxxxx> wrote:
> > On Thu, Jan 19, 2012 at 11:28:56AM -0800, Simon Glass wrote:
> >> The synchronize_rcu() call resulting from making every serial driver
> >> wake-up capable (commit b3b708fa) slows boot down on my Tegra2x system
> >> (with CONFIG_PREEMPT disabled).
> >>
> >> But this is avoidable since it is the device_set_wakeup_enable() and then
> >> subsequence disable which causes the delay. We might as well just make
> >> the device wakeup capable but not actually enable it for wakeup until
> >> needed.
> >>
> >> Effectively the current code does this:
> >>
> >> device_set_wakeup_capable(dev, 1);
> >> device_set_wakeup_enable(dev, 1);
> >> device_set_wakeup_enable(dev, 0);
> >>
> >> We can just drop the last two lines.
> >>
> >> Before this change my boot log says:
> >> [ 0.227062] Serial: 8250/16550 driver, 4 ports, IRQ sharing disabled
> >> [ 0.702928] serial8250.0: ttyS0 at MMIO 0x70006040 (irq = 69) is a Tegra
> >>
> >> after:
> >> [ 0.227264] Serial: 8250/16550 driver, 4 ports, IRQ sharing disabled
> >> [ 0.227983] serial8250.0: ttyS0 at MMIO 0x70006040 (irq = 69) is a Tegra
> >>
> >> for saving of 450ms.
> >
> > You have multiple CPUs running at this point, correct? Before that
> > second CPU starts up, synchronize_rcu() is a no-op.
>
> Yes that's right, although I didn't get different behavior with 'nosmp'.
If you are running CONFIG_PREEMPT=y, that is expected behavior, though
that grace period is a bit on the long side. What is the value of HZ?
On the other hand, if you are running CONFIG_PREEMPT=n on recent kernels,
synchronize_rcu() will be a no-op any time that you are running with
only a single CPU.
The reason for this difference is that with CONFIG_PREEMPT=y, there might
be a preempted RCU reader, and RCU must check for this condition, waiting
for any such reader to complete. In contrast, when CONFIG_PREEMPT=n on a
single-CPU system, the fact that you are executing in synchronize_rcu()
guarantees that there are no running RCU readers, so synchronize_rcu()
need do nothing in that case.
Thanx, Paul
> > The patch looks good to me, but then again, I do not consider myself
> > qualified to have an opinion on the TTY layer. ;-)
>
> Thanks, me neither :-)
>
> Regards,
> Simon
>
> >
> > Thanx, Paul
> >
> >> Suggested-by: Rafael J. Wysocki <rjw@xxxxxxx>
> >> Signed-off-by: Simon Glass <sjg@xxxxxxxxxxxx>
> >> ---
> >> drivers/tty/serial/serial_core.c | 6 +++---
> >> 1 files changed, 3 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c
> >> index c7bf31a..1305618 100644
> >> --- a/drivers/tty/serial/serial_core.c
> >> +++ b/drivers/tty/serial/serial_core.c
> >> @@ -2348,11 +2348,11 @@ int uart_add_one_port(struct uart_driver *drv, struct uart_port *uport)
> >> */
> >> tty_dev = tty_register_device(drv->tty_driver, uport->line, uport->dev);
> >> if (likely(!IS_ERR(tty_dev))) {
> >> - device_init_wakeup(tty_dev, 1);
> >> - device_set_wakeup_enable(tty_dev, 0);
> >> - } else
> >> + device_set_wakeup_capable(tty_dev, 1);
> >> + } else {
> >> printk(KERN_ERR "Cannot register tty device on line %d\n",
> >> uport->line);
> >> + }
> >>
> >> /*
> >> * Ensure UPF_DEAD is not set.
> >> --
> >> 1.7.7.3
> >>
> >
>
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
