[PATCHv3,0/4] Get rid of pm_runtime_irq_safe() for 8250_omap

Message

Tony Lindgren Oct. 15, 2021, 11:26 a.m. UTC

Hi,

Here are v3 patches to get rid of pm_runtime_irq_safe() for the 8250_omap
driver. Based on comments from Andy, Johan and Greg, I improved a bunch of
things as listed below.

For removing the pm_runtime_irq_safe() usage, serial TX is the last
remaining issue. We deal with TX by waking up the port and returning 0
bytes written from write_room() and write() if the port is not available
because of PM runtime autoidle.

This series also removes the dependency to Andy's pending generic serial
layer PM runtime patches, and hopefully makes that work a bit easier :)

Regards,

Tony

Chganges since v2:

- Use locking instead of atomic_t as suggested by Greg

Changes since v1:

- Separated out line discipline patches, n_tty -EAGAIN change I still
  need to retest

- Changed prep_tx() to more generic wakeup() as also flow control needs it

- Changed over to using wakeup() with device driver runtime PM instead
  of write_room()

- Added runtime_suspended flag for drivers and generic serial layer PM
  to use


Tony Lindgren (4):
  serial: core: Add wakeup() and start_pending_tx() for power management
  serial: 8250: Implement wakeup for TX and use it for 8250_omap
  serial: 8250_omap: Require a valid wakeirq for deeper idle states
  serial: 8250_omap: Drop the use of pm_runtime_irq_safe()

 Documentation/driver-api/serial/driver.rst |  9 +++
 drivers/tty/serial/8250/8250_omap.c        | 44 ++++++++++----
 drivers/tty/serial/8250/8250_port.c        | 39 ++++++++++++-
 drivers/tty/serial/serial_core.c           | 68 +++++++++++++++++++++-
 include/linux/serial_core.h                |  3 +
 5 files changed, 149 insertions(+), 14 deletions(-)

Comments

Johan Hovold Oct. 18, 2021, 7:11 a.m. UTC | #1

On Fri, Oct 15, 2021 at 02:26:22PM +0300, Tony Lindgren wrote:
> Hi,
> 
> Here are v3 patches to get rid of pm_runtime_irq_safe() for the 8250_omap
> driver. Based on comments from Andy, Johan and Greg, I improved a bunch of
> things as listed below.
> 
> For removing the pm_runtime_irq_safe() usage, serial TX is the last
> remaining issue. We deal with TX by waking up the port and returning 0
> bytes written from write_room() and write() if the port is not available
> because of PM runtime autoidle.

Oh, there's a lot more than TX that needs fixing... And I believe the
second sentence no longer applies since v1.

> Chganges since v2:
> 
> - Use locking instead of atomic_t as suggested by Greg
> 
> Changes since v1:
> 
> - Separated out line discipline patches, n_tty -EAGAIN change I still
>   need to retest
> 
> - Changed prep_tx() to more generic wakeup() as also flow control needs it
> 
> - Changed over to using wakeup() with device driver runtime PM instead
>   of write_room()
> 
> - Added runtime_suspended flag for drivers and generic serial layer PM
>   to use

Johan