[net-next,00/15] in_interrupt() cleanup, part 2

Message

Sebastian Andrzej Siewior Oct. 27, 2020, 10:54 p.m. UTC

Folks,

in the discussion about preempt count consistency across kernel configurations:

  https://lore.kernel.org/r/[e-mail]/

Linus clearly requested that code in drivers and libraries which changes
behaviour based on execution context should either be split up so that
e.g. task context invocations and BH invocations have different interfaces
or if that's not possible the context information has to be provided by the
caller which knows in which context it is executing.

This includes conditional locking, allocation mode (GFP_*) decisions and
avoidance of code paths which might sleep.

In the long run, usage of 'preemptible, in_*irq etc.' should be banned from
driver code completely.

This is part two addressing remaining drivers except for orinoco-usb.

Sebastian

Comments

Jakub Kicinski Oct. 30, 2020, 6:29 p.m. UTC | #1

On Tue, 27 Oct 2020 23:54:39 +0100 Sebastian Andrzej Siewior wrote:
> Folks,
> 
> in the discussion about preempt count consistency across kernel configurations:
> 
>   https://lore.kernel.org/r/[e-mail]/
> 
> Linus clearly requested that code in drivers and libraries which changes
> behaviour based on execution context should either be split up so that
> e.g. task context invocations and BH invocations have different interfaces
> or if that's not possible the context information has to be provided by the
> caller which knows in which context it is executing.
> 
> This includes conditional locking, allocation mode (GFP_*) decisions and
> avoidance of code paths which might sleep.
> 
> In the long run, usage of 'preemptible, in_*irq etc.' should be banned from
> driver code completely.
> 
> This is part two addressing remaining drivers except for orinoco-usb.

Freescale folks - can I get an ack for merging the last three patches
into net-next?

Jakub Kicinski Oct. 31, 2020, 5:08 p.m. UTC | #2

On Tue, 27 Oct 2020 23:54:39 +0100 Sebastian Andrzej Siewior wrote:
> Folks,
> 
> in the discussion about preempt count consistency across kernel configurations:
> 
>   https://lore.kernel.org/r/[e-mail]/
> 
> Linus clearly requested that code in drivers and libraries which changes
> behaviour based on execution context should either be split up so that
> e.g. task context invocations and BH invocations have different interfaces
> or if that's not possible the context information has to be provided by the
> caller which knows in which context it is executing.
> 
> This includes conditional locking, allocation mode (GFP_*) decisions and
> avoidance of code paths which might sleep.
> 
> In the long run, usage of 'preemptible, in_*irq etc.' should be banned from
> driver code completely.
> 
> This is part two addressing remaining drivers except for orinoco-usb.

Sebastian, thanks for there, I picked up only the Ethernet patches:

5ce7f3f46f6b net: neterion: s2io: Replace in_interrupt() for context detection
dc5e8bfcd12e net: forcedeth: Replace context and lock check with a lockdep_assert()
beca92820dc4 net: tlan: Replace in_irq() usage

Please repost the wireless stuff for Kalle to linux-wireless@vger, 
and repost the fsl stuff separately (our patchwork queue is huge, 
I can't keep this waiting for an ack any longer).