[PATCH v2 0/2] irqchip/gic-v2, v3: Enforce ACK/EOI pairing on IRQ retrigger

[Valentin Schneider]

Hi,

This is v2 of:

  https://lkml.kernel.org/r/20200710145642.28978-1-valentin.schneider@arm.com

the TL;DR being: retriggering of IRQs marked with IRQS_PENDING breaks the
ACK/EOI pairing the GIC spec requires, and this is an attempt to clean it
up; details are in the changelogs.

Testing
=======

I briefly tested this on my Juno r0 by hackishly configuring the RTC
interrupt as edge-sensitive (level interrupts don't get retriggered) and
triggering a wakeup interrupt via

  $ rtcwake -d /dev/rtc0 -s 1 -m mem

along with sprinkling a few printks to make sure things go down the right
way. Note that wakeup from RTC is currently broken on Juno and requires
this patch from Sudeep:

 4df2ef85f0ef ("rtc: pl031: fix set_alarm by adding back call to alarm_irq_enable")

The RTC on my eMAG can't be set up as a wakeup alarm, so I didn't get to
test the retriggering side of things on GICv3 - the rest does seem to still
behave correctly.

Revisions
=========

v1 -> v2
--------

o Trimmed the changelog
o Fixed gic_retrigger() return value (Marc)
o (new patch) Added IRQD_HANDLE_ENFORCE_IRQCTX to all GIC IRQs

Valentin Schneider (2):
  irqchip/gic-v2, v3: Implement irq_chip->irq_retrigger()
  irqchip/gic-v2, v3: Prevent SW resends entirely

 drivers/irqchip/irq-gic-v3.c | 12 +++++++++++-
 drivers/irqchip/irq-gic.c    | 12 +++++++++++-
 2 files changed, 22 insertions(+), 2 deletions(-)

--
2.27.0