Jianmin,
We know that you are busy, Would you like spend some time to give some
feedback?
And I recommend you as maintainer of interrupt controller since you are
author the this piece of code. Also I am willing to become new reviewer
about eiointc driver since VM will use it always.
The eiointc driver is important, it need evolution rather than post the
code and finish job, everything is done.
Regards
Bibo Mao
On 2024/8/26 下午10:46, Huacai Chen wrote:
Hi, Jianmin,
What's your opinion on this series?
Huacai
On Wed, Aug 21, 2024 at 6:11 PM Bibo Mao <maobibo@xxxxxxxxxxx> wrote:
There are four times about EIOINTC_REG_ISR register group access in
eiointc irq handler, in order to get all irq status about 256 interrupt
vectors. It causes four times VM-exits since eiointc register are
software emulated, here multiple interrupt pin routing is introduced
and each 64 interrupt vector is routed to one interrupt pin.
With this method, there will be only on one EIOINTC_REG_ISR register
group acces in irq handler, it will reduce VM-exits.
Bibo Mao (2):
irqchip/loongson-eiointc: Route interrupt parsed from acpi table
irqchip/loongson-eiointc: Add multiple interrupt pin routing support
arch/loongarch/kernel/irq.c | 3 +-
arch/loongarch/kernel/smp.c | 2 +-
drivers/irqchip/irq-loongson-eiointc.c | 66 +++++++++++++++++++++++---
3 files changed, 62 insertions(+), 9 deletions(-)
base-commit: 1fb918967b56df3262ee984175816f0acb310501
--
2.39.3
