Re: [PATCH v2 2/3] hw/nvme: use KVM irqfd when available

[Jinhao Fan]

at 11:34 PM, Keith Busch <kbusch@kernel.org> wrote:

> On Fri, Aug 26, 2022 at 11:12:04PM +0800, Jinhao Fan wrote:
>> Use KVM's irqfd to send interrupts when possible. This approach is
>> thread safe. Moreover, it does not have the inter-thread communication
>> overhead of plain event notifiers since handler callback are called
>> in the same system call as irqfd write.
>> 
>> Signed-off-by: Jinhao Fan <fanjinhao21s@ict.ac.cn>
>> Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
> 
> No idea what's going on here... This one is causing the following assert
> failure with --enable-kvm:
> 
>  qemu-system-x86_64: ../accel/kvm/kvm-all.c:1781: kvm_irqchip_commit_routes: 
> Assertion `ret == 0' failed.
> 
> I find it calls KVM_SET_GSI_ROUTING ioctl with gsi set to KVM_IRQ_ROUTING_MSI,
> and linux kernel returns EINVAL in that case. It's never set that way without
> this patch. Am I the only one seeing this?

nvme_start_ctrl() registers MSI-X masking handlers without checking
irq-eventfd. This causes nvme_kvm_vector_unmask() to be called when it is
not supposed to.