Re: [PATCH v3 2/3] uio_pci_generic: add MSI/MSI-X support

[Avi Kivity <avi@scylladb.com>]

On 10/05/2015 06:11 AM, Greg KH wrote:
> On Sun, Oct 04, 2015 at 11:43:17PM +0300, Vlad Zolotarov wrote:
>> Add support for MSI and MSI-X interrupt modes:
>>     - Interrupt mode selection order is:
>>          INT#X (for backward compatibility) -> MSI-X -> MSI.
>>     - Add ioctl() commands:
>>        - UIO_PCI_GENERIC_INT_MODE_GET: query the current interrupt mode.
>>        - UIO_PCI_GENERIC_IRQ_NUM_GET: query the maximum number of IRQs.
>>        - UIO_PCI_GENERIC_IRQ_SET: bind the IRQ to eventfd (similar to vfio).
>>     - Add mappings to all bars (memory and portio): some devices have
>>       registers related to MSI/MSI-X handling outside BAR0.
>>
>> Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
>> ---
>> New in v3:
>>     - Add __iomem qualifier to temp buffer receiving ioremap value.
>>
>> New in v2:
>>     - Added #include <linux/uaccess.h> to uio_pci_generic.c
>>
>> Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
>> ---
>>   drivers/uio/uio_pci_generic.c   | 410 +++++++++++++++++++++++++++++++++++++---
>>   include/linux/uio_pci_generic.h |  36 ++++
>>   2 files changed, 423 insertions(+), 23 deletions(-)
>>   create mode 100644 include/linux/uio_pci_generic.h
>>
>> diff --git a/drivers/uio/uio_pci_generic.c b/drivers/uio/uio_pci_generic.c
>> index d0b508b..6b8b1789 100644
>> --- a/drivers/uio/uio_pci_generic.c
>> +++ b/drivers/uio/uio_pci_generic.c
>> @@ -22,16 +22,32 @@
>>   #include <linux/device.h>
>>   #include <linux/module.h>
>>   #include <linux/pci.h>
>> +#include <linux/msi.h>
>>   #include <linux/slab.h>
>>   #include <linux/uio_driver.h>
>> +#include <linux/uio_pci_generic.h>
>> +#include <linux/eventfd.h>
>> +#include <linux/uaccess.h>
>>   
>>   #define DRIVER_VERSION	"0.01.0"
>>   #define DRIVER_AUTHOR	"Michael S. Tsirkin <mst@redhat.com>"
>>   #define DRIVER_DESC	"Generic UIO driver for PCI 2.3 devices"
>>   
>> +struct msix_info {
>> +	int num_irqs;
>> +	struct msix_entry *table;
>> +	struct uio_msix_irq_ctx {
>> +		struct eventfd_ctx *trigger;	/* MSI-x vector to eventfd */
> Why are you using eventfd for msi vectors?  What's the reason for
> needing this?
>
> You haven't documented how this api works at all, you are going to have
> to a lot more work to justify this, as this greatly increases the
> complexity of the user/kernel api in unknown ways.
>
>

Of course it has to be documented, but this just follows vfio.

Eventfd is a natural enough representation of an interrupt; both kvm and 
vfio use it, and are also able to share the eventfd, allowing a vfio 
interrupt to generate a kvm interrupt, without userspace intervention, 
and one day without even kernel intervention.