mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Gupta, Nipun" <nipun.gupta@amd.com>
To: Alex Williamson <alex.williamson@redhat.com>
Cc: gregkh@linuxfoundation.org, linux-kernel@vger.kernel.org,
	git@amd.com, pieter.jansen-van-vuuren@amd.com,
	nikhil.agarwal@amd.com, michal.simek@amd.com,
	abhijit.gangurde@amd.com, Shubham Rohila <shubham.rohila@amd.com>
Subject: Re: [PATCH 2/2 v4] vfio-cdx: add ioctl support for bus master enable
Date: Fri, 28 Jul 2023 16:24:31 +0530	[thread overview]
Message-ID: <27dc2387-a27e-3991-c933-5b293d4f8046@amd.com> (raw)
In-Reply-To: <20230726110322.7074371a.alex.williamson@redhat.com>



On 7/26/2023 10:33 PM, Alex Williamson wrote:
> On Wed, 26 Jul 2023 16:32:20 +0530
> Nipun Gupta <nipun.gupta@amd.com> wrote:
> 
>> add VFIO_DEVICE_CDX_CTRL IOCTL to expose control operations for CDX
>> devices to VFIO users. Support Bus master enable and Bus master disable
>> on CDX bus control.
>>
>> Signed-off-by: Shubham Rohila <shubham.rohila@amd.com>
>> Signed-off-by: Nipun Gupta <nipun.gupta@amd.com>
>> ---
>>
>> Changes in v4:
>> - This patch is newly added which uses cdx_set_master() and
>>    cdx_clear_master() APIs.
>>
>>   drivers/vfio/cdx/main.c   | 26 ++++++++++++++++++++++++++
>>   include/uapi/linux/vfio.h | 14 ++++++++++++++
>>   2 files changed, 40 insertions(+)
>>
>> diff --git a/drivers/vfio/cdx/main.c b/drivers/vfio/cdx/main.c
>> index c376a69d2db2..c39a965716f4 100644
>> --- a/drivers/vfio/cdx/main.c
>> +++ b/drivers/vfio/cdx/main.c
>> @@ -98,6 +98,30 @@ static int vfio_cdx_ioctl_get_region_info(struct vfio_cdx_device *vdev,
>>   	return copy_to_user(arg, &info, minsz) ? -EFAULT : 0;
>>   }
>>   
>> +static int vfio_cdx_ioctl_ctrl(struct vfio_cdx_device *vdev,
>> +			       struct vfio_device_cdx_ctrl __user *arg)
>> +{
>> +	unsigned long minsz = offsetofend(struct vfio_device_cdx_ctrl, op);
>> +	struct cdx_device *cdx_dev = to_cdx_device(vdev->vdev.dev);
>> +	struct vfio_device_cdx_ctrl ops;
>> +
>> +	if (copy_from_user(&ops, arg, minsz))
>> +		return -EFAULT;
>> +
>> +	if (ops.argsz < minsz)
>> +		return -EINVAL;
>> +
>> +	switch (ops.op) {
>> +	case VFIO_CDX_CTRL_CLEAR_MASTER:
>> +		cdx_clear_master(cdx_dev);
>> +		return 0;
>> +	case VFIO_CDX_CTRL_SET_MASTER:
>> +		return cdx_set_master(cdx_dev);
>> +	default:
>> +		return -EINVAL;
>> +	}
>> +}
>> +
>>   static long vfio_cdx_ioctl(struct vfio_device *core_vdev,
>>   			   unsigned int cmd, unsigned long arg)
>>   {
>> @@ -112,6 +136,8 @@ static long vfio_cdx_ioctl(struct vfio_device *core_vdev,
>>   		return vfio_cdx_ioctl_get_region_info(vdev, uarg);
>>   	case VFIO_DEVICE_RESET:
>>   		return cdx_dev_reset(core_vdev->dev);
>> +	case VFIO_DEVICE_CDX_CTRL:
>> +		return vfio_cdx_ioctl_ctrl(vdev, uarg);
>>   	default:
>>   		return -ENOTTY;
>>   	}
>> diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
>> index 20c804bdc09c..5f6a58f9f8e2 100644
>> --- a/include/uapi/linux/vfio.h
>> +++ b/include/uapi/linux/vfio.h
>> @@ -1649,6 +1649,20 @@ struct vfio_iommu_spapr_tce_remove {
>>   };
>>   #define VFIO_IOMMU_SPAPR_TCE_REMOVE	_IO(VFIO_TYPE, VFIO_BASE + 20)
>>   
>> +/**
>> + * VFIO_DEVICE_CDX_CTRL - _IO(VFIO_TYPE, VFIO_BASE + 21)
>> + *
>> + * Control CDX device.
>> + * Variable op is set as per the operation required
>> + */
>> +struct vfio_device_cdx_ctrl {
>> +	__u32 argsz;
>> +	__u32 op;
>> +#define	VFIO_CDX_CTRL_SET_MASTER	0	/* Set Bus Master */
>> +#define	VFIO_CDX_CTRL_CLEAR_MASTER	1	/* Clear Bus Master */
>> +};
>> +#define VFIO_DEVICE_CDX_CTRL		_IO(VFIO_TYPE, VFIO_BASE + 21)
>> +
>>   /* ***************************************************************** */
>>   
>>   #endif /* _UAPIVFIO_H */
> 
> This doesn't follow standard conventions for a vfio ioctl.  Not knowing
> how CDX works, is there really a need for this ioctl vs enabling bus
> master when the device is opened and clearing it when closed?  I think
> this is effectively the behavior of vfio-platform.  Otherwise, I assume
> this is not a high frequency operation, would the VFIO_DEVICE_FEATURE
> ioctl be an option?  Thanks,

Thank you for the quick feedback. There are few use-cases where VFIO-cdx 
users need to explicitly trigger bus master enable/disable. As you 
suggest VFIO_DEVICE_FEATURE is better suited in this case, and we will 
move the BME handling in this ioctl.

Regards,
Nipun

> 
> Alex
> 

      reply	other threads:[~2023-07-28 10:54 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-07-26 11:02 [PATCH 1/2 v4] cdx: add support for bus mastering Nipun Gupta
2023-07-26 11:02 ` [PATCH 2/2 v4] vfio-cdx: add ioctl support for bus master enable Nipun Gupta
2023-07-26 11:25   ` Greg KH
2023-07-28 11:26     ` Gupta, Nipun
2023-07-26 17:03   ` Alex Williamson
2023-07-28 10:54     ` Gupta, Nipun [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=27dc2387-a27e-3991-c933-5b293d4f8046@amd.com \
    --to=nipun.gupta@amd.com \
    --cc=abhijit.gangurde@amd.com \
    --cc=alex.williamson@redhat.com \
    --cc=git@amd.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=michal.simek@amd.com \
    --cc=nikhil.agarwal@amd.com \
    --cc=pieter.jansen-van-vuuren@amd.com \
    --cc=shubham.rohila@amd.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®