Re: [PATCH v2 i-g-t 3/4] tests/intel/xe_sriov_vfio: Add open-basic subtest

["Laguna, Lukasz" <lukasz.laguna@intel.com>]

On 3/27/2026 15:06, Marcin Bernatowicz wrote:
> The subtest binds a VF to xe-vfio-pci, opens its VFIO group, and
> verifies basic VFIO access with a minimal group status query. This
> provides a simple smoke test for VFIO device setup.
>
> Signed-off-by: Marcin Bernatowicz <marcin.bernatowicz@linux.intel.com>
> Cc: Adam Miszczak <adam.miszczak@linux.intel.com>
> Cc: Jakub Kolakowski <jakub1.kolakowski@intel.com>
> Cc: Kamil Konieczny <kamil.konieczny@linux.intel.com>
> Cc: Lukasz Laguna <lukasz.laguna@intel.com>

Reviewed-by: Lukasz Laguna <lukasz.laguna@intel.com>

> ---
> v2:
> - Introduce vf_num variable for clarity (Lukasz)
> - Remove unnecessary unbind (Lukasz)
> ---
>   tests/intel/xe_sriov_vfio.c | 136 ++++++++++++++++++++++++++++++++++++
>   1 file changed, 136 insertions(+)
>
> diff --git a/tests/intel/xe_sriov_vfio.c b/tests/intel/xe_sriov_vfio.c
> index fa9026741..6a86b3780 100644
> --- a/tests/intel/xe_sriov_vfio.c
> +++ b/tests/intel/xe_sriov_vfio.c
> @@ -2,6 +2,9 @@
>   /*
>    * Copyright(c) 2026 Intel Corporation. All rights reserved.
>    */
> +#include <linux/vfio.h>
> +#include <limits.h>
> +#include <sys/ioctl.h>
>   
>   #include "drmtest.h"
>   #include "igt_core.h"
> @@ -22,6 +25,9 @@
>    *
>    * SUBTEST: bind-unbind-vfs
>    * Description: Enable VFs and bind/unbind each one to xe-vfio-pci via driver_override.
> + *
> + * SUBTEST: open-basic
> + * Description: Bind VF to xe-vfio-pci and perform minimal VFIO group open and status ioctl.
>    */
>   
>   IGT_TEST_DESCRIPTION("Xe SR-IOV VFIO tests (xe-vfio-pci)");
> @@ -109,6 +115,109 @@ static void bind_unbind_vfs(unsigned int num_vfs)
>   	igt_sriov_disable_vfs(pf_fd);
>   }
>   
> +static bool vf_iommu_group_id_alloc(const char *pci_slot, char **out_group_id)
> +{
> +	char path[PATH_MAX];
> +	char link[PATH_MAX];
> +	ssize_t len;
> +	const char *base;
> +
> +	igt_assert(out_group_id);
> +	*out_group_id = NULL;
> +
> +	snprintf(path, sizeof(path), "/sys/bus/pci/devices/%s/iommu_group", pci_slot);
> +	len = readlink(path, link, sizeof(link) - 1);
> +	if (len < 0)
> +		return false;
> +
> +	link[len] = '\0';
> +	base = strrchr(link, '/');
> +	if (!base || !base[1])
> +		return false;
> +
> +	*out_group_id = strdup(base + 1);
> +	igt_assert(*out_group_id);
> +
> +	return true;
> +}
> +
> +struct vfio_dev_fds {
> +	int container_fd;
> +	int group_fd;
> +	int device_fd;
> +	char *group_id;
> +};
> +
> +static void vfio_open_group(const char *pci_slot, struct vfio_dev_fds *out,
> +			    struct vfio_group_status *out_group_status)
> +{
> +	char group_path[PATH_MAX];
> +	int ret;
> +
> +	igt_assert(out);
> +	*out = (struct vfio_dev_fds){
> +		.container_fd = -1,
> +		.group_fd = -1,
> +		.device_fd = -1,
> +		.group_id = NULL,
> +	};
> +
> +	if (out_group_status)
> +		*out_group_status = (struct vfio_group_status){
> +					.argsz = sizeof(*out_group_status)
> +				    };
> +
> +	if (!vf_iommu_group_id_alloc(pci_slot, &out->group_id))
> +		igt_skip("No IOMMU group (IOMMU disabled or not exposed)\n");
> +
> +	out->container_fd = open("/dev/vfio/vfio", O_RDWR | O_CLOEXEC);
> +	if (out->container_fd < 0)
> +		igt_skip("/dev/vfio/vfio not available\n");
> +
> +	snprintf(group_path, sizeof(group_path), "/dev/vfio/%s", out->group_id);
> +	out->group_fd = open(group_path, O_RDWR | O_CLOEXEC);
> +	igt_require_f(out->group_fd >= 0, "Failed to open %s (%d)\n", group_path, -errno);
> +
> +	if (out_group_status) {
> +		ret = ioctl(out->group_fd, VFIO_GROUP_GET_STATUS, out_group_status);
> +		igt_require_f(ret == 0, "VFIO_GROUP_GET_STATUS failed (%d)\n", -errno);
> +	}
> +}
> +
> +static void vfio_close_device(struct vfio_dev_fds *fds)
> +{
> +	if (!fds)
> +		return;
> +
> +	if (fds->device_fd >= 0)
> +		close(fds->device_fd);
> +
> +	if (fds->group_fd >= 0 && fds->container_fd >= 0)
> +		ioctl(fds->group_fd, VFIO_GROUP_UNSET_CONTAINER, &fds->container_fd);
> +
> +	if (fds->group_fd >= 0)
> +		close(fds->group_fd);
> +	if (fds->container_fd >= 0)
> +		close(fds->container_fd);
> +
> +	free(fds->group_id);
> +
> +	fds->container_fd = -1;
> +	fds->group_fd = -1;
> +	fds->device_fd = -1;
> +	fds->group_id = NULL;
> +}
> +
> +static void vfio_open_basic(const char *pci_slot)
> +{
> +	struct vfio_dev_fds fds;
> +	struct vfio_group_status group_status;
> +
> +	vfio_open_group(pci_slot, &fds, &group_status);
> +	igt_info("VFIO group %s status flags=0x%x\n", fds.group_id, group_status.flags);
> +	vfio_close_device(&fds);
> +}
> +
>   static void open_pf(void)
>   {
>   	int fd;
> @@ -207,6 +316,33 @@ int igt_main()
>   		}
>   	}
>   
> +	igt_describe("Bind VF to xe-vfio-pci and do minimal VFIO open and GET_STATUS.");
> +	igt_subtest("open-basic") {
> +		unsigned int vf_num = 1;
> +		char *slot = NULL;
> +
> +		igt_skip_on_f(igt_kmod_load(XE_VFIO_PCI_MOD, NULL),
> +			      "Failed to load %s\n", XE_VFIO_PCI_MOD);
> +
> +		open_pf();
> +
> +		igt_sriov_disable_driver_autoprobe(pf_fd);
> +		igt_sriov_enable_vfs(pf_fd, vf_num);
> +
> +		igt_require_f(!igt_pci_system_reinit(), "Failed to refresh PCI state\n");
> +		igt_sriov_enable_driver_autoprobe(pf_fd);
> +
> +		slot = vf_pci_slot_alloc(vf_num);
> +
> +		vf_bind_override(vf_num, XE_VFIO_PCI_DRV);
> +
> +		vfio_open_basic(slot);
> +
> +		vf_unbind_override(vf_num);
> +		free(slot);
> +		igt_sriov_disable_vfs(pf_fd);
> +	}
> +
>   	igt_fixture() {
>   		cleanup_pf();
>   		restore_xe_vfio_module();