Re: [PATCH v4 10/15] vfio/ccw: Use vfio_[attach/detach]_device

[Eric Auger <eric.auger@redhat.com>]

Hi Zhenzhong,

On 10/9/23 03:25, Duan, Zhenzhong wrote:
>
>> -----Original Message-----
>> From: Eric Auger <eric.auger@redhat.com>
>> Sent: Monday, October 9, 2023 1:46 AM
>> Subject: Re: [PATCH v4 10/15] vfio/ccw: Use vfio_[attach/detach]_device
>>
>> Hi Zhenzhong,
>> On 10/8/23 12:21, Duan, Zhenzhong wrote:
>>> Hi Eric,
>>>
>>>> -----Original Message-----
>>>> From: Eric Auger <eric.auger@redhat.com>
>>>> Sent: Wednesday, October 4, 2023 11:44 PM
>>>> Subject: [PATCH v4 10/15] vfio/ccw: Use vfio_[attach/detach]_device
>>>>
>>>> Let the vfio-ccw device use vfio_attach_device() and
>>>> vfio_detach_device(), hence hiding the details of the used
>>>> IOMMU backend.
>>>>
>>>> Note that the migration reduces the following trace
>>>> "vfio: subchannel %s has already been attached" (featuring
>>>> cssid.ssid.devid) into "device is already attached"
>>>>
>>>> Also now all the devices have been migrated to use the new
>>>> vfio_attach_device/vfio_detach_device API, let's turn the
>>>> legacy functions into static functions, local to container.c.
>>>>
>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>> Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
>>>>
>>>> ---
>>>>
>>>> v3:
>>>> - simplified vbasedev->dev setting
>>>>
>>>> v2 -> v3:
>>>> - Hopefully fix confusion beteen vbasedev->name, mdevid and sysfsdev
>>>>  while keeping into account Matthew's comment
>>>>  https://lore.kernel.org/qemu-devel/6e04ab8f-dc84-e9c2-deea-
>>>> 2b6b31678b53@linux.ibm.com/
>>>> ---
>>>> include/hw/vfio/vfio-common.h |   5 --
>>>> hw/vfio/ccw.c                 | 122 +++++++++-------------------------
>>>> hw/vfio/common.c              |  10 +--
>>>> 3 files changed, 37 insertions(+), 100 deletions(-)
>>>>
>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>>>> index 12fbfbc37d..c486bdef2a 100644
>>>> --- a/include/hw/vfio/vfio-common.h
>>>> +++ b/include/hw/vfio/vfio-common.h
>>>> @@ -202,7 +202,6 @@ typedef struct {
>>>>     hwaddr pages;
>>>> } VFIOBitmap;
>>>>
>>>> -void vfio_put_base_device(VFIODevice *vbasedev);
>>>> void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
>>>> void vfio_unmask_single_irqindex(VFIODevice *vbasedev, int index);
>>>> void vfio_mask_single_irqindex(VFIODevice *vbasedev, int index);
>>>> @@ -220,11 +219,7 @@ void vfio_region_unmap(VFIORegion *region);
>>>> void vfio_region_exit(VFIORegion *region);
>>>> void vfio_region_finalize(VFIORegion *region);
>>>> void vfio_reset_handler(void *opaque);
>>>> -VFIOGroup *vfio_get_group(int groupid, AddressSpace *as, Error **errp);
>>>> -void vfio_put_group(VFIOGroup *group);
>>>> struct vfio_device_info *vfio_get_device_info(int fd);
>>>> -int vfio_get_device(VFIOGroup *group, const char *name,
>>>> -                    VFIODevice *vbasedev, Error **errp);
>>>> int vfio_attach_device(char *name, VFIODevice *vbasedev,
>>>>                        AddressSpace *as, Error **errp);
>>>> void vfio_detach_device(VFIODevice *vbasedev);
>>>> diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
>>>> index 1e2fce83b0..6ec35fedc9 100644
>>>> --- a/hw/vfio/ccw.c
>>>> +++ b/hw/vfio/ccw.c
>>>> @@ -572,88 +572,15 @@ static void vfio_ccw_put_region(VFIOCCWDevice
>>>> *vcdev)
>>>>     g_free(vcdev->io_region);
>>>> }
>>>>
>>>> -static void vfio_ccw_put_device(VFIOCCWDevice *vcdev)
>>>> -{
>>>> -    g_free(vcdev->vdev.name);
>>>> -    vfio_put_base_device(&vcdev->vdev);
>>>> -}
>>>> -
>>>> -static void vfio_ccw_get_device(VFIOGroup *group, VFIOCCWDevice *vcdev,
>>>> -                                Error **errp)
>>>> -{
>>>> -    S390CCWDevice *cdev = S390_CCW_DEVICE(vcdev);
>>>> -    char *name = g_strdup_printf("%x.%x.%04x", cdev->hostid.cssid,
>>>> -                                 cdev->hostid.ssid,
>>>> -                                 cdev->hostid.devid);
>>>> -    VFIODevice *vbasedev;
>>>> -
>>>> -    QLIST_FOREACH(vbasedev, &group->device_list, next) {
>>>> -        if (strcmp(vbasedev->name, name) == 0) {
>>>> -            error_setg(errp, "vfio: subchannel %s has already been attached",
>>>> -                       name);
>>>> -            goto out_err;
>>>> -        }
>>>> -    }
>>>> -
>>>> -    /*
>>>> -     * All vfio-ccw devices are believed to operate in a way compatible with
>>>> -     * discarding of memory in RAM blocks, ie. pages pinned in the host are
>>>> -     * in the current working set of the guest driver and therefore never
>>>> -     * overlap e.g., with pages available to the guest balloon driver.  This
>>>> -     * needs to be set before vfio_get_device() for vfio common to handle
>>>> -     * ram_block_discard_disable().
>>>> -     */
>>>> -    vcdev->vdev.ram_block_discard_allowed = true;
>>>> -
>>>> -    if (vfio_get_device(group, cdev->mdevid, &vcdev->vdev, errp)) {
>>>> -        goto out_err;
>>>> -    }
>>>> -
>>>> -    vcdev->vdev.ops = &vfio_ccw_ops;
>>>> -    vcdev->vdev.type = VFIO_DEVICE_TYPE_CCW;
>>>> -    vcdev->vdev.name = name;
>>>> -    vcdev->vdev.dev = DEVICE(vcdev);
>>>> -
>>>> -    return;
>>>> -
>>>> -out_err:
>>>> -    g_free(name);
>>>> -}
>>>> -
>>>> -static VFIOGroup *vfio_ccw_get_group(S390CCWDevice *cdev, Error **errp)
>>>> -{
>>>> -    char *tmp, group_path[PATH_MAX];
>>>> -    ssize_t len;
>>>> -    int groupid;
>>>> -
>>>> -    tmp =
>> g_strdup_printf("/sys/bus/css/devices/%x.%x.%04x/%s/iommu_group",
>>>> -                          cdev->hostid.cssid, cdev->hostid.ssid,
>>>> -                          cdev->hostid.devid, cdev->mdevid);
>>>> -    len = readlink(tmp, group_path, sizeof(group_path));
>>>> -    g_free(tmp);
>>>> -
>>>> -    if (len <= 0 || len >= sizeof(group_path)) {
>>>> -        error_setg(errp, "vfio: no iommu_group found");
>>>> -        return NULL;
>>>> -    }
>>>> -
>>>> -    group_path[len] = 0;
>>>> -
>>>> -    if (sscanf(basename(group_path), "%d", &groupid) != 1) {
>>>> -        error_setg(errp, "vfio: failed to read %s", group_path);
>>>> -        return NULL;
>>>> -    }
>>>> -
>>>> -    return vfio_get_group(groupid, &address_space_memory, errp);
>>>> -}
>>>> -
>>>> static void vfio_ccw_realize(DeviceState *dev, Error **errp)
>>>> {
>>>> -    VFIOGroup *group;
>>>>     S390CCWDevice *cdev = S390_CCW_DEVICE(dev);
>>>>     VFIOCCWDevice *vcdev = VFIO_CCW(cdev);
>>>>     S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
>>>> +    VFIODevice *vbasedev = &vcdev->vdev;
>>>>     Error *err = NULL;
>>>> +    char *name;
>>>> +    int ret;
>>>>
>>>>     /* Call the class init function for subchannel. */
>>>>     if (cdc->realize) {
>>>> @@ -663,14 +590,31 @@ static void vfio_ccw_realize(DeviceState *dev, Error
>>>> **errp)
>>>>         }
>>>>     }
>>>>
>>>> -    group = vfio_ccw_get_group(cdev, &err);
>>>> -    if (!group) {
>>>> -        goto out_group_err;
>>>> -    }
>>>> +    name = g_strdup_printf("%x.%x.%04x", vcdev->cdev.hostid.cssid,
>>>> +                           vcdev->cdev.hostid.ssid,
>>>> +                           vcdev->cdev.hostid.devid);
>>>> +    vbasedev->sysfsdev = g_strdup_printf("/sys/bus/css/devices/%s/%s",
>>>> +                                         name,
>>>> +                                         cdev->mdevid);
>>> Hoping not late for you to include this in v5.
>>> I think no need to re-assign sysfsdev as it's a user property, we'd better to
>>> keep the original user value. Also looks a memory leak here.
>> OK I removed it.
>>>> +    vbasedev->ops = &vfio_ccw_ops;
>>>> +    vbasedev->type = VFIO_DEVICE_TYPE_CCW;
>>>> +    vbasedev->name = name;
>>> There will be a potential failure when a second mdev device under
>>> same cssid.ssid.devid attached. We can use cdev->mdevid as name.
>> But this mathes vfio_ccw_get_device() existing code where
>> vcdev->vdev.name = name; and
>> name = g_strdup_printf("%x.%x.%04x", cdev->hostid.cssid,
>>                                  cdev->hostid.ssid,
>>                                  cdev->hostid.devid);
> I suspect this is a bug of the existing code.
Then I would prefer we fix it separately. This patch migrates the
existing code without functional change intended.

>
>> cdev->mdevid is passed as first arg of vfio_attach_device() instead .
> vfio_attach_device() uses cdev->mdevid to get device FD, nothing more.
>
> If we use cssid.ssid.devid as name, then different mdev under same cssid.ssid.devid will have same name, and the second mdev attachment will fail to attach in vfio_attach_device():
>
>     QLIST_FOREACH(vbasedev_iter, &group->device_list, next) {
>         if (strcmp(vbasedev_iter->name, vbasedev->name) == 0) {
>             error_setg(errp, "device is already attached");
>             vfio_put_group(group);
>             return -EBUSY;
>         }
>     }
I get your point but this conversion matches the existing
vfio_ccw_get_device() code:
    char *name = g_strdup_printf("%x.%x.%04x", cdev->hostid.cssid,
                                 cdev->hostid.ssid,
                                 cdev->hostid.devid);
    VFIODevice *vbasedev;

    QLIST_FOREACH(vbasedev, &group->device_list, next) {
        if (strcmp(vbasedev->name, name) == 0) {
            error_setg(errp, "vfio: subchannel %s has already been
attached",
                       name);
            goto out_err;
        }
    }

>
>> i think this also matches
>> https://lore.kernel.org/all/PH7PR11MB67222DD282F98E03095FBA8A92C1A@PH
>> 7PR11MB6722.namprd11.prod.outlook.com/
>> no?
> It doesn't match what Mattew suggested: https://lore.kernel.org/qemu-devel/6e04ab8f-dc84-e9c2-deea-2b6b31678b53@linux.ibm.com/
this was RFC v3. At that time we did not pass any "name" arg to

vfio_attach_device(VFIODevice *vbasedev, AddressSpace *as, Error **errp)
and vbasedev->name was used when calling vfio_get_device() while we now
use cdev->mdevid. Besides Mattew ran some basic tests on PATCH v3:
https://lore.kernel.org/all/33b7803c-f231-d4fb-d9d9-26a097a89e93@redhat.com/
So I would be tempted to leave it as is (without the sysfsdev overwrite
which came from Mattew's suggestion in
https://lore.kernel.org/qemu-devel/6e04ab8f-dc84-e9c2-deea-2b6b31678b53@linux.ibm.com/
). Thanks Eric

>
> Thanks
> Zhenzhong
>