Re: [PATCH v5 07/15] vfio/pci: Introduce vfio_[attach/detach]_device

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

Hi Zhenzhong,
On 10/11/23 03:33, Duan, Zhenzhong wrote:
> Hi Eric,
>
>> -----Original Message-----
>> From: Eric Auger <eric.auger@redhat.com>
>> Sent: Monday, October 9, 2023 5:09 PM
>> Subject: [PATCH v5 07/15] vfio/pci: Introduce vfio_[attach/detach]_device
>>
>> We want the VFIO devices to be able to use two different
>> IOMMU backends, the legacy VFIO one and the new iommufd one.
>>
>> Introduce vfio_[attach/detach]_device which aim at hiding the
>> underlying IOMMU backend (IOCTLs, datatypes, ...).
>>
>> Once vfio_attach_device completes, the device is attached
>> to a security context and its fd can be used. Conversely
>> When vfio_detach_device completes, the device has been
>> detached from the security context.
>>
>> At the moment only the implementation based on the legacy
>> container/group exists. Let's use it from the vfio-pci device.
>> Subsequent patches will handle other devices.
>>
>> We also take benefit of this patch to properly free
>> vbasedev->name on failure.
>>
>> 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: Cédric Le Goater <clg@redhat.com>
>>
>> ---
>>
>> v4 -> v5:
>> - remove vbasedev->name g_free as it is done on instance_finalize
>>
>> v2 -> v3:
>> - added trace_vfio_detach_device
>> - added a comment explaining why we pass @name to vfio_attach_device
>>  although vbasedev->name is populated
>> - free vbasedev->name and detach_device if needed
>> ---
>> include/hw/vfio/vfio-common.h |  3 ++
>> hw/vfio/common.c              | 74 +++++++++++++++++++++++++++++++++++
>> hw/vfio/pci.c                 | 66 +++++++------------------------
>> hw/vfio/trace-events          |  3 +-
>> 4 files changed, 93 insertions(+), 53 deletions(-)
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index c4e7c3b4a7..12fbfbc37d 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -225,6 +225,9 @@ 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);
>>
>> int vfio_kvm_device_add_fd(int fd, Error **errp);
>> int vfio_kvm_device_del_fd(int fd, Error **errp);
>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>> index d8ed432cb6..f4c33c9858 100644
>> --- a/hw/vfio/common.c
>> +++ b/hw/vfio/common.c
>> @@ -2611,3 +2611,77 @@ int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
>>     }
>>     return vfio_eeh_container_op(container, op);
>> }
>> +
>> +static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
>> +{
>> +    char *tmp, group_path[PATH_MAX], *group_name;
>> +    int ret, groupid;
>> +    ssize_t len;
>> +
>> +    tmp = g_strdup_printf("%s/iommu_group", vbasedev->sysfsdev);
>> +    len = readlink(tmp, group_path, sizeof(group_path));
>> +    g_free(tmp);
>> +
>> +    if (len <= 0 || len >= sizeof(group_path)) {
>> +        ret = len < 0 ? -errno : -ENAMETOOLONG;
>> +        error_setg_errno(errp, -ret, "no iommu_group found");
>> +        return ret;
>> +    }
>> +
>> +    group_path[len] = 0;
>> +
>> +    group_name = basename(group_path);
>> +    if (sscanf(group_name, "%d", &groupid) != 1) {
>> +        error_setg_errno(errp, errno, "failed to read %s", group_path);
>> +        return -errno;
>> +    }
>> +    return groupid;
>> +}
>> +
>> +/*
>> + * vfio_attach_device: attach a device to a security context
>> + * @name and @vbasedev->name are likely to be different depending
>> + * on the type of the device, hence the need for passing @name
>> + */
>> +int vfio_attach_device(char *name, VFIODevice *vbasedev,
>> +                       AddressSpace *as, Error **errp)
>> +{
>> +    int groupid = vfio_device_groupid(vbasedev, errp);
>> +    VFIODevice *vbasedev_iter;
>> +    VFIOGroup *group;
>> +    int ret;
>> +
>> +    if (groupid < 0) {
>> +        return groupid;
>> +    }
>> +
>> +    trace_vfio_attach_device(vbasedev->name, groupid);
>> +
>> +    group = vfio_get_group(groupid, as, errp);
>> +    if (!group) {
>> +        return -ENOENT;
>> +    }
>> +
>> +    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;
>> +        }
>> +    }
>> +    ret = vfio_get_device(group, name, vbasedev, errp);
>> +    if (ret) {
>> +        vfio_put_group(group);
>> +    }
>> +
>> +    return ret;
>> +}
>> +
>> +void vfio_detach_device(VFIODevice *vbasedev)
>> +{
>> +    VFIOGroup *group = vbasedev->group;
>> +
>> +    trace_vfio_detach_device(vbasedev->name, group->groupid);
>> +    vfio_put_base_device(vbasedev);
>> +    vfio_put_group(group);
>> +}
>> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
>> index 898296fd54..40ae46266e 100644
>> --- a/hw/vfio/pci.c
>> +++ b/hw/vfio/pci.c
>> @@ -2895,10 +2895,10 @@ static void vfio_populate_device(VFIOPCIDevice
>> *vdev, Error **errp)
>>
>> static void vfio_pci_put_device(VFIOPCIDevice *vdev)
>> {
>> +    vfio_detach_device(&vdev->vbasedev);
>> +
>>     g_free(vdev->vbasedev.name);
>>     g_free(vdev->msix);
>> -
>> -    vfio_put_base_device(&vdev->vbasedev);
>> }
>>
>> static void vfio_err_notifier_handler(void *opaque)
>> @@ -3045,13 +3045,9 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>> {
>>     VFIOPCIDevice *vdev = VFIO_PCI(pdev);
>>     VFIODevice *vbasedev = &vdev->vbasedev;
>> -    VFIODevice *vbasedev_iter;
>> -    VFIOGroup *group;
>> -    char *tmp, *subsys, group_path[PATH_MAX], *group_name;
>> +    char *tmp, *subsys;
>>     Error *err = NULL;
>> -    ssize_t len;
>>     struct stat st;
>> -    int groupid;
>>     int i, ret;
>>     bool is_mdev;
>>     char uuid[UUID_FMT_LEN];
>> @@ -3082,39 +3078,6 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     vbasedev->type = VFIO_DEVICE_TYPE_PCI;
>>     vbasedev->dev = DEVICE(vdev);
>>
>> -    tmp = g_strdup_printf("%s/iommu_group", vbasedev->sysfsdev);
>> -    len = readlink(tmp, group_path, sizeof(group_path));
>> -    g_free(tmp);
>> -
>> -    if (len <= 0 || len >= sizeof(group_path)) {
>> -        error_setg_errno(errp, len < 0 ? errno : ENAMETOOLONG,
>> -                         "no iommu_group found");
>> -        goto error;
>> -    }
>> -
>> -    group_path[len] = 0;
>> -
>> -    group_name = basename(group_path);
>> -    if (sscanf(group_name, "%d", &groupid) != 1) {
>> -        error_setg_errno(errp, errno, "failed to read %s", group_path);
>> -        goto error;
>> -    }
>> -
>> -    trace_vfio_realize(vbasedev->name, groupid);
>> -
>> -    group = vfio_get_group(groupid, pci_device_iommu_address_space(pdev),
>> errp);
>> -    if (!group) {
>> -        goto error;
>> -    }
>> -
>> -    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);
>> -            goto error;
>> -        }
>> -    }
>> -
>>     /*
>>      * Mediated devices *might* operate compatibly with discarding of RAM, but
>>      * we cannot know for certain, it depends on whether the mdev vendor driver
>> @@ -3132,7 +3095,6 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     if (vbasedev->ram_block_discard_allowed && !is_mdev) {
>>         error_setg(errp, "x-balloon-allowed only potentially compatible "
>>                    "with mdev devices");
>> -        vfio_put_group(group);
>>         goto error;
>>     }
>>
>> @@ -3143,17 +3105,17 @@ static void vfio_realize(PCIDevice *pdev, Error
>> **errp)
>>         name = g_strdup(vbasedev->name);
>>     }
>>
>> -    ret = vfio_get_device(group, name, vbasedev, errp);
>> +    ret = vfio_attach_device(name, vbasedev,
>> +                             pci_device_iommu_address_space(pdev), errp);
>>     g_free(name);
>>     if (ret) {
>> -        vfio_put_group(group);
>>         goto error;
>>     }
>>
>>     vfio_populate_device(vdev, &err);
>>     if (err) {
>>         error_propagate(errp, err);
>> -        goto error;
>> +        goto out_detach;
>>     }
>>
>>     /* Get a copy of config space */
>> @@ -3163,7 +3125,7 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     if (ret < (int)MIN(pci_config_size(&vdev->pdev), vdev->config_size)) {
>>         ret = ret < 0 ? -errno : -EFAULT;
>>         error_setg_errno(errp, -ret, "failed to read device config space");
>> -        goto error;
>> +        goto out_detach;
>>     }
>>
>>     /* vfio emulates a lot for us, but some bits need extra love */
>> @@ -3182,7 +3144,7 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     if (vdev->vendor_id != PCI_ANY_ID) {
>>         if (vdev->vendor_id >= 0xffff) {
>>             error_setg(errp, "invalid PCI vendor ID provided");
>> -            goto error;
>> +            goto out_detach;
>>         }
>>         vfio_add_emulated_word(vdev, PCI_VENDOR_ID, vdev->vendor_id, ~0);
>>         trace_vfio_pci_emulated_vendor_id(vbasedev->name, vdev->vendor_id);
>> @@ -3193,7 +3155,7 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     if (vdev->device_id != PCI_ANY_ID) {
>>         if (vdev->device_id > 0xffff) {
>>             error_setg(errp, "invalid PCI device ID provided");
>> -            goto error;
>> +            goto out_detach;
>>         }
>>         vfio_add_emulated_word(vdev, PCI_DEVICE_ID, vdev->device_id, ~0);
>>         trace_vfio_pci_emulated_device_id(vbasedev->name, vdev->device_id);
>> @@ -3204,7 +3166,7 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     if (vdev->sub_vendor_id != PCI_ANY_ID) {
>>         if (vdev->sub_vendor_id > 0xffff) {
>>             error_setg(errp, "invalid PCI subsystem vendor ID provided");
>> -            goto error;
>> +            goto out_detach;
>>         }
>>         vfio_add_emulated_word(vdev, PCI_SUBSYSTEM_VENDOR_ID,
>>                                vdev->sub_vendor_id, ~0);
>> @@ -3215,7 +3177,7 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     if (vdev->sub_device_id != PCI_ANY_ID) {
>>         if (vdev->sub_device_id > 0xffff) {
>>             error_setg(errp, "invalid PCI subsystem device ID provided");
>> -            goto error;
>> +            goto out_detach;
>>         }
>>         vfio_add_emulated_word(vdev, PCI_SUBSYSTEM_ID, vdev->sub_device_id,
>> ~0);
>>         trace_vfio_pci_emulated_sub_device_id(vbasedev->name,
>> @@ -3248,7 +3210,7 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>>     vfio_msix_early_setup(vdev, &err);
>>     if (err) {
>>         error_propagate(errp, err);
>> -        goto error;
>> +        goto out_detach;
>>     }
>>
>>     vfio_bars_register(vdev);
>> @@ -3364,6 +3326,8 @@ out_deregister:
>> out_teardown:
>>     vfio_teardown_msi(vdev);
>>     vfio_bars_exit(vdev);
>> +out_detach:
>> +    vfio_detach_device(vbasedev);
> I just reproduced a regression here with hot-unplug:
>
> (qemu) device_add vfio-pci,host=81:11.0,id=vfio0,bus=root0
> vfio: error disconnecting group 447 from container
> Error: vfio 0000:81:11.0: does not support requested IGD OpRegion feature
>
> Reason is there are some resources only released in finalize(),
> so UNSET_CONTAINER call fails here.
> IMHO, we can drop this vfio_detach_device call and let the same call in
> finalize() do the real work as original.
>
> After fix:
> (qemu) device_add vfio-pci,host=81:11.0,id=vfio0,bus=root0
> Error: vfio 0000:81:11.0: does not support requested IGD OpRegion feature
I just sent
[PATCH] vfio/pci: Remove vfio_detach_device from vfio_realize error path

Please have a check and confirm this is the fix you expected

Thanks!

Eric
>
> Thanks
> Zhenzhong
>> error:
>>     error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->name);
>> }
>> @@ -3371,7 +3335,6 @@ error:
>> static void vfio_instance_finalize(Object *obj)
>> {
>>     VFIOPCIDevice *vdev = VFIO_PCI(obj);
>> -    VFIOGroup *group = vdev->vbasedev.group;
>>
>>     vfio_display_finalize(vdev);
>>     vfio_bars_finalize(vdev);
>> @@ -3385,7 +3348,6 @@ static void vfio_instance_finalize(Object *obj)
>>      * g_free(vdev->igd_opregion);
>>      */
>>     vfio_pci_put_device(vdev);
>> -    vfio_put_group(group);
>> }
>>
>> static void vfio_exitfn(PCIDevice *pdev)
>> diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
>> index 0ba3c5a0e2..8ac13eb106 100644
>> --- a/hw/vfio/trace-events
>> +++ b/hw/vfio/trace-events
>> @@ -37,7 +37,8 @@ vfio_pci_hot_reset_dep_devices(int domain, int bus, int slot,
>> int function, int
>> vfio_pci_hot_reset_result(const char *name, const char *result) "%s hot
>> reset: %s"
>> vfio_populate_device_config(const char *name, unsigned long size, unsigned
>> long offset, unsigned long flags) "Device %s config:\n  size: 0x%lx, offset: 0x%lx,
>> flags: 0x%lx"
>> vfio_populate_device_get_irq_info_failure(const char *errstr)
>> "VFIO_DEVICE_GET_IRQ_INFO failure: %s"
>> -vfio_realize(const char *name, int group_id) " (%s) group %d"
>> +vfio_attach_device(const char *name, int group_id) " (%s) group %d"
>> +vfio_detach_device(const char *name, int group_id) " (%s) group %d"
>> vfio_mdev(const char *name, bool is_mdev) " (%s) is_mdev %d"
>> vfio_add_ext_cap_dropped(const char *name, uint16_t cap, uint16_t offset) "%s
>> 0x%x@0x%x"
>> vfio_pci_reset(const char *name) " (%s)"
>> --
>> 2.41.0