Re: [PATCH v27 07/17] vfio: Register SaveVMHandlers for VFIO device

[Kirti Wankhede <kwankhede@nvidia.com>]

On 10/23/2020 12:21 AM, Alex Williamson wrote:
> On Thu, 22 Oct 2020 16:41:57 +0530
> Kirti Wankhede <kwankhede@nvidia.com> wrote:
> 
>> Define flags to be used as delimiter in migration stream for VFIO devices.
>> Added .save_setup and .save_cleanup functions. Map & unmap migration
>> region from these functions at source during saving or pre-copy phase.
>>
>> Set VFIO device state depending on VM's state. During live migration, VM is
>> running when .save_setup is called, _SAVING | _RUNNING state is set for VFIO
>> device. During save-restore, VM is paused, _SAVING state is set for VFIO device.
>>
>> Signed-off-by: Kirti Wankhede <kwankhede@nvidia.com>
>> Reviewed-by: Neo Jia <cjia@nvidia.com>
>> ---
>>   hw/vfio/migration.c  | 96 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   hw/vfio/trace-events |  2 ++
>>   2 files changed, 98 insertions(+)
>>
>> diff --git a/hw/vfio/migration.c b/hw/vfio/migration.c
>> index 7c4fa0d08ea6..2e1054bf7f43 100644
>> --- a/hw/vfio/migration.c
>> +++ b/hw/vfio/migration.c
>> @@ -8,12 +8,15 @@
>>    */
>>   
>>   #include "qemu/osdep.h"
>> +#include "qemu/main-loop.h"
>> +#include "qemu/cutils.h"
>>   #include <linux/vfio.h>
>>   
>>   #include "sysemu/runstate.h"
>>   #include "hw/vfio/vfio-common.h"
>>   #include "cpu.h"
>>   #include "migration/migration.h"
>> +#include "migration/vmstate.h"
>>   #include "migration/qemu-file.h"
>>   #include "migration/register.h"
>>   #include "migration/blocker.h"
>> @@ -25,6 +28,22 @@
>>   #include "trace.h"
>>   #include "hw/hw.h"
>>   
>> +/*
>> + * Flags to be used as unique delimiters for VFIO devices in the migration
>> + * stream. These flags are composed as:
>> + * 0xffffffff => MSB 32-bit all 1s
>> + * 0xef10     => Magic ID, represents emulated (virtual) function IO
>> + * 0x0000     => 16-bits reserved for flags
>> + *
>> + * The beginning of state information is marked by _DEV_CONFIG_STATE,
>> + * _DEV_SETUP_STATE, or _DEV_DATA_STATE, respectively. The end of a
>> + * certain state information is marked by _END_OF_STATE.
>> + */
>> +#define VFIO_MIG_FLAG_END_OF_STATE      (0xffffffffef100001ULL)
>> +#define VFIO_MIG_FLAG_DEV_CONFIG_STATE  (0xffffffffef100002ULL)
>> +#define VFIO_MIG_FLAG_DEV_SETUP_STATE   (0xffffffffef100003ULL)
>> +#define VFIO_MIG_FLAG_DEV_DATA_STATE    (0xffffffffef100004ULL)
>> +
>>   static inline int vfio_mig_access(VFIODevice *vbasedev, void *val, int count,
>>                                     off_t off, bool iswrite)
>>   {
>> @@ -129,6 +148,69 @@ static int vfio_migration_set_state(VFIODevice *vbasedev, uint32_t mask,
>>       return 0;
>>   }
>>   
>> +/* ---------------------------------------------------------------------- */
>> +
>> +static int vfio_save_setup(QEMUFile *f, void *opaque)
>> +{
>> +    VFIODevice *vbasedev = opaque;
>> +    VFIOMigration *migration = vbasedev->migration;
>> +    int ret;
>> +
>> +    trace_vfio_save_setup(vbasedev->name);
>> +
>> +    qemu_put_be64(f, VFIO_MIG_FLAG_DEV_SETUP_STATE);
>> +
>> +    if (migration->region.mmaps) {
>> +        /*
>> +         * vfio_region_mmap() called from migration thread. Memory API called
>> +         * from vfio_regio_mmap() need it when called from outdide the main loop
>> +         * thread.
>> +         */
> 
> Thanks for adding this detail, maybe refine slightly as:
> 
>    Calling vfio_region_mmap() from migration thread.  Memory APIs called
>    from this function require locking the iothread when called from
>    outside the main loop thread.
> 
> Does that capture the intent?
> 

Ok.

>> +        qemu_mutex_lock_iothread();
>> +        ret = vfio_region_mmap(&migration->region);
>> +        qemu_mutex_unlock_iothread();
>> +        if (ret) {
>> +            error_report("%s: Failed to mmap VFIO migration region: %s",
>> +                         vbasedev->name, strerror(-ret));
>> +            error_report("%s: Falling back to slow path", vbasedev->name);
>> +        }
>> +    }
>> +
>> +    ret = vfio_migration_set_state(vbasedev, VFIO_DEVICE_STATE_MASK,
>> +                                   VFIO_DEVICE_STATE_SAVING);
>> +    if (ret) {
>> +        error_report("%s: Failed to set state SAVING", vbasedev->name);
>> +        return ret;
>> +    }
>> +
>> +    qemu_put_be64(f, VFIO_MIG_FLAG_END_OF_STATE);
>> +
>> +    ret = qemu_file_get_error(f);
>> +    if (ret) {
>> +        return ret;
>> +    }
>> +
>> +    return 0;
>> +}
>> +
>> +static void vfio_save_cleanup(void *opaque)
>> +{
>> +    VFIODevice *vbasedev = opaque;
>> +    VFIOMigration *migration = vbasedev->migration;
>> +
>> +    if (migration->region.mmaps) {
>> +        vfio_region_unmap(&migration->region);
>> +    }
> 
> 
> Are we in a different thread context here that we don't need that same
> iothread locking?
> 

qemu_savevm_state_setup() is called without holding iothread lock and 
qemu_savevm_state_cleanup() is called holding iothread lock, so we don't 
need lock here.

> 
>> +    trace_vfio_save_cleanup(vbasedev->name);
>> +}
>> +
>> +static SaveVMHandlers savevm_vfio_handlers = {
>> +    .save_setup = vfio_save_setup,
>> +    .save_cleanup = vfio_save_cleanup,
>> +};
>> +
>> +/* ---------------------------------------------------------------------- */
>> +
>>   static void vfio_vmstate_change(void *opaque, int running, RunState state)
>>   {
>>       VFIODevice *vbasedev = opaque;
>> @@ -219,6 +301,8 @@ static int vfio_migration_init(VFIODevice *vbasedev,
>>       int ret;
>>       Object *obj;
>>       VFIOMigration *migration;
>> +    char id[256] = "";
>> +    g_autofree char *path = NULL, *oid;
> 
> 
> AIUI, oid must also be initialized as a g_autofree variable.
> 
>>   
>>       if (!vbasedev->ops->vfio_get_object) {
>>           return -EINVAL;
>> @@ -248,6 +332,18 @@ static int vfio_migration_init(VFIODevice *vbasedev,
>>   
>>       vbasedev->migration = migration;
>>       migration->vbasedev = vbasedev;
>> +
>> +    oid = vmstate_if_get_id(VMSTATE_IF(DEVICE(obj)));
>> +    if (oid) {
>> +        path = g_strdup_printf("%s/vfio", oid);
>> +    } else {
>> +        path = g_strdup("vfio");
> 
> 
> If we get here then all vfio devices have the same id string.  Isn't
> that a problem?  Thanks,
> 

Most of the bus types has get_dev_path() callback implemented which is 
called from get_id, so there are very less chance to get here.
With above change, id string we get looks like '0000:00:04.0/vfio', 
trace logs below:
qemu_loadvm_state_section_startfull 61.942 pid=625231 section_id=0x2f 
idstr=b'0000:00:04.0/vfio' instance_id=0x0 version_id=0x1

qemu_loadvm_state_section_startfull 1.242 pid=625231 section_id=0x30 
idstr=b'0000:00:05.0/vfio' instance_id=0x0 version_id=0x1

where '0000:00:04.0'shows location within guest, so that it gets 
preserved and used during resume.

In the worst when it is not present, idstr remains same but instance_id 
changes:

qemu_loadvm_state_section_startfull 54.931 pid=609474 section_id=0x2f 
idstr=b'vfio' instance_id=0x0 version_id=0x1

qemu_loadvm_state_section_startfull 1.180 pid=609474 section_id=0x30 
idstr=b'vfio' instance_id=0x1 version_id=0x1

But there is no other way to know location of the device within guest.

Dave, any suggestions here?

Thanks,
Kirti


> Alex
> 
> 
>> +    }
>> +    strpadcpy(id, sizeof(id), path, '\0');
>> +
>> +    register_savevm_live(id, VMSTATE_INSTANCE_ID_ANY, 1, &savevm_vfio_handlers,
>> +                         vbasedev);
>> +
>>       migration->vm_state = qemu_add_vm_change_state_handler(vfio_vmstate_change,
>>                                                             vbasedev);
>>       migration->migration_state.notify = vfio_migration_state_notifier;
>> diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
>> index 78d7d83b5ef8..f148b5e828c1 100644
>> --- a/hw/vfio/trace-events
>> +++ b/hw/vfio/trace-events
>> @@ -151,3 +151,5 @@ vfio_migration_probe(const char *name, uint32_t index) " (%s) Region %d"
>>   vfio_migration_set_state(const char *name, uint32_t state) " (%s) state %d"
>>   vfio_vmstate_change(const char *name, int running, const char *reason, uint32_t dev_state) " (%s) running %d reason %s device state %d"
>>   vfio_migration_state_notifier(const char *name, const char *state) " (%s) state %s"
>> +vfio_save_setup(const char *name) " (%s)"
>> +vfio_save_cleanup(const char *name) " (%s)"
> 
>