Re: [RFC PATCH v2 1/6] remoteproc: core: Introduce virtio device add/remove functions

[Arnaud POULIQUEN <arnaud.pouliquen@xxxxxxxxxxx>]

Hello Mathieu,

On 1/4/22 8:08 PM, Mathieu Poirier wrote:
> Good morning,
>
> On Wed, Dec 22, 2021 at 09:23:44AM +0100, Arnaud Pouliquen wrote:
> > In preparation of the migration of the management of rvdev in
> > remoteproc_virtio.c, this patch spins off new functions to manage the
> > remoteproc virtio device.
> >
> > The rproc_rvdev_add_device and rproc_rvdev_remove_device will be
> > moved to remoteproc_virtio.c.
> >
> > Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@xxxxxxxxxxx>
> > ---
> > update vs previous revision:
> >   - update according to the rebase from v15-rc1 to v16-rc1
> >   - split patch to introduce rproc_register_rvdev and rproc_unregister_rvdev
> >     function in a separate patch
> > ---
> >   drivers/remoteproc/remoteproc_core.c | 94 +++++++++++++++++-----------
> >   1 file changed, 57 insertions(+), 37 deletions(-)
> >
> > diff --git a/drivers/remoteproc/remoteproc_core.c b/drivers/remoteproc/remoteproc_core.c
> > index 69f51acf235e..d1f1c5c25bd7 100644
> > --- a/drivers/remoteproc/remoteproc_core.c
> > +++ b/drivers/remoteproc/remoteproc_core.c
> > @@ -484,6 +484,61 @@ static int copy_dma_range_map(struct device *to, struct device *from)
> >   	return 0;
> >   }
> >   
> > +static int rproc_rvdev_add_device(struct rproc_vdev *rvdev)
> > +{
> > +	struct rproc *rproc = rvdev->rproc;
> > +	char name[16];
> > +	int ret;
> > +
> > +	snprintf(name, sizeof(name), "vdev%dbuffer", rvdev->index);
> > +	rvdev->dev.parent = &rproc->dev;
> > +	rvdev->dev.release = rproc_rvdev_release;
> > +	dev_set_name(&rvdev->dev, "%s#%s", dev_name(rvdev->dev.parent), name);
> > +	dev_set_drvdata(&rvdev->dev, rvdev);
> > +
> > +	ret = device_register(&rvdev->dev);
> > +	if (ret) {
> > +		put_device(&rvdev->dev);
> > +		return ret;
> > +	}
>
> Registering the device here is a problem...  If device_register() fails
> put_device() is called and we return, only to call device_unregister() on the
> same device in rproc_handle_vdev().
>
> Moreover in rproc_handle_vdev(), device_unregister() is called in the error
> path but device_register() is called here in rproc_rvdev_add_device().  This
> introduces coupling between the two functions, making it hard to maintain from
> hereon.

Very relevant, I need to rework the error management.

> I suggest calling device_register() in rproc_handle_vdev() after
> rproc_rvdev_add_device() has returned successfully.

One of the goals of this patchset is to move the device_register in 
remote_proc_virtio.c
Doing this would not go in this direction.

I need to test but following could be an alternative:
- Call rproc_rvdev_remove_device in rproc_handle_vdev in case of error.
- Remove the put_device in rproc_rvdev_add_device.

=> This would be aligned with patch [6/6] implementation
with rproc_virtio_register_device/rproc_virtio_unregister_device...

Thanks,
Arnaud

> More comments to come tomorrow.
>
> Thanks,
> Mathieu
>
> > +	ret = copy_dma_range_map(&rvdev->dev, rproc->dev.parent);
> > +	if (ret)
> > +		goto free_rvdev;
> > +
> > +	/* Make device dma capable by inheriting from parent's capabilities */
> > +	set_dma_ops(&rvdev->dev, get_dma_ops(rproc->dev.parent));
> > +
> > +	ret = dma_coerce_mask_and_coherent(&rvdev->dev,
> > +					   dma_get_mask(rproc->dev.parent));
> > +	if (ret) {
> > +		dev_warn(&rvdev->dev,
> > +			 "Failed to set DMA mask %llx. Trying to continue... (%pe)\n",
> > +			 dma_get_mask(rproc->dev.parent), ERR_PTR(ret));
> > +	}
> > +
> > +	list_add_tail(&rvdev->node, &rproc->rvdevs);
> > +
> > +	rvdev->subdev.start = rproc_vdev_do_start;
> > +	rvdev->subdev.stop = rproc_vdev_do_stop;
> > +
> > +	rproc_add_subdev(rproc, &rvdev->subdev);
> > +
> > +	return 0;
> > +
> > +free_rvdev:
> > +	device_unregister(&rvdev->dev);
> > +	return ret;
> > +}
> > +
> > +static void rproc_rvdev_remove_device(struct rproc_vdev *rvdev)
> > +{
> > +	struct rproc *rproc = rvdev->rproc;
> > +
> > +	rproc_remove_subdev(rproc, &rvdev->subdev);
> > +	list_del(&rvdev->node);
> > +	device_unregister(&rvdev->dev);
> > +}
> > +
> >   /**
> >    * rproc_handle_vdev() - handle a vdev fw resource
> >    * @rproc: the remote processor
> > @@ -519,7 +574,6 @@ static int rproc_handle_vdev(struct rproc *rproc, void *ptr,
> >   	struct device *dev = &rproc->dev;
> >   	struct rproc_vdev *rvdev;
> >   	int i, ret;
> > -	char name[16];
> >   
> >   	/* make sure resource isn't truncated */
> >   	if (struct_size(rsc, vring, rsc->num_of_vrings) + rsc->config_len >
> > @@ -553,34 +607,10 @@ static int rproc_handle_vdev(struct rproc *rproc, void *ptr,
> >   	rvdev->rproc = rproc;
> >   	rvdev->index = rproc->nb_vdev++;
> >   
> > -	/* Initialise vdev subdevice */
> > -	snprintf(name, sizeof(name), "vdev%dbuffer", rvdev->index);
> > -	rvdev->dev.parent = &rproc->dev;
> > -	rvdev->dev.release = rproc_rvdev_release;
> > -	dev_set_name(&rvdev->dev, "%s#%s", dev_name(rvdev->dev.parent), name);
> > -	dev_set_drvdata(&rvdev->dev, rvdev);
> > -
> > -	ret = device_register(&rvdev->dev);
> > -	if (ret) {
> > -		put_device(&rvdev->dev);
> > -		return ret;
> > -	}
> > -
> > -	ret = copy_dma_range_map(&rvdev->dev, rproc->dev.parent);
> > +	ret = rproc_rvdev_add_device(rvdev);
> >   	if (ret)
> >   		goto free_rvdev;
> >   
> > -	/* Make device dma capable by inheriting from parent's capabilities */
> > -	set_dma_ops(&rvdev->dev, get_dma_ops(rproc->dev.parent));
> > -
> > -	ret = dma_coerce_mask_and_coherent(&rvdev->dev,
> > -					   dma_get_mask(rproc->dev.parent));
> > -	if (ret) {
> > -		dev_warn(dev,
> > -			 "Failed to set DMA mask %llx. Trying to continue... (%pe)\n",
> > -			 dma_get_mask(rproc->dev.parent), ERR_PTR(ret));
> > -	}
> > -
> >   	/* parse the vrings */
> >   	for (i = 0; i < rsc->num_of_vrings; i++) {
> >   		ret = rproc_parse_vring(rvdev, rsc, i);
> > @@ -598,13 +628,6 @@ static int rproc_handle_vdev(struct rproc *rproc, void *ptr,
> >   			goto unwind_vring_allocations;
> >   	}
> >   
> > -	list_add_tail(&rvdev->node, &rproc->rvdevs);
> > -
> > -	rvdev->subdev.start = rproc_vdev_do_start;
> > -	rvdev->subdev.stop = rproc_vdev_do_stop;
> > -
> > -	rproc_add_subdev(rproc, &rvdev->subdev);
> > -
> >   	return 0;
> >   
> >   unwind_vring_allocations:
> > @@ -619,7 +642,6 @@ void rproc_vdev_release(struct kref *ref)
> >   {
> >   	struct rproc_vdev *rvdev = container_of(ref, struct rproc_vdev, refcount);
> >   	struct rproc_vring *rvring;
> > -	struct rproc *rproc = rvdev->rproc;
> >   	int id;
> >   
> >   	for (id = 0; id < ARRAY_SIZE(rvdev->vring); id++) {
> > @@ -627,9 +649,7 @@ void rproc_vdev_release(struct kref *ref)
> >   		rproc_free_vring(rvring);
> >   	}
> >   
> > -	rproc_remove_subdev(rproc, &rvdev->subdev);
> > -	list_del(&rvdev->node);
> > -	device_unregister(&rvdev->dev);
> > +	rproc_rvdev_remove_device(rvdev);
> >   }
> >   
> >   /**
> > --
> > 2.17.1