Re: [PATCH 1/6] hyperv: Make remove callback of hyperv driver void returned

[Dawei Li <set_pte_at@xxxxxxxxxxx>]

On Tue, Dec 06, 2022 at 11:37:26AM +0000, Wei Liu wrote:
> On Mon, Dec 05, 2022 at 11:36:39PM +0800, Dawei Li wrote:
> > Since commit fc7a6209d571 ("bus: Make remove callback return
> > void") forces bus_type::remove be void-returned, it doesn't
> > make much sense for any bus based driver implementing remove
> > callbalk to return non-void to its caller.
> > 
> > This change is for hyperv bus based drivers.
> > 
> > Signed-off-by: Dawei Li <set_pte_at@xxxxxxxxxxx>
> [...]

Hi Wei:
Thanks for the review.

> > -static int netvsc_remove(struct hv_device *dev)
> > +static void netvsc_remove(struct hv_device *dev)
> >  {
> >     struct net_device_context *ndev_ctx;
> >     struct net_device *vf_netdev, *net;
> > @@ -2603,7 +2603,6 @@ static int netvsc_remove(struct hv_device *dev)
> >     net = hv_get_drvdata(dev);
> >     if (net == NULL) {
> >             dev_err(&dev->device, "No net device to remove\n");
> > -           return 0;
> 
> This is wrong. You are introducing a NULL pointer dereference.
Nice catch, will fix it.

> 
> >     }
> >  
> >     ndev_ctx = netdev_priv(net);
> > @@ -2637,7 +2636,6 @@ static int netvsc_remove(struct hv_device *dev)
> >  
> >     free_percpu(ndev_ctx->vf_stats);
> >     free_netdev(net);
> > -   return 0;
> >  }
> >  
> >  static int netvsc_suspend(struct hv_device *dev)
> > diff --git a/drivers/pci/controller/pci-hyperv.c 
> > b/drivers/pci/controller/pci-hyperv.c
> > index ba64284eaf9f..3a09de70d6ea 100644
> > --- a/drivers/pci/controller/pci-hyperv.c
> > +++ b/drivers/pci/controller/pci-hyperv.c
> > @@ -3756,7 +3756,7 @@ static int hv_pci_bus_exit(struct hv_device *hdev, 
> > bool keep_devs)
> >   *
> >   * Return: 0 on success, -errno on failure
> >   */
> 
> This comment is no longer needed in the new world.
> 
> But, are you sure you're modifying the correct piece of code?
> 
> hv_pci_remove is not a hook in the base bus type. It is used in struct
> hv_driver.
> 
> The same comment applies to all other modifications.
Sorry about the confusion.
In short, the point of this commit is making remove() of _driver_ void returned.

For bus-based driver, device removal is implemented as:
1 device_remove() =>
  2 bus->remove() =>
    3 driver->remove()

1 is void. 
For 2, commit fc7a6209d571 ("bus: Make remove callback return void") forces
bus_type::remove be void-returned, which applies for hv_bus(vmbus) too.
So it doesn't make sense for 3(driver->remove) to return non-void, in this case 
it's hv_driver. 

Thanks,
      Dawei

> 
> Thanks,
> Wei.