Re: [PATCH 1/2] eni_vdpa: Fix an error handling path in 'eni_vdpa_probe()'

Re: [PATCH 1/2] eni_vdpa: Fix an error handling path in 'eni_vdpa_probe()'

[Jason Wang]

On Mon, Nov 8, 2021 at 12:15 AM Christophe JAILLET
<christophe.jaillet@wanadoo.fr> wrote:
>
> In the error handling path, a successful 'vp_legacy_probe()' should be
> balanced by a corresponding 'vp_legacy_remove()' call, as already done in
> the remove function.
>
> Add the missing call and update gotos accordingly.
>
> Fixes: e85087beedca ("eni_vdpa: add vDPA driver for Alibaba ENI")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
>  drivers/vdpa/alibaba/eni_vdpa.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/vdpa/alibaba/eni_vdpa.c b/drivers/vdpa/alibaba/eni_vdpa.c
> index 3f788794571a..12b3db6b4517 100644
> --- a/drivers/vdpa/alibaba/eni_vdpa.c
> +++ b/drivers/vdpa/alibaba/eni_vdpa.c
> @@ -501,7 +501,7 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>         if (!eni_vdpa->vring) {
>                 ret = -ENOMEM;
>                 ENI_ERR(pdev, "failed to allocate virtqueues\n");
> -               goto err;
> +               goto err_remove_vp_legacy;
>         }
>
>         for (i = 0; i < eni_vdpa->queues; i++) {
> @@ -513,11 +513,13 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>         ret = vdpa_register_device(&eni_vdpa->vdpa, eni_vdpa->queues);
>         if (ret) {
>                 ENI_ERR(pdev, "failed to register to vdpa bus\n");
> -               goto err;
> +               goto err_remove_vp_legacy;
>         }
>
>         return 0;
>
> +err_remove_vp_legacy:
> +       vp_legacy_remove(&eni_vdpa->ldev);

Won't vp_legacy_remove() be triggered by the put_devic() below?

Thanks

>  err:
>         put_device(&eni_vdpa->vdpa.dev);
>         return ret;
> --
> 2.30.2
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH 1/2] eni_vdpa: Fix an error handling path in 'eni_vdpa_probe()'

[Jason Wang]

On Tue, Nov 9, 2021 at 3:32 AM Christophe JAILLET
<christophe.jaillet@wanadoo.fr> wrote:
>
> Le 08/11/2021 à 06:55, Jason Wang a écrit :
> > On Mon, Nov 8, 2021 at 12:15 AM Christophe JAILLET
> > <christophe.jaillet@wanadoo.fr> wrote:
> >>
> >> In the error handling path, a successful 'vp_legacy_probe()' should be
> >> balanced by a corresponding 'vp_legacy_remove()' call, as already done in
> >> the remove function.
> >>
> >> Add the missing call and update gotos accordingly.
> >>
> >> Fixes: e85087beedca ("eni_vdpa: add vDPA driver for Alibaba ENI")
> >> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> >> ---
> >>   drivers/vdpa/alibaba/eni_vdpa.c | 6 ++++--
> >>   1 file changed, 4 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/vdpa/alibaba/eni_vdpa.c b/drivers/vdpa/alibaba/eni_vdpa.c
> >> index 3f788794571a..12b3db6b4517 100644
> >> --- a/drivers/vdpa/alibaba/eni_vdpa.c
> >> +++ b/drivers/vdpa/alibaba/eni_vdpa.c
> >> @@ -501,7 +501,7 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> >>          if (!eni_vdpa->vring) {
> >>                  ret = -ENOMEM;
> >>                  ENI_ERR(pdev, "failed to allocate virtqueues\n");
> >> -               goto err;
> >> +               goto err_remove_vp_legacy;
> >>          }
> >>
> >>          for (i = 0; i < eni_vdpa->queues; i++) {
> >> @@ -513,11 +513,13 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> >>          ret = vdpa_register_device(&eni_vdpa->vdpa, eni_vdpa->queues);
> >>          if (ret) {
> >>                  ENI_ERR(pdev, "failed to register to vdpa bus\n");
> >> -               goto err;
> >> +               goto err_remove_vp_legacy;
> >>          }
> >>
> >>          return 0;
> >>
> >> +err_remove_vp_legacy:
> >> +       vp_legacy_remove(&eni_vdpa->ldev);
> >
> > Won't vp_legacy_remove() be triggered by the put_devic() below?
>
> Hi, I'm sorry but i don't see how.
>
> My understanding is that:
>    - on "put_device(&eni_vdpa->vdpa.dev);", the corresponding release
> function will be called.
>
>    - This release function is the one recorded in the
> '__vdpa_alloc_device()' function.
>
>    - So it should be 'vdpa_release_dev()'.
>
>    - This function, AFAIU, has no knowledge of 'vp_legacy_remove()' or
> anything that could call it.
>
>
> Unless I misread something or miss something obvious, I don't see how
> 'vp_legacy_remove() would be called.
>
>
> Could you elaborate?

I think the device should release the driver (see
device_release_driver()) during during its deleting.

Thanks

>
> CJ
>
> >
> > Thanks
> >
> >>   err:
> >>          put_device(&eni_vdpa->vdpa.dev);
> >>          return ret;
> >> --
> >> 2.30.2
> >>
> >
> >
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH 2/2] eni_vdpa: Simplify 'eni_vdpa_probe()'

[Jason Wang]

On Mon, Nov 8, 2021 at 12:15 AM Christophe JAILLET
<christophe.jaillet@wanadoo.fr> wrote:
>
> When 'pcim_enable_device()' is used, some resources become automagically
> managed.
> There is no need to call 'pci_free_irq_vectors()' when the driver is
> removed. The same will already be done by 'pcim_release()'.
>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>

Acked-by: Jason Wang <jasowang@redhat.com>

> ---
> Leaving the 'pci_free_irq_vectors()' call is harmless.
> ---
>  drivers/vdpa/alibaba/eni_vdpa.c | 12 ------------
>  1 file changed, 12 deletions(-)
>
> diff --git a/drivers/vdpa/alibaba/eni_vdpa.c b/drivers/vdpa/alibaba/eni_vdpa.c
> index 12b3db6b4517..3db686436628 100644
> --- a/drivers/vdpa/alibaba/eni_vdpa.c
> +++ b/drivers/vdpa/alibaba/eni_vdpa.c
> @@ -450,11 +450,6 @@ static u16 eni_vdpa_get_num_queues(struct eni_vdpa *eni_vdpa)
>         return num;
>  }
>
> -static void eni_vdpa_free_irq_vectors(void *data)
> -{
> -       pci_free_irq_vectors(data);
> -}
> -
>  static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>  {
>         struct device *dev = &pdev->dev;
> @@ -488,13 +483,6 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>         eni_vdpa->vdpa.dma_dev = &pdev->dev;
>         eni_vdpa->queues = eni_vdpa_get_num_queues(eni_vdpa);
>
> -       ret = devm_add_action_or_reset(dev, eni_vdpa_free_irq_vectors, pdev);
> -       if (ret) {
> -               ENI_ERR(pdev,
> -                       "failed for adding devres for freeing irq vectors\n");
> -               goto err;
> -       }
> -
>         eni_vdpa->vring = devm_kcalloc(&pdev->dev, eni_vdpa->queues,
>                                       sizeof(*eni_vdpa->vring),
>                                       GFP_KERNEL);
> --
> 2.30.2
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH 1/2] eni_vdpa: Fix an error handling path in 'eni_vdpa_probe()'

[Michael S. Tsirkin]

On Tue, Nov 09, 2021 at 10:21:44PM +0100, Christophe JAILLET wrote:
> Le 09/11/2021 à 03:54, Jason Wang a écrit :
> > On Tue, Nov 9, 2021 at 3:32 AM Christophe JAILLET
> > <christophe.jaillet@wanadoo.fr> wrote:
> > > 
> > > Le 08/11/2021 à 06:55, Jason Wang a écrit :
> > > > On Mon, Nov 8, 2021 at 12:15 AM Christophe JAILLET
> > > > <christophe.jaillet@wanadoo.fr> wrote:
> > > > > 
> > > > > In the error handling path, a successful 'vp_legacy_probe()' should be
> > > > > balanced by a corresponding 'vp_legacy_remove()' call, as already done in
> > > > > the remove function.
> > > > > 
> > > > > Add the missing call and update gotos accordingly.
> > > > > 
> > > > > Fixes: e85087beedca ("eni_vdpa: add vDPA driver for Alibaba ENI")
> > > > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > > > > ---
> > > > >    drivers/vdpa/alibaba/eni_vdpa.c | 6 ++++--
> > > > >    1 file changed, 4 insertions(+), 2 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/vdpa/alibaba/eni_vdpa.c b/drivers/vdpa/alibaba/eni_vdpa.c
> > > > > index 3f788794571a..12b3db6b4517 100644
> > > > > --- a/drivers/vdpa/alibaba/eni_vdpa.c
> > > > > +++ b/drivers/vdpa/alibaba/eni_vdpa.c
> > > > > @@ -501,7 +501,7 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> > > > >           if (!eni_vdpa->vring) {
> > > > >                   ret = -ENOMEM;
> > > > >                   ENI_ERR(pdev, "failed to allocate virtqueues\n");
> > > > > -               goto err;
> > > > > +               goto err_remove_vp_legacy;
> > > > >           }
> > > > > 
> > > > >           for (i = 0; i < eni_vdpa->queues; i++) {
> > > > > @@ -513,11 +513,13 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> > > > >           ret = vdpa_register_device(&eni_vdpa->vdpa, eni_vdpa->queues);
> > > > >           if (ret) {
> > > > >                   ENI_ERR(pdev, "failed to register to vdpa bus\n");
> > > > > -               goto err;
> > > > > +               goto err_remove_vp_legacy;
> > > > >           }
> > > > > 
> > > > >           return 0;
> > > > > 
> > > > > +err_remove_vp_legacy:
> > > > > +       vp_legacy_remove(&eni_vdpa->ldev);
> > > > 
> > > > Won't vp_legacy_remove() be triggered by the put_devic() below?
> > > 
> > > Hi, I'm sorry but i don't see how.
> > > 
> > > My understanding is that:
> > >     - on "put_device(&eni_vdpa->vdpa.dev);", the corresponding release
> > > function will be called.
> > > 
> > >     - This release function is the one recorded in the
> > > '__vdpa_alloc_device()' function.
> > > 
> > >     - So it should be 'vdpa_release_dev()'.
> > > 
> > >     - This function, AFAIU, has no knowledge of 'vp_legacy_remove()' or
> > > anything that could call it.
> > > 
> > > 
> > > Unless I misread something or miss something obvious, I don't see how
> > > 'vp_legacy_remove() would be called.
> > > 
> > > 
> > > Could you elaborate?
> > 
> > I think the device should release the driver (see
> > device_release_driver()) during during its deleting.
> 
> Hi, I still don't follow the logic and I don't understand how
> 'vp_legacy_remove()' will finely be called, but it is not that important.
> 
> If it's fine for you, it's fine for me :)
> 
> CJ

So pls post just patch 2?

> > 
> > Thanks
> > 
> > > 
> > > CJ
> > > 
> > > > 
> > > > Thanks
> > > > 
> > > > >    err:
> > > > >           put_device(&eni_vdpa->vdpa.dev);
> > > > >           return ret;
> > > > > --
> > > > > 2.30.2
> > > > > 
> > > > 
> > > > 
> > > 
> > 
> > _______________________________________________
> > Virtualization mailing list
> > Virtualization@lists.linux-foundation.org
> > https://lists.linuxfoundation.org/mailman/listinfo/virtualization
> > 

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH 1/2] eni_vdpa: Fix an error handling path in 'eni_vdpa_probe()'

[Michael S. Tsirkin]

On Thu, Nov 25, 2021 at 07:34:02AM +0100, Christophe JAILLET wrote:
> Le 25/11/2021 à 00:58, Michael S. Tsirkin a écrit :
> > On Tue, Nov 09, 2021 at 10:21:44PM +0100, Christophe JAILLET wrote:
> > > Le 09/11/2021 à 03:54, Jason Wang a écrit :
> > > > On Tue, Nov 9, 2021 at 3:32 AM Christophe JAILLET
> > > > <christophe.jaillet@wanadoo.fr> wrote:
> > > > > 
> > > > > Le 08/11/2021 à 06:55, Jason Wang a écrit :
> > > > > > On Mon, Nov 8, 2021 at 12:15 AM Christophe JAILLET
> > > > > > <christophe.jaillet@wanadoo.fr> wrote:
> > > > > > > 
> > > > > > > In the error handling path, a successful 'vp_legacy_probe()' should be
> > > > > > > balanced by a corresponding 'vp_legacy_remove()' call, as already done in
> > > > > > > the remove function.
> > > > > > > 
> > > > > > > Add the missing call and update gotos accordingly.
> > > > > > > 
> > > > > > > Fixes: e85087beedca ("eni_vdpa: add vDPA driver for Alibaba ENI")
> > > > > > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > > > > > > ---
> > > > > > >     drivers/vdpa/alibaba/eni_vdpa.c | 6 ++++--
> > > > > > >     1 file changed, 4 insertions(+), 2 deletions(-)
> > > > > > > 
> > > > > > > diff --git a/drivers/vdpa/alibaba/eni_vdpa.c b/drivers/vdpa/alibaba/eni_vdpa.c
> > > > > > > index 3f788794571a..12b3db6b4517 100644
> > > > > > > --- a/drivers/vdpa/alibaba/eni_vdpa.c
> > > > > > > +++ b/drivers/vdpa/alibaba/eni_vdpa.c
> > > > > > > @@ -501,7 +501,7 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> > > > > > >            if (!eni_vdpa->vring) {
> > > > > > >                    ret = -ENOMEM;
> > > > > > >                    ENI_ERR(pdev, "failed to allocate virtqueues\n");
> > > > > > > -               goto err;
> > > > > > > +               goto err_remove_vp_legacy;
> > > > > > >            }
> > > > > > > 
> > > > > > >            for (i = 0; i < eni_vdpa->queues; i++) {
> > > > > > > @@ -513,11 +513,13 @@ static int eni_vdpa_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> > > > > > >            ret = vdpa_register_device(&eni_vdpa->vdpa, eni_vdpa->queues);
> > > > > > >            if (ret) {
> > > > > > >                    ENI_ERR(pdev, "failed to register to vdpa bus\n");
> > > > > > > -               goto err;
> > > > > > > +               goto err_remove_vp_legacy;
> > > > > > >            }
> > > > > > > 
> > > > > > >            return 0;
> > > > > > > 
> > > > > > > +err_remove_vp_legacy:
> > > > > > > +       vp_legacy_remove(&eni_vdpa->ldev);
> > > > > > 
> > > > > > Won't vp_legacy_remove() be triggered by the put_devic() below?
> > > > > 
> > > > > Hi, I'm sorry but i don't see how.
> > > > > 
> > > > > My understanding is that:
> > > > >      - on "put_device(&eni_vdpa->vdpa.dev);", the corresponding release
> > > > > function will be called.
> > > > > 
> > > > >      - This release function is the one recorded in the
> > > > > '__vdpa_alloc_device()' function.
> > > > > 
> > > > >      - So it should be 'vdpa_release_dev()'.
> > > > > 
> > > > >      - This function, AFAIU, has no knowledge of 'vp_legacy_remove()' or
> > > > > anything that could call it.
> > > > > 
> > > > > 
> > > > > Unless I misread something or miss something obvious, I don't see how
> > > > > 'vp_legacy_remove() would be called.
> > > > > 
> > > > > 
> > > > > Could you elaborate?
> > > > 
> > > > I think the device should release the driver (see
> > > > device_release_driver()) during during its deleting.
> > > 
> > > Hi, I still don't follow the logic and I don't understand how
> > > 'vp_legacy_remove()' will finely be called, but it is not that important.
> > > 
> > > If it's fine for you, it's fine for me :)
> > > 
> > > CJ
> > 
> > So pls post just patch 2?
> 
> Patch 2/2 should apply cleanly with or without patch 1/2.
> Do I really need to send a v2 just for dropping the first patch? :/
> 
> CJ

It is preferable, yes. A patch in isolation is not the same as a patch
in the series.

> > 
> > > > 
> > > > Thanks
> > > > 
> > > > > 
> > > > > CJ
> > > > > 
> > > > > > 
> > > > > > Thanks
> > > > > > 
> > > > > > >     err:
> > > > > > >            put_device(&eni_vdpa->vdpa.dev);
> > > > > > >            return ret;
> > > > > > > --
> > > > > > > 2.30.2
> > > > > > > 
> > > > > > 
> > > > > > 
> > > > > 
> > > > 
> > > > _______________________________________________
> > > > Virtualization mailing list
> > > > Virtualization@lists.linux-foundation.org
> > > > https://lists.linuxfoundation.org/mailman/listinfo/virtualization
> > > > 
> > 
> > 

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization