[PATCH] PCI: endpoint: Fix WARN() when an endpoint driver is removed

[PATCH] PCI: endpoint: Fix WARN() when an endpoint driver is removed

[Yoshihiro Shimoda]

Add pci_epc_nop_release() for epc->dev.release. Otherwise,
WARN() happened when a PCIe endpoint driver is removed.

 Device 'e65d0000.pcie-ep' does not have a release() function, it is broken and must be fixed. See Documentation/core-api/kobject.rst.
 WARNING: CPU: 0 PID: 139 at drivers/base/core.c:2232 device_release+0x78/0x8c

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
---
 drivers/pci/endpoint/pci-epc-core.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/pci/endpoint/pci-epc-core.c b/drivers/pci/endpoint/pci-epc-core.c
index 3bc9273d0a08..cb533821b072 100644
--- a/drivers/pci/endpoint/pci-epc-core.c
+++ b/drivers/pci/endpoint/pci-epc-core.c
@@ -746,6 +746,10 @@ void devm_pci_epc_destroy(struct device *dev, struct pci_epc *epc)
 }
 EXPORT_SYMBOL_GPL(devm_pci_epc_destroy);
 
+static void pci_epc_nop_release(struct device *dev)
+{
+}
+
 /**
  * __pci_epc_create() - create a new endpoint controller (EPC) device
  * @dev: device that is creating the new EPC
@@ -779,6 +783,7 @@ __pci_epc_create(struct device *dev, const struct pci_epc_ops *ops,
 	device_initialize(&epc->dev);
 	epc->dev.class = pci_epc_class;
 	epc->dev.parent = dev;
+	epc->dev.release = pci_epc_nop_release;
 	epc->ops = ops;
 
 	ret = dev_set_name(&epc->dev, "%s", dev_name(dev));
-- 
2.25.1

Re: [PATCH] PCI: endpoint: Fix WARN() when an endpoint driver is removed

[Manivannan Sadhasivam]

On Tue, Jun 21, 2022 at 09:11:47PM +0900, Yoshihiro Shimoda wrote:
> Add pci_epc_nop_release() for epc->dev.release. Otherwise,
> WARN() happened when a PCIe endpoint driver is removed.
> 
>  Device 'e65d0000.pcie-ep' does not have a release() function, it is broken and must be fixed. See Documentation/core-api/kobject.rst.
>  WARNING: CPU: 0 PID: 139 at drivers/base/core.c:2232 device_release+0x78/0x8c
> 

Hmm... So the core is correctly throwing the warning here but just using a NOP
release function is not the right solution IMO.

Currently, pci_epc_destroy() is handling all resource release including freeing
the memory for epc. But I think the "kfree(epc)" should be moved to this
release callback. Reason is quite obvious. Until the release() callback gets
called, there would be a "epc" device instance floating around and we should not
free it until then.

Thanks,
Mani

> Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
> ---
>  drivers/pci/endpoint/pci-epc-core.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/pci/endpoint/pci-epc-core.c b/drivers/pci/endpoint/pci-epc-core.c
> index 3bc9273d0a08..cb533821b072 100644
> --- a/drivers/pci/endpoint/pci-epc-core.c
> +++ b/drivers/pci/endpoint/pci-epc-core.c
> @@ -746,6 +746,10 @@ void devm_pci_epc_destroy(struct device *dev, struct pci_epc *epc)
>  }
>  EXPORT_SYMBOL_GPL(devm_pci_epc_destroy);
>  
> +static void pci_epc_nop_release(struct device *dev)
> +{
> +}
> +
>  /**
>   * __pci_epc_create() - create a new endpoint controller (EPC) device
>   * @dev: device that is creating the new EPC
> @@ -779,6 +783,7 @@ __pci_epc_create(struct device *dev, const struct pci_epc_ops *ops,
>  	device_initialize(&epc->dev);
>  	epc->dev.class = pci_epc_class;
>  	epc->dev.parent = dev;
> +	epc->dev.release = pci_epc_nop_release;
>  	epc->ops = ops;
>  
>  	ret = dev_set_name(&epc->dev, "%s", dev_name(dev));
> -- 
> 2.25.1
> 

-- 
மணிவண்ணன் சதாசிவம்

RE: [PATCH] PCI: endpoint: Fix WARN() when an endpoint driver is removed

[Yoshihiro Shimoda]

Hi Manivannan,

> From: Manivannan Sadhasivam, Sent: Tuesday, June 21, 2022 11:21 PM
> 
> On Tue, Jun 21, 2022 at 09:11:47PM +0900, Yoshihiro Shimoda wrote:
> > Add pci_epc_nop_release() for epc->dev.release. Otherwise,
> > WARN() happened when a PCIe endpoint driver is removed.
> >
> >  Device 'e65d0000.pcie-ep' does not have a release() function, it is broken and must be fixed. See
> Documentation/core-api/kobject.rst.
> >  WARNING: CPU: 0 PID: 139 at drivers/base/core.c:2232 device_release+0x78/0x8c
> >
> 
> Hmm... So the core is correctly throwing the warning here but just using a NOP
> release function is not the right solution IMO.
> 
> Currently, pci_epc_destroy() is handling all resource release including freeing
> the memory for epc. But I think the "kfree(epc)" should be moved to this
> release callback. Reason is quite obvious. Until the release() callback gets
> called, there would be a "epc" device instance floating around and we should not
> free it until then.

I understood it. Especially, we enable CONFIG_DEBUG_KOBJECT_RELEASE=y, the release()
callback will be called after pci_epc_destroy() exited. So, I'll move "kfree(epc)" to
the release function.

Best regards,
Yoshihiro Shimoda

> Thanks,
> Mani
> 
> > Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
> > ---
> >  drivers/pci/endpoint/pci-epc-core.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/pci/endpoint/pci-epc-core.c b/drivers/pci/endpoint/pci-epc-core.c
> > index 3bc9273d0a08..cb533821b072 100644
> > --- a/drivers/pci/endpoint/pci-epc-core.c
> > +++ b/drivers/pci/endpoint/pci-epc-core.c
> > @@ -746,6 +746,10 @@ void devm_pci_epc_destroy(struct device *dev, struct pci_epc *epc)
> >  }
> >  EXPORT_SYMBOL_GPL(devm_pci_epc_destroy);
> >
> > +static void pci_epc_nop_release(struct device *dev)
> > +{
> > +}
> > +
> >  /**
> >   * __pci_epc_create() - create a new endpoint controller (EPC) device
> >   * @dev: device that is creating the new EPC
> > @@ -779,6 +783,7 @@ __pci_epc_create(struct device *dev, const struct pci_epc_ops *ops,
> >  	device_initialize(&epc->dev);
> >  	epc->dev.class = pci_epc_class;
> >  	epc->dev.parent = dev;
> > +	epc->dev.release = pci_epc_nop_release;
> >  	epc->ops = ops;
> >
> >  	ret = dev_set_name(&epc->dev, "%s", dev_name(dev));
> > --
> > 2.25.1
> >
> 
> --
> மணிவண்ணன் சதாசிவம்