Re: [PATCH -next] PCI: dra7xx: Add missing of_node_put() in dra7xx_pcie_init_irq_domain()

Re: [PATCH -next] PCI: dra7xx: Add missing of_node_put() in dra7xx_pcie_init_irq_domain()

[Bjorn Helgaas]

[+cc Shawn, Heiko, Michal, Soren, Bharat, Rob H, Frank, devicetree@vger]

On Sat, Nov 12, 2016 at 12:39:01PM +0530, Kishon Vijay Abraham I wrote:
> Hi,
> 
> On Saturday 12 November 2016 03:08 AM, Bjorn Helgaas wrote:
> > On Mon, Oct 17, 2016 at 02:54:37PM +0000, Wei Yongjun wrote:
> >> From: Wei Yongjun <weiyongjun1-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
> >>
> >> This node pointer is returned by of_get_next_child() with refcount
> >> incremented in this function. of_node_put() on it before exitting
> >> this function on error.
> >>
> >> This is detected by Coccinelle semantic patch.
> >>
> >> Signed-off-by: Wei Yongjun <weiyongjun1-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
> > 
> > Kishon, this looks correct to me, so I applied it to pci/host-dra7xx for
> > v4.10.  Let me know if you have any issue with it.
> > 
> >> ---
> >>  drivers/pci/host/pci-dra7xx.c | 1 +
> >>  1 file changed, 1 insertion(+)
> >>
> >> diff --git a/drivers/pci/host/pci-dra7xx.c b/drivers/pci/host/pci-dra7xx.c
> >> index 9595fad..79297e9 100644
> >> --- a/drivers/pci/host/pci-dra7xx.c
> >> +++ b/drivers/pci/host/pci-dra7xx.c
> >> @@ -177,6 +177,7 @@ static int dra7xx_pcie_init_irq_domain(struct pcie_port *pp)
> >>  					       &intx_domain_ops, pp);
> >>  	if (!pp->irq_domain) {
> >>  		dev_err(dev, "Failed to get a INTx IRQ domain\n");
> >> +		of_node_put(pcie_intc_node);
> 
> I think of_node_put should be used for both the error case and non-error case.

Hmm, OK.  I don't know what the rules are.  Certainly if we made these
drivers modular, I don't think we'd want to leak these references
every time we unload/reload the module.  Should we do the put
immediately here, or in the module remove path, or ...?

Adding other driver and DT folks for comment.

I dropped these patches for now (dra7xx, rockchip, xilinx-nwl,
xilinx).
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH -next] PCI: dra7xx: Add missing of_node_put() in dra7xx_pcie_init_irq_domain()

[Heiko Stuebner]

Am Montag, 14. November 2016, 15:19:28 CET schrieb Bjorn Helgaas:
> [+cc Shawn, Heiko, Michal, Soren, Bharat, Rob H, Frank, devicetree@vger]
> 
> On Sat, Nov 12, 2016 at 12:39:01PM +0530, Kishon Vijay Abraham I wrote:
> > Hi,
> > 
> > On Saturday 12 November 2016 03:08 AM, Bjorn Helgaas wrote:
> > > On Mon, Oct 17, 2016 at 02:54:37PM +0000, Wei Yongjun wrote:
> > >> From: Wei Yongjun <weiyongjun1@huawei.com>
> > >> 
> > >> This node pointer is returned by of_get_next_child() with refcount
> > >> incremented in this function. of_node_put() on it before exitting
> > >> this function on error.
> > >> 
> > >> This is detected by Coccinelle semantic patch.
> > >> 
> > >> Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
> > > 
> > > Kishon, this looks correct to me, so I applied it to pci/host-dra7xx for
> > > v4.10.  Let me know if you have any issue with it.
> > > 
> > >> ---
> > >> 
> > >>  drivers/pci/host/pci-dra7xx.c | 1 +
> > >>  1 file changed, 1 insertion(+)
> > >> 
> > >> diff --git a/drivers/pci/host/pci-dra7xx.c
> > >> b/drivers/pci/host/pci-dra7xx.c
> > >> index 9595fad..79297e9 100644
> > >> --- a/drivers/pci/host/pci-dra7xx.c
> > >> +++ b/drivers/pci/host/pci-dra7xx.c
> > >> @@ -177,6 +177,7 @@ static int dra7xx_pcie_init_irq_domain(struct
> > >> pcie_port *pp)> >> 
> > >>  					       &intx_domain_ops, pp);
> > >>  	
> > >>  	if (!pp->irq_domain) {
> > >>  	
> > >>  		dev_err(dev, "Failed to get a INTx IRQ domain\n");
> > >> 
> > >> +		of_node_put(pcie_intc_node);
> > 
> > I think of_node_put should be used for both the error case and non-error
> > case.
> Hmm, OK.  I don't know what the rules are.  Certainly if we made these
> drivers modular, I don't think we'd want to leak these references
> every time we unload/reload the module.  Should we do the put
> immediately here, or in the module remove path, or ...?
> 
> Adding other driver and DT folks for comment.

I think the function above (dra7xx_pcie_init_irq_domain) can do the 
of_node_put at its end in all cases as suggested by Kishon.

of_get_next_child() will increment the refcount
	irq_domain_add_linear()
		__irq_domain_add() also increments the refcount


irq_domain_remove() will decrement the refcount

So it should be safe to decrement the refcount in 
dra7xx_pcie_init_irq_domain() in all cases as the irq-domain internals will 
always keep it above 1 as long as the node is used.


Heiko