Re: [PATCH v5 13/16] cxl/pci: Add error handler for CXL PCIe Port RAS errors

[Gregory Price]

On Fri, Feb 07, 2025 at 01:23:06PM -0600, Bowman, Terry wrote:
> 
> 
> On 2/7/2025 2:01 AM, Gregory Price wrote:
> > On Tue, Jan 07, 2025 at 08:38:49AM -0600, Terry Bowman wrote:
> >> +static void __iomem *cxl_pci_port_ras(struct pci_dev *pdev)
> >> +{
> >> +	struct cxl_port *port;
> >> +
> >> +	if (!pdev)
> >> +		return NULL;
> >> +
> >> +	if ((pci_pcie_type(pdev) == PCI_EXP_TYPE_ROOT_PORT) ||
> >> +	    (pci_pcie_type(pdev) == PCI_EXP_TYPE_DOWNSTREAM)) {
> >> +		struct cxl_dport *dport;
> >> +		void __iomem *ras_base;
> >> +
> >> +		port = find_cxl_port(&pdev->dev, &dport);
> >> +		ras_base = dport ? dport->regs.ras : NULL;
> > I'm fairly certain dport can come back here uninitialized, you
> > probably want to put this inside the `if (port)` block and 
> > pre-initialize dport to NULL.
> Right, it can. NULL dport check here covers this, no?:
> 
> 		ras_base = dport ? dport->regs.ras : NULL;

dport can be garbage (whatever is on the stack) at this check because
nothing ever intializes it to NULL.

> 
> Terry
> 
> >> +		if (port)
> >> +			put_device(&port->dev);
> >> +		return ras_base;
> > You can probably even simplify this down to something like
> >
> > 		struct_cxl_dport *dport = NULL;
> >
> > 		port = find_cxl_port(&pdev->dev, &dport);
> > 		if (port)
> > 			put_device(&port->dev);
> >
> > 		return dport ? dport->regs.ras : NULL;
> >
> >
> > ~Gregory
>