On 08/10/2016 12:35 AM, Andrew Donnellan wrote:
if (phb->controller_ops.release_device) phb->controller_ops.release_device(dev); + + if (unlikely(!phb)) + pr_warn("%s: PCI device %s has null PHB; refcount bug!", + __func__, dev_name(&dev->dev)); /* WARN_ON ahead */This should probably go before trying to dereference phb->controller_ops above?
You're right; I misplaced this check; will fix that. Just a bit of explanation/history: While trying to understand why I didn't hit that null dereference when I initially hit the WARN_ON (the reason for the 'small change' in the commit description), I found that back then I checked for 'pci_dev->sysdata' directly (not 'phb' -- early stages of the patch). The former indeed was null, as it didn't inherit 'pci_bus->sysdata' on pseries, but the code uses 'phb = dev->bus->sysdata' (and this was not null as pci_bus->sysdata was actually set). -- Mauricio Faria de Oliveira IBM Linux Technology Center
