Re: [drivers/pci] Possible memleak in pci_bus_set_aer_ops

[Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>]

On 3/13/24 7:40 PM, Kuppuswamy Sathyanarayanan wrote:
> On 3/13/24 7:20 PM, Zijie Zhao wrote:
>> Dear PCI Developers,
>>
>> We are curious whether the function `pci_bus_set_aer_ops` might have a memory leak.
>>
>> The function is https://elixir.bootlin.com/linux/v6.8/source/drivers/pci/pcie/aer_inject.c#L297
>> and the relevant code is
>> ```
>> static int pci_bus_set_aer_ops(struct pci_bus *bus)
>> {
>> 	struct pci_ops *ops;
>> 	struct pci_bus_ops *bus_ops;
>> 	unsigned long flags;
>>
>> 	bus_ops = kmalloc(sizeof(*bus_ops), GFP_KERNEL);
>> 	if (!bus_ops)
>> 		return -ENOMEM;
>> 	ops = pci_bus_set_ops(bus, &aer_inj_pci_ops);
>> 	spin_lock_irqsave(&inject_lock, flags);
>> 	if (ops == &aer_inj_pci_ops)
>> 		goto out;
>> 	pci_bus_ops_init(bus_ops, bus, ops);
>> 	list_add(&bus_ops->list, &pci_bus_ops_list);
>> 	bus_ops = NULL;
>> out:
>> 	spin_unlock_irqrestore(&inject_lock, flags);
>> 	kfree(bus_ops);
>> 	return 0;
>> }
>> ```
>>
>> Here if the goto statement does not jump to `out`, the `bus_ops` will be assigned with `NULL` and then `kfree(bus_ops)` will not free the allocated memory.
>>
>> Please kindly correct us if we missed any key information. Looking forward to your response!
> I think it is a valid issue that needs to be fixed. If you would like, please send a patch to fix it.

Sorry, I misread it. I think it is not a issue. For a valid case, the bus_ops is
added to pci_bus_ops_list, which is freed in module exit function. Ignore
my previous comments.

>
>> Best,
>> Zijie
>>
-- 
Sathyanarayanan Kuppuswamy
Linux Kernel Developer