Re: [PATCH 02/13] PCI: prevent duplicate slot names

From: Kenji Kaneshige
Date: Wed Sep 10 2008 - 22:45:14 EST

Hi Alex-san,

Thank you very much for your continuous effort.
I have one comment about the new design. Please see below.

>> I found another problem in pci_hp_register() that would be more complex
to fix than above-mentioned problem. Here is a pci_hp_register() with
all of your patch applied.

int pci_hp_register(struct hotplug_slot *slot, struct pci_bus *bus, int slot_nr,
const char *name)

* No problems if we call this interface from both ACPI_PCI_SLOT
* driver and call it here again. If we've already created the
* pci_slot, the interface will simply bump the refcount.
pci_slot = pci_create_slot(bus, slot_nr, name);
if (IS_ERR(pci_slot))
return PTR_ERR(pci_slot);

if (pci_slot->hotplug) {
dbg("%s: already claimed\n", __func__);
return -EBUSY;

slot->pci_slot = pci_slot;
pci_slot->hotplug = slot;

* Allow pcihp drivers to override the ACPI_PCI_SLOT name.
if (strcmp(kobject_name(&pci_slot->kobj), name)) {
result = kobject_rename(&pci_slot->kobj, name);
if (result) {
return result;


If name duplication was detected in pci_create_slot(), it renames the
slot name like 'N-1' and return successfully. Even though slot's kobject
name was registered as 'N-1', 'name' array still have 'N' at this point.
So the following 'if' statement becomes true unexpectedly.

* Allow pcihp drivers to override the ACPI_PCI_SLOT name.
if (strcmp(kobject_name(&pci_slot->kobj), name)) {

Then pci_hp_register() attempt to rename the slot name with 'N' again
by calling kobject_rename(), but it fails because there already exists
kobject with name 'N'. As a result, pci_hp_register() will fail.

Yes, you are right, that is a problem.

I've taken the following approach:

- the above code is providing a mechanism to allow a
_hotplug_ driver to override a _detection_ driver slot

- in other words, we only have to worry about the case
when a _detection_ driver was loaded before a _hotplug_

- we can ignore the case where another _hotplug_ driver
was loaded first, because we'll already return -EBUSY.

- so, to figure out if a _detection_ driver has already
been loaded, we check to see if the pci_dev already has
a valid pci_slot pointer.

We need to take into account that the hotplug slot can be empty.
In this case, we cannot do this check because pci_dev doesn't
exist, I think.

- if yes, then we later check to see if the existing slot
name matches the requested slot name from the hotplug

- if the hotplug driver is requesting a different name,
then we use a new interface, pci_rename_slot() which
will safely attempt to rename the slot without name

I'll send out the patch set soon, it would be great if you could
test it for me, since I don't have systems with duplicate slot


I also don't have the system with duplicate slot names. So I use
the debug patch that emulates this kind of system for testing.

Kenji Kaneshige

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at