Re: [PATCH v2] PCI: pciehp: Check link state before accessing device during removal

From: Bjorn Helgaas
Date: Thu Dec 11 2014 - 16:00:33 EST

On Thu, Dec 11, 2014 at 1:45 PM, Guenter Roeck <groeck@xxxxxxxxxxx> wrote:
> On Thu, Dec 11, 2014 at 01:26:47PM -0700, Bjorn Helgaas wrote:
>> On Thu, Dec 11, 2014 at 9:38 AM, Guenter Roeck <groeck@xxxxxxxxxxx> wrote:
>> > On Wed, Dec 10, 2014 at 05:26:30PM -0700, Bjorn Helgaas wrote:
>> >> On Thu, Nov 20, 2014 at 02:54:24PM -0800, Rajat Jain wrote:

>> >> If we do need it (and it looks like most or all hotplug drivers copied it),
>> >> isn't there still a race? Can't we have the following sequence?
>> >>
>> >> - pciehp_check_link_active() # returns true
>> >> - Link goes down
>> >> - pci_read_config_byte() # fails because link is down
>> >>
>> > I would guess so. Question is how to address it. Read the configuration byte
>> > first, then check if the link is down ? Check if link is still up after reading
>> > the configuration byte ? Add a note that there may be a potential race condition
>> > and do nothing until it is actually seen ?
>> I think we should just read PCI_BRIDGE_CONTROL and look for a 0xff
>> value. That's not a legal value for the register, so if we see it, it
>> should be pretty safe to assume the link is down or the device is not
>> present at all.
> Something like
> if (bctl != 0xff && (bctl & PCI_BRIDGE_CTL_VGA)) {
> in addition to Rajat's changes ?
> I think it would be good to keep the change Rajat proposed, ie to check
> the link state instead of presence. Question then is if you'd want a new
> revision of Rajat's patch or another patch on top of it with the bctl
> related change.

Why do we need the link state or the presence check? It seems like
those are sort of a 90% solution, and doing them provides the illusion
of value but without real value. If we think that checking for 0xff
is a 100% solution, we should rely on that and not bother with
anything else.

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