On Tue, May 25, 2021 at 4:18 PM Bjorn Helgaas <helgaas@xxxxxxxxxx> wrote:
>
> Capitalize "Add" in the subject.
>
> On Tue, Apr 27, 2021 at 01:51:39PM -0400, Jim Quinlan wrote:
> > The shutdown() call is similar to the remove() call except the former does
> > not need to invoke pci_{stop,remove}_root_bus(), and besides, errors occur
> > if it does.
>
> This doesn't explain why shutdown() is necessary. "errors occur"
> might be a hint, except that AFAICT, many similar drivers do invoke
> pci_stop_root_bus() and pci_remove_root_bus() (several of them while
> holding pci_lock_rescan_remove()), without implementing .shutdown().
>
> It is ... unfortunate that there's such a variety of implementations
> here. I don't believe these driver differences are all necessary
> consequences of hardware differences.
What's correct here?
This was on my radar and something we should get in front of. It's
only a matter of time until everyone is converting their drivers to
modules since that is now the Android WayTM. My plan here was to
register a devm hook to call pci_stop_root_bus() and
pci_remove_root_bus(). That way every driver doesn't have to implement
a remove() hook. Though maybe they all need to quiesce themselves
anyways.
Rob
