Re: [RFC PATCH v5 7/8] PCI/PCIe: add "pci=nopciehp" to disable PCIe native hotplug

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Sat, Jan 19, 2013 at 09:56:24AM +0800, Yijing Wang wrote:
> 于 2013-01-19 1:35, Bjorn Helgaas 写道:
> > On Fri, Jan 18, 2013 at 9:07 AM, Jiang Liu <liuj97@gmail.com> wrote:
> >> If user specifies "pci=nopciehp" on kernel boot command line, OSPM
> >> won't claim PCIe native hotplug service from firmware and no PCIe
> >> port devices will be created for PCIe native hotplug service.
> > 
> > Why do we need this option?
> > 
> > If I understand correctly, there are machines where it *looks* like we
> > should use pciehp, but pciehp doesn't work because we don't get the
> > interrupts we expect.  On those machines, we have to use acpiphp
> > instead.  It seems like many Dell XPS laptops have this issue with
> > ExpressCard slots, e.g.,
> > https://bugzilla.kernel.org/show_bug.cgi?id=40802 .
> 
> What about use modprobe pciehp pciehp_poll_mode=1?
> If just cannot receive the interrupt, maybe this module parameter will fix it.

You can't add a new option that you now force hardware that was working
with a different module to now define and use.  It needs to be
"automatic".

> > If you want "pci=nopciehp" as a way for users to deal with this
> > problem by forcing the use of acpiphp, I object.  Windows manages to
> > make these slots work without having users do anything special, so we
> > should be able to do it, too.
> 
> In fact, pcie native hotplug may not be implemented perfectly,

Oh, we know that is true, it's the problem here.  Fixing BIOSes would be
nice, but we can't do that :(

greg k-h