[Qemu-devel] Re: [PATCH v6 07/12] pcie/hotplug: introduce pushing attention button command

[Michael S. Tsirkin]

On Thu, Oct 21, 2010 at 12:46:33PM +0900, Isaku Yamahata wrote:
> On Wed, Oct 20, 2010 at 12:00:11PM +0200, Michael S. Tsirkin wrote:
> > On Wed, Oct 20, 2010 at 05:18:56PM +0900, Isaku Yamahata wrote:
> > > glue pcie_push_attention_button command.
> > > 
> > > Signed-off-by: Isaku Yamahata <address@hidden>
> > 
> > So as a high level command, I think we need to
> > think about how to tie this into pci_add/pci_del.
> > Right?
> 
> Maybe or maybe not.
> I'm not sure it's a good idea to tie it to pci_add/pci_del
> in the first place.
> The specification says only that pushing the button is just to
> initiate the hot-plug operation. It means only a notification.
> The spec says nothing about the concrete action from OS when
> the button is pushed. It's up to OS.
> 
> For example.
> OS may start to probe the slot.
> OS may start to quiescence the device.
> OS is allowed to ignore the notification. 
> OS may propagate the notification to the management software,
> and it would pop up the dialog to the user for further operation.
> 

pci_add/pci_del really behave in the same way. There's no way to force
the OS to respond. We only use ACPI for now, but for express I expect
standard interfaces will work better long term.

> > As a low level command, this is not really useful unless
> > there is an event on LED status change and a way
> > to get info on LED status.
> > Right?
> 
> No. Guest OS can provide users those infos, and Linux does
> via sysfs. So there already is a way to know LED status.
> This is the reason why LED status change event stuff has low
> priority in my TODO list.


But that's in the guest.

-- 
MST