Re: [Qemu-devel] [PATCHv2 0/2] pci: allow PCI bus slots to be marked as reserved

[Mark Cave-Ayland]

On 16/07/17 08:28, Marcel Apfelbaum wrote:

>>> As in prev version, other than the minor comment
>>> on replacing "if (...) return true; else return false"
>>> with the actual value, I am OK with it.
>>
>> Okay great! So change pci_bus_devfn_available() to something like this?
>>
>> static bool pci_bus_devfn_available(PCIBus *bus, int devfn)
>> {
>>      return !(bus->devices[devfn]);
>> }
>>
> 
> Yes, thanks.

Great, I'll send a v3 with the above modification shortly.

>>> I believe Michael asked to see the series using this feature,
>>> can you add a link to it, or post it with the dependency on this one?
>>
>> Sure, I posted the link earlier in the week:
>> https://lists.nongnu.org/archive/html/qemu-devel/2017-07/msg03045.html.
>> Or more specifically:
>>
>> https://lists.nongnu.org/archive/html/qemu-devel/2017-07/msg03041.html
>>
> 
> Hi Michael,
> Can you please have a look to the above patch using
> the 'reserved slots' patch?

I should add that if this patch isn't suitable for freeze then it's
likely I'll drop the sun4u patchset for 2.10 since otherwise using
-device places devices on the PCI root bus.

For existing users this means that the machine will start up with their
existing command lines but any extra devices added on the command line
via -device, e.g. virtio will be silently broken because the interrupt
routing is incorrect.

At least with this patch the user will get an error, and thus it's easy
to update the wiki to explain that moving forward it is necessary to add
"bus=pciB" for extra devices to function correctly.


ATB,

Mark.