Re: [Qemu-devel] [Qemu-ppc][PATCH v5 1/4] Add usb option in machine options

[Alexander Graf]

On 08.07.2012, at 16:46, Li Zhang <address@hidden> wrote:

> On Sat, Jul 7, 2012 at 12:03 AM, Alexander Graf <address@hidden> wrote:
>> 
>> On 06.07.2012, at 17:58, Andreas Färber wrote:
>> 
>>> Am 06.07.2012 15:43, schrieb Alexander Graf:
>>>> 
>>>> On 02.07.2012, at 07:25, address@hidden wrote:
>>>> 
>>>>> From: Li Zhang <address@hidden>
>>>>> 
>>>>> pSeries machine needs to enable USB to add a USB
>>>>> keyboard or USB mouse. -usb option won't be used in
>>>>> the future, and machine options are a better way to
>>>>> enable USB.
>>>>> 
>>>>> So this patch is to add USB option to machine options
>>>>> (-machine type=pseries,usb=on/off) to enable/disable
>>>>> USB controller. And machines will get the machine option
>>>>> and create a USB controller if USB is on.
>>>>> 
>>>>> By the way, USB is on by default on pSeries machine.
>>>>> So USB controller should be turned off explicitly through
>>>>> the command line for "-nodefault" case as the following:
>>>>> -machine type=pseries,usb=off.
>>>>> 
>>>>> Signed-off-by: Li Zhang <address@hidden>
>>>>> Reviewed-by: Andreas Färber <address@hidden>
>>>>> ---
>>>>> hw/spapr.c    |   11 +++++++++++
>>>>> qemu-config.c |    4 ++++
>>>>> 2 files changed, 15 insertions(+)
>>>>> 
>>>>> diff --git a/hw/spapr.c b/hw/spapr.c
>>>>> index 81c9343..973de1b 100644
>>>>> --- a/hw/spapr.c
>>>>> +++ b/hw/spapr.c
>>> [...]
>>>>> @@ -710,6 +712,15 @@ static void ppc_spapr_init(ram_addr_t ram_size,
>>>>>       spapr_vscsi_create(spapr->vio_bus);
>>>>>   }
>>>>> 
>>>>> +    machine_opts = qemu_opts_find(qemu_find_opts("machine"), 0);
>>>>> +    if (machine_opts) {
>>>>> +        add_usb = qemu_opt_get_bool(machine_opts, "usb", true);
>>>>> +    }
>>>>> +
>>>>> +    if (add_usb) {
>>>>> +        pci_create_simple(QLIST_FIRST(&spapr->phbs)->host_state.bus,
>>>>> +                          -1, "pci-ohci");
>>>>> +    }
>>>> 
>>>> Didn't I ask you to extract this out to generic code? I don't want to have 
>>>> a "usb" machine opt that is only available for -M pseries.
>>> 
>>> Isn't the mapping from "usb=on" to device-level actions
>>> machine-specific? We have ohci, uhci, ehci, xhci to choose from. And the
>>> bus to place it on is machine-specific, too.
>>> 
>>> So did you rather mean adding usb= awareness to more machines? If we
>>> generalize usb=on to usb=none/ohci/... plus some usbbus= we get -device.
>> 
>> I was thinking of replacing usb_enabled with
>> 
>> static inline int usb_enabled(bool default) {
>>    machine_opts = qemu_opts_find(qemu_find_opts("machine"), 0);
>>    if (machine_opts) {
>>        return qemu_opt_get_bool(machine_opts, "usb", default);
>>    }
>>    return default;
>> }
>> 
>> Whereas -usb in vl.c would set machine.usb = on. Then all users of 
>> usb_enabled would automatically be converted to the new syntax and we could 
>> even use the same logic in spapr.c.
>> 
> 
> I see, you mean global varible usb_enabled is still used, and when
> -usb is passed to qemu command line,
> use_enabled will be assigned by the above function and set machine.usb=on.
> 
> So all the places where usb_enabled is used still can work correctly.

I mean convert usb_enabled into a function that checks machine.usb :). The only 
global state should live in machine opts.


Alex