On Wed, May 27, 2020 at 06:40:05PM -0700, Ranjani Sridharan wrote:
> On Wed, 2020-05-27 at 21:12 -0300, Jason Gunthorpe wrote:
> > On Wed, May 27, 2020 at 01:18:35PM -0700, Ranjani Sridharan wrote:
> > > On Wed, 2020-05-20 at 09:56 -0300, Jason Gunthorpe wrote:
> > > > On Wed, May 20, 2020 at 12:02:26AM -0700, Jeff Kirsher wrote:
> > > > > +static const struct virtbus_dev_id sof_ipc_virtbus_id_table[]
> > > > > = {
> > > > > + {"sof-ipc-test"},
> > > > > + {},
> > > > > +};
> > > > > +
> > > > > +static struct sof_client_drv sof_ipc_test_client_drv = {
> > > > > + .name = "sof-ipc-test-client-drv",
> > > > > + .type = SOF_CLIENT_IPC,
> > > > > + .virtbus_drv = {
> > > > > + .driver = {
> > > > > + .name = "sof-ipc-test-virtbus-drv",
> > > > > + },
> > > > > + .id_table = sof_ipc_virtbus_id_table,
> > > > > + .probe = sof_ipc_test_probe,
> > > > > + .remove = sof_ipc_test_remove,
> > > > > + .shutdown = sof_ipc_test_shutdown,
> > > > > + },
> > > > > +};
> > > > > +
> > > > > +module_sof_client_driver(sof_ipc_test_client_drv);
> > > > > +
> > > > > +MODULE_DESCRIPTION("SOF IPC Test Client Driver");
> > > > > +MODULE_LICENSE("GPL v2");
> > > > > +MODULE_IMPORT_NS(SND_SOC_SOF_CLIENT);
> > > > > +MODULE_ALIAS("virtbus:sof-ipc-test");
> > > >
> > > > Usually the MODULE_ALIAS happens automatically rhough the struct
> > > > virtbus_dev_id - is something missing in the enabling patches?
> > >
> > > Hi Jason,
> > >
> > > Without the MODULE_ALIAS, the driver never probes when the virtual
> > > bus
> > > device is registered. The MODULE_ALIAS is not different from the
> > > ones
> > > we typically have in the platform drivers. Could you please give me
> > > some pointers on what you think might be missing?
> >
> > Look at how the stuff in include/linux/mod_devicetable.h works and do
> > the same for virtbus
> It looks like include/linux/mod_devicetable.h has everything needed for
> virtbus already.
> >
> > Looks like you push a MODALIAS= uevent when creating the device and
> > the generic machinery does the rest based on the matching table, once
> > mod_devicetable.h and related is updated. But it has been a long time
> > since I looked at this..
>
> This is also done with uevent callback in the bus_type definition for
> the virtual_bus.
>
> Is your expectation that with the above changes, we should not be
> needing the MODULE_ALIAS() in the driver?
Yes, it should not be needed if you did everything properly in
mod_devicetable.h
thanks,
greg k-h
