On Wed, Mar 20, 2013 at 12:46 PM, Arnd Bergmann [off-list ref] wrote:

On Wednesday 20 March 2013, Fabio Porcedda wrote:

quoted

On Wed, Mar 20, 2013 at 11:20 AM, Arnd Bergmann [off-list ref] wrote:

quoted

On Wednesday 20 March 2013, Fabio Porcedda wrote:

quoted

I think we can check inside the  deferred_probe_work_func()
if the dev->probe function pointer is equal to platform_drv_probe_fail().

I think it's too late by then, because that would only warn if we try to probe
it again, but when platform_driver_probe() does not succeed immediately, it

Maybe you mean "does succeed immediately" ?

I mean in this code (simplified for the sake of discussion)

int __init_or_module platform_driver_probe(struct platform_driver *drv,
                int (*probe)(struct platform_device *))
{
        int retval, code;

        drv->probe = probe;
        retval = code = platform_driver_register(drv);

        drv->probe = NULL;
        if (code = 0 && list_empty(&drv->driver.p->klist_devices.k_list))
                retval = -ENODEV;
        drv->driver.probe = platform_drv_probe_fail;

        if (code != retval)
                platform_driver_unregister(drv);
        return retval;
}

we assume that all devices are bound to drivers during the call to
platform_driver_register, and if the device list is empty afterwards,
we unregister the driver and will never get to the deferred probing
stage.

Thanks for the explanation, I understand now that is not that simple.

I was hoping it was easier.

Regards
--
Fabio Porcedda