Hi,
On 08-01-18 13:47, Johan Hovold wrote:
On Mon, Jan 08, 2018 at 01:34:11PM +0100, Johan Hovold wrote:
On Mon, Dec 25, 2017 at 09:50:45PM +0100, Hans de Goede wrote:
ACPI enumerated serdev-controllers do not have an ACPI companion, the ACPI
companion belongs to the serdev-device child of the serdev-controller, not
to the controller itself. This was causing serdev_uevent to always return
-ENODEV when called on a serdev-controller leading to errors like these:
kernel: serial serial0: uevent: failed to send synthetic uevent
being logged. This commit modifies serdev_uevent to directly return 0
when called on an ACPI enumerated serdev-controller fixing this.
Note: I do not think that setting a modalias on a devicetree enumerated
serdev-controller makes sense either. So perhaps the !dev->of_node part of
the check can be dropped too, but I'm not entirely sure that doing this
on devicetree too is correct.
Yes, generating modaliases for controllers doesn't make any sense as
they are not bound to any drivers.
I've prepared a patch, on top of yours, which moves the modalias
handling over to the device-type (rather than bus) code instead. Note
that this includes the modalias attributes, which we shouldn't be
creating for controllers either.
Signed-off-by: Hans de Goede <hdegoede@xxxxxxxxxx>
Acked-by: Johan Hovold <johan@xxxxxxxxxx>
Thank you for the review.
So this can go in as a minimal fix for the ACPI uevents errors, and then
the modalias generation can be dropped for controllers entirely by my
series after.
Sounds good to me. Note directly going with the complete fix works for
me to if people prefer that.
Regards,
Hans
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
