On 25/01/2016 15:01, Javier Martinez Canillas wrote:

Hello John,

On Mon, Jan 25, 2016 at 10:46 AM, John Crispin [off-list ref] wrote:

quoted

On 25/01/2016 14:25, Javier Martinez Canillas wrote:

quoted

On Mon, Jan 25, 2016 at 10:19 AM, Javier Martinez Canillas
[off-list ref] wrote:

[snip]

quoted

In fact, the kernel is currently not matching the compatible, it is
only matching because you provided a .of_compatible is provided in the
mfd_cell.

Sorry my English was a bit off in this paragraph...

I tried to say that OF does not traverse MFD sub-devices and lookups a
device driver that matches the compatible automatically since a MFD
device is not a bus. Currently it is only trying to match a compatible
string because the mfd_cell has a .of_compatible set so an of_node is
assigned on mfd_add_device().

But it is failing to match because no OF device table is provided and
the platform bus match callback is falling back to the driver .name to
match so the compatible is not really used as Mark said.

Best regards,
Javier

Hi,

just so i am sure to have understood properly. i just need to drop the
compatible string from [1/2] and resend. if this is the case i will fix
the mt6397 binding doc while at it.

And you will also need to remove the .of_compatible =
"mediatek,mt6323-regulator" from patch "[PATCH V2 4/4] mfd: mediatek:
add MT6323 support to MT6397 driver" since otherwise an
MODALIAS=of:foo will be reported instead of an MODALIAS=platform:foo

But if I were you, I would keep the MFD driver and DT binding as they
are and provide a .id_table and .of_match_table to the mt6323
regulator platform driver.

I'll write patches for the mt6397 regulator driver adding those tables
since it has the same issue and you can see what I mean.

quoted

        John

Best regards,
Javier

Hi Javier,

fine i'll do that then. thanks for the elaborate explanation.

	John