On Thu, Mar 17, 2011 at 11:42:19AM -0700, David Daney wrote:

On 03/17/2011 11:25 AM, Grant Likely wrote:

quoted

On Thu, Mar 17, 2011 at 12:18:49PM +0000, Alan Cox wrote:

quoted

On Wed, 16 Mar 2011 18:26:06 -0700
David Daney[off-list ref]  wrote:

quoted

The 8250 driver is a bit weird in that in addition to supporting
platform devices, extra devices can be added by calling
serial8250_register_port().

The problem is that if we call serial8250_register_port() before the
driver is initialized Bad Things happen (we dereference NULL
pointers).

There doesn't seem to be a general way to know if a driver has been
initialized

I've had a bigger dig into this. I think the correct answer is probably
"always go via platform devices or similar". That *is* the notifier in
the kernel of today. serial8250_register_port ultimately should I think
ultimatly become an internal helper.

+1

Depending on serial8250_register_port() definitely the wrong thing to
do for platform support code.  It would be better to figure out how to
get the dt bits you need into 8250.c or of_serial.c.

IMHO, of_serial.c is no better than my board/chip specific code that
calls serial8250_register_port().

Really what would be ideal would be a hook to add a
dev.platform_data pointer to the appropriate struct
plat_serial8250_port when the platform device is created.  It is
possible that the platform bus notifiers could be used for this.  We
would also want to have a way to add an of_device_id to those
recognized by 8250.c

If we did that, serial8250_probe() would automatically do the right thing.

Take a look at the way arch/powerpc/platforms/512x/pdm360ng.c uses a
notifier for amending a platform_device with additional data..

g.