Best way to use pcf50633

[Andy Green]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

| As I understand mach-gta02.c and such similar board specific sources in
| the kernel know exactly about every piece of hardware that's wired to it
| and it wouldn't have to work with any other device, ever. So, there's
| nothing wrong in using the pcf50633 at a register level from within
| mach-gta02.c.
|
| Please point me to examples that are in favour or opposition with
| this idea. I really want to know how to deal with such devices from
| within platform specific code.

I'll let others give their answers to this actual question, but it puts
me in mind of another related issue that can be asked at the same time
- -- the device tree ordering stuff relies on callbacks added on
successful probe also serviced at mach-gta02.c.

I think these device creation callbacks, or some equivalent technique
used to trigger child device creation to enforce the relationship for
suspend-resume as well as raceless device creation is generically
useful, what should we do about it, implement it generically on good
return from probe in the normal kernel side?

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkj4YKEACgkQOjLpvpq7dMqj0ACfZpGaSo0bz/nDcQ+wczEfq0sd
QgYAnRTJj9945TnS8UPkmz3BY9hCjz5t
=zyIw
-----END PGP SIGNATURE-----