[PATCH 1/2] I2C: Convert PCF50606 to I2C device

[Andy Green]

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

Werner Almesberger wrote:
> Andy Green wrote:
>> But the Glamo is dead at that point,
> 
> So you can't even dump its registers ? Okay, that would limit the
> usefulness of such a debug interrupt.

Glamo won't complete any transaction, just stiffs the machine with nWAIT
down.  It just means --->

>> fundamentally a sign it was broken either by the suspend actions on it
>> or the resume attempt on it.

...for example that the PLL that runs the bus unit never came back up,
any bus unit transaction will just be insane then.

> Do you think much time passes between making whatever error causes
> the hang and the moment the chip hangs for real ? If not, the stack
> trace from the AUX-dump handler might reveal where it happened.

Sometimes yes, it panics in the resume code when I force nWAIT
deasserted.  Other times it survives the resume code fine and dies the
first time framebuffer code tries to flash the cursor.  The datasheet
mentions 4ms after reset before you can "access registers", dunno if
that is bound up in it and the fake hard resets.

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

iEYEARECAAYFAkjnd9gACgkQOjLpvpq7dMpIfQCeLhP3Bvrj5Qxc3HEXVCKzDhu/
DQkAn2XPfLJYzaDlzO9N0keJKWSsJ03k
=CYn8
-----END PGP SIGNATURE-----