[PATCH 2/2] fix-pcf50633-only-do-platform-callback-once-per-event.patch

[Andy Green]

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

Somebody in the thread at some point said:

|         /* clear batfull */
|         reg_set_bit_mask(pcf, PCF50633_REG_MBCC1,
|                                 PCF50633_MBCC1_AUTORES,
|                                 0);
|         reg_set_bit_mask(pcf, PCF50633_REG_MBCC1,
|                                 PCF50633_MBCC1_RESUME,
|                                 PCF50633_MBCC1_RESUME);
|         reg_set_bit_mask(pcf, PCF50633_REG_MBCC1,
|                                 PCF50633_MBCC1_AUTORES,
|                                 PCF50633_MBCC1_AUTORES);
|
| From the datasheet I know that resume has no effect when autores=1. Is
that a
| trick to have an effect? It looks from the reports that charging stops
and
| doesn't restart that we didn't have autoresume on (wildly guessing
without
| knowing the MBC at all...)?

Yes, it clears autores just to use resume, then unconditionally sticks
it back on.  From the dumps I'm saw chgena=0 during the bad time, but
autores=1, so it should have retried.  I would have guessed it can be
USB current limit is set for SUSPEND mode so it doesn't allowing
anything in from USB, but the reporter also showed "charger_type" /sys
was correctly showing Charger 1A / mode 1A, which is correct and that
comes direct from MBCC7.

I'll stick these last two patches on stable and unless a new idea comes
wait for feedback from people with the problem on tomorrow's kernel to
see if it helped.

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

iEYEARECAAYFAkiGTiAACgkQOjLpvpq7dMotGQCeJuuIiH6iHSpemfEDoYScPpys
W7IAni0QK5c/TxSlOssU5+sRnr0CDZeb
=fsCg
-----END PGP SIGNATURE-----