power management regressing?

[Andy Green]

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

Somebody in the thread at some point said:
| Andy Green wrote:
|> Somebody in the thread at some point said:
|> | It seems power management is much worse this week than last.
|> |
|> | 1) Audio is not returned to the proper state on resume.
|> | 2) GSM radio serial link not returned to proper state.
|> | 3) Screen flashes occasionally as though it is being awakened at times
|> | then goes back to black.
|> | 4) audio pops.
|> | 5) Sometimes doesn't wake up at all and I have to pull USB and battery
|> | to recover. No other way works.
|> |
|> | These issues were not present about a wee ago. I have made sure I do
|> not
|> | have merge issues with the andy tree.
|>
|> Not really seeing these, although I don't particularly test the GSM
|> side.  If what you propose is exactly as you say it, you are really
|> saying that these four patches
|>
|> 4 days ago
|>     Andy Green
|>     fix-reduce-wake-reasons-in-pcf50633.patch
|>
|> 4 days ago
|>     Andy Green
|>     add-resume-reason-sysfs.patch
|>
|> 7 days ago
|>     Andy Green
|>     fix-motion-sensor-corruption.patch
|>
|> 7 days ago
|>     Holger Freyther
|>     Fixup hang on resume caused by the s3c2410 touch screen ...
|>
|> made all that trouble?  It seems more likely some of these are longer
|> term issues other people see, eg, pulseaudio bug is floating around for
|> audio issue, hang in pcf50633 when suspended for a while, etc.
|>
|> What config are you using?
|>
|> -Andy
| Well one of the problems is that the LED is being seriously over-tweaked
| on resume. There should be no need to turn it on, then off, then on yet
| again if you are a gta02. Plus there is a timing issue where the
| gta02_jbt6k74_resuming is expected to be called after pcf50633_resume.
| This does not appear to be the case if you awake from something like a
| ring.
|
| The attached changes make my screen reliably turn back on.

I will study your patch tomorrow, but can I ask again about which config
you are using?  Suspend / resume issues are exquisitely sensitive to the
config.

Why should stuttering the backlight make a problem with backlight not
returning in the end, if we always end with turning it on?  Seems
something else like the variable ordering of pcf50633 resume is to
blame, maybe that is implicated in Holger's delayed resume issue.

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

iEYEARECAAYFAkg5JGQACgkQOjLpvpq7dMqZqgCfdlF40F6RmziiXbWvw3tilKXh
BvwAn2mHDSQUDSNuzauIMhd5oUfhOa5L
=w9Tv
-----END PGP SIGNATURE-----