[gta04] Meeting minutes for GTA04 discussion 20080411

Carsten Haitzler (The Rasterman) raster at openmoko.org
Fri Apr 18 10:10:24 CEST 2008


On Fri, 18 Apr 2008 08:39:48 +0100 Andy Green <andy at openmoko.com> babbled:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Somebody in the thread at some point said:
> | On Fri, 18 Apr 2008 08:11:50 +0100 Andy Green <andy at openmoko.com> babbled:
> |
> |> We're talking about this MPU taking over the duty of scanning the
> |> touchscreen during this time and detecting a touch and coordinating wake
> |> from suspend, specifically if the MPU should start ramping up the
> |> backlight during the time the CPU is resuming.
> |
> | yeah - we need both i suspect. we need the pmu to be able to handle
> touhscreen
> | and bring up bakclight OR be able to turn that behavior off (it should be
> 
> Yes this seems to be a general theme that we are OK with to allow
> software decision about MPU actions or do it direct.  Then we are super
> flexible.
> 
> | software controlled in the kenrel). userspace needs to be able to tell the
> | kenrel:
> |
> | * shut down touchscreen and generate no interupts at all
> | * enable touchscreen (interrupts) (and wake up cpu when they happen)
> | * disable balckight turn-on on wakeup
> | * enable backlight turn-on on wakeup
> 
> OK.
> 
> | DEFAULT (imho) would be:
> | * enable touchscreen interrupts
> | * enable backlight on wakeup
> 
> OK, but there will be some period ~1s where the user sees white and / or
> black display from LCM as backlight ramps up but X did not give video
> back yet.  We can see what it looks like and decide.

if we leave this to x + usersapce, then x can turn on the bakclight ONCe it has
filled the framebuffer. in fact x itself can just not turn it on, but wait for
the "screensaver" app to turn it on after wakup from screenblank (ie
screensaver app will actually do the drawing and once it has drawn the screen
it can turn backlight on. since x is pipelined and 1 clients operations are
executed in sequence, the bl shouldnt go on until x has dumped all the draw
requests to the screen).

> | the rest then is userspace (xserver and then above that wm/desktop and
> apps)
> | making all the higher level decisions
> 
> Yes that is OK if it is understood that "decision" is limited to they
> are "leaving a will" about what they want to happen when they are in a coma.

yup. thats all we need. for non-intelligent userpace the kernel defaults will
mean things wake up on touchscreen touch and backlight is on so u see stuff.
when userspace is smart enough to handle this - it can ask the kernel to do
less/nothing and know it will deal with it. :)

> - -Andy
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
> 
> iEYEARECAAYFAkgIUD4ACgkQOjLpvpq7dMoKPQCeLuMxZ8BW5HHPNGSAM0u/Sie1
> r7EAoIzC16TA1YyvS7ixG6syD+jt15Fo
> =yw1V
> -----END PGP SIGNATURE-----


-- 
Carsten Haitzler (The Rasterman) <raster at openmoko.org>




More information about the openmoko-kernel mailing list