development tree for U-Boot: Changes to 'andy'

[git at git.openmoko.org]

Rebased ref, commits from common ancestor:
commit 53c9bd51b52f3b41d71fc2dd823e4ff73af01e20
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:14:15 2008 +0100

    deadbat-safe-voltage.patch
    
    This patch finally enables the last bit of Matt's "safe battery voltage"
    patch. Before the other changes in this series of patches, the
    wait_for_power logic didn't really work anyway, but now we need this
    protection.
    
    I've lowered the safe voltage from 3.6V to 3.4V, since the system seems
    to boot fine with 3.4V. A smaller threshold decreases the time the
    system spends refusing to do anything but to charge.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 1e1410ba46c624e545ed1703986db59f73e77406
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:14:11 2008 +0100

    deadbat-alternate.patch
    
    When there is no battery or the battery is in its cut-off state, even the
    slightest charge (once the capacitors on the battery rail are filled) will
    drive the battery voltage up to VB_SYS.
    
    The battery voltage monitor therefore only reports garbage. We work around
    this problem by alternating the LED with the charger, so that the battery
    only gets charged half of the time. We then measure right at the end of a
    LED cycle, i.e., when the battery voltage has had time to return to the
    level that really represents the battery state.
    
    (Note: battery voltage drops quickly, so giving it one whole second to
    drain is very much on the safe side.)
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit b663ea63563e4c6b19f788477692bc09a3d7978d
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:14:07 2008 +0100

    deadbat-gta02v5-noled.patch
    
    The LEDs of GTA02v5 draw an excessive amount of current, which can sometimes
    push us out of our current budget while running from 100mA. This patch turns
    the blinking LED off if we're on GTA02v5.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 9f788dcc9c6761ec338f7a172911e1e32dd1c3a8
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:14:03 2008 +0100

    fix-early-upll.patch
    
    During early initialization, the UPLL only ran at 3/4 speed.
    
    This is swiftly corrected in board_init, but better do it right the
    first time.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit d96e96d5eec2fd13de91f5183bfbd4ca683609b2
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:13:59 2008 +0100

    deadbat-cpu-200.patch
    
    Raise the CPU clock (FCLK) from 200MHz to 400MHz only after we've established
    that we have a good power source.
    
    That way, the little bit of processing that are done in wait_for_power cause
    less spikes in current consumption, which decreases the risk of running out
    of power while waiting for USB signaling or similar.
    
    We also save about 7mA of static consumption.
    
    We need to do this awkward two-step process, once in board_init and then
    again in board_late_init, since various clocks that depend on PCLK are
    calculated after board_init (and start.S uses yet another set of different
    settings).
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 967bd3baccc104d5657846ba4ed66c669f498f8a
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:13:55 2008 +0100

    deadbat-cpu-idle.patch
    
    This patch makes the CPU enter IDLE mode while waiting in wait_for_power.
    
    To wake up from IDLE mode, we need an interrupt. Note that this interrupt
    only needs to be signaled to the CPU, but the CPU does not have to execute
    it. However, it has to clear the interrupt or, upon returning from the
    interrupt handler, the system will immediately be interrupted by the same
    interrupt.
    
    We have the following sources of external events:
    
    - battery insertion / battery voltage rises above Vth(batok)
    - USB insertion (giving us 100mA)
    - adapter insertion (giving us 1A)
    - USB host signals availability of 500mA
    - 1Hz tick
    
    USB signaling is already handled by an interrupt handler, so we just use
    what's there. USB insertion and the 1Hz tick cause PMU interrupts, so we
    need to enable the interrupt path from the PMU. Adapter insertion is
    handled as part of USB insertion.
    
    We don't wake up from IDLE on battery insertion or when the battery
    voltage rises above the threshold. Instead, we just poll for it on each
    1Hz tick.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 6d0431712c076a832f462054adcf2b8201130c5a
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:13:50 2008 +0100

    deadbat-defer-glamo.patch
    
    This patch defers initialization of the Glamo core, and in particular
    enabling the clocks, until we're sure to have good power. This reduces
    the power budget we need in early boot by about 15mA.
    
    This patch also cleans up the remains of related code that has already
    been moved to a later boot phase.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit dfed9cc17ea65a6fe0448f45a12bed950f402ede
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:13:46 2008 +0100

    deadbat-soft-start.patch
    
    This patch sets the maximum USB fast charge current to 100mA instead of
    1A. This is an attempt to reduce the inrush current of the battery and
    circuits directly attached to the battery.
    
    Without change, the comparably large capacitors on subsystems that are
    directly supplied from the battery would cause a sharp drop of VB_SYS,
    which prompted the PMU to reset.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 20716cb52c3eee4d418ee50640ee78ff850ad051
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:13:40 2008 +0100

    deadbat-defer-ldo.patch
    
    This patch defers the enabling of LDO regulators that are not needed during
    early boot. (In fact, LDO1, LDO2, and LDO5 are not needed while in u-boot
    at all.) These are:
    
    - LDO1 (GSENSOR_3V3)
    - LDO2 (CODEC_3V3)
    - LDO5 (RF_3V)
    - LDO6 (LCM_3V)
    
    Enabling them after we've established that we have a good power source
    reduces the USB current budget by about 40mA.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 579930fbe92f36a50babf830e8373d161382c067
Author: Werner Almesberger <werner at openmoko.org>
Date:   Tue Jul 15 12:13:34 2008 +0100

    deadbat-wait-silently.patch
    
    In wait_for_power, we switch the AUX LED on and off every second. There's
    no real need to noisily report this on the console.
    
    Signed-off-by: Werner Almesberger <werner at openmoko.org>

commit 96611475add0ea5019d693cfa088a8343701546b
Author: Andy Green <andy at openmoko.com>
Date:   Sun Jun 29 21:28:32 2008 +0100

    uboot-fix-OOCCTL-reset-phase.patch
    
    System reset comes up too early according to how OOCTL
    has been set.  Change to come up in activation state 4
    not 1.
    
    Signed-off-by: Andy Green <andy at openmoko.com>

commit 78b9f3d351436ef14fc38a96082bf9299a3932e9
Author: Andy Green <andy at openmoko.com>
Date:   Sun May 18 12:55:48 2008 +0100

    add-find-wake-reason.patch
    
    Currently we don't take care about the resume reason.
    This patch captures it into the spare GSTATUS 4 register
    and so passes it up into Linux.
    
    Signed-off-by: Andy Green <andy at openmoko.com>

commit 4ec60973cf2e028de905f6ea1f81ef99ca60d834
Author: michael <trimarchi at gandalf.sssup.it>
Date:   Fri May 9 14:35:10 2008 +0100

    nand_skip_bad.patch
    
    Change the skip bad block. Avoid false positives by only checking
    the beginning of a block and check the second page
    
    Signed-off-by: michael <trimarchi at gandalf.sssup.it>

commit 19a477b3d04148f0142058c45f7e1a34dfd412db
Author: warmcat <andy at warmcat.com>
Date:   Fri Apr 11 14:45:54 2008 +0100

    fix-rtc_get_return.patch
    
    rtc_get has changed to have an int return, not void any more.
    
    Signed-off-by: Andy Green <andy at openmoko.com>

commit a5a7ebd4fb672ff4c950d5b4bd8e2540484e0161
Author: warmcat <andy at warmcat.com>
Date:   Fri Apr 11 14:45:54 2008 +0100

    uboot-fix-lcm-flicker.patch

commit 7e36139ffa3b4bc59c2f0d3a6aeee0c0185d7a2f
Author: Andy Green <andy at openmoko.com>
Date:   Fri Apr 11 14:45:54 2008 +0100

    uboot-force-backlight-on.patch
    
    "Special feature" of pcf50633 -- a documented feature though --
    means that you may need to perform a little sequence of actions to
    get the backlight to come back on.
    
    This patch does the sequence in U-Boot so that we no longer have
    the sticky "off" state for the backlight if you reset while the
    backlight was down in Linux.
    
    Signed-off-by: Andy Green <andy at openmoko.com>

commit 1ebdcad82c9c37bae7a7ee86c10b0bbff83d53ff
Author: warmcat <andy at warmcat.com>
Date:   Fri Apr 11 14:45:54 2008 +0100

    stop-internetworking-warning.patch
    
    Stops compiler warning noise on build

commit 8c17f29e213115598308dee9e50450cdf8b6c5eb
Author: warmcat <andy at warmcat.com>
Date:   Fri Apr 11 14:45:54 2008 +0100

    Build scripts