kernel.git
7 years agoar6000: minimise possibility of race in ar6000_ioctl_siwscan andy-tracking github/andy-tracking
Paul Fertser [Sat, 6 Mar 2010 12:44:10 +0000 (15:44 +0300)]
ar6000: minimise possibility of race in ar6000_ioctl_siwscan

This stupid patch should make driver not do NULL pointer dereference when
there's a race between scan and removing of the device. As clumsy as the
rest of the driver, this is not a proper solution, of course.

Screw the anonymous dev who's written the original code!

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

7 years agoar6000: send userspace an event on disconnection
Paul Fertser [Sun, 21 Feb 2010 09:52:58 +0000 (12:52 +0300)]
ar6000: send userspace an event on disconnection

Without this userspace gets confused and might fail to obtain a DHCP
lease, bring the interface up etc etc.

This place is definetely not the best or most intuitive to add this code
but when added to the real disconnect callback it messes wpa_supplicant
completely because hardware signals disconnection right after it was
asked to associate (and signals association the next moment).

The particular problem this patch solves is lack of DISCONNECTED event
during suspend/resume; because of it "wpa_cli -a" doesn't relaunch its
script and the interface is not brought up automatically after resume.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

7 years agogta02_packaging_defconfig: disable mtd write verification
Paul Fertser [Wed, 6 Jan 2010 17:22:03 +0000 (20:22 +0300)]
gta02_packaging_defconfig: disable mtd write verification

This option significantly impacts write speeds making raw write speed 2x
slower and mtdblock device write speed 20% slower. Turn off since it's not
of real benefit anyway and can be even harmful when used with ubifs with
subpages writes enabled.

Big thanks to Gennady for reporting and testing.

Reported-by: Gennady Kupava <gb@bsdmn.com>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

7 years agomfd: glamo: Enable FIFO stage for the LCD engine's memory access
Thomas White [Thu, 26 Nov 2009 08:55:33 +0000 (11:55 +0300)]
mfd: glamo: Enable FIFO stage for the LCD engine's memory access

By avoiding conflicts of memory access inside Glamo, this doubles the
speed of internal memory access.

Signed-off-by: Thomas White <taw@bitwiz.org.uk>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

7 years agoinput: lis302dl: fix the resume path
Michael Trimarchi [Thu, 26 Nov 2009 00:51:19 +0000 (01:51 +0100)]
input: lis302dl: fix the resume path

This fix the resume path of the accelerometer. If the accelerometer
is not power up, restore to initial status. Move the REG_CTRL1 to the
last one position avoid writing on the register when the device is
in power down mode.

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoglamo-spi: missing semicolon after MODULE_AUTHOR()
Martin Jansa [Wed, 4 Nov 2009 19:09:03 +0000 (20:09 +0100)]
glamo-spi: missing semicolon after MODULE_AUTHOR()

Fix compilation error due to missing semicolon

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoGTA01/GTA02: disable android drivers in default configs
Radek Polak [Fri, 2 Oct 2009 18:54:29 +0000 (20:54 +0200)]
GTA01/GTA02: disable android drivers in default configs

Android drivers are not needed for standard linux systems. We can save
memory by disabling them.

Android low memory killer (CONFIG_ANDROID_LOW_MEMORY_KILLER) also
prevents using swap by killing applications before swapping can start
on non android systems.

Reported-by: Jim Morris <morris@wolfman.com>
Signed-off-by: Radek Polak <psonek2@seznam.cz>
Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoFix s3c-adc suspend
Vasily Khoruzhick [Fri, 2 Oct 2009 01:58:18 +0000 (20:58 -0500)]
Fix s3c-adc suspend

Fix for a bug that shows when the s3c2410 TS driver requests
a conversion from the s3c-adc driver and the machine goes into suspend.
In this case the touchscreen stops working.

Note: Nelson edited the original patch with a few small changes.

Reported-by: Radek Polak <psonek2@seznam.cz>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoFix small typo
Nelson Castillo [Thu, 24 Sep 2009 22:37:14 +0000 (17:37 -0500)]
Fix small typo

s/though/thought/

Reported-by: Rask Ingemann Lambertsen <ccc94453@vip.cybercity.dk>
Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoFix wrong commit
Nelson Castillo [Wed, 23 Sep 2009 07:44:08 +0000 (02:44 -0500)]
Fix wrong commit

I'm quite sorry I sent the commit that crashes GTA01.

The fix is quite short. Fixing is easier than reverting and
fixing conflicts.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoRequire CONFIG_S3C24XX_ADC for GTA01/GTA02
Nelson Castillo [Wed, 23 Sep 2009 07:01:37 +0000 (02:01 -0500)]
Require CONFIG_S3C24XX_ADC for GTA01/GTA02

Select CONFIG_S3C24XX_ADC by defult for GTA01/GTA02.
Somebody will hate me if I don't send this patch soon.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoGTA01: Disable glamo
Nelson Castillo [Tue, 22 Sep 2009 10:23:50 +0000 (05:23 -0500)]
GTA01: Disable glamo

Remove glamo from defconfig.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoRemove sort call from group filter
Nelson Castillo [Tue, 22 Sep 2009 10:23:43 +0000 (05:23 -0500)]
Remove sort call from group filter

This patch applies upstream feedback to the group filter.
The algorithms are equivalent, thus we will get the same
results after applying this patch.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoFix Andy's email in filters.
Nelson Castillo [Tue, 22 Sep 2009 10:23:36 +0000 (05:23 -0500)]
Fix Andy's email in filters.

He is not reachable at the OM address.
We need the right email if we send upstream.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agos3_ts: cleanups
Nelson Castillo [Tue, 22 Sep 2009 10:23:28 +0000 (05:23 -0500)]
s3_ts: cleanups

General cleanups. Now we pass checkpatch.pl.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoMake s3c TS driver use s3c-adc API
Nelson Castillo [Fri, 18 Sep 2009 06:45:39 +0000 (01:45 -0500)]
Make s3c TS driver use s3c-adc API

I had a patch by Vasily Khoruzhick <anarsoul@gmail.com>
in the linux-arm-kernel as a guide for some of the changes.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agos3c24xx: introduce s3c-adc delay
Vasily Khoruzhick [Tue, 22 Sep 2009 10:23:12 +0000 (05:23 -0500)]
s3c24xx: introduce s3c-adc delay

Without this patch we cannot make the s3c2410/s3c2440 touchscreen
driver work.

Is this delay device-dependant?
Is there a better way to do this?

We specify the delay value (S3C2410_ADCDLY).

Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoS3C: ADC: Expose number of remaining conversions to convert callback
Nelson Castillo [Tue, 22 Sep 2009 10:23:04 +0000 (05:23 -0500)]
S3C: ADC: Expose number of remaining conversions to convert callback

This patch allow us to efficiently modify the number of
remaining conversions from the client side. This us useful
when we do not know in advance how many conversions we will
need or when we need to cancel pending conversions.

This change is simple enough to be compatible with existing
code that can just define the new pointer in the callback
and ignore it.

Sample usage:

http://tinyurl.com/s3c2410-ts-c (function stylus_adc_action).

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>
Signed-off-by: Ben Dooks <ben-linux@fluff.org>

8 years agoS3C: ADC: Fix lines with more than 80 chars in adc.h
Nelson Castillo [Tue, 22 Sep 2009 10:22:57 +0000 (05:22 -0500)]
S3C: ADC: Fix lines with more than 80 chars in adc.h

Small cleanup.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>
[ben-linux@fluff.org: rewrote subject]
Signed-off-by: Ben Dooks <ben-linux@fluff.org>

8 years agoRemove dependence on old netdev operations
Michael Trimarchi [Tue, 22 Sep 2009 17:36:37 +0000 (19:36 +0200)]
Remove dependence on old netdev operations

Remove the dependence to the OLD NET DEV operation

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>

8 years agogta01: when charging the current reported should be negative
Paul Fertser [Thu, 27 Aug 2009 08:19:12 +0000 (12:19 +0400)]
gta01: when charging the current reported should be negative

Reported-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50606: fix RTC alarm
Arnaud Patard [Thu, 27 Aug 2009 07:46:41 +0000 (11:46 +0400)]
pcf50606: fix RTC alarm

This makes wake up on RTC alarm work properly, ported from rtc-pcf50633
commits 4caf79de95c26495e7cdc8204023d97598f887d2 and
c3e4e22fb0c3e1d82f66e67f6592949e48f3995a.

Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: move messages to appropriate log levels
Arnaud Patard [Thu, 27 Aug 2009 07:39:33 +0000 (11:39 +0400)]
pcf50633: move messages to appropriate log levels

Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50606: move messages to appropriate log levels
Arnaud Patard [Thu, 27 Aug 2009 07:37:49 +0000 (11:37 +0400)]
pcf50606: move messages to appropriate log levels

Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoRevert "AR6000: move low-level cleanup from ar6000_destroy to ar6000_close"
Paul Fertser [Sun, 30 Aug 2009 07:55:48 +0000 (11:55 +0400)]
Revert "AR6000: move low-level cleanup from ar6000_destroy to ar6000_close"

This reverts commit 9c4451ff31b937a478f3d3eabef30b71cbe12b12.

This commit made wifi unusable after ifconfig down and sometimes after
unbinding.

8 years agoFix max voltage of LDO1
Sven Rebhan [Sun, 23 Aug 2009 15:17:23 +0000 (17:17 +0200)]
Fix max voltage of LDO1

This patch fixes a typo in the voltage setting of LDO1.
In the function mangle_pmu_pdata_by_system_rev() the value of min_uV
is set twice, while the value of max_uV stays at the initialized
value. This leaves us with .constraints.min_uV = 3300000 and
.constraints.max_uV = 1300000, which is probably not what we want.

This patch is only required on andy-tracking.

Signed-off-by: Sven Rebhan <odinshorse@googlemail.com>

8 years agoGTA02: add support for dumb battery driver
Paul Fertser [Sun, 26 Jul 2009 14:09:13 +0000 (18:09 +0400)]
GTA02: add support for dumb battery driver

This allows to use gta01 battery driver on gta02.

echo bq27000-battery.0 > /sys/bus/platform/drivers/bq27000-battery/unbind
modprobe gta01_battery

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta02: use a dedicated function to query usb online status
Paul Fertser [Wed, 29 Jul 2009 19:56:48 +0000 (23:56 +0400)]
gta02: use a dedicated function to query usb online status

We need to use a dedicated function because we can't get charger
status from an irq context and that is required by usb gadgets.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: query charger status directly
Paul Fertser [Wed, 29 Jul 2009 14:24:39 +0000 (18:24 +0400)]
pcf50633: query charger status directly

Current scheme is fragile and is likely to go off sync, especially on
batfull->adapter charging automatic MBC transition.

Query the status bit every time we need it instead.

We need to export another function to query for USB presence because
we can't read anything from PCF50633 (via I2C) inside irq context and
that is needed by usb gadgets.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: properly reenable charging when the supply conditions change
Paul Fertser [Wed, 29 Jul 2009 14:09:20 +0000 (18:09 +0400)]
pcf50633: properly reenable charging when the supply conditions change

If chgmod == BATFULL, setting chgena has no effect. Datasheet says we
need to set resume instead but when autoresume is used resume doesn't
work. Clear and set chgena instead.

This enables a user to force charging by re-plugging USB even when the
charger entered Battery Full mode, might be handy before a long trip.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: get rid of charging restart software auto-triggering
Paul Fertser [Wed, 29 Jul 2009 13:39:52 +0000 (17:39 +0400)]
pcf50633: get rid of charging restart software auto-triggering

After reaching Battery Full condition MBC state machine switches back
into charging mode when the battery voltage falls below 96% of a
battery float voltage. The voltage drop in Li-Ion batteries is
marginal (1-2%) till about 80% of its capacity - which means, after a
BATFULL, charging won't be restarted until 75-80%.

That is a desired behaviour recommended by battery manufacturers,
don't mess with it.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: fix wrong define for 10bit ADC mode
Paul Fertser [Mon, 27 Jul 2009 21:09:04 +0000 (01:09 +0400)]
pcf50633: fix wrong define for 10bit ADC mode

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: revise locking for ADC
Paul Fertser [Mon, 27 Jul 2009 20:58:48 +0000 (00:58 +0400)]
pcf50633: revise locking for ADC

Current implementation is prone to races, this patch attempts to remove all
but one (in pcf50633_adc_sync_read).

The idea is that we need to guard the queue access only on inserting and
removing items. If we insert and there're no more items in the queue it
means that the last irq already happened and we need to trigger ADC
manually. If not, then the next conversion will be triggered by the irq
handler upon completion of the previous.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agopcf50633: use a dedicated workqueue for irq processing
Paul Fertser [Mon, 27 Jul 2009 20:41:15 +0000 (00:41 +0400)]
pcf50633: use a dedicated workqueue for irq processing

Using the default kernel "events" workqueue causes problems with
synchronous adc readings if initiated from some task on the same
workqueue.

I had a deadlock trying to use pcf50633_adc_sync_read from a
power_supply class driver because the reading was initiated from the
workqueue and it waited for the irq processing to complete (to get the
result) and that was put on the same workqueue.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta02: charging_restart_interval should no longer be used, count on MBC auto-restart
Paul Fertser [Wed, 29 Jul 2009 14:32:52 +0000 (18:32 +0400)]
gta02: charging_restart_interval should no longer be used, count on MBC auto-restart

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta01_battery: add charge_{now,full} properties
Paul Fertser [Thu, 30 Jul 2009 20:47:41 +0000 (00:47 +0400)]
gta01_battery: add charge_{now,full} properties

Some userspace (notably Enlightment's batget) ignores "capacity"
completely if there're no other properties. Expose a fake value
(850mAh, typical for BL-5C) to make them happy.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta01_battery: generalise the driver, add capacity estimation
Paul Fertser [Sun, 26 Jul 2009 14:00:29 +0000 (18:00 +0400)]
gta01_battery: generalise the driver, add capacity estimation

Check if machine-specific hooks exist before calling, this allows to use
this driver with gta02 or any other device that is able to provide battery
voltage readings.

Report an estimate of capacity percentage left, based on discharge curves.
Strictly speaking, this is a violation of power_supply class specs but
since this is the only driver for dumb batteries and userspace lacks any
library to do the estimation itself, we have to stuff it here.

Please keep in mind that this estimation is by no means accurate, usually
cell phones have only 5 bars to represent charge status.

Credit goes to Joerg Reisenweber and SpeedEvil for helping with gathering
and analysing the data.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta01_battery: fix unloading bug
Paul Fertser [Sun, 26 Jul 2009 13:56:54 +0000 (17:56 +0400)]
gta01_battery: fix unloading bug

Driver data must be properly registered after allocation

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta02: fix compile without CHARGER_PCF50633
Lars-Peter Clausen [Sun, 26 Jul 2009 18:51:48 +0000 (20:51 +0200)]
gta02: fix compile without CHARGER_PCF50633

gta02_pmu_force_shutdown is only defined when CONFIG_CHARGER_PCF50633
is set, but it's used unconditionally.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>

8 years agobq27000: Return -ENODEV in bq27000_battery_get_property if the battery is not
Lars-Peter Clausen [Tue, 21 Jul 2009 19:02:57 +0000 (21:02 +0200)]
bq27000: Return -ENODEV in bq27000_battery_get_property if the battery is not

If the battery is not present hdq_read will always return an error.
If the drivers knows that the battery is not present the correct thing to do is
to return -ENODEV instead of passing the error on.
Do this for all properties except POWER_SUPPLY_PROP_PRESENT.

The power supply sysfs expects us to do so, else it won't generate a proper
uevent file.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>

8 years agoGTA01: fix current and voltage reporting units
Paul Fertser [Sun, 5 Jul 2009 18:53:25 +0000 (22:53 +0400)]
GTA01: fix current and voltage reporting units

According to power_supply class specification all voltages and currents
must be reported in uV and uA correspondingly.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoASoC: Fix wm8753 register cache size and initialization
Lars-Peter Clausen [Thu, 2 Jul 2009 21:04:16 +0000 (23:04 +0200)]
ASoC: Fix wm8753 register cache size and initialization

Register cache space was not being allocated for the final register,
causing bugs when it was used.  Allocate space for it.

Also ensure that the final register is displayed in sysfs.

[Commit message rewritten to document actual issue. -- broonie]

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

8 years agoASoC: Fix register cache initialisation for WM8753
Mark Brown [Fri, 3 Jul 2009 08:33:39 +0000 (09:33 +0100)]
ASoC: Fix register cache initialisation for WM8753

The wrong register cache variable was being used to provide the size for
the memcpy(), resulting in a copy of only a void * of data.

Reported-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

8 years agoGTA0[12]: add nls 866, 1251 to defconfigs
Paul Fertser [Wed, 24 Jun 2009 08:29:37 +0000 (12:29 +0400)]
GTA0[12]: add nls 866, 1251 to defconfigs

These codepages are still widely used by inferior operating systems to
represent cyrillic letters in filenames, both in fat filesystem and by some
samba-servers.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoRNDIS: crude hack to fix oops on connecting to WXP SP2
Paul Fertser [Thu, 11 Jun 2009 06:16:14 +0000 (10:16 +0400)]
RNDIS: crude hack to fix oops on connecting to WXP SP2

F4t discovered that his freerunner panics on connect to his windows box.
This crude hack workarounds the problem.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agogta02: move debugging messages to the appropriate levels
Paul Fertser [Fri, 12 Jun 2009 22:04:18 +0000 (02:04 +0400)]
gta02: move debugging messages to the appropriate levels

Clean up debugging messages so that we don't see any output with loglevel=4
(default for Qi). This avoids slowing down suspend/resume by slow fb
output.

Checkpatch barks on this patch but i guess most of that debugging would
have to be changed prior to upstream submission anyway.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoCONFIG_BATTERY_BQ27000_HDQ selects CONFIG_HDQ_GPIO_BITBANG
Luca Capello [Tue, 9 Jun 2009 23:23:59 +0000 (01:23 +0200)]
CONFIG_BATTERY_BQ27000_HDQ selects CONFIG_HDQ_GPIO_BITBANG

This is because of commit

  1f06c28eaa6e77384a366c3d2ab1ccdbb8a807e0
  The BQ27000 Battery requires HDQ.

Otherwise the battery is not available at all.

Signed-off-by: Luca Capello <luca@pca.it>

8 years agogta01: asoc: fix lm4857 control
Paul Fertser [Wed, 10 Jun 2009 11:39:38 +0000 (15:39 +0400)]
gta01: asoc: fix lm4857 control

Commit 4eaa9819dc08d7bfd1065ce530e31b18a119dcaf changed semantics of
private_value member of kcontrol. This resulted in inability to control
amplifier and subsequently in very low output volume.

Tested-by: Johannes Schauer <josch@pyneo.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoGTA0[12]: make headset button report PLAY pressed on press
Paul Fertser [Mon, 8 Jun 2009 19:46:53 +0000 (23:46 +0400)]
GTA0[12]: make headset button report PLAY pressed on press

As discussed on the ML, currently headset button reports "KEY_PAUSE
released" on press and "pressed" on release. For "pause" semantics it makes
sense but if someone thinks of that as of just another button, it'd be good
to have "pressed/released" events consistent with button press/release.

Luca Capello <luca@pca.it> suggested to change the button event to KEY_PLAY
and to invert the state to make it more consistent.

This trivial patch inverts button events for GTA03 KEY_PLUS and KEY_MINUS,
i'm not sure how undesirable that is. I can prepare an alternative that
preserves current behaviour on GTA03 per request.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoGTA02: gps power management bugfix
Paul Fertser [Fri, 5 Jun 2009 07:11:04 +0000 (11:11 +0400)]
GTA02: gps power management bugfix

After suspending with GPS powered on power_on sysfs node reports 0 and GPS
is actually unpowered but "echo 1 > power_on" has no effect. This patch
fixes it.

Reported and analised by Timo Juhani Lindfors <timo.lindfors@iki.fi>,
ticket #2293.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoGTA01, GTA02: remove RTC_DEBUG from defconfigs
Paul Fertser [Thu, 4 Jun 2009 20:01:15 +0000 (00:01 +0400)]
GTA01, GTA02: remove RTC_DEBUG from defconfigs

This option is meant for debugging and is not any more of interest to
ordinary users.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoGTA01: disable android wakelocks
Paul Fertser [Thu, 4 Jun 2009 20:10:08 +0000 (00:10 +0400)]
GTA01: disable android wakelocks

Enabling wakelocks leads to immediate suspend after resume. See
ce6840da7184043a09dd57d4c201a32ef7b6c083 for clarifications.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoGTA02: bq27000 report current charge
Paul Fertser [Thu, 4 Jun 2009 19:40:52 +0000 (23:40 +0400)]
GTA02: bq27000 report current charge

Using HAL for E's battery gadget highlighted an oddity: the kernel exposed
last full charge property but didn't expose current charge property. This
resulted in the wrong computation of current battery capacity by E's gadget
(and probably other programs as well).

This patch exposes a corresponding bq27000 register to make E battery
gadget happy (it is still not showing correct values because of bugs in HAL
resulting in 3 batteries (apm emulation and usb supply being bogus here)
instead of one).

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoImprove battery status notification
Michael Trimarchi [Sat, 23 May 2009 20:54:00 +0000 (17:54 -0300)]
Improve battery status notification

This patch adds a call to cancel_delayed_work before a call
to schedule_delayed_work.

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>
Signed-off-by: Daniel Willmann <daniel@totalueberwachung.de>

8 years agogta01: fix jbt6k74 spi registration
Paul Fertser [Sun, 24 May 2009 11:19:58 +0000 (15:19 +0400)]
gta01: fix jbt6k74 spi registration

Without this jbt6k74 is not seen by the system and consequently
display doesn't work with Qi.

Rebasing to .28 introduced a regression (probably in
a6af63e826b79ae1ba2d8cbfa66805cc2dfe367c) resulting in jbt6k74 no longer
manageable by the kernel. This was unnoticed before because u-boot does its
own LCM initialization. With Qi it resulted in a white screen.

This patch is fixing it by making spi host bus number (unspecified,
defaults to 0) and spi device (jbt6k74) bus number match.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years ago Add a vbus polling function for use by the udc driver.
Mike Westerhof [Thu, 14 May 2009 01:15:54 +0000 (20:15 -0500)]
Add a vbus polling function for use by the udc driver.

Signed-off-by: Mike Westerhof <mwester@dls.net>

8 years ago This patch adds the option to poll vbus status to the s3c24xx udc driver.
Mike Westerhof [Thu, 14 May 2009 01:11:06 +0000 (20:11 -0500)]
This patch adds the option to poll vbus status to the s3c24xx udc driver.

    Machines that do not route the USB vbus signal to a GPIO can use this
    mechanism as an alternate way for the udc driver to detect the usb
    connect/disconnect state, which is useful (for example) to enable the
    link up/down state to be determined for the g_ether gadget.

Signed-off-by: Mike Westerhof <mwester@dls.net>

8 years agousb gadget: fix ethernet link reports to ethtool
Jonathan McDowell [Thu, 26 Mar 2009 07:45:27 +0000 (00:45 -0700)]
usb gadget: fix ethernet link reports to ethtool

The g_ether USB gadget driver currently decides whether or not there's a
link to report back for eth_get_link based on if the USB link speed is
set. The USB gadget speed is however often set even before the device is
enumerated. It seems more sensible to only report a "link" if we're
actually connected to a host that wants to talk to us. The patch below
does this for me - tested with the PXA27x UDC driver.

Signed-off-by: Jonathan McDowell <noodles@earth.li>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

8 years agoAvoid null pointer access on early call to pcf50633_mbc_get_status
Nelson Castillo [Fri, 15 May 2009 04:46:41 +0000 (23:46 -0500)]
Avoid null pointer access on early call to pcf50633_mbc_get_status

Avoid a kernel panic during an early call
to the pcf50633_mbc_get_status function.

8 years agoGlamo MFD: MFD_GLAMO_SPI_GPIO requires SPI_BITBANG to build
Rask Ingemann Lambertsen [Wed, 6 May 2009 00:41:24 +0000 (02:41 +0200)]
Glamo MFD: MFD_GLAMO_SPI_GPIO requires SPI_BITBANG to build

   This patch fixes a linking failure in the Glamo GPIO SPI bitbang support.

Signed-off-by: Rask Ingemann Lambertsen <rask@sygehus.dk>

8 years agoGlamo MFD: Fix indentation of Kconfig, convert spaces to tabs
Rask Ingemann Lambertsen [Wed, 6 May 2009 00:36:00 +0000 (02:36 +0200)]
Glamo MFD: Fix indentation of Kconfig, convert spaces to tabs

   This patch fixes some indentation which was off by one and converts
spaces to tabs.

Signed-off-by: Rask Ingemann Lambertsen <rask@sygehus.dk>

8 years agoGTA01: Add "present" property to power_supply
Nelson Castillo [Tue, 28 Apr 2009 23:58:05 +0000 (18:58 -0500)]
GTA01: Add "present" property to power_supply

Small patch to add /sys/class/power_supply/battery/present.
You must not run a GTA01 without battery thus it is always 1.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoCall unbind() hook of the usb gadget so that it can properly release its resources
Timo Juhani Lindfors [Tue, 28 Apr 2009 17:22:46 +0000 (12:22 -0500)]
Call unbind() hook of the usb gadget so that it can properly release its resources

Call unbind() hook of the gadget so that it can properly release its
resources. This fixes bug http://docs.openmoko.org/trac/ticket/2240
("removing/reloading g_ether not working"). All other instances of
usb_gadget_unregister_driver in Linux already call unbind() so it is
bit odd s3c2410_udc.c did not do so.

Signed-off-by: Timo Juhani Lindfors <timo.lindfors@iki.fi>

8 years agoUpdate status battery
Michael Trimarchi [Thu, 23 Apr 2009 07:12:14 +0000 (09:12 +0200)]
Update status battery

This patch adds the call to the worker in
bq27000_battery_external_power_changed. Now (un)plugging the USB cable
effects the battery status soon. I don't know if it is possible call
the status change directly.

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>
Signed-off-by: Daniel Willmann <daniel@totalueberwachung.de>

8 years agoThis patch take the min platform data array size and keys array size.
Michael Trimarchi [Tue, 21 Apr 2009 16:23:42 +0000 (11:23 -0500)]
This patch take the min platform data array size and keys array size.

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>

8 years agoglamo_fb: make Xglamo workaround conditional
Nelson Castillo [Tue, 21 Apr 2009 16:18:43 +0000 (11:18 -0500)]
glamo_fb: make Xglamo workaround conditional

Distributions that use Xorg will not need the workaround at
all and now they can disable it.

Updated configurations to enable it by default.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoglamo_fb: Small cleanup
Nelson Castillo [Tue, 21 Apr 2009 16:18:33 +0000 (11:18 -0500)]
glamo_fb: Small cleanup

~ Cleanup of a redundant call I introduced before.
~ Removed useless assignation.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoMake Xglamo workaround opt-out. Fixes #2255.
Nelson Castillo [Tue, 21 Apr 2009 02:39:56 +0000 (21:39 -0500)]
Make Xglamo workaround opt-out. Fixes #2255.

To disable the workaround at run-time run:

echo 0 > /sys/class/i2c-adapter/i2c-0/0-0073/pcf50633-regltr.9/glamo3362.0/glamo-fb.0/xglamo_hack

I guess I should add a compile option now to completely disable the workaround but let's call
this an enhancement and let's close the bug now.

By default the workaround is enabled.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>
Reported-by: Timo Juhani Lindfors <timo.lindfors@iki.fi>

8 years agoDisable EXT4 / enable reiserfs
Nelson Castillo [Mon, 20 Apr 2009 06:33:29 +0000 (01:33 -0500)]
Disable EXT4 / enable reiserfs

Disable EXT4 in gta02_moredrivers_defconfig.
Enable reiserfs3 in gta02_moredrivers_defconfig and gta02_packaging_defconfig.

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>
Reported-by: W.Kenworthy <billk@iinet.net.au>

8 years agoremove-ext4-from-gta02_packaging_defconfig.patch
Mike (mwester) [Mon, 20 Apr 2009 03:06:00 +0000 (22:06 -0500)]
remove-ext4-from-gta02_packaging_defconfig.patch

Removes ext4 support from the packaging defconfig; there is little
value in having this built into the kernel, and a bug in the current
2.6.29-rc3 kernel results in the kernel attempting to mount an ext2
rootfs as ext4, which fails.

Signed-off-by: Mike Westerhof <mwester@dls.net>

8 years agoRevert "Revert "Fixed the LCM driver state transition delay and pixel clock issue""
Nelson Castillo [Sun, 19 Apr 2009 19:41:33 +0000 (14:41 -0500)]
Revert "Revert "Fixed the LCM driver state transition delay and pixel clock issue""

This reverts commit aa2dbc74bed8cf60e591c654ba1882011b9a8ece.

Reverting the revert. Sorry for the noise.

Check:

http://lists.openmoko.org/pipermail/openmoko-kernel/2009-April/010023.html

8 years agoRevert "Fixed the LCM driver state transition delay and pixel clock issue"
Nelson Castillo [Sun, 19 Apr 2009 16:16:56 +0000 (11:16 -0500)]
Revert "Fixed the LCM driver state transition delay and pixel clock issue"

This reverts commit bd4b7e8e84ab43a13a4620b001d52d373c2122b3 which brought
WSOD back for some devices.

It was reported that the problem does not happen with Qi and it
seem it is good news. See:

https://docs.openmoko.org/trac/ticket/2274

Conflicts:

drivers/video/display/jbt6k74.c

8 years agoRevert "update gta01 config based on gta02 config"
Nelson Castillo [Thu, 16 Apr 2009 14:18:51 +0000 (09:18 -0500)]
Revert "update gta01 config based on gta02 config"

This reverts commit 962780e3631255fc5423192d0762f30a1fa45239.

Nothing really wrong with it but the resulting kernel was
too big for the GTA01 NAND flash.

GTA01 boots.

8 years agogta02: fix re-enabling backlight with <12 values
Paul Fertser [Thu, 16 Apr 2009 12:04:14 +0000 (16:04 +0400)]
gta02: fix re-enabling backlight with <12 values

If brightness value was set to anything less than 12, the backlight wasn't
re-enabled after unblanking the display. Closes ticket #2262.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoOHCI: Disabled additional debugging
Paul Fertser [Wed, 15 Apr 2009 21:37:55 +0000 (01:37 +0400)]
OHCI: Disabled additional debugging

Disabled additional debugging introduced
in 70b36104ab8c03f5be6d03344214d09447d9d4a8

8 years agomake touchscreen filter symbols const static in mach-gta01.c and mach-gta02.c
Nelson Castillo [Wed, 15 Apr 2009 05:21:33 +0000 (00:21 -0500)]
make touchscreen filter symbols const static in mach-gta01.c and mach-gta02.c

Signed-off-by: Nelson Castillo <arhuaco@freaks-unidos.net>

8 years agoToo many info messages for apm battery support
Tim Niemeyer [Tue, 14 Apr 2009 17:07:26 +0000 (19:07 +0200)]
Too many info messages for apm battery support

Changing dev_info => dev_dbg.

Signed-off-by: Tim Niemeyer <tim.niemeyer@mastersword.de>

8 years agobetter naming for filter_configuration
Tim Niemeyer [Tue, 14 Apr 2009 17:07:25 +0000 (19:07 +0200)]
better naming for filter_configuration

this affects gta01 and gta02, now we can enable both in the .config

Signed-off-by: Tim Niemeyer <tim.niemeyer@mastersword.de>

8 years agoupdate gta01 config based on gta02 config
Tim Niemeyer [Tue, 14 Apr 2009 17:07:27 +0000 (19:07 +0200)]
update gta01 config based on gta02 config

Signed-off-by: Tim Niemeyer <tim.niemeyer@mastersword.de>

8 years agoFix kernel oops when starting wpa_supplicant.
Mike Westerhof [Tue, 14 Apr 2009 07:10:02 +0000 (02:10 -0500)]
Fix kernel oops when starting wpa_supplicant.
This closes bug #2267.

Signed-off-by: Mike Westerhof <mwester@dls.net>

8 years agofbcon: Cleaned unnecessary debugging output (fix bug 2265)
Nicolas Dufresne [Mon, 13 Apr 2009 14:43:30 +0000 (10:43 -0400)]
fbcon: Cleaned unnecessary debugging output (fix bug 2265)

Cleaned unnecessary debugging output.

Signed-off-by: Nicolas Dufresne <nicolas.dufresne@gmail.com>

8 years agoRe-enable USBHOST attribute
Michael Trimarchi [Wed, 8 Apr 2009 13:50:02 +0000 (15:50 +0200)]
Re-enable USBHOST attribute

Re-enable usbmode attribute.
It was disabled by 70b36104ab8c03f5be6d03344214d09447d9d4a8.

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>

8 years agoCleanup USB host code of s3c2410
Michael Trimarchi [Wed, 8 Apr 2009 13:48:15 +0000 (15:48 +0200)]
Cleanup USB host code of s3c2410

Cleanup code of ohci-s3c2410.

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>

8 years agoCorrect platform s3c-ohci device name, makes bluetooth work
Paul Fertser [Tue, 7 Apr 2009 20:51:35 +0000 (00:51 +0400)]
Correct platform s3c-ohci device name, makes bluetooth work

Not sure what is the correct naming scheme and if MODULE_ALIAS in
drivers/usb/host/ohci-s3c2410.c is needed at all (if yes, probably it
should be modified too).

Signed-off-by: Paul Fertser <fercerpav@gmail.com>

8 years agoASoC: Clean up coding style issues in GTA02
Mark Brown [Fri, 3 Apr 2009 11:36:26 +0000 (12:36 +0100)]
ASoC: Clean up coding style issues in GTA02

No substantial changes, this just tidies up a bunch of coding style
issues that ought to be fixed up before merge (which I'll do when the
GTA02 machine support is queued for merge).

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

8 years agoFixed the LCM driver state transition delay and pixel clock issue
Nicolas Dufresne [Sat, 4 Apr 2009 18:15:48 +0000 (14:15 -0400)]
Fixed the LCM driver state transition delay and pixel clock issue

    Space transition between sleep-in and sleep-out of at least
    120 ms as mentionned in the spec. Also, made sure the LCM
    is moved into sleep state before the pixel clock is turned off.
    To do so, I've modified glamo-fb to send FB_BLANK_POWERDOWN before
    turning off the pixel clock. Also fixed various violation of coding
    style. This patch should reduce the risk of white screen.

Signed-off-by: Nicolas Dufresne <nicolas.dufresne@gmail.com>

8 years agoAR6000: move low-level cleanup from ar6000_destroy to ar6000_close before-nelson-commits
Werner Almesberger [Thu, 2 Apr 2009 11:12:12 +0000 (08:12 -0300)]
AR6000: move low-level cleanup from ar6000_destroy to ar6000_close

A network device is supposed to disable its internals in the "close"
function. Doing so affords us protection against various races,
including the ioctl vs. rfkill conflict reported by Michael.

Signed-off-by: Werner Almesberger <werner@openmoko.org>
Reported-by: Michael Trimarch <michael@panicking.kicks-ass.org>

8 years agortc-pcf50606: Fix month off-by-one error
Werner Almesberger [Wed, 1 Apr 2009 02:23:05 +0000 (23:23 -0300)]
rtc-pcf50606: Fix month off-by-one error

According to the PCF50606 manual, the issues resoved by commit
cc1663fc922c03feb0d7bbb8b18d62fbac0128de also exists there.

Signed-off-by: Werner Almesberger <werner@openmoko.org>
Reported-by: Rask Ingemann Lambertsen <rask@sygehus.dk>

8 years agoAR6000: support GTA02-specific rfkill only if building for GTA02
Werner Almesberger [Tue, 31 Mar 2009 07:47:06 +0000 (04:47 -0300)]
AR6000: support GTA02-specific rfkill only if building for GTA02

This is loosely based on a patch by Ivan Petrov.

Signed-off-by: Werner Almesberger <werner@openmoko.org>
Reported-by: Ivan Petrov <ivan_p@hotbox.ru>

8 years agoThis patch brings suspend/resume back to GTA01
Tim Niemeyer [Fri, 27 Mar 2009 17:10:36 +0000 (18:10 +0100)]
This patch brings suspend/resume back to GTA01

"pdata" is NULL on GTA01 and you oops.
Revised patch, this time with if(..) instead of #ifdef

Signed-off-by: Tim Niemeyer <tim.niemeyer@mastersword.de>

8 years agoremove sysfs-links on remove
Tim Niemeyer [Fri, 27 Mar 2009 16:27:53 +0000 (17:27 +0100)]
remove sysfs-links on remove

Signed-off-by: Tim Niemeyer <tim.niemeyer@mastersword.de>

8 years agoAR6000: Corrected 'vendor/device IDs'.
Werner Almesberger [Tue, 31 Mar 2009 07:16:32 +0000 (04:16 -0300)]
AR6000: Corrected 'vendor/device IDs'.

IDs 0 and 1 are AR6002, not AR6001. This is based on a patch by
Ivan Petrov, but it differs in leaving the IDs in
include/linux/mmc/sdio_ids.h

Signed-off-by: Werner Almesberger <werner@openmoko.org>
Reported-by: Ivan Petrov <ivan_p@hotbox.ru>

8 years agoAR6000: revert MMC busy status check
Werner Almesberger [Tue, 31 Mar 2009 07:02:04 +0000 (04:02 -0300)]
AR6000: revert MMC busy status check

e2c0650efa751a6a2220618695fa41a2a5e7d23c introduced a platform-specific
hack to check if the MMC driver and hardware handle busy signaling from
the device properly.

Since this seems to be the case, we can revert this hack now.

Signed-off-by: Werner Almesberger <werner@openmoko.org>

8 years agoMinor AR6000 cleanup
Werner Almesberger [Tue, 31 Mar 2009 03:49:46 +0000 (00:49 -0300)]
Minor AR6000 cleanup

This patch cleans up a little after Ivan's netif_queue_stop fix.

Signed-off-by: Werner Almesberger <werner@openmoko.org>

8 years agoAR6000 netif_queue_stop non stop, Bug?
ivan_p@hotbox.ru [Thu, 26 Mar 2009 12:03:03 +0000 (15:03 +0300)]
AR6000 netif_queue_stop non stop, Bug?

This patch resolves the following issue:
http://lists.openmoko.org/pipermail/openmoko-kernel/2009-March/009643.html

Changed: prevent rescheduling network queue at interface opened/connected.
Removed: wake network queue at transmit complete.
Added: wake network queue at packet queue limit not reached.

Signed-off-by: Ivan Petrov <ivan_p@hotbox.ru>

8 years agoFix ar6000_do_activate return value
Werner Almesberger [Tue, 31 Mar 2009 00:06:02 +0000 (21:06 -0300)]
Fix ar6000_do_activate return value

... and check the return value in ar6000_activate.

[ Werner: changed ___FUNCTION__ to __func__ ]

Signed-off-by: Michael Trimarchi <michael@panicking.kicks-ass.org>
Signed-off-by: Werner Almesberger <werner@openmoko.org>

8 years agoFix cpu_is_s3c2442() returning false for an S3C2442B
Rask Ingemann Lambertsen [Mon, 30 Mar 2009 22:28:09 +0000 (00:28 +0200)]
Fix cpu_is_s3c2442() returning false for an S3C2442B

   An S3C2442B would be detected as S3C2440 by the cpu_is_s3cxxxx()
macros. This patch fixes it.

Signed-off-by: Rask Ingeman Lambertsen <rask@sygehus.dk>

8 years agoCodingStyle with copy-src also fixed
Werner Almesberger [Thu, 26 Mar 2009 07:56:25 +0000 (04:56 -0300)]
CodingStyle with copy-src also fixed

Signed-off-by: Tim Niemeyer <reddog@mastersword.de>

8 years agoFix trivial whitespace slippage.
Werner Almesberger [Thu, 26 Mar 2009 07:52:29 +0000 (04:52 -0300)]
Fix trivial whitespace slippage.

Signed-off-by: Werner Almesberger <werner@openmoko.org>

8 years agoClean up pcf50606_client_dev_register
tim.niemeyer@mastersword.de [Tue, 24 Mar 2009 10:11:20 +0000 (11:11 +0100)]
Clean up pcf50606_client_dev_register

platform_device_add_data copies the data, so no need for
kmalloc.

Signed-off-by: Tim Niemeyer <reddog@mastersword.de>