]> Gitweb @ Texas Instruments - Open Source Git Repositories - git.TI.com/gitweb - android-sdk/kernel-video.git/log
android-sdk/kernel-video.git
10 years agoarm: dts: dra7-evm: Add pinmux configs needed for display
Archit Taneja [Thu, 30 May 2013 15:51:07 +0000 (21:21 +0530)]
arm: dts: dra7-evm: Add pinmux configs needed for display

DSS requires pinmux configurations for the VOUT1 DPI lines and I2C2 as it
controls a PCF 8575 IO expander needed by the HDMI level shifter.

Add pinmux configurations for them.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoHACK: pinctrl: pinctrl single: Make pinctrl-single init early
Archit Taneja [Thu, 30 May 2013 11:39:05 +0000 (17:09 +0530)]
HACK: pinctrl: pinctrl single: Make pinctrl-single init early

Change pinctrl-single driver's initcall level from module initcall to arch
initcall. This prevents dependent drivers not supporting deferred probe
mechanism from failing.

This is done for omapdss which doesn't support deferred probe yet. Deferred
probe for omapdss is complicated as init order between omapdss and it's users
(omapfb or omapdrm) is important, and deferred probe complicates that.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoOMAPDSS:HDMI: Change PLL calculations
Andy Gross [Wed, 29 May 2013 07:38:06 +0000 (13:08 +0530)]
OMAPDSS:HDMI: Change PLL calculations

This patch modifies the algorithms used to determine the correct PLL
settings for the HDMI TMDS output.  Instead of using fixed REGM2 values
we search for a valid combination of REGN, REGM, REGM2 that satisfy the
constraints for the requested frequency.

The computation involves looping through the values of REGN and trying to
find an adequate REGM and REGM2 (if applicable).  This results in the lowest
value for REGN that will satisfy the frequency request.

This patch also modifies the API used for the hdmi_compute_pll function.
If a valid PLL combination cannot be found, we return an error back to the
caller.

Signed-off-by: Andy Gross <andy.gross@ti.com>
Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: hdmi: fix deepcolor mode configuration
Archit Taneja [Mon, 6 May 2013 15:36:06 +0000 (17:36 +0200)]
omapdss: hdmi: fix deepcolor mode configuration

The hdmi driver supports deep color modes. However, there is a specific check
which doesn't allow 1080p timings or higher to be shown 36 bit deep color mode.

The more generic check would be to check if the pixel clock in a deep color mode
is less than the maximum pixel clock the TV overlay manager can support for the
given OMAP. Make this change.

Also, make sure that hdmi.dssdev is assigned by the end of probe, this param is
needed by the deepcolor sysfs attribute file.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoARM: dts: DRA7x: Add DMM bindings
Archit Taneja [Mon, 6 May 2013 08:37:54 +0000 (10:37 +0200)]
ARM: dts: DRA7x: Add DMM bindings

Add Dynamic Memory Manager (DMM) bindings for DRA7x device.

Signed-off-by: Archit Taneja <archit@ti.com>
Conflicts:

arch/arm/boot/dts/dra7.dtsi

10 years agoomapdrm: hack: Assign managers/channel to outputs in a more trivial way
Archit Taneja [Mon, 6 May 2013 08:38:33 +0000 (10:38 +0200)]
omapdrm: hack: Assign managers/channel to outputs in a more trivial way

3.8 kernel doesn't have 'dispc_channel' field for omapdss outputs, do a simple
assignment of channels for outputs for now.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agogpu: drm: omap: Use bitmaps for placement
Andy Gross [Thu, 2 May 2013 11:23:59 +0000 (13:23 +0200)]
gpu: drm: omap: Use bitmaps for placement

Modified Tiler placement to utilize bitmaps for bookkeeping and
all placement algorithms.  This resulted in a substantial savings
in time for all Tiler reservation and free operations.  Typical
savings are in the range of 28% decrease in time taken with larger
buffers showing a 80%+ decrease.

Signed-off-by: Andy Gross <andy.gross@ti.com>
10 years agodrm/omap: Fix and improve crtc and overlay manager correlation
Archit Taneja [Tue, 26 Mar 2013 13:45:19 +0000 (19:15 +0530)]
drm/omap: Fix and improve crtc and overlay manager correlation

The omapdrm driver currently takes a config/module arg to figure out the number
of crtcs it needs to create. We could create as many crtcs as there are overlay
managers in the DSS hardware, but we don't do that because each crtc eats up
one DSS overlay, and that reduces the number of planes we can attach to a single
crtc.

Since the number of crtcs may be lesser than the number of hardware overlay
managers, we need to figure out which overlay managers to use for our crtcs. The
current approach is to use pipe2chan(), which returns a higher numbered manager
for the crtc.

The problem with this approach is that it assumes that the overlay managers we
choose will connect to the encoders the platform's panels are going to use,
this isn't true, an overlay manager connects only to a few outputs/encoders, and
choosing any overlay manager for our crtc might lead to a situation where the
encoder cannot connect to any of the crtcs we have chosen. For example, an
omap5-panda board has just one hdmi output. If num_crtc is set to 1, with the
current approach, pipe2chan will pick up the LCD2 overlay manager, which cannot
connect to the hdmi encoder at all. The only manager that could have connected
to hdmi was the TV overlay manager.

Therefore, there is a need to choose our overlay managers keeping in mind the
panels we have on that platform. The new approach iterates through all the
available panels, creates encoders and connectors for them, and then tries to
get a suitable overlay manager to create a crtc which can connect to the
encoders.

We use the dispc_channel field in omap_dss_output to retrieve the desired
overlay manager's channel number, we then check whether the manager had already
been assigned to a crtc or not. If it was already assigned to a crtc, we assume
that out of all the encoders which intend use this crtc, only one will run at a
time. If the overlay manager wan't assigned to a crtc till then, we create a
new crtc and link it with the overlay manager.

This approach just looks for the best dispc_channel for each encoder. On DSS HW,
some encoders can connect to multiple overlay managers. Since we don't try
looking for alternate overlay managers, there is a greater possibility that 2
or more encoders end up asking for the same crtc, causing only one encoder to
run at a time.

Also, this approach isn't the most optimal one, it can do either good or bad
depending on the sequence in which the panels/outputs are parsed. The optimal
way would be some sort of back tracking approach, where we improve the set of
managers we use as we iterate through the list of panels/encoders. That's
something left for later.

Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
10 years agodrm/omap: fix modeset_init if a panel doesn't satisfy omapdrm requirements
Archit Taneja [Tue, 26 Mar 2013 13:45:18 +0000 (19:15 +0530)]
drm/omap: fix modeset_init if a panel doesn't satisfy omapdrm requirements

modeset_init iterates through all the registered omapdss devices and has some
initial checks to see if the panel has a driver and the required driver ops for
it to be usable by omapdrm.

The function bails out from modeset_init if a panel doesn't meet the
requirements, and stops the registration of the future panels and encoders which
come after it, that isn't the correct thing to do, we should go through the rest
of the panels. Replace the 'return's with 'continue's.

Signed-off-by: Archit Taneja <archit@ti.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
10 years agodrm/omap: Take a fb reference in omap_plane_update()
Archit Taneja [Tue, 9 Apr 2013 12:26:00 +0000 (15:26 +0300)]
drm/omap: Take a fb reference in omap_plane_update()

When userspace calls SET_PLANE ioctl, drm core takes a reference of the fb and
passes control to the update_plane op defined by the drm driver.

In omapdrm, we have a worker thread which queues framebuffers objects received
from update_plane and displays them at the appropriate time.

It is possible that the framebuffer is destoryed by userspace between the time
of calling the ioctl and apply-worker being scheduled. If this happens, the
apply-worker holds a pointer to a framebuffer which is already destroyed.

Take an extra refernece/unreference of the fb in omap_plane_update() to prevent
this from happening. A reference is taken of the fb passed to update_plane(),
the previous framebuffer (held by plane->fb) is unreferenced. This will prevent
drm from destroying the framebuffer till the time it's unreferenced by the
apply-worker.

This is in addition to the exisitng reference/unreference in update_pin(),
which is taken for the scanout of the plane's current framebuffer, and an
unreference the previous framebuffer.

Signed-off-by: Archit Taneja <archit@ti.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
10 years agodrm/omap: move out of staging
Rob Clark [Mon, 11 Feb 2013 17:43:09 +0000 (12:43 -0500)]
drm/omap: move out of staging

Now that the omapdss interface has been reworked so that omapdrm can use
dispc directly, we have been able to fix the remaining functional kms
issues with omapdrm.  And in the mean time the PM sequencing and many
other of that open issues have been solved.  So I think it makes sense
to finally move omapdrm out of staging.

Signed-off-by: Rob Clark <robdclark@gmail.com>
10 years agoRevert "gpu: drm: omap: Use bitmaps for placement"
Archit Taneja [Thu, 2 May 2013 09:19:29 +0000 (11:19 +0200)]
Revert "gpu: drm: omap: Use bitmaps for placement"

This reverts commit 2857ec212485e407354d03be48cf00f03437c098.

10 years agoomapdss: hdmi: Add an i2c adapter node as a possible DT phandle
Archit Taneja [Tue, 30 Apr 2013 12:08:29 +0000 (14:08 +0200)]
omapdss: hdmi: Add an i2c adapter node as a possible DT phandle

The hdmi driver supports reading edid either via HDMI IP's ddc in the case
of omap4. We support a bitbanged gpio i2c mode for omap5. For Vayu, the pins
muxed with HDMI IP's ddc are the I2C2 lines, so we provide a mode where we allow
the i2c adapter to be passed as a phandle by DT.

Add a DT node for hdmi similar to that as of omap5-uevm. The difference on Vayu
evm is that it the hdmi ddc lines are the I2C2 line instead of bitbanged gpio
lines. There is a also a PCF level shifter on I2C2 which is used to drive LS_OE
and CT_HPD pins in the TPS level shifter. The HPD gpio is connected directly to
a Vayu GPIO bank.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoHACK: omapdss: HDMI hacks for DRA7x
Archit Taneja [Tue, 30 Apr 2013 16:50:29 +0000 (18:50 +0200)]
HACK: omapdss: HDMI hacks for DRA7x

- Vayu evm has a demux which decides whether i2c2 lines go to the hdmi level
shifter(and finally to the panel) or if they are in their normal state(where
i2c2 is connected to many slaves, one of which is a gpio expander to configure
the LS_OE and CT_HPD pins of the level shifter). This demux is controlled by
a mcasp based gpio. There is no mcasp-gpio driver, neither there is a clean
way to represent such a pin in DT. So the bit is toggled as a hack in the
hdmi driver itself.

- On DRA7x, hdmi phy state transition checks are failing even though hdmi phy is
transitioning correctly, ignore the phy transition checks for now.

These need to be cleaned up later.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: hdmi: Add support for DRA7xx SoCs
Archit Taneja [Mon, 22 Apr 2013 13:19:17 +0000 (18:49 +0530)]
omapdss: hdmi: Add support for DRA7xx SoCs

DRA7xx DSS has the same HDMI IP as OMAP5, add OMAPDSS version checks where
necessary.

HDMI DPLL on DRA7xx can be accessed via DSS address space only when
DSS_PLL_CONTROL in controle module is configured correctly. Do this
configuration here. This is hacky and should be moved to somewhere in the
prcm driver.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoHACK: omap_hwmod: enable DESHDCP clock before DSS module is enabled
Archit Taneja [Tue, 16 Apr 2013 06:05:21 +0000 (11:35 +0530)]
HACK: omap_hwmod: enable DESHDCP clock before DSS module is enabled

enable DESHDCP clock very early in omap_hwmod_setup_all to make sure it's set
before DSS hwmods are setup. Find a better way to do this.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoarm: dra7-evm dts: Add lcd panel support
Archit Taneja [Tue, 30 Apr 2013 12:02:45 +0000 (14:02 +0200)]
arm: dra7-evm dts: Add lcd panel support

Add DT nodes for the dpi lcd panel and the TLC i2c client.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agogpio: pcf857x: Convert to DT mode
Archit Taneja [Mon, 22 Apr 2013 09:13:59 +0000 (14:43 +0530)]
gpio: pcf857x: Convert to DT mode

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: displays: Add tfcs9700 DPI panel driver
Archit Taneja [Tue, 16 Apr 2013 10:14:37 +0000 (15:44 +0530)]
omapdss: displays: Add tfcs9700 DPI panel driver

TFCS9700 is a 800x480 DPI LCD panel found on the Vayu display daughter card.
The card also contains an i2c brightness controller/gpio expander(TLCxxxx) which
drives a pwm signal for the panel and also provides some gpios to configure some
of the panel pins.

Add the DPI panel driver as an omapdss driver, and add the i2c controller as a
client driver in the same file for now. In the longer run, we would have a
separate mfd driver for the TLC chip.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoarm: dra7 dtsi: add DSS DT node
Archit Taneja [Tue, 30 Apr 2013 11:55:34 +0000 (13:55 +0200)]
arm: dra7 dtsi: add DSS DT node

Add a DT node for DSS, this is similar to the DT nodes in omap4/5, the only
difference is that there are multpiple instances of DPI, and they match with
the dra7xx-dpi driver.

Signed-off-by: Archit Taneja <archit@ti.com>
Conflicts:

arch/arm/boot/dts/dra7.dtsi

10 years agoomapdss: Add a DPI driver for DRA7xx SoC
Archit Taneja [Thu, 4 Apr 2013 12:09:35 +0000 (17:39 +0530)]
omapdss: Add a DPI driver for DRA7xx SoC

Create a DPI driver for DRA7xx DSS, the only difference it has compared to the
OMAP DPI driver is that it uses dss core DPLLs as an alternative clock source
and not the DSI PLLs.

These drivers can be merged later on when DSI PLL code is made more generic and
is moved to dss core.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: dpi: create common DPI ops for panel drivers
Archit Taneja [Mon, 18 Mar 2013 08:50:08 +0000 (14:20 +0530)]
omapdss: dpi: create common DPI ops for panel drivers

This allows panel drivers to be independent of the underlying DPI IP, i.e,
whether it's OMAP DPI or DRA DPI.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: features: Add a dss_features struct for DRA7xx SoC
Archit Taneja [Thu, 4 Apr 2013 12:07:31 +0000 (17:37 +0530)]
omapdss: features: Add a dss_features struct for DRA7xx SoC

Add a new features struct for DRA7xx. The connections between outputs and
managers in DRA7xx differ a lot from OMAP5, so we add a new feat struct for it.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: Add dss dplls to dss family
Archit Taneja [Fri, 5 Apr 2013 06:10:06 +0000 (11:40 +0530)]
omapdss: Add dss dplls to dss family

DRA7x has video DPLLs similar to that of the DSI PLLs on OMAP. However, they
aren't a part of DSI, and belong to the DSS core/ DSS family of registers. Tie
the video DPLLs to the dss core platform device.

Borrow the code from DSI to power on, configure and lock the PLLs. The apis
provided will be used by the DRA7x DPI driver to get a desired pixel clock.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: dss: Make overlay manager source selection possible for multiple DPI IPs
Archit Taneja [Thu, 4 Apr 2013 14:34:12 +0000 (20:04 +0530)]
omapdss: dss: Make overlay manager source selection possible for multiple DPI IPs

DRA7x has 3 DPI IPs, out of these DPI1 has a configurable source while the
others don't. Create a func which allows us to select sources for different DPI
instances.

A new dss feat struct is needed for DRA7x, add it.
Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: Add DRA7XX as an OMAPDSS version
Archit Taneja [Thu, 4 Apr 2013 14:19:41 +0000 (19:49 +0530)]
omapdss: Add DRA7XX as an OMAPDSS version

DRA7xx has a slightly modified version of the DSS on OMAP5. Create a new OMAPDSS
version for it. Use omap5 structs for inititalizations in dss features, dss and
dispc.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoomapdss: dss: add module_id param to dpi_select_source
Archit Taneja [Thu, 4 Apr 2013 14:27:38 +0000 (19:57 +0530)]
omapdss: dss: add module_id param to dpi_select_source

DRA7x SoC has 3 DPI instances. DPI1 instance can receive it's pixels from either
LCD1, LCD2, LCD3 or TV vverlay Manager's video ports. The remaining two DPI
instances can receive pixels only from LCD2 and LCD3 respectively.

Add an argument which describes which DPI instance we are referring to when
chosing it's overlay manager.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoMerge remote-tracking branch 'av-feat-tree/omap5_audio_video-3.8.y' into vayu_exp
Archit Taneja [Thu, 30 May 2013 13:36:08 +0000 (19:06 +0530)]
Merge remote-tracking branch 'av-feat-tree/omap5_audio_video-3.8.y' into vayu_exp

Conflicts:
arch/arm/boot/dts/omap5-sevm.dts
arch/arm/boot/dts/omap5-uevm.dts
arch/arm/boot/dts/omap5.dtsi
arch/arm/configs/omap2plus_defconfig
arch/arm/mach-omap2/omap-mpuss-lowpower.c
arch/arm/mach-omap2/omap_hwmod_33xx_data.c

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoMerge remote-tracking branch 'rnayak/platform-base-3.8.y' into pm-linux-3.8.y
Tero Kristo [Wed, 29 May 2013 13:18:10 +0000 (16:18 +0300)]
Merge remote-tracking branch 'rnayak/platform-base-3.8.y' into pm-linux-3.8.y

Conflicts:
arch/arm/mach-omap2/dpll3xxx.c

Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: dts: Remove the non existent l3_main_3 entry
Rajendra Nayak [Wed, 29 May 2013 11:55:32 +0000 (17:25 +0530)]
ARM: DRA7: dts: Remove the non existent l3_main_3 entry

l3_main_3 does not exist on dra7xx devices.

We see the below errors at boot because of it..

[    0.144775] platform ocp.2: Cannot lookup hwmod 'l3_main_3'
[    0.145050] omap_l3_noc ocp.2: couldn't find resource 0

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: dpll: No freqsel on DRA7
Rajendra Nayak [Wed, 29 May 2013 11:33:18 +0000 (17:03 +0530)]
ARM: DRA7: dpll: No freqsel on DRA7

With the growing number of devices which *do not* support freqsel,
stop extending the checks and have checks based on the only device
(343x) family that supports it.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: dpll: Lock GMAC PLL at boot
Rajendra Nayak [Wed, 29 May 2013 11:13:15 +0000 (16:43 +0530)]
ARM: DRA7: dpll: Lock GMAC PLL at boot

Lock GMAC PLL as per the Vayu PLL spec v0.3.2. Without
the GMAC PLL locked, we see the below hwmod init failures
at boot

[    1.254333] clock: dpll_gmac_ck failed transition to 'locked'
[    1.256835] omap_hwmod: pruss1: cannot be enabled for reset (3)
[    2.415313] clock: dpll_gmac_ck failed transition to 'locked'
[    2.417724] omap_hwmod: pruss2: cannot be enabled for reset (3)

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: dpll: Lock ABE PLL according to ATL needs
Rajendra Nayak [Wed, 29 May 2013 10:56:33 +0000 (16:26 +0530)]
ARM: DRA7: dpll: Lock ABE PLL according to ATL needs

The ATL (Audio Tracking Logic) module within DRA7xx requires a
precise 44.1Kz clock. The only possible source for it that can
supply the clock was found to be ABE PLL.

So lock the ABE PLL at 361.2670Mhz, so ATL module can derive the
precise 44.1Khz out of it.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: id: Change control_id_code register address
Sricharan R [Tue, 28 May 2013 09:29:29 +0000 (14:59 +0530)]
ARM: DRA7: id: Change control_id_code register address

The CONTROL_ID_CODE register used for the device identification
is moved to CTRL_MODULE_WAKEUP partition. So updating the
register address for the same here.

Signed-off-by: Sricharan R <r.sricharan@ti.com>
10 years agothermal: consider emul_temperature while computing trend
J Keerthy [Mon, 27 May 2013 13:12:44 +0000 (18:42 +0530)]
thermal: consider emul_temperature while computing trend

In case emulated temperature is in use, using the trend
provided by driver layer can lead to bogus situation.
In this case, debugger user would set a temperature value,
but the trend would be from driver computation.

To avoid this situation, this patch changes the get_tz_trend()
to consider the emulated temperature whenever that is in use.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
[Backported to 3.8]
Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoARM: OMAP5: Add select tuples to enable CPUFREQ on OMAP5
J Keerthy [Thu, 16 May 2013 10:15:47 +0000 (15:45 +0530)]
ARM: OMAP5: Add select tuples to enable CPUFREQ on OMAP5

10 years agoARM: DRA7XX: PM: add soc_is_dra7xx check to take care of freqsel absence
J Keerthy [Thu, 2 May 2013 11:32:37 +0000 (17:02 +0530)]
ARM: DRA7XX: PM: add soc_is_dra7xx check to take care of freqsel absence

add soc_is_dra7xx check to take care of freqsel absence.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoARM: dts: dra7: add clock nodes for CPU and link avs_mpu regulator to cpu
J Keerthy [Fri, 26 Apr 2013 09:13:59 +0000 (14:43 +0530)]
ARM: dts: dra7: add clock nodes for CPU and link avs_mpu regulator to cpu

Add clock nodes for CPU and link avs_mpu regulator to cpu.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoArm: DRA7XX: Clock: Update dpll_mpu_ck_ops
J Keerthy [Wed, 24 Apr 2013 17:44:35 +0000 (23:14 +0530)]
Arm: DRA7XX: Clock: Update dpll_mpu_ck_ops

Update dpll_mpu_ck_ops so as to have a custom dpll_set_rate
on the similar lines of OMAP5. The bit fields and the rates
are exactly the same hence reusing the same functions as that
of OMAP5.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoARM: DTS: DRA7: Link the actual TPS659038 regulators to
J Keerthy [Fri, 26 Apr 2013 13:36:41 +0000 (19:06 +0530)]
ARM: DTS: DRA7: Link the actual TPS659038 regulators to

Link the actual TPS659038 regulators to AVS class 0 regulators.

Referred Visio-Power_Diagram 20121109b.pdf a visio diagram
showing which SMPS feeds on to which voltage domain on DRA7.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoARM: DTS: DRA7: Add avs class 0 regulator nodes
J Keerthy [Fri, 26 Apr 2013 08:56:42 +0000 (14:26 +0530)]
ARM: DTS: DRA7: Add avs class 0 regulator nodes

Add avs class 0 regulator nodes.

Referred:

DRA7xx_ES1.0_NDA_TRM_vC.pdf for the Efuse register addresses.
DMInputs-Voltages-2013-04-05.xlsx for the latest OPP voltages.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoregulator: ti-avs-class0: use regulator name based on device instance
Nishanth Menon [Tue, 30 Apr 2013 19:51:06 +0000 (14:51 -0500)]
regulator: ti-avs-class0: use regulator name based on device instance

Don't use the same regulator descriptor for all regulator instance,
as there will be a name conflict. Instead pick up the device name and
create the descriptor.

As part of this change, save on multiple kzallocs by moving the
descriptor allocation as part of tiavs_class0_data allocation and drop
the need for an temp descriptor.

Signed-off-by: Nishanth Menon <nm@ti.com>
10 years agoregulator: ti-avs-class0: dont reserve efuse memory offsets
Nishanth Menon [Tue, 30 Apr 2013 19:48:47 +0000 (14:48 -0500)]
regulator: ti-avs-class0: dont reserve efuse memory offsets

devm_request_and_ioremap reserves the efuse offsets for AVS class0,
However, these efuse offsets might be used by other drivers for
information from the same efuse offset ranges as well. hence just
remap the range with devm_ioremap_nocache and read the required data
out.

Signed-off-by: Nishanth Menon <nm@ti.com>
10 years agocpufreq: cpufreq-cpu0: defer probe when regulator is not ready(v2)
Nishanth Menon [Fri, 5 Apr 2013 04:21:48 +0000 (04:21 +0000)]
cpufreq: cpufreq-cpu0: defer probe when regulator is not ready(v2)

upstream posted https://patchwork.kernel.org/patch/2505311/

With commit 1e4b545, regulator_get will now return -EPROBE_DEFER
when the cpu0-supply node is present, but the regulator is not yet
registered.

It is possible for this to occur when the regulator registration
by itself might be defered due to some dependent interface not yet
instantiated. For example: an regulator which uses I2C and GPIO might
need both systems available before proceeding, in this case, the
regulator might defer it's registration.

However, the cpufreq-cpu0 driver assumes that any un-successful return
result is equivalent of failure.

When the regulator_get returns failure other than -EPROBE_DEFER, it
makes sense to assume that supply node is not present and proceed with
the assumption that only clock control is necessary in the platform.

With this change, we can now handle the following conditions:
a) cpu0-supply binding is not present, regulator_get will return
appropriate error result, resulting in cpufreq-cpu0 driver controlling
just the clock.
b) cpu0-supply binding is present, regulator_get returns
-EPROBE_DEFER, we retry resulting in cpufreq-cpu0 driver registering
later once the regulator is available.
c) cpu0-supply binding is present, regulator_get returns
-EPROBE_DEFER, however, regulator never registers, we retry until
cpufreq-cpu0 driver fails to register pointing at device tree
information bug. However, in this case, the fact that cpufreq-cpu0
operates with clock only when the DT binding clearly indicates need of
a supply is a bug of it's own.
d) cpu0-supply gets an regulator at probe - cpufreq-cpu0 driver
controls both the clock and regulator

Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Acked-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Nishanth Menon <nm@ti.com>
10 years agoregulator: core: return err value for regulator_get if there is no DT binding
Nishanth Menon [Tue, 16 Apr 2013 21:45:16 +0000 (16:45 -0500)]
regulator: core: return err value for regulator_get if there is no DT binding

commit 1e4b545cdd93318379c6b1fb0a99536fa3260f53 upstream.

commit 6d191a5fc7a969d972f1681e1c23781aecb06a61
(regulator: core: Don't defer probe if there's no DT binding for a supply)

Attempted to differentiate between regulator_get() with an actual
DT binding for the supply and when there is none to avoid unnecessary
deferal.
However, ret value supplied by regulator_dev_lookup() is being
ignored by regulator_get(). So, exit with the appropriate return value.

Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
10 years agoARM/dts: dra7-evm: Enable tps659038 on i2c1 bus
J Keerthy [Mon, 27 May 2013 11:18:13 +0000 (16:48 +0530)]
ARM/dts: dra7-evm: Enable tps659038 on i2c1 bus

Enable tps659038 on i2c1 bus.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoarm: dts: dra7: add I2C devices to dra7 DeviceTree file
Sourav Poddar [Fri, 25 Jan 2013 14:49:49 +0000 (16:49 +0200)]
arm: dts: dra7: add I2C devices to dra7 DeviceTree file

Add all 5 I2C instances to dra7.dtsi file.
Also populate the i2c nodes in dra board file.

Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoarm: omap2plus_defconfig: enable DRA752 thermal support by default
Eduardo Valentin [Wed, 24 Apr 2013 16:18:14 +0000 (12:18 -0400)]
arm: omap2plus_defconfig: enable DRA752 thermal support by default

Make DRA752 thermal support enabled on omap2plus_defconfig

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agoarm: mach-omap2: flag DRA7 as having bandgap
Eduardo Valentin [Wed, 24 Apr 2013 16:18:13 +0000 (12:18 -0400)]
arm: mach-omap2: flag DRA7 as having bandgap

Make sure bandgap feature is available for building DRA7 chips
support.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agoarm: dts: add bandgap entry for DRA752 chips
Eduardo Valentin [Wed, 24 Apr 2013 16:18:12 +0000 (12:18 -0400)]
arm: dts: add bandgap entry for DRA752 chips

Add bandgap DT entry.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agostaging: ti-soc-thermal: add DT example for DRA752 chip
Eduardo Valentin [Wed, 24 Apr 2013 16:18:11 +0000 (12:18 -0400)]
staging: ti-soc-thermal: add DT example for DRA752 chip

Update documentation by adding an example for DRA752 on DT description.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agostaging: ti-soc-thermal: remove external heat while extrapolating hotspot
Eduardo Valentin [Tue, 23 Apr 2013 14:48:12 +0000 (10:48 -0400)]
staging: ti-soc-thermal: remove external heat while extrapolating hotspot

For boards that provide a PCB sensor close to SoC junction
temperature, it is possible to remove the cumulative heat
reported by the SoC temperature sensor.

This patch changes the extrapolation computation to consider
an external sensor in the extrapolation equations.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agothermal: expose thermal_zone_get_temp API
Eduardo Valentin [Tue, 23 Apr 2013 14:48:11 +0000 (10:48 -0400)]
thermal: expose thermal_zone_get_temp API

This patch exports the thermal_zone_get_temp API so that driver
writers can fetch temperature of thermal zones managed by other
drivers.

Acked-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Durgadoss R <durgadoss.r@intel.com>
Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agothermal: introduce thermal_zone_get_zone_by_name helper function
Eduardo Valentin [Tue, 23 Apr 2013 14:48:10 +0000 (10:48 -0400)]
thermal: introduce thermal_zone_get_zone_by_name helper function

This patch adds a helper function to get a reference of
a thermal zone, based on the zone type name.

It will perform a zone name lookup and return a reference
to a thermal zone device that matches the name requested.
In case the zone is not found or when several zones match
same name or if the required parameters are invalid, it will return
the corresponding error code (ERR_PTR).

Acked-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Durgadoss R <durgadoss.r@intel.com>
Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agostaging: ti-soc-thermal: add dra752 chip to device table
Eduardo Valentin [Tue, 23 Apr 2013 14:52:16 +0000 (10:52 -0400)]
staging: ti-soc-thermal: add dra752 chip to device table

Add support to TI dra752 chips by adapting the driver
device table.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agostaging: ti-soc-thermal: add thermal data for DRA752 chips
Eduardo Valentin [Wed, 10 Apr 2013 21:12:05 +0000 (17:12 -0400)]
staging: ti-soc-thermal: add thermal data for DRA752 chips

This patch adds the thermal data for TI DRA752 chips.
In this change it includes (autogen):
. Register offset definitions
. Bitfields and masks for all registers
. Conversion table

Also, the thermal limits, thresholds and extrapolation
rules are included. The extrapolation rule is simply
add +2C as margin.

All 5 sensors, MPU, GPU, CORE, DSPEVE and IVA, are defined
and exposed. Only MPU has cooling device.

Signed-off-by: Eduardo Valentin <eduardo.valentin@ti.com>
10 years agoregulator: introduce TI AVS class 0 driver
Nishanth Menon [Thu, 18 Apr 2013 15:05:35 +0000 (10:05 -0500)]
regulator: introduce TI AVS class 0 driver

OMAP5 technology based processors such as DRA7XX[1] has an simplified AVS
strategy called SmartReflex[2] Class 0. Texas Instrument's nomenclature for
this is: Class - 0: Production Test Calibration

This, in it's simplistic form implies that the voltage value for each OPP
is now encoded into an efuse location.

On entering an OPP, the voltage value to be selected is no longer the
traditional nominal voltage, but the voltage meant from the efuse offset
encoded in millivolts. Each device will have it's own unique voltage for a
given OPP. Therefore, it is not possible to encode a range of voltage
representing an OPP voltage.

DRA processors may be powered using various PMICs - I2C based ones such
as TPS659039 or SPI / GPIO controlled ones as well.

cpufreq/devfreq driver which controls voltage and frequency pairs traditionally
used:
cpufreq/devfreq --> PMIC regulator
                \-> clock framework
This opens up a few issues:
a) PMIC regulator is designed for platforms that maynot use SmartReflex class 0
   based SoCs, encoding the efuse offsets into every possible PMIC regulator
   driver is practically in-efficient.
b) Voltage values are not known a-priori to be encoded into DTB as they are
   device specific.

Another option might have been to start off with the nominal voltages in
DT and then fixup the voltage entries in the OPP table by reading the
eFuse information.

However, such an approach introduces formidable constraints:
A) OPP tables once added are not modifiable beyond enable/disable - that
was the entire design constraint we wanted on OPP tables
B) We need some indexing value for uniquely identifying what to do for
an given OPP voltage - if OPP voltage by itself changes, how do we index
various subsystems? - read abb/avs class 2 based class 1.
C) We have n voltage domains and m devices per domain. - we no longer
want to maintain an centralized OPP table.
D) We also need to support multiple platforms which have "ganged voltage
rails" each domain with variable voltages.
E) Modification of the OPP data itself is hard - considering that we do
not have a centralized OPP repository, architecture independent drivers
like cpufreq-cpu0 driver does the "adding of OPPs".

Options:
a) modify arch generic drivers such as cpufreq-cpu0 driver with an OMAP
specific an efuse offset and OMAP specific handling is a bad idea.
b) we are able to overlay OPP table entried *before* cpufreq driver
comes active
- I am not sure we have ability to do that in kernel yet
- I am not even sure we can modify the dtb being loaded by all
bootloaders we may have (thinking beyond u-boot).

To simplify this, we introduce:
cpufreq/devfreq --> SmartReflex Class 0 regulator --> PMIC regulator
                \-> clock framework

Class 0 Regulator has information of translating the "nominal voltage" into
voltage value stored in efuse offset.
Example encoding:
uVolts mVolt  --> stored as 16 bit hex value of mV
975000 975 --> 0x03CF
1075000 1075 --> 0x0433
1200000 1200 --> 0x04B0

[1] http://www.ti.com/lit/ds/sprt659/sprt659.pdf
[2] http://www.ti.com/lit/wp/swpy015a/swpy015a.pdf

Signed-off-by: Nishanth Menon <nm@ti.com>
10 years agoregulator: introduce regulator chain locking scheme
Grygorii Strashko [Mon, 8 Apr 2013 15:55:04 +0000 (18:55 +0300)]
regulator: introduce regulator chain locking scheme

In some cases the regulators may be organized in a chain like:
  ->regA->regB->regC
where regA is supplier for regB and regB is supplier for regC.

Currently it would be possible to reconfigure regA and regC at same time
form different contexts, because each regulator has it own mutex.
But in some cases, the only the whole chain is allowed be reconfigured
because of dependencies between regulators - to change regB
configuration the regA need to be updated first.

Hence, introduce regulator chain locking scheme to lock whole Regulator
chain in case if any part of it has been accessed from outside. To
achieve this goal the root Regulator (which has no supply defined, like
regA) in chain is used to protect the whole chain.

In addition, such locking scheme allows to have access to the supplier
regulator API from inside child's (consumer) regulator API.

Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
10 years agoregulators: abstract locking out into helper functions
Grygorii Strashko [Mon, 8 Apr 2013 15:55:03 +0000 (18:55 +0300)]
regulators: abstract locking out into helper functions

Create locking helpers for the regulator mutex. These helpers will be
needed for the next patch which introduces Regulator chain locking
scheme.

Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
10 years agortc: palmas: Enable device tree support for Palmas RTC.
J Keerthy [Wed, 24 Apr 2013 06:20:57 +0000 (11:50 +0530)]
rtc: palmas: Enable device tree support for Palmas RTC.

Enable device tree support for Palmas RTC.
The Documentation is updated here:
Documentation/devicetree/bindings/rtc/palmas-rtc.txt

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agortc: palmas: Add RTC driver Palmas series PMIC
Laxman Dewangan [Wed, 24 Apr 2013 06:20:56 +0000 (11:50 +0530)]
rtc: palmas: Add RTC driver Palmas series PMIC

TI Palmas series PMIC support the RTC and alarm functionality.
Add RTC driver with alarm support for this device.

This patch is picked from the mainline kernel.
https://patchwork.kernel.org/patch/1926921/
Commit id: 0101e53cbb1c653738de8dfddc80c0faa5a4fd39

Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
[j-keerthy@ti.com: backport from upstream 0101e53cbb1c653738de8dfddc80c0faa5a4fd39]
Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agortc: Revert rtc-palmas driver for palmas PMIC
J Keerthy [Wed, 24 Apr 2013 06:20:55 +0000 (11:50 +0530)]
rtc: Revert rtc-palmas driver for palmas PMIC

This reverts commit 5e3935c6e9d8d1bd05d39fde55b913f85a1b42a4.
(rtc: rtc-palmas a new driver for palmas PMIC). Removing
the intermediate version of the driver to absorb the upstreamed
driver.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agomfd: palmas: Add rtc irq number as irq resource for palmas-rtc
Laxman Dewangan [Wed, 24 Apr 2013 06:20:54 +0000 (11:50 +0530)]
mfd: palmas: Add rtc irq number as irq resource for palmas-rtc

Palma RTC is capable of generating alarm interrupt. Pass the alarm interrupt
as IRQ_RESOURCE for palmas-rtc sub device driver so that rtc driver can get
irq as platform_get_irq().

Also pass the irq domain in mfd_add_devices() to properly offset the irqs for
sub devices. This is needed when adding device through DT.

This patch is absorbed from the Mainline kernel:
http://lkml.indiana.edu/hypermail/linux/kernel/1301.0/00559.html
Commit id: a36516b016f52f0f6e5284025e3487f63b4be33b

Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
[j-keerthy@ti.com: backport from upstream a36516b016f52f0f6e5284025e3487f63b4be33b]
Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agomfd: palmas: Add APIs to access the Palmas' registers
Laxman Dewangan [Wed, 24 Apr 2013 06:20:53 +0000 (11:50 +0530)]
mfd: palmas: Add APIs to access the Palmas' registers

Palmas register set is divided into different blocks (base and offset)
and hence different i2c addresses. The i2c address offsets are derived
from base address of block of registers.

Add inline APIs to access the Palma's registers which takes the base of
register block and register offset. The i2c address offset is  derived
from the base address of register blocks.

This patch is picked up from Mainline kernel:
https://patchwork.kernel.org/patch/1926911/
Commit id: 60c185f059c88ad4b9b170b1f9322e3adcccca07

Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
[j-keerthy@ti.com: backport from upstream 60c185f059c88ad4b9b170b1f9322e3adcccca07]
Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoARM/dts: add dtsi file for TPS659038 device
J Keerthy [Mon, 27 May 2013 11:01:58 +0000 (16:31 +0530)]
ARM/dts: add dtsi file for TPS659038 device

Add dtsi file for TPS659038 device.

The Documents referred:
1) TPS659039-Q1 POWER MANAGEMENT UNIT (PMU) FOR PROCESSOR Data Manual
2) TPS659038/39-Q1 Functional Register Descriptions.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agoregulators: palmas: Add TPS659038 support
J Keerthy [Wed, 24 Apr 2013 06:20:51 +0000 (11:50 +0530)]
regulators: palmas: Add TPS659038 support

Add TPS659038 support. The TPS659038 belongs to PALMAS family.
The only difference between TPS659038 and TWL6035 w.r.t
regulators is the SMPS10 is not present in TPS659038.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agomfd: Palmas: Add TPS659038 PMIC support
J Keerthy [Mon, 27 May 2013 10:55:00 +0000 (16:25 +0530)]
mfd: Palmas: Add TPS659038 PMIC support

The Patch adds TPS659038 PMIC support in the palmas mfd driver.
The TPS659038 has almost the same registers as of the earlier
supported variants of PALMAS family such as the TWL6035.

The critical differences between TPS659038 and TWL6035 being:

1) TPS659038 has nothing related to battery charging and back up battery stuff.
2) TPS659038 does not have does not have SMPS10(Boost) step up convertor.
3) TPS659038 does not have Battery detection and anything related to battery.
4) TPS659038 has something more than Palmas in one case it is called
   Sync clock Functionality.
5) can use an external crystal unit or 16MHz oscillator to generate clock.
6) SD card detection, Battery presence detection, Vibrator, USB OTG are missing
   when compared to TWL6035.

Hence some of the IRQs are reserved in case of TPS659038 but the
remaining are bit exact compared to TWL6035.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agomfd: Palmas: Introduce palmas_pmic_data structure
J Keerthy [Wed, 24 Apr 2013 06:20:49 +0000 (11:50 +0530)]
mfd: Palmas: Introduce palmas_pmic_data structure

Introduce palmas_pmic_data structure so that Palmas variants
like TWL6035, TWL6037, TPS65913 can populate data specific
to them. This enables driver to support variants.

Signed-off-by: J Keerthy <j-keerthy@ti.com>
10 years agomfd: DT bindings for the palmas family MFD (v10)
Graeme Gregory [Wed, 24 Apr 2013 06:20:48 +0000 (11:50 +0530)]
mfd: DT bindings for the palmas family MFD (v10)

Add the various binding files for the palmas family of chips.
There is a top level MFD binding then a seperate binding for IP blocks on chips.

This patch is picked up from v10 of the palmas patch set:
https://patchwork.kernel.org/patch/2321171/

Signed-off-by: Graeme Gregory <gg@slimlogic.co.uk>
[j-keerthy@ti.com: backport from v10 of the palmas patch set:
https://patchwork.kernel.org/patch/2321171/]
Signed-off-by: J Keerthy <j-keerthy@ti.com>
Signed-off-by: Ian Lartey <ian@slimlogic.co.uk>
10 years agoMerge remote-tracking branch 'rnayak/platform-base-vayu-3.8.y' into vayu-pm-linux...
Tero Kristo [Mon, 27 May 2013 14:06:42 +0000 (17:06 +0300)]
Merge remote-tracking branch 'rnayak/platform-base-vayu-3.8.y' into vayu-pm-linux-3.8.y

Conflicts:
arch/arm/mach-omap2/clockdomain.h
arch/arm/mach-omap2/control.h
arch/arm/mach-omap2/omap-hotplug.c
arch/arm/mach-omap2/omap-mpuss-lowpower.c
arch/arm/mach-omap2/pm_omap4plus.c

Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: Prevent any low power entry in suspend
Rajendra Nayak [Mon, 27 May 2013 10:16:44 +0000 (15:46 +0530)]
ARM: DRA7: Prevent any low power entry in suspend

On DRA7 we do not support low power for MPU in either suspend or
cpuidle.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: Add the build support
R Sricharan [Thu, 7 Feb 2013 10:51:46 +0000 (16:21 +0530)]
ARM: DRA7: Add the build support

Adding the build support required for DRA7xx soc
in to omap2+ config.

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoARM: DRA7: Add PRM dev_inst module offset in global_warm_reset function
R Sricharan [Fri, 19 Apr 2013 09:13:15 +0000 (14:43 +0530)]
ARM: DRA7: Add PRM dev_inst module offset in global_warm_reset function

Currently omap4_prminst_global_warm_reset function is invoked by
OMAP4/5 socs upon a system reboot command. Add the right PRM module
device_inst offset for DRA_7xx devices, to reuse it for
DRA7xx as well.

Signed-off-by: Sricharan R <r.sricharan@ti.com>
10 years agoARM: DRA7: Enable framework initializations
Ambresh K [Wed, 27 Mar 2013 06:15:09 +0000 (11:45 +0530)]
ARM: DRA7: Enable framework initializations

Initialise the power, HWMOD, clock domains and the clock tree.

Signed-off-by: Ambresh K <ambresh@ti.com>
10 years agoARM: DRA7: hwmod data: Create initial DRA7XX SOC hwmod data
Ambresh K [Tue, 26 Mar 2013 09:09:34 +0000 (14:39 +0530)]
ARM: DRA7: hwmod data: Create initial DRA7XX SOC hwmod data

Adding the hwmod data for DRA7XX platforms.

Signed-off-by: Ambresh K <ambresh@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: clock data: Add DRA7XX full clock tree and headers
Ambresh K [Tue, 12 Feb 2013 13:42:54 +0000 (19:12 +0530)]
ARM: DRA7: clock data: Add DRA7XX full clock tree and headers

Add the clock tree related data for DRA7XX platforms.

Signed-off-by: Ambresh K <ambresh@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: powerdomain data: Add DRA7XX data and update the header
Ambresh K [Tue, 26 Mar 2013 08:03:55 +0000 (13:33 +0530)]
ARM: DRA7: powerdomain data: Add DRA7XX data and update the header

Add the data file to describe all power domains inside the DRA7XX soc.

Signed-off-by: Ambresh K <ambresh@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: clockdomain data: Add DRA7XX data and update the header
Ambresh K [Tue, 26 Mar 2013 07:59:02 +0000 (13:29 +0530)]
ARM: DRA7: clockdomain data: Add DRA7XX data and update the header

Add the data file to describe all clock domains inside the DRA7XX SOC

Signed-off-by: Ambresh K <ambresh@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: PRCM: Add DRA7XX local MPU PRCM regsiters
Ambresh K [Mon, 25 Feb 2013 13:11:21 +0000 (18:41 +0530)]
ARM: DRA7: PRCM: Add DRA7XX local MPU PRCM regsiters

Add the PRCM MPU registers for DRA7XX platforms

Signed-off-by: Ambresh K <ambresh@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
[rnayak@ti.com: Added PRCM partition information]
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: CM: Add DRA7XX register and bitfield files
Ambresh K [Sat, 13 Apr 2013 12:47:03 +0000 (18:17 +0530)]
ARM: DRA7: CM: Add DRA7XX register and bitfield files

Add the new defines for DRA7XX CM registers.

Signed-off-by: Ambresh K <ambresh@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
10 years agoARM: DRA7: PRM: Add DRA7XX register and bitfield files
Ambresh K [Sat, 13 Apr 2013 12:43:55 +0000 (18:13 +0530)]
ARM: DRA7: PRM: Add DRA7XX register and bitfield files

Add the new defines for DRA7xx prm module registers.

Signed-off-by: Ambresh K <ambresh@ti.com>
10 years agoARM: DRA7: dts: Add the dts files for dra7 SoC and dra7-evm board
R Sricharan [Thu, 7 Feb 2013 10:44:00 +0000 (16:14 +0530)]
ARM: DRA7: dts: Add the dts files for dra7 SoC and dra7-evm board

The DRA7xx is a high-performance, infotainment application
device, based on enhanced OMAP architecture integrated on a
28-nm technology.

DRA75x, DRA74x belong to the DRA7xx family.
DRA75x and DRA74x are dual core CORTEX A15 SOCs with GIC used
for interrupt handling and with an integrated L2 cache
controller.

Adding the minimum device tree files required for
DRA7xx based socs.

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoARM: DRA7: PM: Avoid all SAR save on dra7
Rajendra Nayak [Fri, 3 May 2013 10:04:40 +0000 (15:34 +0530)]
ARM: DRA7: PM: Avoid all SAR save on dra7

Get rid of all assumptions about always having a sar base on *all*
omap4+ platforms. We dont need one on DRA7.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: Add static dep from EMIF to MPU
Rajendra Nayak [Fri, 24 May 2013 14:26:53 +0000 (19:56 +0530)]
ARM: DRA7: Add static dep from EMIF to MPU

Enable the static dependency between EMIF and MPU.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: OMAP2+: Powerdomains: Remove the hard requirement to always have a voltdm assoca...
Rajendra Nayak [Thu, 25 Apr 2013 06:36:25 +0000 (12:06 +0530)]
ARM: OMAP2+: Powerdomains: Remove the hard requirement to always have a voltdm assocaited to a pwrdm

The powerdomain framework expects all powerdomains to be asoociated with
a corresponding voltage domain. For some SoCs (like the DRA7 family) which
does not have a Voltage controller/Voltage Processor (neither the SR I2C
bus to communicate with the PMIC) there is no need for a Powerdomain to have
a voltage domain association (since they are really non scalable, even
though the volatge domains exist in place).

Extend the arch operations to add an api which the powerdomain core can
then use to identify if a voltdm lookup and association for a powerdomain
is really needed.

Signed-off-by: Rajendra Nayak <rnayak@ti.com>
10 years agoARM: DRA7: Kconfig: Increase the default gpio count
R Sricharan [Thu, 7 Feb 2013 12:13:35 +0000 (17:43 +0530)]
ARM: DRA7: Kconfig: Increase the default gpio count

DRA7xx has 8 GPIO banks so that there are 32x8 = 256 GPIOs.
In order for the gpiolib to detect and initialize these
additional GPIOs and other TWL GPIOs, ARCH_NR_GPIO is set
to 512 instead of present 256.

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoARM: DRA7: board-generic: Add device tree support
R Sricharan [Thu, 7 Feb 2013 08:23:05 +0000 (13:53 +0530)]
ARM: DRA7: board-generic: Add device tree support

Adding the minimal support for DRA7xx soc with
device tree.

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoARM: DRA7: timer: Add clocksource, clockevent support
R Sricharan [Thu, 7 Feb 2013 07:55:39 +0000 (13:25 +0530)]
ARM: DRA7: timer: Add clocksource, clockevent support

Reusing all the OMAP5 timer code here.
So just adding SOC_DRA7XX check to enable on single
platform build.

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoARM: DRA7: Add minimal support for DRA7xx SOC
R Sricharan [Thu, 7 Feb 2013 04:06:27 +0000 (09:36 +0530)]
ARM: DRA7: Add minimal support for DRA7xx SOC

The DRA7xx is a high-performance, infotainment application
device, based on enhanced OMAP architecture integrated on a
28-nm technology.

The architecture is designed for advanced graphical HMI and
Navigation, Digital and Analog Radio, Rear Seat Entertainment
and Multimedia playback, providing Advanced Driver Assistance
integration capabilities with Video analytics support, and
best-in-class CPU performance, video, image, and graphics
processing.

DRA75x, DRA74x belong to the DRA7xx family.
DRA75x and DRA74x are dual core CORTEX A15 SOCs with GIC used
for interrupt handling and with an integrated L2 cache
controller.

In this patch:
  Most of the minimal support code is reused from OMAP5.
  - A new early init function is added.
  - The machine specific header is updated for base address
    changes.
  - Makefile updates
  - hwmod file is updated to include the common functions.
  - Updated the sram size

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoARM: DRA7: id: Add cpu detection support for DRA7xx based socs
R Sricharan [Wed, 6 Feb 2013 14:55:40 +0000 (20:25 +0530)]
ARM: DRA7: id: Add cpu detection support for DRA7xx based socs

The DRA7xx is a high-performance, infotainment application device,
based on enhanced OMAP architecture. DRA7xx family is composed
of DRA75x and DRA74x devices.

Adding the DRA752 ES1.0 cpu revision detection support.

Signed-off-by: R Sricharan <r.sricharan@ti.com>
10 years agoarm: omap2+: omap_device: remove no_idle_on_suspend
Sourav Poddar [Wed, 13 Mar 2013 15:02:36 +0000 (20:32 +0530)]
arm: omap2+: omap_device: remove no_idle_on_suspend

Remove "no_idle_on_suspend" check, since respective
driver should be able to prevent idling of an
omap device whenever required.

Driver's can get same behavior by just returning -EBUSY
from their ->runtime_suspend only during suspend.

Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Rajendra nayak <rnayak@ti.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
10 years agoarm: omap2+: serial: remove no_console_suspend support
Sourav Poddar [Wed, 13 Mar 2013 15:02:36 +0000 (20:32 +0530)]
arm: omap2+: serial: remove no_console_suspend support

"no_console_suspend" is no longer handled in platform file,
Since the omap serial driver is now adapted to prevent
console UART idleing during suspend.

Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Rajendra nayak <rnayak@ti.com>
Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
10 years agodriver: serial: omap: prevent runtime PM for "no_console_suspend"
Sourav Poddar [Wed, 13 Mar 2013 15:02:36 +0000 (20:32 +0530)]
driver: serial: omap: prevent runtime PM for "no_console_suspend"

The driver manages "no_console_suspend" by preventing runtime PM
during the suspend path, which forces the console UART to stay awake.

Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
10 years agodriver: tty: serial: Move "uart_console" def to core header file.
Sourav Poddar [Tue, 16 Apr 2013 13:05:03 +0000 (18:35 +0530)]
driver: tty: serial: Move "uart_console" def to core header file.

Move "uart_console" definition to serial core header file, so that it can be
used by serial drivers.
Get rid of the uart_console defintion from mpc52xx_uart driver.

Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Rajendra nayak <rnayak@ti.com>
Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
10 years agoMerge branch 'platform-base-3.8.y' of git://github.com/hvaibhav/am335x-linux into...
Rajendra Nayak [Fri, 26 Apr 2013 08:43:40 +0000 (14:13 +0530)]
Merge branch 'platform-base-3.8.y' of git://github.com/hvaibhav/am335x-linux into platform-base-3.8.y

10 years agoARM: dts: AM33XX: Set pinmux for clkout2 pad used for clock output
Vaibhav Hiremath [Mon, 22 Apr 2013 11:00:12 +0000 (16:30 +0530)]
ARM: dts: AM33XX: Set pinmux for clkout2 pad used for clock output

xdma_event_intr1.clkout2 pad can be used to source clock
from either 32K OSC or any of the PLL (except MPU) outputs.
On the existing AM335x based boards (EVM, EVM-SK and Bone),
this pad is used to feed the clock to audio codes.

So, this patch configures the pinmux to get clkout2 on the pad.

Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
10 years agoARM: OMAP AM33XX: clock data: Enable clkout2 as part of init
Vaibhav Hiremath [Mon, 22 Apr 2013 10:55:24 +0000 (16:25 +0530)]
ARM: OMAP AM33XX: clock data: Enable clkout2 as part of init

clkout2 comes out on the pad and is being used by various
external on-board peripherals like, Audio codecs and stuff.
So enable the clkout2 by default during init sequence itself.

Also, add the missing entry of "clkout2_ck" to the clock table.

Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
10 years agoARM: dts: AM33XX: Add support for AM335x BeagleBone-Black
Vaibhav Hiremath [Mon, 22 Apr 2013 10:22:05 +0000 (15:52 +0530)]
ARM: dts: AM33XX: Add support for AM335x BeagleBone-Black

Add DT support for AM335X based (>PG2.x) BeagleBone-Black
board (am335x-boneblack.dts), it is different than original
Bone in terms of on-board peripherals -

- On board HDMI A/V support
- On board eMMC support
- DDR3 support (4Gbit)
- Removal of FTDI chip - So no on board JTAG

This patch adds support all the peripherals reused and
supported on older Bone platforms.

Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
10 years agoomapdss: hdmi: select I2C_ALGOBIT in Kconfig
Archit Taneja [Thu, 25 Apr 2013 12:46:57 +0000 (18:16 +0530)]
omapdss: hdmi: select I2C_ALGOBIT in Kconfig

The hdmi driver uses i2c bitbanging to read edid for OMAP5. The core hdmi
driver(hdmi.c) adds an i2c adpater based on the i2c bitbanging driver available
in the kernel.

This CONFIG should be selected only for OMAP5 HDMI, but since the adapter is
added in hdmi.c, it's select for both OMAP4 and OMAP5 HDMI for now. This can
be clenaed up later.

This prevents a build break when DRM is not selected, DRM core selects
I2C_ALGOBIT by default, so the compile error isn't noticed when omapdrm is
built.

Signed-off-by: Archit Taneja <archit@ti.com>
10 years agoOMAPDSS: HDMI: do not set HDMI_TXPHY_POWER_CTRL for OMAP5
Fairuz WAN ISMAIL [Tue, 23 Apr 2013 09:04:55 +0000 (14:34 +0530)]
OMAPDSS: HDMI: do not set HDMI_TXPHY_POWER_CTRL for OMAP5

Do not set register field HDMI_TXPHY_POWER_CTRL.LDOVOLTAGE for OMAP5 as the bits
are reserved. Setting these bits lead to the PHY power state reporting as not
transitioning.

Signed-off-by: Fairuz WAN ISMAIL <f-wanismail@ti.com>
Signed-off-by: Archit Taneja <archit@ti.com>