If CONFIG_BUG is not enabled, BUG() does not stop the execution. Many
places in code expect the execution to stop, and this causes compiler
warnings about uninitialized variables and returning from a non-void
function without a return value.
This patch fixes the warnings by initializing the variables and
returning properly after BUG() lines. However, the behaviour is still
undefined after the BUG, but this is the choice the user makes when
using CONFIG_BUG=n.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Commit 05dd0f5308 ("OMAPDSS: DISPC: Update
Accumulator configuration for chroma plane") adds
dispc_ovl_set_accu_uv() function that sets the accu, but the function
only handles YUV and NV12 modes, and BUGs otherwise.
The patch also adds a call to the function, but unfortunately the place
of call was such that the mode could be other than YUV or NV12, thus
crashing the driver.
This patchs moves the call to a slightly later spot, at which point only
YUV and NV12 modes are handled.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: Chandrabhanu Mahapatra <cmahapatra@ti.com>
There is a problem related to DSS FIFO thresholds and power management
on OMAP3. It seems that when the full PM hits in, we get underflows. The
core reason is unknown, but after experiments it looks like only
particular FIFO thresholds work correctly.
This bug is related to an earlier patch, which added special FIFO
threshold configuration for OMAP3, because DSI command mode output
didn't work with the normal threshold configuration.
However, as the above work-around worked fine for other output types
also, we currently always configure thresholds in this special way on
OMAP3. In theory there should be negligible difference with this special
way and the standard way. The first paragraph explains what happens in
practice.
This patch changes the driver to use the special threshold configuration
only when the output is a manual update display on OMAP3. This does
include RFBI displays also, and although it hasn't been tested (no
boards using RFBI) I suspect the similar behaviour is present there
also, as the DISPC side should work similarly for DSI command mode and
RFBI.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: Joe Woodward <jw@terrafix.co.uk>
When using the MERAM the LCDC line size needs to be programmed with a
MERAM-specific value different than the real frame buffer pitch. Fix it.
Reported-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: stable@vger.kernel.org # for 3.4
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
DSI supports interleaving of command mode packets during the HSA, HFP, HBP and
BLLP blanking intervals in a video mode stream. This is useful as a user may
want to read or change the configuration of a panel without stopping the video
stream.
On OMAP DSI, we can queue HS or LP command mode packets in the TX FIFO, and
the DSI HW takes care of interleaving this data during the one of the blanking
intervals. The DSI HW needs to be programmed with the maximum amount of data
that can be interleaved in a particular blanking period. A blanking period
cannot be used to send command mode data for it's complete duration, there is
some amount of time required for the DSI data and clock lanes to transition
to the desired LP or HS state.
Based on the state of the lanes at the beginning and end of the blanking period,
we have different scenarios, with each scenario having a different value of time
required to transition to HS or LP. Refer to the section 'Interleaving Mode' in
OMAP TRM for more info on the scenarios and the equations to calculate the time
required for HS or LP transitions.
We use the scenarios which takes the maximum time for HS or LP transition, this
gives us the minimum amount of time that can be used to interleave command mode
data. The amount of data that can be sent during this minimum time is calculated
for command mode packets both in LP and HS. These are written to the registers
DSI_VM_TIMING4 to DSI_VM_TIMING6.
The calculations don't take into account the time required of transmitting BTA
when doing a DSI read, or verifying if a DSI write went through correctly. Until
these latencies aren't considered, the behaviour of DSI is unpredictable when
a BTA is interleaved during a blanking period. Enhancement of these calculations
is a TODO item.
The calculations are derived from DSI parameter calculation tools written by
Sebastien Fagard <s-fagard@ti.com>
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
DISPC has two accumulator registers DISPC_VIDp_ACCU_0 and DISPC_VIDp_ACCU_1 each
with horizontal and vertical bit fields. The bit fields can take values in the
range of -1024 to 1023. Based on bit field values DISPC decides on which one out
of 8 phases the filtering starts. DISPC_VIDp_ACCU_0 is used for progressive
output and for interlaced output both DISPC_VIDp_ACCU_0 and DISPC_VIDp_ACCU_1
are used.
The current accumulator values in DISPC scaling logic for chroma plane takes
default values for all color modes and rotation types. So, the horizontal and
vertical up and downsampling accumulator bit field values have been updated for
better performance.
Signed-off-by: Chandrabhanu Mahapatra <cmahapatra@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Conflicts:
drivers/mmc/host/sdhci-esdhc-imx.c
drivers/net/ethernet/freescale/fec.c
drivers/spi/spi-imx.c
drivers/tty/serial/imx.c
This resolves dependencies between the pinctrl and clock changes
in imx.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
The THRESHOLDS register configures thresholds for two interrupts, but
these interrupts are not used in the DIU driver. An early version of the
driver may have used the "lines before vsync" interrupt, which requires
the LS_BF_VS of THRESHOLDS to be initialized.
Unfortunately, the initialization of this register does not do a
read-modify-write to set only LS_BF_VS. On the MPC8610, the value
written is correct. On other chips, like the P1022, the value overwrites
some reserved bits. This results in a performance drop on the P1022.
Since the default value is acceptable as-is on all SOCs, we should just
avoid touching this register.
Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Rework the interrupt handler so the RX_DONE, FIFO_EMPTY interrupts are
properly detected. This prevents missing the interrupts when there are
other bits set in the INTSRC register than just RX_DONE and FIFO_EMPTY.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Donghwa Lee <dh09.lee@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Make sure all resources are initialized before interrupt handler is
registered. Pass full platform device name to request_irq() so it
can be distinguished which device has requested an interrupt in cases
there are multiple instances in the system.
Also enable voltage regulators regardless of they have been enabled
by bootloader or not, to make sure other drivers using same regulators
don't disable them unexpectedly.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Donghwa Lee <dh09.lee@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Use proper PM ops from struct dev_pm_ops rather than the deprecated ones.
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Donghwa Lee <dh09.lee@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Latest SuperH HDMI uses not only HDMI Core Register (HTOP0)
but also HDMI Control Register (HTOP1).
This patch adds HDMI Control Register support.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Latest SuperH HDMI allows 32bit access only.
But the data is 8bit. So, we can keep compatibility by switching 8/32 bit access.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
This patch uses pr_debug instead of printk to allow dynamic debugging.
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Remove unnecessary headers from the file.
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Release bridge, as done on all other paths out of the function.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Pull networking fixes from David S. Miller:
1) Since we do RCU lookups on ipv4 FIB entries, we have to test if the
entry is dead before returning it to our caller.
2) openvswitch locking and packet validation fixes from Ansis Atteka,
Jesse Gross, and Pravin B Shelar.
3) Fix PM resume locking in IGB driver, from Benjamin Poirier.
4) Fix VLAN header handling in vhost-net and macvtap, from Basil Gor.
5) Revert a bogus network namespace isolation change that was causing
regressions on S390 networking devices.
6) If bonding decides to process and handle a LACPDU frame, we
shouldn't bump the rx_dropped counter. From Jiri Bohac.
7) Fix mis-calculation of available TX space in r8169 driver when doing
TSO, which can lead to crashes and/or hung device. From Julien
Ducourthial.
8) SCTP does not validate cached routes properly in all cases, from
Nicolas Dichtel.
9) Link status interrupt needs to be handled in ks8851 driver, from
Stephen Boyd.
10) Use capable(), not cap_raised(), in connector/userns netlink code.
From Eric W. Biederman via Andrew Morton.
11) Fix pktgen OOPS on module unload, from Eric Dumazet.
12) iwlwifi under-estimates SKB truesizes, also from Eric Dumazet.
13) Cure division by zero in SFC driver, from Ben Hutchings.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (26 commits)
ks8851: Update link status during link change interrupt
macvtap: restore vlan header on user read
vhost-net: fix handle_rx buffer size
bonding: don't increase rx_dropped after processing LACPDUs
connector/userns: replace netlink uses of cap_raised() with capable()
sctp: check cached dst before using it
pktgen: fix crash at module unload
Revert "net: maintain namespace isolation between vlan and real device"
ehea: fix losing of NEQ events when one event occurred early
igb: fix rtnl race in PM resume path
ipv4: Do not use dead fib_info entries.
r8169: fix unsigned int wraparound with TSO
sfc: Fix division by zero when using one RX channel and no SR-IOV
openvswitch: Validation of IPv6 set port action uses IPv4 header
net: compare_ether_addr[_64bits]() has no ordering
cdc_ether: Ignore bogus union descriptor for RNDIS devices
bnx2x: bug fix when loading after SAN boot
e1000: Silence sparse warnings by correcting type
igb, ixgbe: netdev_tx_reset_queue incorrectly called from tx init path
openvswitch: Release rtnl_lock if ovs_vport_cmd_build_info() failed.
...
A single patch from Alexander Shiyan <shc_work@mail.ru>:
* clps711x/cleanup:
ARM: clps711x: Using a single definition for the PHYS and VIRT registers offset
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Using a single definition for the physical and virtual address register for all
variants boards clps711x. This patch also includes the use of a single function
clps_read/write in some units.
Signed-off-by: Alexander Shiyan <shc_work@mail.ru>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Implement the DSS device driver audio support interface in the HDMI
panel driver and generic driver. The implementation relies on the
IP-specific functions that are defined at DSS probe time.
A mixed locking strategy is used. The panel's mutex is used when
the state of the panel is queried as required by the audio functions.
The audio state is protected using a spinlock as users of DSS HDMI
audio functionality might start/stop audio while holding a spinlock.
The mutex and the spinlock are held and released as needed by each
individual function to protect the panel state and the audio state.
Although the panel's audio_start functions does not check whether
the panel is active, the audio _ENABLED state can be reached only
from audio_enable, which does check the state of the panel. Also,
if the panel is ever disabled, the audio state will transition
to _DISABLED. Transitions are always protected by the audio lock.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
As the hdmi_lock mutex is inside the hdmi struct, rename to simply
"lock". This is only a change in the name. There are not changes
in functionality.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
As of today, the only know user of the DSS HDMI audio support is
ASoC. Hence, it makes sense to remap the speaker order to match
the ALSA speaker order. In the future, a dynamic mapping mechanism
may be implemented.
Remapping is needed as the HDMI speaker order is FL/FR/LFE/C/RL/RR/
RLC-FLC/RRC-FLC while the ALSA order is FL/FR/RL/RR/C/LFE/SL/SR.
Refer to CEA-861 Section 6.6.2 for further details.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
The generic HDMI driver does not need to know about the specific
settings of a given IP. Hence, it just passes the audio configuration
and the IP library parses such configuration and sets the IP
accordingly. This patch introduces an IP-specific audio configuration
function.
Also, this patch implements the audio config function for OMAP4. The
DMA, format and core config functions are no longer exposed to the
generic HDMI driver as they are IP-specific.
The audio configuration function caters for 16-bit through 24-bit
audio samples with sample rates from 32kHz and up to 192kHz as well
as up to 8 audio channels.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Add support for more sample rates when calculating N and CTS. This
covers all the audio sample rates that an HDMI source is allowed
to transmit according to the HDMI 1.4a specification.
Also, reorganize the logic for the calculation when using deep color.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
The N and CTS parameters are relevant to all HDMI implementations and
not specific to a given IP. Hence, the calculation is relocated
into the generic HDMI driver.
Also, deep color is not queried but it is still considered in the
calculation of N. This is to be changed when deep color functionality is
implemented in the driver.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Utilize a snd_aes_iec958 struct to write the parameters of the IEC-60958
channel status word into the HDMI IP registers. Hence, the user of the
driver has full control of what parameters are written in the word.
Also, some of the parameters of the I2S structure have been removed
as they are actually IEC-60958 parameters.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Instead of having OMAPDSS HDMI audio functionality depending on the
ASoC HDMI audio driver, use a new config option so that
potential users, including ASoC, may select if needed.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Decouple the enable/disable operation of the HDMI audio wrapper from
audio start/stop. Otherwise, an audio FIFO underflow may occur. The
audio wrapper enablement must be done after configuration and
before audio playback is started.
Signed-off-by: Axel Castaneda Gonzalez <x0055901@ti.com>
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
According to the most up-to-date documentation from Texas Instruments,
the configuration of High Bitrate Audio is not possible. Also, it is
not possible to set polarity of the I2S Word Select signal. This patch
removes the invalid settings.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Instead of having its own definitions for CEA-861 and IEC-60958, the HDMI
driver should use those provided by ALSA. This patch removes the definitions
that are already provided by ALSA.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Remove the ASoC OMAP HDMI audio codec. The goal of removing the codec
is to, in subsequent patches, give way to the implementation of the HDMI
audio support using the DSS device driver audio interface. This
approach will expose the HDMI audio functionality to any interested entity.
In a separate patch, ASoC will use this new approach to expose HDMI audio
to ALSA.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
To improve readability, split the video_enable HDMI IP operation
into two separate functions for enabling and disabling video.
The video_enable function is also modified to return an error value.
While there, update these operations for the OMAP4 IP accordingly.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
To improve readability, split the audio_enable HDMI IP operation
into two separate functions for enabling and disabling audio.
The audio_enable function is also modified to return an error value.
While there, update these operations for the OMAP4 IP accordingly.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
The Beagleboard xM gpio used for TFP410 powerdown is connected through
an I2C attached chip which means setting the GPIO can sleep. Code that
calls tfp410_power_on/off holds a mutex, so sleeping should be fine.
Signed-off-by: Russ Dill <Russ.Dill@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Move the platform-data based display device initialization into a
separate function, so that we may later add of-based initialization.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
We currently use the id of the dsi platform device (dsidev->id) as the
DSI hardware module ID. This works because we assign the ID manually in
arch/arm/mach-omap2/display.c at boot time.
However, with device tree the platform device IDs are automatically
assigned to an arbitrary number, and we can't use it.
Instead of using dsidev->id during operation, this patch stores the
value of dsidev->id to a private field of the dsi driver at probe(). The
future device tree code can thus set the private field with some other
way.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Now that each output driver creates their own display devices, the
output drivers can also initialize those devices.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Currently the higher level omapdss platform driver gets the list of
displays in its platform data, and uses that list to create the
omap_dss_device for each display.
With DT, the logical way to do the above is to list the displays under
each individual output, i.e. we'd have "dpi" node, under which we would
have the display that uses DPI. In other words, each output driver
handles the displays that use that particular output.
To make the current code ready for DT, this patch modifies the output
drivers so that each of them creates the display devices which use that
output. However, instead of changing the platform data to suit this
method, each output driver is passed the full list of displays, and the
drivers pick the displays that are meant for them. This allows us to
keep the old platform data, and thus we avoid the need to change the
board files.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
We currently have a two ways to set a "default panel device" for dss, to
which the overlays are connected when the omapdss driver is loaded:
- in textual format (name of the display) as cmdline parameter
- as a pointer to the panel device from board file via pdata
The current code handles this in a bit too complex way by using both of
the above methods during runtime. However, with DT we don't have pdata
anymore, so the code handling the second case won't work anymore. The
current code has also the problem that it modifies the platform_data.
This patch simplifies the code a bit by using the pointer method only
inside the probe function, and stores the name of the panel device. This
way we only need to handle the textual format during operation and also
avoid modifying the platform_data.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Now that the core.c doesn't fail if output driver's init fails, we can
change the uses of platform_driver_register to platform_driver_probe.
This will allow us to use __init in the following patches.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Instead of having an ugly #ifdef mess in the core.c for creating debugfs
files, add a dss_debugfs_create_file() function that the dss drivers
can use to create the debugfs files.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Initialize and uninitialize the output drivers by using arrays of
pointers to the init/uninit functions. This simplifies the code
slightly.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Now that the omapdss_core device is the parent for all other dss
devices, we don't need to use the dss_runtime_get/put anymore. Instead,
enabling omapdss_core will happen automatically when a child device is
enabled.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
We currently have separate device/driver for each DSS HW module. The DPI
and SDI outputs are more or less parts of the DSS or DISPC hardware
modules, but in SW it makes sense to represent them as device/driver
pairs similarly to all the other outputs. This also makes sense for
device tree, as each node under dss will be a platform device, and
handling DPI & SDI somehow differently than the rest would just make the
code more complex.
This patch modifies the dpi.c and sdi.c to create drivers for the
platform devices.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The platform devices for omapdss, dss and dispc drivers are always
present, so we can use platform_driver_probe instead of
platform_driver_register.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
For unknown reasons we seem to have a return in each of the omapdss's
uninit functions, which is a void function.
Remove the returns.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The omapdss pdata handling is a mess. This is more evident when trying
to use device tree for DSS, as we don't have platform data anymore in
that case. This patch cleans the pdata handling by:
- Remove struct omap_display_platform_data. It was used just as a
wrapper for struct omap_dss_board_info.
- Pass the platform data only to omapdss device. The drivers for omap
dss hwmods do not need the platform data. This should also work better
for DT, as we can create omapdss device programmatically in generic omap
boot code, and thus we can pass the pdata to it.
- Create dss functions for get_ctx_loss_count and dsi_enable/disable_pads
that the dss hwmod drivers can call.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The DSI driver uses dsi_get_dsidev_id() to get the ID number for the DSI
instance. However, there were a few places where dsidev->id was used
instead of the function. Fix those places to use the function.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
To ease device tree adaptation in the future, rewrite TFP410 platform
data handling to be done inside probe(), so that probe() is the only
place where we need to handle the DT/pdata choice.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
omapfb_parse_vram_param()'s check for end pointer returned from
simple_strtoul() is wrong, causing the code to bug if the second or
later vram parameters are non-parseable, for example
"omapfb.vram=0:2M,:5M".
However, even in that case the code will most likely bail out with
-EINVAL in the following checks, so the bug is probably not a fatal one.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reported-by: Hein Tibosch <hein_tibosch@yahoo.es>
Currently when multiple overlays are active, OMAPFB_SETUP_PLANE fails.
Instead of failing, allow it to configure the first overlay as if there
was only one overlay, the remaining ones will have to be configured in
other ways (sysfs).
This allows overlay-controlling programs (like video players) to function
properly when framebuffer is cloned to another display (like TV).
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
VENC output type (composite/svideo) doesn't have to be fixed by board
wiring, it is possible to also provide composite signal through svideo
luminance connector (software enabled), which is what pandora does.
Having to recompile the kernel for users who have TV connector types
that don't match default board setting is very inconvenient, especially
for users of a consumer device, so add support for switching VENC output
type at runtime over a new sysfs file output_type.
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
In 2009 Philip Reiser notied that a few users of netlink connector
interface needed a capability check and added the idiom
cap_raised(nsp->eff_cap, CAP_SYS_ADMIN) to a few of them, on the premise
that netlink was asynchronous.
In 2011 Patrick McHardy noticed we were being silly because netlink is
synchronous and removed eff_cap from the netlink_skb_params and changed
the idiom to cap_raised(current_cap(), CAP_SYS_ADMIN).
Looking at those spots with a fresh eye we should be calling
capable(CAP_SYS_ADMIN). The only reason I can see for not calling capable
is that it once appeared we were not in the same task as the caller which
would have made calling capable() impossible.
In the initial user_namespace the only difference between between
cap_raised(current_cap(), CAP_SYS_ADMIN) and capable(CAP_SYS_ADMIN) are a
few sanity checks and the fact that capable(CAP_SYS_ADMIN) sets
PF_SUPERPRIV if we use the capability.
Since we are going to be using root privilege setting PF_SUPERPRIV seems
the right thing to do.
The motivation for this that patch is that in a child user namespace
cap_raised(current_cap(),...) tests your capabilities with respect to that
child user namespace not capabilities in the initial user namespace and
thus will allow processes that should be unprivielged to use the kernel
services that are only protected with cap_raised(current_cap(),..).
To fix possible user_namespace issues and to just clean up the code
replace cap_raised(current_cap(), CAP_SYS_ADMIN) with
capable(CAP_SYS_ADMIN).
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Patrick McHardy <kaber@trash.net>
Cc: Philipp Reisner <philipp.reisner@linbit.com>
Acked-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: Andrew G. Morgan <morgan@kernel.org>
Cc: Vasiliy Kulikov <segoon@openwall.com>
Cc: David Howells <dhowells@redhat.com>
Reviewed-by: James Morris <james.l.morris@oracle.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This was defined in asm/pdc.h which needs to include asm/page.h for
__PAGE_OFFSET. This leads to an include loop so that page.h eventually will
include pdc.h again. While this is no problem because of header guards, it is
a problem because some symbols may be undefined. Such an error is this:
In file included from include/linux/bitops.h:35:0,
from include/asm-generic/getorder.h:7,
from arch/parisc/include/asm/page.h:162,
from arch/parisc/include/asm/pdc.h:346,
from arch/parisc/include/asm/processor.h:16,
from arch/parisc/include/asm/spinlock.h:6,
from arch/parisc/include/asm/atomic.h:20,
from include/linux/atomic.h:4,
from include/linux/sysfs.h:20,
from include/linux/kobject.h:21,
from include/linux/device.h:17,
from include/linux/eisa.h:5,
from arch/parisc/kernel/pci.c:11:
arch/parisc/include/asm/bitops.h: In function ‘set_bit’:
arch/parisc/include/asm/bitops.h:82:2: error: implicit declaration of function ‘_atomic_spin_lock_irqsave’ [-Werror=implicit-function-declaration]
arch/parisc/include/asm/bitops.h:84:2: error: implicit declaration of function ‘_atomic_spin_unlock_irqrestore’ [-Werror=implicit-function-declaration]
Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Conflicts:
drivers/video/omap2/displays/panel-taal.c
Merge OMAP DSS related board file changes. The branch will also be
merged through linux-omap tree to solve conflicts.
The functions calc_fclk_five_taps() and check_horiz_timing_omap3() use the
function dispc_mgr_get_device() to get the omap_dss_device pointer to which
the manager is connected, the width of the panel is derived from that.
The manager's timing is stored in it's private data in APPLY. This contains
the latest timings applied to the manager. Pass these timings to
dispc_ovl_setup() and use them in the above functions. Remove the function
dispc_mgr_get_device() as it isn't used any more.
Signed-off-by: Archit Taneja <archit@ti.com>
The pixel clock rate for the TV manager is calculated by checking the device
type connected to the manager, and then requesting the VENC/HDMI interface for
the pixel clock rate.
Remove the use of omap_dss_device pointer from here by checking which interface
generates the pixel clock by reading the DSS_CTRL.VENC_HDMI_SWITCH bit.
Signed-off-by: Archit Taneja <archit@ti.com>
The omap_dss_device pointer declared in dss_ovl_setup_fifo() isn't used. Remove
the pointer variable declaration and it's assignment.
Signed-off-by: Archit Taneja <archit@ti.com>
The DPI and HDMI interfaces use their 'set_timing' functions to take in a new
set of timings. If the panel is disabled, they do not disable and re-enable
the interface. Currently, the manager timings are applied in hdmi_power_on()
and dpi_set_mode() respectively, these are not called by set_timings if the
panel is disabled.
When checking overlay and manager data, the DSS driver uses the last applied
manager timings, and not the timings held by omap_dss_device struct. Hence,
there is a need to apply the new manager timings even if the panel is disabled.
Apply the manager timings if the panel is disabled. Eventually, there should be
one common place where the timings are applied independent of the state of the
panel.
Signed-off-by: Archit Taneja <archit@ti.com>
In order to check the validity of overlay and manager info, there was a need to
use the omap_dss_device struct to get the panel resolution. The manager's
private data in APPLY now contains the manager timings. Hence, we don't need to
rely on the display resolution any more.
Pass the manager's timings in private data to dss_mgr_check(). Remove the need
to pass omap_dss_device structs in the functions which check for the validity
of overlay and manager parameters.
Signed-off-by: Archit Taneja <archit@ti.com>
If a manager is disabled, there is no guarantee at any point in time that all
it's parameters are configured. There is always a chance that some more
parameters are yet to be configured by a user of DSS, or by DSS itself.
However, when the manager is enabled, we can be certain that all the parameters
have been configured, as we can't enable a manager with an incomplete
configuration. Therefore, if a manager is disabled, don't check for the validity
of it's parameters or the parameters of the overlays connected to it. Only check
once it is enabled. Add a check in dss_check_settings_low() to achieve the same.
Signed-off-by: Archit Taneja <archit@ti.com>
Create a function dss_mgr_check_timings() which wraps around the function
dispc_mgr_timings_ok(). This is mainly a clean up to hide dispc functions
from interface drivers.
dss_mgr_check_timings() is added in the function dss_mgr_check(), it currently
takes the timings maintained in the omap_dss_device struct. This would be later
replaced by the timings stored in the manager's private data.
Make dss_mgr_check_timings() and dispc_mgr_timings_ok() take a const
omap_video_timings pointer since these functions just check the timings.
Signed-off-by: Archit Taneja <archit@ti.com>
Replace the function dispc_mgr_set_timings() with dss_mgr_set_timings() in the
interface drivers. The latter function ensures that the timing related DISPC
registers are configured according to the shadow register programming model.
Remove the call to dispc_mgr_go() in dpi_set_timings() as the manager's go bit
is set by dss_mgr_set_timings().
Signed-off-by: Archit Taneja <archit@ti.com>
DISPC manager size and DISPC manager blanking parameters(for LCD managers)
follow the shadow register programming model. Currently, they are programmed
directly by the interface drivers.
To configure manager timings using APPLY, there is a need to introduce extra
info flags for managers, similar to what is done for overlays. This is needed
because timings aren't a part of overlay_manager_info struct configured by a
user of DSS, they are configured internally by the interface or panel drivers.
Add dirty and shadow_dirty extra_info flags for managers, update these flags
at the appropriate places. Rewrite the function extra_info_update_ongoing()
slightly as checking for manager's extra_info flags can simplify the code a bit.
Create function dss_mgr_set_timings() which applies the new manager timings to
extra_info.
Signed-off-by: Archit Taneja <archit@ti.com>
Fake VSYNC support is a hack and has some bugs in it. It isn't used by any user
of DSS. Remove Fake VSYNC support. For DSI command mode and RFBI panels, a user
of DSS should wait for the completion of a frame by using the panel driver's
sync op.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The wrong bit field was being updated in DSS_CTRL when trying to configure the
clock source of DSI2 functional clock. Use the correct bit field based on the
dsi module number.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The HDMI core register offset macros aren't defined in ascending order of their
values, some of the offset macros are also redefined. The same issues occur when
these core registers are dumped.
Clean up the ordering of HDMI core registers and remove repeated registers in
the definition in ti_hdmi_4xxx_ip.h and in ti_hdmi_4xxx_core_dump().
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The function ti_hdmi_4xxx_core_dump has some bugs, the following mention the
bugs and the solutions:
- The macros DUMPCORE and DUMPCOREAV in ti_hdmi_4xxx_core_dump() use
hdmi_pll_base() for the offsets needed to calculate register addresses, use
functions hdmi_core_sys_base() amd hdmi_av_base() to calculate the correct
offsets for CORE_SYS and CORE_AV registers.
- Many of the CORE_AV registers use the DUMPCORE macro, and hence the register
addresses are calculated incorrectly. Rename the current DUMPCOREAV macro as
DUMPCOREAV2 as it takes 2 arguments to dump indexed CORE_AV registers, create
a new macro called DUMPCOREAV which is now used for dumping non-indexed
CORE_AV registers.
Thanks to Ancy Tom <ancytom@gmail.com> for pointing out the issues.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
In preparation for device tree, this patch changes how the DSI pins are
configured. The current configuration method is only doable with board
files and the configuration data is OMAP specific.
This patch moves the configuration data to the panel's platform data,
and the data can easily be given via DT in the future. The configuration
data format is also changed to a generic one which should be suitable
for all platforms.
The new format is an array of pin numbers, where the array items start
from clock + and -, then data1 + and -, and so on. For example:
{
0, // pin num for clock lane +
1, // pin num for clock lane -
2, // pin num for data1 lane +
3, // pin num for data1 lane -
...
}
The pin numbers are translated by the DSI driver and used to configure
the hardware appropriately.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
The reset GPIO for Taal panel driver is currently requested in the
4430sdp board file. This patch moves the gpio request/free into the Taal
driver, where it should be.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Now that the tfp410 driver has been renamed in the code, this patch
finishes the renaming by renaming the files.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
The driver for the TFP410 DPI-to-DVI chip was named quite badly as "DVI
panel driver". This patch renames the code to use tfp410 name for the
driver.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Now that the panel-dvi driver handles the PD (power-down) GPIO, we can
remove the custom PD handling from the board files.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
The driver for the TFP410 chip should handle the power-down signal of
the chip, instead of the current way of handling it in the board files.
This patch adds power_down_gpio into the device's platform data, and
adds the necessary code in the driver to request and handle the GPIO.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
* fix to Kconfig to make it fit within 80 line characters,
* two bootup fixes (AMD 8-core and with PCI BIOS),
* cleanup code in a Xen PV fb driver,
* and a crash fix when trying to see non-existent PTE's
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
iQEcBAABAgAGBQJPp/iaAAoJEFjIrFwIi8fJAFsH/1NVxvOAHyzyczU49U/1vi8T
d67kIb8fFHni6HO7BiBkuM8DricGQnDhP7uC1n9waWf8jRiYsPTAbesyedTLbQos
SLfzpsLWKilJOxWCf17cBnm6i9ScQn1ioJ6h3jFzHgNCXnvvAVYqfKHW0V6HTErH
JL0eb4+asiZgXNSeac1gabQlai6LuBzMWaFgzRGY+hDnCQhkdQfDkD7X5zEhUUmH
jUmtSxRx+5LkfelwRb2kHhI5j58ilOEa7YLZFQc3C+2+bUvgsG9vJDsQ3jwFaGDn
cryfRY9WJXxgcXqk1ClOnk9U9PGzRc48gdLVLhYuLsIvUWN7RzgRlBMsH33Gq9M=
=kjvX
-----END PGP SIGNATURE-----
Merge tag 'stable/for-linus-3.4-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen
Pull xen fixes from Konrad Rzeszutek Wilk:
- fix to Kconfig to make it fit within 80 line characters,
- two bootup fixes (AMD 8-core and with PCI BIOS),
- cleanup code in a Xen PV fb driver,
- and a crash fix when trying to see non-existent PTE's
* tag 'stable/for-linus-3.4-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/Kconfig: fix Kconfig layout
xen/pci: don't use PCI BIOS service for configuration space accesses
xen/pte: Fix crashes when trying to see non-existent PGD/PMD/PUD/PTEs
xen/apic: Return the APIC ID (and version) for CPU 0.
drivers/video/xen-fbfront.c: add missing cleanup code
When exynos_mipi_update_cfg() is called, mipi dsi registers were become
sw reset. So, It needs to enable interrupt again.
Signed-off-by: Donghwa Lee <dh09.lee@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
This patch replaces the code for getting an number from a
userspace buffer by a simple call to kstrou8_from_user.
This makes it easier to read and less error prone.
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Acked-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
SiS_DRAMType is const and only used by sisfb_post_300_rwtest which is
marked __devinit we can annotate SiS_DRAMType with __devinitconst and
move it into the file scope in order to not have it created on the
stack.
This patch decreases the compiled module size by about 100bytes.
And since hardcoded values are bad we use ARRAY_SIZE for determining
the size of SiS_DRAMType ;)
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
This patch removes the unused structs SiS_SDRDRAM_TYPE and SiS_DDRDRAM_TYPE
from init.h
These are not used anywhere so we can delete them.
The SiS_DRAMType is identically defined in sis_main.c and only used
there so we can remove it here.
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Functions not referenced outside of a source file should be marked
static to prevent them from being exposed globally.
Quiets the sparse warnings:
warning: symbol 'mb862xx_intr' was not declared. Should it be static?
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Functions not referenced outside of a source file should be marked
static to prevent them from being exposed globally.
Quiets the sparse warnings:
warning: symbol 'mb862xx_i2c_stop' was not declared. Should it be static?
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Functions not referenced outside of a source file should be marked
static to prevent them from being exposed globally.
Quiets the sparse warning:
warning: symbol 'fb_deferred_io_page' was not declared. Should it be static?
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Jaya Kumar <jayalk@intworks.biz>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Functions not referenced outside of a source file should be marked
static to prevent them from being exposed globally.
Quiets the sparse warnings:
warning: symbol 'ufx_handle_damage' was not declared. Should it be static?
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Steve Glendinning <steve.glendinning@smsc.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Since s3c_fb_missing_pixclock() is called from s3c_fb_set_rgb_timing()
which is used in the suspend/resume paths it can't be marked __devinit
as this could result in it being discarded after boot.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (GNU/Linux)
iQEcBAABAgAGBQJPpvY9AAoJEHm+PkMAQRiGpEoIAJgbu+Y8gITnBK/wh9O6zy3S
5jie5KK4YWdbJsvO58WbNr3CyVIwGIqQ2dUZLiU59aBVLarlGw8xor0MmW+cZwhp
6fBHaf0qDYAV0MZjD+mnnExOiCRyISa2lPmsfu9dAWywh5KGe6/oAP6/qcXIyok3
KZyl3qQf4ENpaZPHwZPXCEkUvtuyHgNiszN+QXEadA3s19Ot4VGe9A3VGw+GNrSm
JqFIq3acQAbKa5BYaqf7TQC02v2FI7//eqt6QHxTqbE6a7LGbTvLfX3HlJ2mnfqa
1R6QHhM4y4OZDHbaMT2raHZ8WuLXzhehJzhP8Co7AHFOKwVKOb5XbcUr2RrukMU=
=HkMd
-----END PGP SIGNATURE-----
Merge tag 'v3.4-rc6' into drm-intel-next
Conflicts:
drivers/gpu/drm/i915/intel_display.c
Ok, this is a fun story of git totally messing things up. There
/shouldn't/ be any conflict in here, because the fixes in -rc6 do only
touch functions that have not been changed in -next.
The offending commits in drm-next are 14415745b2..1fa611065 which
simply move a few functions from intel_display.c to intel_pm.c. The
problem seems to be that git diff gets completely confused:
$ git diff 14415745b2..1fa611065
is a nice mess in intel_display.c, and the diff leaks into totally
unrelated functions, whereas
$git diff --minimal 14415745b2..1fa611065
is exactly what we want.
Unfortunately there seems to be no way to teach similar smarts to the
merge diff and conflict generation code, because with the minimal diff
there really shouldn't be any conflicts. For added hilarity, every
time something in that area changes the + and - lines in the diff move
around like crazy, again resulting in new conflicts. So I fear this
mess will stay with us for a little longer (and might result in
another backmerge down the road).
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The operations in the subsequent error-handling code appear to be also
useful here.
Acked-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
[v1: Collapse some of the error handling functions]
[v2: Fix compile warning]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
DISPC_FCLK is incorrectly used as functional clock of DISPC in scaling
calculations. So, DISPC_CORE_CLK replaces as functional clock of DISPC.
DISPC_CORE_CLK is derived from DISPC_FCLK divided by an independent DISPC
divisor LCD.
Signed-off-by: Chandrabhanu Mahapatra <cmahapatra@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
In OMAP3 DISPC video overlays suffer from some undocumented horizontal position
and timing related limitations leading to SYNCLOST errors. Whenever the image
window is moved towards the right of the screen SYNCLOST errors become
frequent. Checks have been implemented to see that DISPC driver rejects
configuration exceeding above limitations.
This code was successfully tested on OMAP3. This code is written based on code
written by Ville Syrjälä <ville.syrjala@nokia.com> in Linux OMAP kernel. Ville
Syrjälä <ville.syrjala@nokia.com> had added checks for video overlay horizontal
timing and DISPC horizontal blanking length limitations.
Signed-off-by: Chandrabhanu Mahapatra <cmahapatra@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
In OMAP3 and OMAP4, the DISPC Scaler can downscale an image up to 4 times, and
up to 2 times in OMAP2. However, with predecimation, the image can be reduced
to 16 times by fetching only the necessary pixels in memory. Then this
predecimated image can be downscaled further by the DISPC scaler.
The pipeline is configured to use a burst of size 8 * 128 bits which consists
of 8 mini bursts of 16 bytes each. So, horizontal predecimation more than 16
can lead to complete discarding of such mini bursts. L3 interconnect may
handover the bus to some other initiator and inturn delay the fetching of
pixels leading to underflows. So, maximum predecimation limit is fixed at 16.
Based on the downscaling required, a prior calculation of predecimation values
for width and height of an image is done. Since, Predecimation reduces quality
of an image higher priorty is given to DISPC Scaler for downscaling.
This code was successfully tested on OMAP2, OMAP3 and OMAP4. Horizontal and
vertical predecimation worked fine except for some synclost errors due to
undocumented errata in OMAP3 which are fixed later and skewed images were seen
on OMAP2 and OMAP3 during horizontal predecimation which will be addressed in
the future patches.
This code is based on code written by Lajos Molnar <lajos@ti.com> who had added
predecimation support for NV12/YUV/rotated/SDMA buffers.
Signed-off-by: Chandrabhanu Mahapatra <cmahapatra@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Daniel Vetter writes:
A new drm-intel-next pull. Highlights:
- More gmbus patches from Daniel Kurtz, I think gmbus is now ready, all
known issues fixed.
- Fencing cleanup and pipelined fencing removal from Chris.
- rc6 residency interface from Ben, useful for powertop.
- Cleanups and code reorg around the ringbuffer code (Ben&me).
- Use hw semaphores in the pageflip code from Ben.
- More vlv stuff from Jesse, unfortunately his vlv cpu is doa, so less
merged than I've hoped for - we still have the unused function warning :(
- More hsw patches from Eugeni, again, not yet enabled fully.
- intel_pm.c refactoring from Eugeni.
- Ironlake sprite support from Chris.
- And various smaller improvements/fixes all over the place.
Note that this pull request also contains a backmerge of -rc3 to sort out
a few things in -next. I've also had to frob the shortlog a bit to exclude
anything that -rc3 brings in with this pull.
Regression wise we have a few strange bugs going on, but for all of them
closer inspection revealed that they've been pre-existing, just now
slightly more likely to be hit. And for most of them we have a patch
already. Otherwise QA has not reported any regressions, and I'm also not
aware of anything bad happening in 3.4.
* tag 'drm-intel-next-2012-04-23' of git://people.freedesktop.org/~danvet/drm-intel: (420 commits)
drm/i915: rc6 residency (fix the fix)
drm/i915/tv: fix open-coded ARRAY_SIZE.
drm/i915: invalidate render cache on gen2
drm/i915: Silence the change of LVDS sync polarity
drm/i915: add generic power management initialization
drm/i915: move clock gating functionality into intel_pm module
drm/i915: move emon functionality into intel_pm module
drm/i915: move drps, rps and rc6-related functions to intel_pm
drm/i915: fix line breaks in intel_pm
drm/i915: move watermarks settings into intel_pm module
drm/i915: move fbc-related functionality into intel_pm module
drm/i915: Refactor get_fence() to use the common fence writing routine
drm/i915: Refactor fence clearing to use the common fence writing routine
drm/i915: Refactor put_fence() to use the common fence writing routine
drm/i915: Prepare to consolidate fence writing
drm/i915: Remove the unsightly "optimisation" from flush_fence()
drm/i915: Simplify fence finding
drm/i915: Discard the unused obj->last_fenced_ring
drm/i915: Remove unused ring->setup_seqno
drm/i915: Remove fence pipelining
...
Tomi Valkeinen