Merge Linux 3.7-rc4 to get fixes for CMA.

This reverts commit 14840b9a.

The commit breaks audio, and a new version will be applied later.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

When doing a manual update in dss_mgr_start_update, we clear the shadow dirty
flags. Although there isn't any harm in clearing them. The need to clear them
out here should never arrive.

When applying configurations for a manual update manager, we never do any
register writes, i.e, calls to dss_mgr_write_regs and dss_mgr_write_regs_extra
never happen while applying. We do all these writes only when we call
dss_mgr_start_update. Hence, there is never a time when the shadow registers
are dirty.

Remove the call to mg_clear_shadow_dirty.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

The bool was_updating is never really used for anything. It is set to the
current value of mp->updating, but not used anywhere. Remove this variable.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

An overlay's channel out field isn't a shadow register. The TRM says that it's
taken into effect immediately. This understanding was missing and channel out
was treated as a shadow parameter, and in overlay's private data as extra info.

Program channel out bits directly in dss_ovl_set_manager(). In order to do this
safely, we need to be totally sure that the overlay is disabled in hardware. For
auto update managers, we can assume that the overlay was truly disabled at
dss_ovl_unset_manager() through the wait_pending_extra_info_updates() call.
However, when unsetting manager for an overlay that was previously connected to
a manager in manual update, we can't be sure if the overlay is truly disabled.
That is, op->enabled might not reflect the actual state of the overlay in
hardware. The older manager may require a manual update transfer to truly
disable the overlay. We expect the user of OMAPDSS to take care of this, in
OMAPDSS, we make sure that an overlay's manager isn't unset if there if
extra_info is still dirty for that overlay.

The wrong understanding of channel out bits also explains the reason why we see
sync lost when changing an overlay's manager which was previously connected to a
manual update manager. The following sequence of events caused this:

- When we disable the overlay, no register writes are actually done since the
  manager is manual update, op->enabled is set to false, and the
  extra_info_dirty flag is set. However, in hardware, the overlay is still
  enabled in both shadow and working registers.

- When we unset the manager, the software just configures the overlay's manager
  to point to NULL.

- When we set the overlay to a new manager(which is in auto update) through
  dss_ovl_set_manager, the check  for op->enabled passes, the channel field in
  extra info is set to the new manager. When we do an apply on this manager,
  the new channel out field is set in the hardware immediately, and since the
  overlay enable bit is still set in hardware, the new manager sees that the
  overlay is enabled, and tries to retrieve pixels from it, this leads to sync
  lost as it might be in the middle of processing a frame when we set the
  channel out bit.

The solution to this was to ensure that user space does another update after
disabling the overlay, this actually worked because the overlay was now truly
disabled, and an immediate write to channel out didn't impact since the manager
saw the new overlay as disabled, and doesn't try to retrieve pixels from it.

Remove channel as an extra_info field. Make dss_ovl_unset_manager more strict
about the overlay being disabled when detaching the manager. For overlays
connected to a manual update manager, unset_manager fails if we need another
update to disable the overlay.

We still need to a manual update to ensure the overlay is disabled to get change
the overlay's manager. We could work on doing a dummy update by using DISPC's
capability to gate the different video port signals. This is left for later.

Remove the comment about the sync lost issue.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

When calculating row and pixel increments for graphics and video pipes, we need
to consider the dimensions of the input frame to know how to read from the
buffer. Hence, we need to calculate these parameters from the input to the
pipeline.

For writeback, the row and pixel increments need to be calculated based on the
output of the writeback pipeline, i.e, the dimensions of the frame after
scaling. Ensure that dispc driver uses values of out_width and out_height when
calling calc_dma/calc_tiler_rotation_offset.

For graphics and video pipes, the original code passed the original height as
frame_height to calc_dma_rotation_offset, and not the predecimated height. This
is left as it is for now. We need to figure out why pre decimated height isn't
needed.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Since writeback writes to a buffer instead of reading from one, predecimation
doesn't make sense for it. Configure the width and height predecimation limits
to 1 if the plane is writeback.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

dispc_ovl_calc_scaling_44xx() doesn't work correctly for writeback. There are
two issues with it:

- the function tries to calculate pixel clock for the input plane using
  dispc_plane_pclk_rate(), calling this with writeback as input plane results in
  a BUG(), this function shouldn't be called for writeback at all. Fix this by
  calculating pixel clock only when we are not in mem to mem mode.

- the maximum input_width is the product of the downscale ratio supported and
  the and the given output_width. This was calculated incorrectly by dividing
  output_width with maxdownscale. Fix this.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Normally the omapdss driver gets the regulators using the regulator
names assigned for omapdss. However, in an effort to get a minimal DSS
support for DT enabled kernel on selected boards, we will add omapdss
devices and platform data the old way even for DT kernel. This causes
the problem that omapdss cannot find the regulators using omapdss's
regulator names.

This patch creates a temporary workaround for DSI and HDMI by trying to
get the regulators also using native OMAP4 regulator names.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

This function is now used in the driver init path to handle
probe errors properly. Thus, it may be possible to use this function
outside the exit path.
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Reported-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Fix sparse warning:

drivers/video/omap2/dss/dispc.c:3320:6: warning: symbol
'dispc_dump_irqs' was not declared. Should it be static?
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reported-by: Fengguang Wu <fengguang.wu@intel.com>

Creating the accessory devices, such as audio, from the HDMI driver
allows to regard HDMI as a single entity with audio an display
functionality. This intends to follow the design of drivers such
as MFD, in which a single entity handles the creation of the accessory
devices. Such devices are then used by domain-specific drivers; audio in
this case.

Also, this is in line with the DT implementation of HDMI, in which we will
have a single node to describe this feature of the OMAP SoC.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

It could be possible that the DMA port differs accross diferent HDMI IPs. Thus,
add an IP-specific function to obtain the address offset and size of the DMA
data port.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

The display must be uninitialized in order to free the requested GPIOs.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Do not blindly assume that the panel could be initialized.

While there, group mutex initialization at a single place.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Do not assume blindly that the DSS driver was registered successfully.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Using devm_request_and_ioremap provides better memory handling and
improves readability.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Minor cleanup to give to the resource variable a more proper name.
Signed-off-by: Ricardo Neri <ricardo.neri@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

The return value of wait_for_completion_timeout() is always
>= 0 with unsigned int type.

So the condition "ret < 0" or "ret >= 0" is pointless.
Signed-off-by: liu chuansheng <chuansheng.liu@intel.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

The return value of wait_for_completion_timeout() is always
>= 0 with unsigned int type.

So the condition "ret < 0" or "ret >= 0" is pointless.
Signed-off-by: liu chuansheng <chuansheng.liu@intel.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

check_horiz_timing_omap3() has a variable named 'DS'. i386 uses DS name
for something else, causing a compilation error. As 'DS' is not a very
good local variable name in the first place, let's change it to 'ds',
fixing the issue.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Merge omapdss patches to enable using DSI PLL for DPI output.

We currently get the decision whether to use PRCM or DSI PLL clock for
DPI from the board file. This is not a good way to handle it, and it
won't work with device tree.

This patch changes DPI to always use DSI PLL if it's available.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

The SoCs that have DSI module should have a working DSI PLL. However,
some rare boards have not connected the powers to the DSI PLL.

This patch adds a function that tries to power up the DSI PLL, and
reports if that doesn't succeed. DPI uses this function to fall back to
PRCM clocks if DSI PLL doesn't work.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Instead of using dpi_use_dsi_pll() to check if dsi pll is to be used, we
can just check if dpi.dsidev != NULL.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

dss.c currently exposes functions to configure the dispc source clock
and lcd source clock. There are configured separately from the output
drivers.

However, there is no safe way for the output drivers to handle dispc
clock, as it's shared between the outputs. Thus, if, say, the DSI driver
sets up DSI PLL and configures both the dispc and lcd clock sources to
that DSI PLL, the resulting dispc clock could be too low for, say, HDMI.

Thus the output drivers should really only be concerned about the lcd
clock, which is what the output drivers actually use. There's lot to do
to clean up the dss clock handling, but this patch takes one step
forward and removes the use of dss_select_dispc_clk_source() from the
output drivers.

After this patch, the output drivers only configure the lcd source
clock. On omap4+ the dispc src clock is never changed from the default
PRCM source. On omap3, where the dispc and lcd clocks are actually the
same, setting the lcd clock source sets the dispc clock source.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

We don't currently set the dss fck when starting up. This is not a
problem, as we setup the fck later when configuring the pixel clocks. Or
this is how it was for omap2, for the rest of the omaps this may not be
so.

For DSI, HDMI and also for DPI when using DSI PLL, we don't need to
change the dss fck, and thus it may be left unconfigured. Usually the
dss fck is already setup fine by default, but we can't trust this.

This patch sets the dss fck to maximum at probe time.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

dss_calc_clock_rates() was removed earlier as it was not used, but it is
needed for DSI PLL calculations, so this patch adds it back.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

It looks like on many OMAP versions powers for both HSClk and HSDiv to
be enabled to have a functional HSDiv.

This patch fixes the issue by forcing both powers on.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

The DSI PLL and HSDivider can be used to generate the pixel clock for
LCD overlay manager, which then goes to DPI output. On the DPI output
pin the voltage of the signal is shifted from the OMAP's internal
minimal voltage to 1.8V range. The shifting is not instant, and the
higher the clock frequency, the less time there is to shift the signal
to nominal voltage.

If the HSDivider's divider is greater than one and odd, the resulting
pixel clock does not have 50% duty cycle. For example, with a divider of
3, the duty cycle is 33%.

When combining high frequency (in the area of 140MHz+) and non-50% duty
cycle, it has been observed the the shifter does not have enough time to
shift the voltage enough, and this leads to bad signal which is rejected
by monitors.

As a workaround this patch makes the divider calculation skip all odd
dividers when the required pixel clock is over 100MHz. The limit of
100MHz is a guesstimate.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

DPI may use DSI PLL, so it depends on DSI. However, currently DPI driver
is added first, which causes DPI initialization to fail when it tries to
get the DSI PLL.

This patch changes the init order to fix this.

A better solution would be to separate DSI PLL and DSI drivers. They
have dependencies, though, but we could still have DSI PLL as an
independent entity that we could initialize before any of the output
drivers.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>

Merge omapdss miscellaneous patches.

Pull NFS client bugfixes from Trond Myklebust:

 - Fix a bunch of deadlock situations:
   * State recovery can deadlock if we fail to release sequence ids
     before scheduling the recovery thread.
   * Calling deactivate_super() from an RPC workqueue thread can
     deadlock because of the call to rpc_shutdown_client.

 - Display the device name correctly in /proc/*/mounts

 - Fix a number of incorrect error return values:
   * When NFSv3 mounts fail due to a timeout.
   * On NFSv4.1 backchannel setup failure
   * On NFSv4 open access checks

 - pnfs_find_alloc_layout() must check the layout pointer for NULL

 - Fix a regression in the legacy DNS resolved

* tag 'nfs-for-3.7-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFS4: nfs4_opendata_access should return errno
  NFSv4: Initialise the NFSv4.1 slot table highest_used_slotid correctly
  SUNRPC: return proper errno from backchannel_rqst
  NFS: add nfs_sb_deactive_async to avoid deadlock
  nfs: Show original device name verbatim in /proc/*/mount{s,info}
  nfsv3: Make v3 mounts fail with ETIMEDOUTs instead EIO on mountd timeouts
  nfs: Check whether a layout pointer is NULL before free it
  NFS: fix bug in legacy DNS resolver.
  NFSv4: nfs4_locku_done must release the sequence id
  NFSv4.1: We must release the sequence id when we fail to get a session slot
  NFS: Wait for session recovery to finish before returning

Pull thermal management & ACPI update from Zhang Rui,

Ho humm.  Normally these things go through Len.  But it's just three
small fixes, I guess I can pull directly too.

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/rzhang/linux:
  exynos4_tmu_driver_ids should be exynos_tmu_driver_ids.
  ACPI video: Ignore errors after _DOD evaluation.
  thermal: solve compilation errors in rcar_thermal

Pull i2c embedded fixes from Wolfram Sang:
 "Two patches are usual stuff.

  The bigger patch is needed to correct a wrong decision made in this
  merge window.  We hoped to get the PIOQUEUE mode in the mxs driver
  working with DMA, but it turned out to be too broken (leading to data
  loss), so we now think it is best to remove it entirely and work only
  with DMA now.  The patch should be in 3.7.  IMO, so users never get
  the chance to use both modes in parallel."

* 'i2c-embedded/for-current' of git://git.pengutronix.de/git/wsa/linux:
  i2c: tegra: set irq name as device name
  i2c-nomadik: Fixup clock handling
  i2c: mxs: remove broken PIOQUEUE support

Pull drm fixes from Dave Airlie:
 "Scattered selection of fixes:

   - radeon: load detect fixes from SuSE/AMD
   - intel: misc i830, sdvo regression, vesafb kickoff ums fix
   - exynos: maintainers entry update + fixes
   - udl: fix stride scanout issue

  it's slightly bigger than I'd probably like, but nothing looked
  dangerous enough to hold off on."

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/udl: fix stride issues scanning out stride != width*bpp
  drm/radeon: add load detection support for ext DAC on R200 (v2)
  DRM/radeon: For single CRTC GPUs move handling of CRTC_CRT_ON to crtc_dpms().
  DRM/Radeon: Fix TV DAC Load Detection for single CRTC chips.
  DRM/Radeon: Clean up code in TV DAC load detection.
  drm/radeon: fix ATPX function documentation
  drivers/gpu/drm/radeon/evergreen_cs.c: Remove unnecessary semicolon
  DRM/Radeon: On DVI-I use Load Detection when EDID is bogus.
  DRM/Radeon: Fix primary DAC Load Detection for RV100 chips.
  DRM/Radeon: Fix Load Detection on legacy primary DAC.
  drm: exynos: removed warning due to missing typecast for mixer driver data
  drm/exynos: add support for ARCH_MULTIPLATFORM
  MAINTAINERS: Add git repository for Exynos DRM
  drm/exynos: fix display on issue
  drm/i915: Only kick out vesafb if we takeover the fbcon with KMS
  drm/i915: be less verbose about inability to provide vendor backlight
  drm/i915: clear the entire sdvo infoframe buffer
  drm/i915: VGA needs to be on pipe A on i830M
  drm/i915: fix overlay on i830M

Pull networking fixes from David Miller:
 "First post-Sandy pull request"

 1) Fix antenna gain handling and initialization of chan->max_reg_power
    in wireless, from Felix Fietkau.

 2) Fix nexthop handling in H.232 conntrack helper, from Julian
    Anastasov.

 3) Only process 80211 mesh config header in certain kinds of frames,
    from Javier Cardona.

 4) 80211 management frame header length needs to be validated, from
    Johannes Berg.

 5) Don't access free'd SKBs in ath9k driver, from Felix Fietkay.

 6) Test for permanent state correctly in VXLAN driver, from Stephen
    Hemminger.

 7) BNX2X bug fixes from Yaniv Rosner and Dmitry Kravkov.

 8) Fix off by one errors in bonding, from Nikolay ALeksandrov.

 9) Fix divide by zero in TCP-Illinois congestion control.  From Jesper
    Dangaard Brouer.

10) TCP metrics code says "Yo dawg, I heard you like sizeof, so I did a
    sizeof of a sizeof, so you can size your size" Fix from Julian
    Anastasov.

11) Several drivers do mdiobus_free without first doing an
    mdiobus_unregister leading to stray pointer references.  Fix from
    Peter Senna Tschudin.

12) Fix OOPS in l2tp_eth_create() error path, it's another danling
    pointer kinda situation.  Fix from Tom Parkin.

13) Hardware driven by the vmxnet driver can't handle larger than 16K
    fragments, so split them up when necessary.  From Eric Dumazet.

14) Handle zero length data length in tcp_send_rcvq() properly.  Fix
    from Pavel Emelyanov.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (38 commits)
  tcp-repair: Handle zero-length data put in rcv queue
  vmxnet3: must split too big fragments
  l2tp: fix oops in l2tp_eth_create() error path
  cxgb4: Fix unable to get UP event from the LLD
  drivers/net/phy/mdio-bitbang.c: Call mdiobus_unregister before mdiobus_free
  drivers/net/ethernet/nxp/lpc_eth.c: Call mdiobus_unregister before mdiobus_free
  bnx2x: fix HW initialization using fw 7.8.x
  tcp: Fix double sizeof in new tcp_metrics code
  net: fix divide by zero in tcp algorithm illinois
  net: sctp: Fix typo in net/sctp
  bonding: fix second off-by-one error
  bonding: fix off-by-one error
  bnx2x: Disable FCoE for 57840 since not yet supported by FW
  bnx2x: Fix no link on 577xx 10G-baseT
  bnx2x: Fix unrecognized SFP+ module after driver is loaded
  bnx2x: Fix potential incorrect link speed provision
  bnx2x: Restore global registers back to default.
  bnx2x: Fix link down in 57712 following LFA
  bnx2x: Fix 57810 1G-KR link against certain switches.
  ixgbe: PTP get_ts_info missing software support
  ...

When sending data into a tcp socket in repair state we should check
for the amount of data being 0 explicitly. Otherwise we'll have an skb
with seq == end_seq in rcv queue, but tcp doesn't expect this to happen
(in particular a warn_on in tcp_recvmsg shoots).
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Reported-by: Giorgos Mavrikas <gmavrikas@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

vmxnet3 has a 16Kbytes limit per tx descriptor, that happened to work
as long as we provided PAGE_SIZE fragments.

Our stack can now build larger fragments, so we need to split them to
the 16kbytes boundary.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: jongman heo <jongman.heo@samsung.com>
Tested-by: jongman heo <jongman.heo@samsung.com>
Cc: Shreyas Bhatewara <sbhatewara@vmware.com>
Reviewed-by: Bhavesh Davda <bhavesh@vmware.com>
Signed-off-by: Shreyas Bhatewara <sbhatewara@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>