psb_video: remove OSPM_GL3_CACHE_ISLAND when not enabled

When setting the island up for OSPM_GL3_CACHE_ISLAND the code was not
protected when the config is not enabled

Signed-off-by: Daniel Charles <daniel.charles@intel.com>
Reviewed-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Sean V Kelley <sean.v.kelley@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Fix BUG_ON() triggering in drm_vblank_put()

We're calling drm_vblank_get() too late. It must be done before the
xchg() to avoid psb_flip_complete() from calling drm_vblank_put() before
drm_vblank_get(). This could trigger a a BUG_ON() in drm_vblank_get().

Earlier the pvr sync callbacks were usually executed with interrupts
disabled, which made the problem extremely unlikely to happen.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: ossync: Make sync_list and sync_lock static

drivers/staging/mrst/pvr/services4/srvkm/env/linux/ossync.c:30:1: warning: symbol 'sync_lock' was not declared. Should it be static?
drivers/staging/mrst/pvr/services4/srvkm/env/linux/ossync.c:31:1: warning: symbol 'sync_list' was not declared. Should it be static?

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: Remove an unused variable

drivers/staging/mrst/drv/psb_page_flip.c: In function 'psb_intel_crtc_page_flip':
drivers/staging/mrst/drv/psb_page_flip.c:193: warning: unused variable 'ret'

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: psb_gtt: Fix smatch warnings

drivers/staging/mrst/drv/psb_gtt.c:54 psb_gtt_takedown(4) warn: variable dereferenced before check 'pg' (see line 52) (smatch)
drivers/staging/mrst/drv/psb_gtt.c:343 psb_gtt_mm_init(10) warn: variable dereferenced before check 'pg' (see line 336) (smatch)

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: Silence a compiler warning

drivers/staging/mrst/drv/psb_page_flip.c: In function 'sync_callback':
drivers/staging/mrst/drv/psb_page_flip.c:102: warning: 'reg_offset' may be used uninitialized in this function

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: ossync: Avoid double kmalloc()

Allow the user to embed the pending_sync structure inside its own
structure, and use container_of() in the callback.

This means we can now get by with one kmalloc() per flip, instead
of two.

As PVRSRVCallbackOnSync() can no longer fail, change the return
type to void.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: ossync: Execute sync callbacks outside the spinlock

No need to hold the spinlock while executing the sync callbacks.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: ossync: Use spin_lock_irq() in PVRSRVCallbackOnSync()

The code already calls kmalloc(..., GFP_KERNEL), so interrupts can't be
off anyway.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: ossync: Make flags parameter to PVRSRVCallbackOnSync() unsigned

It's a bitmask, so it should be unsigned.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: ossync: Rewrite sync counter comparisons

Make the sync counter comparisons handle wraparound. Also reorganize
the code a bit to make it more readable.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: overlay: Use msleep(1) in ovl_wait()

Having ovl_wait() react immediately to hardware state changes is not
critical, so use msleep(1) instead of cpu_relax().

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: overlay: Use double buffering for overlay registers and add suspend/resume hooks

The existing suspend/resume code that fiddles with the OVADD registers
simply does not work. Replace it with something sensible.

Allocate two pages for the registers and ping-pong between them. This
allows update_plane() to do its thing without waiting for the previous
update to complete. The previous update, if not completed, will simply
be overwritten by a new update.

If the hardware is powered down when update_plane() is called we can
simply skip programming OVADD, and instead wait for the reusme hook
to do it next time the hardware is woken up. It's also theoretically
possible that the hardware gets woken up just after we've pointed the
driver to the new register page, in which case we can simply skip
programming OVADD in update_plane(), since we know the resume hook
already did it for us.

There is a slight concern whether the hardware fetches the register
contents from memory in an atomic fashion wrt. the OVADD register
writes. That is, if the hardware can still be fetching the registers
from the old page after OVADD is pointed to the new page, there
might be trouble. So far I did not observe any ill effects. The
hardware really lacks any kind of sensible status + atomic flip
registers, so if it turns out there are problems we may need to
poll the pipe scanline counter to avoid updating OVADD too close
to the time when the hardware fetches the registers from memory.
But so far that doesn't seem necessary.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: overlay: Flush posted writes to the OVADD registers

We generally want the overlay update to happen without much delay, so
read the register back to make sure posted writes are flushed.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: overlay: Use jiffies based timeout in ovl_wait()

Wait at most 100 ms for overlay updates.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: overlay: Use set_memory_wc() instead of vmap()

x86 has set_memory_*() API so no need to vmap(). Also change from UC
to WC to make register updates, coeff tables in particular, more
efficient. There's a wmb() already in the code, and AFAICS that
should be enough to flush the WC buffers.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: psb_gtt: Remove the rw semaphore

The read/write semaphore is only taken in read mode, so it
doesn't actually do anything meaningful. Remove it.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: Remove dead code

Nuke a big chunk if #if 0'ed code.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

staging: mrst: Fix error handling in psbfb_create

The two goto labels for unwinding on error were inverted. Swap them
around.

Fixes a smatch error:
drivers/staging/mrst/drv/psb_fb.c:421 psbfb_create(82) error: we previously assumed 'fb' could be null (see line 368) (smatch)

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

drm: Fix __user sparse warnings

Several pointers and casts were missing __user annotations.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

drm: plane: mutex_unlock() was missing

Unlock the mode_config mutex if drm_plane_init() fails.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Install drm_fourcc.h

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Fix zorder handling while overlay updates are pending

Configuring the Z order may potentially need to touch both overlays
simultaneosly. So ovl_wait() needs to be performed for both overlays.
Do that before touching any state to allow set_plane_opts() to fail
in a clean fashion.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Need to wait for overlay in set_plane_opts()

Touching the overlay register memory while the overlay is fetching (or
primed to) the data. Do an ovl_wait() to make sure no updates are
pending.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Add overlay Z order support

Allow the Z order of the overlay to be selected.

Note that there is something strange going on though. It seems changing
the Z order also affects destination color keying. When Z order is
configured one way destination color keying works, but when the Z order
is flipped destination color keying no longer is effective.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Add overlay color keying and constant alpha

Both source and destination color keying can be enabled for the
overlays.

Also the overlay constant alpha blending factor can be changed.
Note that the overlay constant alpha isn't all that useful, though.
What you usually want is to control graphics alpha. That will need
to be handled through the CRTC somehow (a new ioctl possibly) until
such time when the scanout functionality can be ripped out of the
CRTC object. When that happens the graphics plane would be just another
drm plane.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Add overlay CSC matrix and chroma siting settings

The used color space conversion matrix can be toggled between BT.601 and
BT.709 versions. Also the siting of chroma samples can now be selected.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Add overlay color correction settings

Allow the overlay brightness, contrast, hue and saturation to be
adjusted. Also allows one to select the input range (JPEG vs. MPEG)
for the color space conversion.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Add support for Medfield video overlays

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Use drm_framebuffer_check()

Call drm_framebuffer_check() to verify that the requested fb layout is
sane.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Add alignment argument to psb_gtt_map_pvr_memory()

The video overlay register block must be aligned on a 64K boundary.
Allow the caller of psb_gtt_map_pvr_memory() to specify the desired
alignment.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add drm_chroma_phase_offsets() utility function

This function is is there to help driver writers.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: plane: Add plane options ioctl

Add a new ioctl DRM_IOCTL_MODE_PLANE_OPTS which is used to configure
various settings for the plane.

I left out gamma correction thinking that it could be added using a
separate ioctl, since there's already a gamma ioctl for CRTCs.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add drm_calc_{hscale, vscale}() utility functions

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add struct drm_region and assorted utility functions

struct drm_region represents a two dimensional region. The utility
functions are there to help driver writers.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add drm_framebuffer_check() utility function

This function performs a battery of sanity checks on the requested
framebuffer layout. Drivers can call it from their fb_create hook.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add drm_format_{horz, vert}_chroma_subsampling() utility functions

These functions return the chroma subsampling factors for the specified
pixel format.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add drm_format_plane_cpp() utility function

This function returns the bytes per pixel value based on the pixel
format and plane index.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add drm_format_num_planes() utility function

This function returns the number of planes used by a specific pixel
format.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Check that the requested pixel format is valid

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Handle duplicate FOURCCs

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Replace pitch with pitches[] in drm_framebuffer

Otherwise each driver would need to keep the information inside
their own framebuffer object structure. Also add offsets[]. BOs
on the other hand are driver specific, so those can be kept in
driver specific structures.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: plane: Check that the fb pixel format is supported by the plane

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: plane: Make 'formats' parameter to drm_plane_init() const

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: plane: Check crtc coordinates against integer overflows in setplane ioctl

Help drivers a little by guaranteeing that crtc_x+crtc_w and
crtc_y+crtc_h don't overflow.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: plane: Check source coordinates

Make sure the source coordinates stay within the buffer.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: fourcc: Use __u32 instead of u32

drm_fourcc.h can be included from user space so appropriate types.

Also fix some sparse errors while were at it.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: plane: Clear plane.crtc and plane.fb after disable_plane()

These are the only indication to user space that the plane was disabled.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Redefine pixel formats

Name the formats as DRM_FORMAT_X instead of DRM_FOURCC_X. Use consistent
names, especially for the RGB formats. Component order and byte order are
now strictly specified for each format.

The RGB format naming follows a convention where the components names
and sizes are listed from left to right, matching the order within a
single pixel from most significant bit to least significant bit.

The YUV format names vary more. For the 4:2:2 packed formats and 2
plane formats use the fourcc. For the three plane formats the
name includes the subsampling information using the standard
subsampling notation.

The fourccs for for all the RGB formats and some of the YUV formats
I invented myself. The idea was that looking at just the fourcc you
get some idea what the format is about without having to decode it
using some external reference.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: Add a missing ')'

The code happened to compile because the flag wasn't actually used yet.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: add an fb creation ioctl that takes a pixel format v5

To properly support the various plane formats supported by different
hardware, the kernel must know the pixel format of a framebuffer object.
So add a new ioctl taking a format argument corresponding to a fourcc
name from the new drm_fourcc.h header file.  Implement the fb creation
hooks in terms of the new mode_fb_cmd2 using helpers where the old
bpp/depth values are needed.

v2: create DRM specific fourcc header file for sharing with libdrm etc
v3: fix rebase failure and use DRM fourcc codes in intel_display.c and
    update commit message
v4: make fb_cmd2 handle field into an array for multi-object formats
    pull in Ville's fix for the memcpy in drm_plane_init
    apply Ville's cleanup to zero out fb_cmd2 arg in drm_mode_addfb
v5: add 'flags' field for interlaced support (from Ville)

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reviewed-by: Rob Clark <rob.clark@linaro.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

drm: add plane support v3

Planes are a bit like half-CRTCs.  They have a location and fb, but
don't drive outputs directly.  Add support for handling them to the core
KMS code.

v2: fix ABI of get_plane - move format_type_ptr to the end
v3: add 'flags' field for interlaced support (from Ville)

Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reviewed-by: Rob Clark <rob.clark@linaro.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

staging: mrst: Return ERR_PTR() from fb_create hook

In case of an error, the caller expects ERR_PTR() to be
used instead of returning NULL.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: don't touch port control twice in a row

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: use REG_BIT_WAIT for waiting bits to flip

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: unconditionally enable display

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: fix pipe/plane enable disable

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: only change the device ready bit, don't touch others

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: reg and field helpers

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: give the panel more time to wake up after power on

The panel I2C writes would fail if done too soon after power on. The panel
spec is not explicit on how early I2C can be used, but 50 ms was clearly
too little.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: tc35876x: soft reset the LCD controller after configuration

Soft reset the bridge chip LCD controller after configuration according to
the chip bring up and programming guide.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: switch panel power off/on when blanking/unblanking

Add the missing panel power off to save power when the display is blanked.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

HACK: gfx: display: add display I2C device

The display I2C device (bus 2, addr 0x60) is not present in
firmware. Workaround by creating the device by ourselves.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: add driver for CMI LCD panel I2C

The CMI LCD panel has an I2C interface for some configuration. Add driver
for it.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: put DSI lanes to ULPS before disabling pipe

Apparently disabling the pipe does not succeed if DSI lanes are not put to
ULPS before that.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: add Android early suspend support

Use early suspend to switch display off, and late resume to switch display
on. Use runtime PM to automatically handle enabling and disabling hardware.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: drop gl3 enable kernel command line and module parameter

The support for enabling or disabling GL3 via module or command line
parameter is broken (as it only defines whether it's initialized or
not). Also, the gfx userspace depends on this being one or the other, and
it can't be switched at will. Just remove the parameter.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: remove module and early parameter to toggle runtime PM

There's no point in having a module or early parameter to toggle the
driver's runtime PM on and off, since it can be done dynamically through
the sysfs /sys/devices/.../power/control files.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: clean up PCI suspend/resume

Don't mess with the PCI power states. The PCI core will take care of all
this for us. Clean up the functions accordingly.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: fix and clean runtime PM code

Fix and clean up runtime PM code. Call pm_runtime_put_noidle() in init, and
pm_runtime_get_noresume() in deinit, according to the PCI runtime PM model
(see pci-driver.c). Wake up hardware in runtime resume.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: remove unused module parameter rtpm/gfxrtdelay

The rtpm (internally gfxrtdelay) module parameter is not used for
anything. Remove it.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: remove the remains of unused procfs support

The gfx driver procfs has not been enabled or used for a while, and parts
of it are obsolete and bitrotten. Remove what's left of the procfs
support. In the unlikely event that the interface is ever needed in the
future, it should be implemented through sysfs anyway.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: remove suspicious runtime PM related code

Remove suspicious runtime PM related code sprinkled here and there in the
driver. This is in preparation of eventually properly implementing runtime
PM.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: reduce the use of global variables

Do not use gpDrmDevice (a pointer to the struct drm_device) where it's
possible to derive it from other information. This is a step towards
removing gpDrmDevice altogether.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: remove legacy pm interface

Drivers should not support both legacy and new pm framework.

Signed-off-by: Kristen Carlson Accardi <kristen@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: tc35876x: remove device creation hack

The I2C device is now present in firmware. Remove the device creation hack.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: tc35876x: fix null pointer dereference in panel off

Initialize platform data pointer before use.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: fix locking of the firmare trace debugfs entry

Fix regression introduced by:
3233b3a1 - gfx: pvr: debugfs: replace test_and_set_bit with spinlock+flag

Wrong spinlock was used..

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: make firmware trace output IMG compatible

Make the order of words in the firmware trace record match that of the
original IMG driver.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: reduce loglevel of 'fw state not available' msg

This isn't an error, but an indication that user space wasn't built to
provide this information.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: display: Avoid NULL pointer deference

Framebuffer memory might be created without syncobject. Page fliping
assumed that all framebuffers have syncobject which oopses because of
NULL pointer deference.

That kind of memory can't be protected from tearing using
syncobject in kernel.

Signed-off-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: hide false positive warnings include dir warnings

Currently if you try to build PVR driver using W=1 and O=... you'll get
a lot of warnings like:

cc1: warning: drivers/staging/mrst/pvr/include4: No such file or directory [enabled by default]

The root of the problem is kbuild. For every non-absolute -I path in
cflags kbuild adds one more option with $(srctree) prefix even if the
dir doesn't exists.

Let's workaround the issue with absolute paths.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Tested-by: Imre Deak <imre.deak@intel.com>

gfx: build driver using top level Makefile and drop medfield/Makefile

Simplify build by removing medfield/Makefile and building the driver using
top level Makefile. Clean up the Makefile.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Imre Deak <imre.deak@intel.com>
Acked-by: Artem Bityutskiy <Artem.Bityutskiy@linux.intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: Move ioctl number check before first use

Driver is using user provided ioctl number before checking if it is in
valid range. That makes it possible to force kernel to read memory past
the end of ioctl information array.

Signed-off-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Acked-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: fix regression in user debug request

Fix the regression during user debug request IOCTL introduced by:
7214f0fe418 - gfx: pvr: move core debugging functions to a separate file

Spotted-and-tested-by: Pauli Nieminen <pauli.nieminen@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: add hwrec_debugfs entries

These entries provide details about an SGX HW recovery event. In general
the hwrec_xxx entry will provide a snapshot of the corresponding state
as it was at the time of the last recovery, while the xxx entry will
give the current state.

At the moment we have hwrec_event and hwrec_sgx_registers. hwrec_event
is special in that it can be used to step to the next HW recovery event.
After it's opened and read all other hwrec_* entries will provide the
state for the same event. A second open on hwrec_event will reset all
hwrec_* entries, in preparation for the next event.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: debugfs: replace test_and_set_bit with spinlock+flag

The busy variable is not really a bitmap, using a spinlock with a busy
flag better expresses the purpose.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: add sgx_save_registers_no_pwron

Needed by the upcoming patch.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: add sgx_{read/write}_reg

Needed by the upcoming patch.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: make sure power is on during SGX reset

There is a race where after calling the SGX power-on function, it's possible
that the device will be powered-off through another IOCTL or and SGX
interrupt, before we get to perform the actual reset operation. To prevent
this, keep the power-lock for the duration of the reset.

Note that the current PVR API for powering on/off the devices is suboptimal.
Powering on a device requires acquiring the 'power lock', which will be
re-tried for a given amount of time, after which the driver will give up.
Instead of this we should have a refcount based solution and get rid of
the above timeout mechanism.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: move core debugging functions to a separate file

We'll add more of similar functionality in the future, so having a
separate file for them makes things cleaner.

No functional change.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: drv: Fix mode after all drm clients have exited

In case crtc is disabled surfaceflinger will fail to set mode
correct after surfaceflinger restart. Problem is assumbtion in HAL code
that KMS has set prober mode for internal display.

But drm disables crtc when surfaceflinger exists leaving its own memory
as framebuffer that is freed by kernel in process exit handling. After
all process have exited KMS driver should restore mode and fb to initial
state set by fbcon.

Adding the missing state restoration lets surfaceflinger initialize
correctly after restart.

Signed-off-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: drv: tc35876x: Fix physical display size information

The physical size of display is hardcoded in display driver. The
hardcoded value for lvds bridge is returned from tmd panel which is
phone size display.

Userspace reading the incorrect display size would calculate wrong dpi
for display which would results to incorrectly sized UI.

To provide correct information for lvds bridge driver needs separate
panel info function for lvds bridge.

Signed-off-by: Pauli Nieminen <pauli.nieminen@linux.intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: add debugfs entry to read sgx firmware trace

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: add debugfs entry to reset sgx

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: export HWRecoveryResetSGX

Needed by the next patch.

 I8c4219ef0c0773471c3aa147ebee90c04d0e71e6
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: refactor pvr_get_sgx_dev_info

Needed by the upcoming patches.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: refactor fw state dumping code

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: pvr: add Kconfig option to dump fw trace to console

Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

gfx: drv: fix display backlight PWM duty cycle setting

Fix broken PWM calculation. The PWM register accepts values between
0x00...0x63, corresponding to pulse duty cycles of 0...99%.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Tested-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>