linux-stable.git/drivers/gpu/drm/omapdrm, branch v4.2.7

drm/omap: replace ALIGN(PAGE_SIZE) by PAGE_ALIGN

2015-07-02T12:58:07+00:00

use mm.h definition

Cc: David Airlie 
Cc: Tomi Valkeinen 
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Fabian Frederick 
Signed-off-by: Tomi Valkeinen

drm/omap: fix align_pitch() for 24 bits per pixel

2015-07-02T12:58:06+00:00

align_pitch() uses ALIGN() to ensure the pitch is aligned to SGX's
requirement of 8 pixels. However, ALIGN() expects the alignment value to
be a power of two, which is not the case for 24 bits per pixels.

Use roundup() instead, which works for all alignments.

This fixes the error seen with 24 bits per pixel modes:

"buffer pitch (2176 bytes) is not a multiple of pixel size (3 bytes)"

Signed-off-by: Tomi Valkeinen

drm/omap: fix omap_gem_put_paddr() error handling

2015-07-02T12:58:06+00:00

If tiler_unpin() call in omap_gem_put_paddr() fails,
omap_gem_put_paddr() will immediately stop processing and return an
error.

This patch remoes that error checking, and also removes
omap_gem_put_paddr()'s return value, because:

 * The caller of omap_gem_put_paddr() can do nothing if an error
   happens, so it's pointless to return an error value

 * If tiler_unpin() fails, the GEM object will possibly be left in an
   undefined state, where the DMM mapping may have been removed, but the
   GEM object still thinks everything is as it should be, leading to
   crashes later.

 * There's no point in returning an error from a "free" call, as the
   caller can do nothing about it. So it's better to clean up as much as
   possible.

Signed-off-by: Tomi Valkeinen 
Acked-by: Laurent Pinchart

drm/omap: fix omap_framebuffer_unpin() error handling

2015-07-02T12:58:06+00:00

omap_framebuffer_unpin() check the return value of omap_gem_put_paddr()
and return immediately if omap_gem_put_paddr() fails.

This patch removes the check for the return value, and also removes the
return value of omap_framebuffer_unpin(), because:

 * Nothing checks the return value of omap_framebuffer_unpin(), and even
   something did check it, there's nothing the caller can do to handle
   the error.

 * If a omap_gem_put_paddr() fails, the framebuffer's other planes will
   be left unreleased. So it's better to call omap_gem_put_paddr() for
   all the planes, even if one would fail.

Signed-off-by: Tomi Valkeinen 
Acked-by: Laurent Pinchart

drm/omap: increase DMM transaction timeout

2015-07-02T12:58:06+00:00

The DMM driver uses a timeout of 1 ms to wait for DMM transaction to
finish. While DMM should always finish the operation within that time,
the timeout is rather strict. Small misbehavior of the system (e.g. an
irq taking too long) could trigger the timeout.

As the DMM is a critical piece of code for display memory management,
let's increase the timeout to 100 ms so that we are less likely to fail
a memory allocation in case of system misbehaviors. 100 ms is just a
guess of a reasonably large timeout. The HW should accomplish the task
in less than 1 ms.

Signed-off-by: Tomi Valkeinen

drm/omap: check that plane is inside crtc

2015-07-02T12:57:58+00:00

DRM allows planes to be partially off-screen, but DSS hardware does not.
This patch adds the necessary check to reject plane configs if the plane
is not fully inside the crtc.

Signed-off-by: Tomi Valkeinen 
Acked-by: Laurent Pinchart

drm/omap: return error if dma_alloc_writecombine fails

2015-06-24T06:53:26+00:00

On a platform with no TILER (e.g. omap3, am43xx), when the user wants to
allocate buffer with flag OMAP_BO_SCANOUT, the buffer needs to be
allocated with dma_alloc_writecombine. For some reason the driver does
not return an error if that alloc fails, instead it continues without
backing memory. This leads to errors later when the user tries to use
the buffer.

This patch makes the driver return an error if dma_alloc_writecombine
fails.

Signed-off-by: Tomi Valkeinen 
Acked-by: Laurent Pinchart

drm: omapdrm: new vblank and event handling

2015-06-12T20:30:47+00:00

Rework the crtc event/flip_wait system as follows:

- If we enable a crtc (full modeset), we set omap_crtc->pending and
  register vblank irq.

- If we need to set GO bit (page flip), we do the same but also set the
  GO bit.

- On vblank we unregister the irq, clear the 'pending' flag, send vblank
  event to userspace if crtc->state->event != NULL, and wake up
  'pending_wait' wq.

- In omap_atomic_complete() we wait for the 'pending' flag to get reset
  for all enabled crtcs  using 'pending_wait' wq.

The above ensures that we send the events to userspace in vblank, and
that after the wait in omap_atomic_complete() everything for the
affected crtcs has been completed.

Signed-off-by: Tomi Valkeinen 
Signed-off-by: Laurent Pinchart

drm: omapdrm: merge omap_crtc_flush and omap_crtc_atomic_flush

2015-06-12T19:52:52+00:00

omap_crtc_atomic_flush() is the only user of omap_crtc_flush(), so just
move the code from omap_crtc_flush() to omap_crtc_atomic_flush().

Signed-off-by: Tomi Valkeinen 
Acked-by: Laurent Pinchart

drm: omapdrm: add lock for fb pinning

2015-06-12T19:52:51+00:00

Before atomic modesetting omap_framebuffer_pin() and
omap_framebuffer_unpin() were always called with modesetting locks
taken. With atomic modesetting support this is no longer the case, and
we need locking to protect the pin_count and the paddr, as multiple
threads may pin the same fb concurrently.

This patch adds a mutex to struct omap_framebuffer, and uses it in
omap_framebuffer_pin() and omap_framebuffer_unpin().

Signed-off-by: Tomi Valkeinen 
Acked-by: Laurent Pinchart