packaging: gbs: use upstream-git branch

Change-Id: Iad079eca3d7764aca795571af780abcc37d17094
Signed-off-by: Philippe Coval <philippe.coval@open.eurogiciel.org>

xf86xv.h cannot be included without first including xorg-server.h.

Without this the build fails on systems with the latest glibc,
throwing this error:

In file included from /usr/include/string.h:634:0,
                 from /usr/include/xorg/os.h:53,
                 from /usr/include/xorg/misc.h:115,
                 from /usr/include/xorg/screenint.h:50,
                 from /usr/include/xorg/scrnintstr.h:50,
                 from /usr/include/xorg/xvdix.h:55,
                 from /usr/include/xorg/xf86xv.h:32,
                 from vmwgfx_overlay.c:38:
/usr/include/xorg/os.h:579:1: error: expected identifier or '(' before '__extension__'
strndup(const char *str, size_t n);

This is caused by HAVE_STRNDUP not being set (it is set from xorg-server.h),
causing os.h to redefine it.

Change-Id: I3a8b19e1393733620cb4cce10811efec45445129
Signed-off-by: Stefan Dirsch <sndirsch@suse.de>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Acked-by: Thomas Hellstrom <thellstrom@vmware.com>
Origin: upstream

packaging: Bump to 13.0.2 for Tizen

Change-Id: I0d1df26075cf43d5bd6d4155b3d21f247d602626
Signed-off-by: Geoffroy Van Cutsem <geoffroy.vancutsem@intel.com>

xf86-video-vmware should not try to build into a pure wayland platform..

This package is designed for an X server platform.

In a system released, based exclusively on Wayland platform, it should be in a excluded status.

Tizen devel  | Tizen release
_____________|______________
succeeded    | succeeded
excluded     | excluded
broken       |
failed       |
unresolvable |
blocked      |
locked       |
note:
    - The disabled status is only
      allowed for OBS administration.

Change-Id: I63dc5fc9d502d6ce5eb1449495f9f30ca6d560ca
Signed-off-by: Ronan Le Martret <ronan@fridu.net>

resetting manifest requested domain to floor

packaging: Initial packaging on 12.0.2 for Tizen

Change-Id: I16c78d3875db5fd7957c0faa5389f4fc182751db

Bump the version number to 13.0.2

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Fix compilation on newer X servers

Found by Bryan Lee. Later versions of X.org turns dispMode pointers
into const upsetting gcc, turn them into size_t's instead.

Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>

vmware/vmwgfx: Always allocate shared hardware surfaces.

Hardware surfaces are all likely to be shared at some point, and we *really*
don't want to change a hardware surface that is bound as a drm framebuffer.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Require libdrm 2.4.38 to build XMir.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Fix build errors and warnings

A previous commit and the hosted merge unfortunately brought in some build
errors / warnings on early X servers.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Block DMA to prime surfaces for now

Since there is currently no _good_ way to get the surface format of a
prime surface, we block DMA to these surfaces; we don't know if our
software data is compatible with the surface format.

This patch also makes sure that there is a hardware surface backing the
drawable we copy from.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Enable direct dmas

Enable direct dmas instead of using the xa-provided dma functionality.
This saves a bounce-buffer software copy of all dma'd contents.

This also implies that all drawables with mixed software / hardware contents
will use a kernel buffer for software rendering.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

vmwgfx: Add support for XWayland

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Add support for XMir v2.

Use the hosted infrastructure to add support for XMir.
Helpers go in vmwgfx_saa.c.

v2: Added comments for the helpers, and added a
vmwgfx_flush_dri2 to be executed when coming back from vt switch.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Add an infrastructure to be able to run hosted under a compositor v2

Figure out what's needed both for XMir and XWayland and make a common
driver structure out of it.

v2: Added a lot of comments. No code change.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Fix compile breakage on XA version 1.

A recent revert brought back code that assumed XA version 2.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: trivial

vmwgfx: Really allow XA version 2.

When XA starts to correctly advertise version 2, we didn't allow it.
Fix this.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

Revert "vmwgfx: Get rid of device-specific DMA code"

This reverts commit 45b2457516a9db4bd1d60fbb24a1efbe2d9dd932.

Reverting this because using direct DMA for shared pixmaps should
boost performance. It should be usable both in a composited environment
and when running hosted. Need to retest the DMA flaws I saw before disabling
this code.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Remove stray 1

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Support also XA version 1 v2

We need to support also XA version 1, since we want to be backwards
compatible with older mesa releases.
Unfortunately, the intended way of detecting XA major version at
compile-time was broken on mesa 10 so we need a workaround that tests
for XA version 2 at config time.

v2: Update the test for XA version 2.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Fix some compilation warnings and indentations

Most of the compilation warnings we're seeing are due to bad xorg headers,
but fix the ones we can fix in the driver. Some indentation fixes as well.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: handle changes of DamageUnregister API in 1.14.99.2

Fix is inspired from the intel driver.

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Get rid of device-specific DMA code

It's rarely used and things seem to work well enough on top of XA.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>

vmwgfx: Implement textured video completely on top of XA.

Remove device-specific hacks. This may increase resource usage a little
on old hardware revisions, but we don't need separate code paths on
different hardware revisions.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>

vmwgfx: Avoid HW operations when not master

Note that for DRI2, a dri2_copy_region becomes a NOP when not master.
Additionally, all dri2 operations that lead to a potential kernel
access will return FALSE.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: update for XA API changes

Signed-off-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Tested-by: Jakob Bornecrantz <jakob@vmware.com>

Bump the version number to 13.0.1

vmwgfx: Use myGlyphs to fix crashes

Tested-by: Christian Hesse <mail@eworm.de>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Loïc Yhuel <loic.yhuel@gmail.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

Kill mibstore

It was a noop for at least 5 years and it has been removed.

Signed-off-by: Zack Rusin <zackr@vmware.com>

Bump the version number to 13.0.0

New abi and various fixes

Signed-off-by: Zack Rusin <zackr@vmware.com>

xf86-video-vmware 12.0.99.901

Internal release for a round of QE testing

Signed-off-by: Zack Rusin <zackr@vmware.com>

vmwgfx: Guard against null rotate pixmap in shadow destroy

Reviewed-by: Thomas Hellström <thellstrom@vmware.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: avoid warnings about shadowing 'i'

Signed-off-by: Dave Airlie <airlied@redhat.com>

vmware: drop infoFromScreen inline in favour of new interface

This also should fix the build regression introduced with the fix
for this macro.

Signed-off-by: Dave Airlie <airlied@redhat.com>

vmware: Fix up some warnings after new compat API change.

Fixes multiple instances of these warnings on x86_64:
../../src/vmware.h:180:5: warning: implicit declaration of function 'xf86ScreenToScrn' [-Wimplicit-function-declaration]
../../src/vmware.h:180:5: warning: nested extern declaration of 'xf86ScreenToScrn' [-Wnested-externs]
../../src/vmware.h:180:5: warning: return makes pointer from integer without a cast [enabled by default]
Function `xf86ScreenToScrn' implicitly converted to pointer at ../../src/vmware.h:180

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Robert Hooker <sarvatt@ubuntu.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: port vmware driver to new compat API

This is a port of the vmware driver to the new compat API.

Tested-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: set the XA_FLAG_SHARED flag for composite dest surfaces

This fixes a failed assertion in the gallium/svga driver in the
svga_texture_get_handle() function.  The texture resource wasn't
getting created with the PIPE_BIND_SHARED flag so the !cachable
assertion would fail and the X session would abort.  This didn't
happen with release builds.

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

Bump version number for the release

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmware/legacy: Fix crash on Xserver 1.12.0

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>

Fix a memory corruption due to a redundant free.

We use the empty rect purely as a placeholder and never
initialize it with its own memory so lets not try to free
it.
Spotted by Dave Airlie.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

Bump version number for the release

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

config: move AC_SYS_LARGEFILE in autoconf init section

There is no need for this statement to be conditionally included.
It adds support for large files without taking anything away.

http://www.gnu.org/software/autoconf/manual/autoconf.html

Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

config: include saa and vmwgfx subdirs in the tarball

Use AM_CONDITIONAL. Automake knows what to distribute.
It needs to be able to navigate down the subdirs to find what
needs to be included in the tarball.

To test reliably, create a tarball and expand it into a separate
directory and build with xatracker. Distcheck will not detect
missing code when such code is configured not to build.

The content of a tarball *must* always be identical, regardless
of the configuration options used or on which platform it was configured.

Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

Bump version number for the release

Signed-off-by: Zack Rusin <zackr@vmware.com>

vmware/legacy: Fix initial mode size v2

Commit "vmwlegacy: Make the default be a minium of 800x600..."
read the dimensions from the incorrect register pair.

v2: Also treat initial widths and heights larger than the maximum
values as an error and revert back to the minimum.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

saa: Try to avoid readbacks during polyFillRect

This significantly improves performance of shaped windows on top of 3D
contents

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>

vmware: Kill warnings on Xserver 1.12

Compile-tested on servers 1.0 through 1.12.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

vmwgfx: Kill a deprecation warning on server 1.12

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

vmwgfx: Fix up swap[sl] for server 1.12

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

vmwgfx: Avoid including a library header and use pixman for type conversion

The PictTransform type is a typedef of a pixman type.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

Bump version number for release

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

Make the driver build pass "make distcheck"

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Adapt to new XA version number v2

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx,saa: Fix a bunch of compilation warnings

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwlegacy: Make the default be a minimum of 800x600

This patch and the corresponding vmwgfx patch formalizes both drivers
on the same behviour. The minimum is 800x600 (unless the max size is
smaller). This makes it unnecessary to check against VRAM size since
it is always at least large enough to fit the max values.

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Make sure the driver builds on old servers

Move things around a bit so the driver compiles with
servers back to Xserver 1.0.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Update Manpage

Update the Manpage with new options and a brief modesetting- and 3D
description.

Signed-off-by: Thomas Hellström <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

Merge branch 'vmwgfx_branch'

saa, vmwgfx: Fix saa_copy_composite

The traditional accelerated copy methods aren't format aware.
Make saa copy format aware,and pass formats on to the driver copy function
if available. If the driver can't handle format conversions it needs to
return FALSE.

This fixes format confusion in the copy composite fastpath.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

saa: Always read back hw contents on poly_fill_rect

The damage region may be much larger than the actual area
filled, so make sure we read back any hw contents in the damage region
since we use it to mark sw dirty.
Revisit this in the future since it may increase readbacks in an
undesired way. There are workarounds.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

saa: Only try to accelerate pixmaps that are driver pixmaps

Trying to accelerate scratch pixmaps doesn't work currently,
and I'm not sure it's a big benefit either.

This is in line with what EXA does.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrants <jakob@vmware.com>

vmwgfx: Reduce system resource usage with textured video

Use U and V textures and corresponding dma buffers that match
the size of the input UV data rather than using textures and DMA buffers
that match the Y data. This simplifies data copy and reduces the amount of
image data transfered to the host by up to 50% (YV12).

It also reduces dma buffer- and surface memory usage correspondingly.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Hook up XV color conversion control attributes

This enables the standard XV color conversion control attributes
contrast, scaling, saturation and hue.

Currently the attribute change takes effect on the next image only.
If we ever implement a ReputImage callback, we can use it to update
the image immediately on attribute change.

The attribute ranges follow those used by the proprietary Nvidia driver.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Adapt to new interface of xa_yuv_planar_blit

Also make it easier to modify the color conversion matrix based on
yuv encoding properties.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Fix MessageType members that are incorrectly typed as Bool

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Add an option to enable rendercheck mode

Rendercheck mode is used to enable acceleration of all supported composite
operations, regardless of the current data location. The mode is off by
default, but could be turned on to test the hardware composite operation
implementation. Due to excessive data migration, rendercheck mode will be
slow.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Adapt to the dri state tracker sending depth instead of bpp

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Don't register back buffers as "hw_is_dri2_fronts" v2

That is reserved for dri2 front or fake front pixmaps that need
immediate flushing of software contents to hardware.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Include drm.h from correct path

Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Make the drm device name persistent

The DRI2 code doesn't copy the device name, but assumes the storage
is persistent.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Alan Hourihane <alanh@vmware.com>

vmwgfx: Get the drm device name from the open file descriptor

Rather than hardcoding it.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Use config options to optionally set GUI topology

Use option "StaticXinerama" or "GuiLayout" to set initial GUI topology,
similar to how these options are used with the legacy driver.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwlegacy: Factor out code that can be common for vmwgfx and vmwlegacy

In this case it's the parsing of the gui / statix xinerama topology.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwlegacy: Accept option "GuiLayout" as a synonym for "staticXinerama"

It's more user-friendly and in line with what it's being used for in the
vmwgfx driver.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Update library dependencies

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Pre-init fixes

A bunch of fixes that makes the fallback from vmwgfx to vmwlegacy easier.

1) Allow only 16 and 24 bit depths with vmwgfx. We may revisit this later.
2) Try to reclaim some resources if pre_init fails.
3) Print error messages so the user understands what happens on fallback.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Update the vmware register definition

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmware: Hook up vmwgfx to be tried before falling back to legacy

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwlegacy: Factor out bootstrap functions

Preparing for choosing driver path in preinit instead of using
shim loader.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwlegacy: Fix up Xinerama / panoramiX replies.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwlegacy: Fix device fifo communication

This fixes two issues with the device fifo communication:
1) Idle the fifo before initializing it. If the fifo is already up and
processing data due to an uncleanly shut down server, and init could
otherwise confuse the device.

2) Mark fifo memory volatile when writing to it and make sure commands
are written before telling the device they are available.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Make sure we don't render when VT switched

When compiz is enabled, we may try to DMA sw rendered contents of dri2
drawables to surfaces while VT switched. I'm not sure how this could be
happening, but queue the DMA until the next block handler.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Alan Hourihane <alanh@vmware.com>

vmwgfx: Fix rendering errors when rotating / unrotating a screen.

The dirty trackers weren't correctly translated to
pending presents / updates when a pixmap was bound as a scanout.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrants <jakob@vmware.com>

vmwgfx: A number of damage fixes

A damage tracker was sometimes attached to a pixmap in the middle of an
operation. That meant that damage was never registered for that operation
which caused dirty-tracking to fail. Instead, add a damage tracker as
soon as we have storage attached to the pixmap.

Also, when a pixmap was resized, the dirty region trackers could become
stale, and cause DMA operation extending outside of the pixmap with
server crashes as a result. This commit fixes that.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Add an option to use a scanout surface

With this option enabled, we compose all contents (both sw and hw) in
the guest before presenting it. Just like the old xorg state tracker did
when 3D was enabled. It's not as bad as the old xorg state tracker was for
interactivity, but there's a small but noticeable lag when moving windows.

It's mainly intended for testing purposes, like testing the surface
dirty paths in the kernel module.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Be correct when setting up damage tracking

When setting up damage tracking, previously the code would guess wether
sw or hw was currently dirty. The caller always have that information,
so pass it to the damage tracking setup function.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Add an option for direct presents

With this option set to true, accelerated copies to a scanout pixmap will
be performed as hardware presents instead of copies to an intermediate
3D surface backing the scanout pixmap, followed by a present.

Depending on the application this might be a performance boost, but since
it might trigger device software readbacks in other situations, for example
dri2 copy front->fake_front, it might be a performance hog in other situations,
so disable it by default. (Before this commit it was enabled by default).

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Fix acceleration options.

Introduce relevant acceleration options and fix up the log printout.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Alan Hourihane <alanh@vmware.com>

vmwgfx: Remove a couple of unused options

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Alan Hourihane <alanh@vmware.com>

vmwgfx: Don't do present readbacks per crtc

The kernel now takes care of doing this the right way;
no need to duplicate that functionality.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Fall back to sw cursors if needed

If there is a risc that we need two simultaneous cursors,
(two outputs showing the same contents, at least one of them explicit),
fall back to sw cursor.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Fix crtc position of transformed fb

The fb allocated for transformed data always matches the scanout region,
so set the crtc origin to (0,0)

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Hook up vmwarectrl to the gui layout ioctl

Requires drm 2.3.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Update the connector type name array

Also map connector types that are not in the array to "Unknown".

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

vmwgfx: Save the dri2 requested depth since it may differ from the drawable
depth.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

Build fix for ABI version 12

This results in the vmware driver assuming there is only one PCI domain,
which I think is true for everywhere this is expected to run anyway.

Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>

Fix build failures resulting from changes to the swaps and swapl macros in recent xorg-server

Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>

xf86-video-vmware 11.0.99.901

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

Use dixLookupWindow instead of LookupWindow on Xorg 1.5 & later

The LookupWindow function was deprecated in xserver commit ed75b056511ccb4
and removed during the Xorg 1.11 merge window by commit 82a8677d9175732.

Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>