v2: Fix dma buffer validation on resource pinning.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NSinclair Yeh <syeh@vmware.com>

This patch address the following underlying issues with SurfaceDMA

* SurfaceDMA command does not work in a 2D VM, but we can wrap a
  proxy surface around the same DMA buffer and use the SurfaceCopy
  command which does work in a 2D VM.

* Wrapping a DMA buffer with a proxy surface also gives us an
  added optimization path for the case when the DMA buf
  dimensions match the mode.  In this case, the DMA buf can
  be pinned as the display surface, saving an extra copy.
  This only works in a 2D VM because we won't be doing any
  rendering operations directly to the display surface.

v2
* Moved is_dmabuf_proxy field to vmw_framebuffer_surface
* Undone coding style changes
* Addressed other issues from review
Signed-off-by: NSinclair Yeh <syeh@vmware.com>
Reviewed-by: NThomas Hellstrom <thellstrom@vmware.com>

Add support for the screen target device interface.
Add a getparam parameter and bump minor to signal availability.
Signed-off-by: NSinclair Yeh <syeh@vmware.com>
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>

For certain surface copies, we don't have a user space handle for
the destination surface.  In such cases, we are going to trust that
our caller is giving us the right surface ID.

To do this case, we created a quirk flag that may be useful
in the future for handling other cases.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: NSinclair Yeh <syeh@vmware.com>

Signed-off-by: NSinclair Yeh <syeh@vmware.com>
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>

Refactored vmw_gb_surface_define_ioctl() and made the surface
definition part a separate function.  This way other parts of vmwgfx
can use it to allocate kernel-visible GB surfaces.
Signed-off-by: NSinclair Yeh <syeh@vmware.com>
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>

For screen targets it appears we need to pin surfaces while they are bound
as screen targets, so add a small interface to do that.

v2: Always increase pin_count on pin.
v3: Add missing reservation sem.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NSinclair Yeh <syeh@vmware.com>

Add command buffer support.
Currently we don't implement preemption or fancy error handling.
Tested with a couple of mesa-demos, compiz/unity and viewperf maya-03.

v2:
- Synchronize with pending work at command buffer manager takedown.
- Add an interface to flush the current command buffer for latency-critical
  command batches and apply it to framebuffer dirtying.

v3:
- Minor fixes of definitions and typos to address reviews.
- Removed new or moved branch predictor hints.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NSinclair Yeh <syeh@vmware.com>

This commit reworks device initialization so that we always enable the
FIFO at driver load, deferring SVGA enable until either first modeset
or fbdev enable.
This should always leave the fifo properly enabled for render- and
control nodes.
In addition,
*) We disable the use of VRAM when SVGA is not enabled.
*) We simplify PM support so that we only throw out resources on hibernate,
not on suspend, since the device keeps its state on suspend.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NSinclair Yeh <syeh@vmware.com>

Fixes a case where we call vmw_fifo_idle() from within a wait function with
task state !TASK_RUNNING, which is illegal.

In addition, make the locking fine-grained, so that it is performed once
for every read- and write operation. This is of course more costly, but we
don't perform much register access in the timing critical paths anyway. Instead
we have the extra benefit of being sure that we don't forget the hw lock around
register accesses. I think currently the kms code was quite buggy w r t this.

This fixes Red Hat Bugzilla Bug 1180796

Cc: stable@vger.kernel.org
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

Use the new fence interface on vmwgfx too.
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>

---
Changes since v1:
Fix a sleeping function called from invalid context in enable_signaling.

Only one type was ever used. This is needed to simplify the fence
support in the next commit.
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>

No point in converting timespecs back and forth.
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: NJohn Stultz <john.stultz@linaro.org>

Contrary to the host-backed shader interface that has a per-context
name-space for shaders, the compat shader namespace was per client
(or rather, per file). Fix this so that the compat shader namespace is per
context, and at the same time, make command buffer managed context resource
management generic.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

Signal availability of prime fd reference ioctls and render nodes.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>

Don't use a per-master semaphore (ttm lock) for reservation protection, but
rather a per-device semaphore. This is needed since clients connecting using
render nodes aren't master aware.

The ttm lock used should probably be replaced with a reader-write semaphore
once the function down_xx_interruptible() is available.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>

Backing mob contents is propagated to user-space, so make sure backing
mobs are cleared when allocated. This also accidently fix rendering errors
with celestia when emulating legacy mode.

Also update driver date.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>

This patch queries the register SVGA_REG_MOB_MAX_SIZE for the
maximum size of a single mob.
Signed-off-by: NCharmaine Lee <charmainel@vmware.com>
Reviewed-by: NThomas Hellstrom <thellstrom@vmware.com>

When a context is first referenced in the command stream, make sure that all
scrubbed (as a result of eviction) bindings are re-emitted. Also make sure that
all bound resources are put on the resource validate list.

This is needed for legacy emulation, since legacy user-space drivers will
typically not re-emit shader bindings. It also removes the requirement for
user-space drivers to re-emit render-target- and texture bindings.

Makes suspend and hibernate now also work with legacy user-space drivers on
guest-backed devices.

v2: Don't rebind on legacy devices.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

Command stream legacy shader creation and destruction is replaced by
NOPs in the command stream, and instead guest-backed shaders are created
and destroyed as part of the command validation process.

v2: Removed some stray debug messages.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

Update otable definitions and modify the otable setup code
accordingly.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>

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

Only scrub context bindings when a bound resource is destroyed, or when
the MOB backing the context is unbound.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

The device is no longer capable of scrubbing context bindings of resources
that are bound when destroyed.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

Also bump minor to signal a GB-aware kernel module
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

This ioctl enables inter-process synchronization of buffer objects,
which is needed for mesa Guest-Backed objects.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>

Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

Perform a translation of legacy query commands should they occur
in the command stream.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

Also do basic consistency checking.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

To bind a buffer object as a MOB, just validate it as a MOB
memory type. We are reusing the GMRID manager, although we create a new
instance of it to manage MOB ids and tomake sure we don't exceed
the maximum amount of MOB pages.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>

Implement MOB setup, binding and unbinding, but don't hook up to
TTM yet.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

In the future, Scanout buffers need not be backed by VRAM and
the two definitions will differ.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NBrian Paul <brianp@vmware.com>
Reviewed-by: NZack Rusin <zackr@vmware.com>

With dma compliance / IOMMU support added to the driver in kernel 3.13,
the dma addresses can exceed 44 bits, which is what we support in
32-bit mode and with GMR1.
So in 32-bit mode and optionally in 64-bit mode, restrict the dma
addresses to 44 bits, and strip the old GMR1 code.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>
Cc: stable@vger.kernel.org

I've killed them a long time ago in drm/i915, let's get rid of this
remnant of shared drm core days for good.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Also request kernel ttm_buffer objects for buffer objects that obviously
aren't visible to user-space, and save some device address space.

The accounting was broken in a couple of ways:
1) We did not differentiate between user dma buffers and kernel dma buffers.
2) The ttm_bo_acc_size function is broken in that it
a) Doesn't take into account the size of the optional dma address array,
b) Doesn't take into account the fact that drivers typically embed the
ttm_tt structure.

This needs to be fixed in ttm, but meanwhile provide a vmwgfx-specific
function to do the job.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

Also provide a completely dumb dma-buf ops implementation.
Once we have other virtual dma-buf aware devices, we need to provide
something better.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

The code handles three different cases:
1) physical page addresses. The ttm page array is used.
2) DMA subsystem addresses. A scatter-gather list is used.
3) Coherent pages. The ttm dma pool is used, together with the dma_ttm
array os dma_addr_t
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NJakob Bornecrantz <jakob@vmware.com>

So if we survey kms drivers there's a bunch of things they commonly do
in ->lastclose
- delayed processing of vga switcheroo requests (i915, nouveau,
  radeon)
- force-restoring the fbcon (most)
- resetting a bunch properties to make fbcon work better (omap)
- disabling all outputs (vmwgfx)

In short besides the semantically important vga switcheroo stuff they
all try very hard to keep fbcon working in case X dies.

But none of them try to not do this at driver unload time safe for
vmwgfx, and digging through logs I couldn't find any reason for why
vmwgfx is special.

Since ->firstopen has lots of potential for abuse with kms drivers
(like delaying driver setup to pamper over races in the load sequence)
it's imo very much worth it to remove this logic so that we can
stop using the ->firstopen callback for kms drivers.

Also module unloading is rather a debug feature and developers should
know how to restore the display to a sane configuration.

Cc: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>